Edinburgh Tram (Line One) Bill Committee Consideration Stage Phase 1 Responses to the undertakings to the Preliminary Stage Report

Transcription

1 APPENDIX D Response 6 8 November 2005 Edinburgh Tram (Line One) Bill Committee Consideration Stage Phase 1 Responses to the undertakings to the Preliminary Stage Report Question The Committee remains concerned that it has seen only some evidence that the highway DAM and the public transport Dam have been validated. Therefore the Committee requests more detailed evidence, before Final Stage, that both the highway and public transport DAM models are validated to an industry standard. Response Executive Summary 1 The Promoter has carefully reviewed the transport model forecasts for the Edinburgh tram patronage and revenue. To assist the Committee s understanding of the complex issues relating to the model calibration and validation a short Executive Summary has been provided, followed by a more detailed response with cross references to technical notes that are provided within a series of Annexes to this response. 2 In 2000 the City of Edinburgh Council took the decision to allocate significant resources to develop a strategic transport model that would forecast future changes in land uses and travel demand in Edinburgh. The hierarchical modelling suite would integrate and be linked with a detailed lower tier public transport and highway model that would enable the benefits of the tram to be assessed. 3 The local area model was based on the Scottish Executive s model for central Scotland, known as the Central Scotland Transport Model (CSTM3) which was developed and audited by independent transportation consultants in A number of schemes have been appraised and promoted using CSTM3 or models based on the CSTM model or data, as follows: Upper Forth Crossing at Kincardine. Stirling-Alloa-Kincardine railway. Central Scotland Transportation Corridor Studies (CSTCS) Multi-Modal Study. M80. M8 Upgrade.

2 APPENDIX D M8 Design, Build, Finance, Operate (DBFO). M74 Completion. Airdrie-Bathgate railway line. Clyde Corridor. A701 Corridor Multi-Modal Study (Penicuik to Straiton, Edinburgh). 4 The local area transport models for both the public transport and highway network were based on this model. For the Edinburgh tram assessment, the 2001 base year model demand was forecast from 1997 and changes to the network/public transport services incorporated, in accordance with normal practice. 5 The above models together form the hierarchical modelling suite that was developed to assess the revenue and patronage forecasts for the Edinburgh Tram. It is normal practice to use transport models that are less than 5 years old in the assessment of schemes without updating the model, as the timescale associated with recalibrating a model means that, at best, the original model survey data will be about two years old. 6 Examination of the validation of the 2001 local area model indicates that the modelled flows and public transport patronage is lower than observed flow by about 10%. This may reflect a number of factors including the daily variation in traffic flows and that the update was based on the long term forecasts of growth for a model of central Scotland, whereas recent growth in Edinburgh has been greater than predicted. Thus, the forecasts are likely to err on the conservative side. 7 By using the original base model the model data was consistent with both the Scottish Executive s model CSTM3 and the associated extensive transport survey database. The model is also consistent with the models used for the assessment of the proposed congestion charging scheme in Edinburgh. The use of a consistent model enables a number of schemes to be evaluated on a common basis, including congestion charging, Edinburgh tram network, Edinburgh Airport Rail Link (EARL) as well as other developments. The results from these unadjusted model runs have been consistently presented to the Committee. 8 To confirm that the level of the patronage and demand forecast for line 1 and 2 were broadly correct, the forecasts were compared against the observed patronage on existing schemes and found to be comparable. This approach is in accordance with the recommendations subsequently published by the National Audit Office s report on the assessment of tram systems. 9 Transport models that are between 5 and 10 years old are often updated using a standard model adjustment technique known as matrix estimation

3 APPENDIX D which enables the model flows to be adjusted to more closely correspond to observed flows (for more details reference should be made to the section below describing this technique). In response to the Committee s request relating to the model calibration and validation, the public transport model calibration was adjusted using this technique. The revised model when assessed against the national validation criteria was found to be much improved. 10 To summarise, the above indicates that the model forecasts are, in line with current practice, possibly slightly conservative and adjustments would, on balance, improve the case for tram. This conclusion has been supported by sensitivity tests that were undertaken using the adjusted public transport model. Furthermore, the recent detailed modelling work undertaken to assess demand at Edinburgh airport for the EARL project demonstrates that the airport related demand in the current model is conservative. 11 The Promoter, tie Ltd and tie s advisors are confident that the demand forecasts to model the patronage and revenue for the tram network are sufficiently robust. Furthermore, on the basis of benchmarking of observed tram networks within Britain, the forecasts are broadly comparable and therefore considered to be realistic. Detailed Technical Response Aspects of the Model Calibration and Validation Reviewed within this response 12 This response seeks to demonstrate to the Committee that the transport modelling suite applied in the assessment of Edinburgh Tramline 1 is robust and appropriate for the scheme development and appraisal through to Parliamentary approval. The following sets out, in brief terms (with the support of detailed Annexes), the structure of the modelling suite and how it has been integrated within the tram design and appraisal procedures (the model provenance ), demonstrates its robustness (focusing on the Detailed Assignment Models). Model Provenance 13 A substantial amount of information was provided at the Preliminary stage and it is not the intention to repeat that here. However, the following provides a brief summary of the modelling tools used and their provenance to provide background and context for what follows.

4 APPENDIX D 14 The decision was taken by CEC in 2000 to make a significant financial commitment to develop a state of the art hierarchical transport modelling suite over a period of two years. The development of the LUTI 1 and DAM 2 modelling framework commenced in December 2000, and was essentially complete by mid The model development was to employ the most current and best available data and on that basis, the Central Scotland Transport Model (CSTM3), developed with a Base Year of 1997, was employed as the starting point for model development. 15 The Edinburgh model consists of three sub-models forming a hierarchical transport modelling suite: a land use model; a strategic macroscopic transport model; and a detailed assignment model (highway and public transport). The land use model is run initially and forecasts future land use patterns on the basis of accessibility. The strategic ( global or higher tier) model and detailed ( local or lower tier) models are run iteratively until the forecast travel demand patterns stabilise and reach equilibrium. Each of the models is now briefly described in turn. a. i. Land Use Model. This forecasts the changes in future land uses and total travel demand. The model forecasts the most advantageous geographic location ( land use ) for every forecast type of development to maximise the overall level of accessibility associated with the development. The Structure Plan sets out the allocated areas for each type of development. To determine the optimum location for each land use, a weighted measure of accessibility of the specific opportunities required for the land use (for example, accessibility to employment, workforce, services, education, leisure and shopping facilities, etc) is calculated for every location and the most accessible locations identified until all demand has been satisfied. The model also outputs the forecast change in travel demand between areas. b. ii. Strategic Macroscopic Transport Model. Simulates at a coarse geographical level the change in travel demand by mode throughout the day, for a high degree of demand segmentation (for example, broken down by person type, origin purpose, destination purpose, time of travel, etc). The model was based on an extensive survey programme in 2001 for highway and public transport demands as well 1 Land Use Transport Interaction model, used to forecast the strategic impact of transport interventions, such as the tram. The model is composed of two sub-models, a Land Use Model and a strategic macroscopic transport model. 2 Detailed Assignment Models used to understand the detailed impact of the transport intervention on bus and tram passenger demand and private vehicle traffic flows.

5 APPENDIX D as Scottish Household Survey data and other market research surveys (stated preference and revealed preference surveys which may be considered to be akin to before and after studies). c. iii Detailed Assignment Model (DAM). Assigns at a fine geographical level the travel demand forecast by the strategic macroscopic model to the detailed public transport and highway network in the morning and evening peak hour and average interpeak hour. 16 CSTM3 was developed in the late 1990s for the Scottish Executive to provide a consistent and robust tool for developing and appraising major transport projects in central Scotland. The development of CSTM3 utilised a range of existing and new data, as follows: Strathclyde Integrated Transport Model (SITM) (extensive highway and public transport surveys in 1992) Existing model of Edinburgh (VIPS); Bus and rail passenger counts/origin & destination surveys carried out on routes and screenlines to the west of Edinburgh in Autumn 1996 (Forth TRIP); Bus and rail origin/destination data from surveys undertaken in various towns around the edge of the CSTM3 Final Model study area in Autumn 1997; and Rail and bus passenger counts undertaken in Edinburgh in autumn The PT surveys undertaken were as follows: Forth Trip: (comprising on-bus counts, bus stop counts, on-bus questionnaires, rail surveys). CSTM3: (comprising rail surveys and bus surveys). 18 The highway surveys undertaken were as follows: Roadside interview surveys (including origin and destination surveys, journey purpose, etc). Strategic Screenlines traffic counts. Key link screenlines traffic counts. Two-way multi-point screenlines traffic counts. Journey times. Separate validation counts. 19 The development of the Edinburgh transport model employed industry standard tools and techniques (in accord with the Department for Transport procedures described in the Design Manual for Roads and Bridges and using recognised transport modelling software). The model was subject to extensive validation using industry standard methods and criteria and this was set out in the model reports previously provided to the Committee.

6 APPENDIX D 20 The LUTI and DAM models were developed for a 2001 Base Year. The CSTM3 model was employed in forecast mode to obtain the initial demand matrices, which were then further developed for use in the LUTI model. The final LUTI model was validated against observed data and confirmed that it could robustly replicate 2001 flows. 21 The DAM models were produced using the aforementioned 2001 demand data and the transport network from CSTM3. As the CSTM3 model simulated transport movements throughout central Scotland, the detailed geographical area of the model was reduced to encompass an area slightly greater than that affected by the tram system (Edinburgh, Lothian and Fife), although longer distance journeys continued to be modelled as well. The highway network, public transport services and travel demands were updated to reflect changes during the intervening period of four years between 1997 and Some additional detail was added in the areas along the proposed tram corridors (to ensure that the model was able to robustly forecast the tram patronage) including: disaggregation of zones (the use of smaller zones along the corridor improve accuracy); increased network and junctions details (the detailed coding for both the public transport and highway network was increased); and bus timetable and service information was updated to In summary, both LUTI and DAM were developed using the best available data (as described above) using industry standard tools and techniques. The strategic model, LUTI, was explicitly validated to 2001 flows, whilst the DAM model validation rested on the good calibration and validation of the source CSTM3 model that was completed 4 years earlier, in accordance with industry practices. The 2001 DAM model was updated to reflect any changes to the transport network (highway improvements and public transport service and timetable changes) and the growth in demand in travel forecast using industry standard procedures. This was in line with normal model development procedures that recognize that for scheme evaluation it is appropriate to use models that are less than 5 years old. 23 It is normal practice to undertake the design and assessment of a project such as the Edinburgh Tram on the basis of a single transport model to ensure that any design variations are evaluated on a comparable and consistent basis. At subsequent stages in the scheme assessment, the model should be reviewed, refined and may be updated as appropriate to ensure that the results are robust. Transport models are rarely updated more frequently due to the long lead time and high costs associated with the development of new models. For example the commission to enhance and update the CSTM model was awarded in 2001 (now known as the Transport Model for Scotland TMfS) but the model development and audit was not completed until 2005.

7 APPENDIX D 24 Overall, this report demonstrates that that the demand forecasting undertaken for Edinburgh Tram utilised the best data and modelling tools available at the time. The modelling tools were developed with a Base Year of 2001 and were applied from 2003 onwards as part of the tramline development. As the development and appraisal of such major schemes can take several years, it is both appropriate and cost effective to review and update models only at key stages in the assessment. Model robustness 25 The concerns expressed by the Committee are associated with the validation of the DAM models, these being the primary source of the demand forecasts for the tram and the associated impacts on the wider public transport network and on the highway network. On that basis, the following sets out a range of analyses which have been undertaken to address those concerns and demonstrate the robustness of the DAM models. 26 The data used to validate the model was surveyed in 2001 and consisted of in-excess of 50 traffic counts along an inner and outer cordon, plus a northsouth and east-west screenline. 27 The following issues are discussed below: validation criteria; performance of CSTM3 and other models (MVA notes 30/32/33); review of DAM validation (LUTI 2001 data) (MVA note 31); matrix estimation; and benchmarking of demand forecasts. Government guidance on validation criteria 28 Before discussing the robustness of the DAM models, it is considered worthwhile to set out the guidelines for acceptability of model validation. It must be emphasized that these are guidelines, not absolute prescriptive targets that are required to be met for every model. The validation targets and reporting should reflect the robustness of the original survey data as well as the new data and the spatial distribution of discrepancies (i.e. the validation is more acceptable if the poorer validation is further away from the main focus of the model). 29 Guidelines for highway models are set out in a Department for Transport (DfT) publication Design Manual for Roads and Bridges (DMRB) and reproduced in Annex A. This contains criteria based primarily on two measures: the percentage difference between modelled and observed flows, and the value of a measure which incorporates both relative and absolute

8 APPENDIX D errors (GEH statistic). The criteria set out the proportion of comparisons that should pass the target (the DMRB Criteria are set out in Annex A). 30 Although there were no definitive guidelines relating to public transport model calibration and validation at the time of the model development, guidelines for public transport models were set out in June 2005 in the DfT guidance Road Traffic and Public Transport Assignment Modelling, Transport Analysis Guidance Unit This states that modelled flows at individual sites should be within 25% of observed; at screenlines the difference should be 15%. Use of the GEH statistic is not mentioned, although it is often employed by practitioners, given its usefulness. Validation of 1997 CSTM3 and benchmarking the DAM models 31 The formal CSTM3 model reports have been provided to the Committee. A summary version of this has been produced focusing on the DAM model area; this is contained in Annex B (MVA note 30). Key validation data relating to the robustness of the model validation described within this response has been abstracted from Annex B and C and is presented in Table 1. This shows that whilst the model does not fully meet the strict assignment validation guidelines published by DfT, it was considered sufficiently robust by the consultants who developed the model, the client, the Scottish Executive and their auditors, SIAS Ltd. For public transport, virtually all of the screenline comparisons are well within the guidelines of 15%; of the 120 individual screenlines listed, only 14 do not meet the criteria. Overall, CSTM3 in the DAM model area is shown to have high and acceptable levels of validation. 32 A number of schemes have been appraised and promoted using CSTM3 or models based on the CSTM model or data, as follows: Upper Forth Crossing at Kincardine. Stirling-Alloa-Kincardine railway. Central Scotland Transportation Corridor Studies (CSTCS) Multi-Modal Study. M80. M8 Upgrade. M8 Design, Build, Finance, Operate (DBFO). M74 Completion. Airdrie Bathgate railway line. Clyde Corridor. A701 Corridor Multi-Modal Study (Penicuik to Straiton, Edinburgh). 33 A range of models for which the Promoter s consultants have access to have been reviewed to understand how CSTM3 compares with other models. Annex C contains notes setting out such reviews for public transport and highway models, respectively MVA notes 32 and 33. For public transport,

9 APPENDIX D eight models are summarised alongside the results from the DAM area of CSTM3. Overall, CSTM3 performs extremely well and is the third best placed model. For highway, some 20 models were reviewed and CSTM3 is well within the overall range of validation achieved. These reviews confirm that few, if any, transportation models meet a strict interpretation of the guidelines and that in comparison with other models, CSTM3 is very well validated. The DAM models cordoned out from CSTM3 have a high level of validation that is significantly above industry standards compared with similar transportation models DAM model validation 34 The highway and public transport DAM models were calibrated and validated in The tram scheme assessments were started in 2001 and, as this was for a period of less than 5 years, the 2001 DAM base models were updated by coding changes to the highway and public transport networks between 1997 and 2001, and forecasting the travel demand using industry standard techniques, as incorporated within CSTM3. The DAM models were not recalibrated nor were they subject to matrix estimation in order to minimise the level of matrix adjustment to the calibrated model.

10 APPENDIX D Table 1. Detailed Assignment Model (DAM) Validation Statistics Model and date Traffic Count Database GEH (% sites within range) <5 <10 >10 Public Transport DAM Model 1997 CSTM 1997 Calibration & validation Bus counts Rail counts CSTM 2001 Validation All Matrix Estimation 2001 Validation All Benchmark Models Bus counts Rail counts Highways DAM Model 1997 CSTM full area 1997 Calibration & validation AM IP PM CSTM DAM area 2001 Validation AM IP PM CSTM DAM area 2001 Validation All Benchmark Models AM IP PM Notes: Benchmarks models statistics are taken from published Model Development Reports Figures abstracted from Annexes A and B 35 Two sets of data are available to review the robustness of the DAM models: traffic counts undertaken in 2001 in association with the LUTI model development and bus counts undertaken in The 2001 survey count datasets have been compared to the modelled 2001 flows and the results are set out in Annex D (MVA note 31). 36 Model validation is normally undertaken immediately after the model calibration and involves comparing the model simulation data against an independent database that was not used during the original model calibration. The purpose of the validation is to use survey data that is independent of the

11 APPENDIX D model validation to confirm that the small adjustments made to the model during the calibration stage were valid changes. 37 The application of the LUTI traffic database to validate the forecast DAM models, 4 years after its calibration, is unusual as it is normal industry practice to assume that calibrated models remain robust for a period of about 5 years. 38 The LUTI 2001 data comprises four sets of count data: two sets forming cordons around the city, essentially at the Edinburgh By-Pass and around the city centre respectively, and two sets of counts looking at east-west and north south movements within the city. Data is available for highway (car) demand and for bus demand. In summary, modelled car demand across screenlines is generally lower than observed at some locations, by up to 30%. Most of the associated GEH statistics are in excess of 10, against a guideline target of under 4. For bus demand screenlines, around half of the modelled flows are within acceptable guidelines, with some flows falling only marginally outside. Again the trend is for modelled flows being lower than observed. 39 More detailed analysis of the highway modelling by individual count site shows that around a third meet the validation criteria. Looking at the bus demand, many of the flows above 1000 passengers/hr are within the accepted guidelines of +/- 25%, with many of the poorer performing comparisons pertaining to low flows where the guidance accepts that these will be less well validated. In summary, the model is within the ranges of model validation accepted within other studies, as set out in the discussion above of how other models perform. 40 The growth forecasts for Edinburgh in CSTM3 may be low in CSTM as this is a national model designed for modelling interurban schemes and the level of economic growth in Edinburgh and Lothians has generally been greater than elsewhere in Scotland during the period It has not been possible to benchmark the level of the Edinburgh DAM 2001 validation with the level of validation of comparable models several years after their original calibration as no published reports could be located. Table 1 indicates that when the 2001 DAM model validation for both the PT and highway model (at a point 4 years after it was calibrated) is compared against the calibration of similar models that were validated immediately after their calibration, the DAM model validation continues to be within the range of calibration that was considered acceptable for other models. 42 Given the level of validation of the DAM models, it is worth setting out the implications for the forecasts of tram patronage and benefits. In general, both highway and public transport demand in the DAM base year model is low and therefore the forecasting process is considered conservative. The

12 APPENDIX D assessment is also based on the relative difference between the reference network (the calibrated base network, updated with committed future changes) and tram network, so any difference that results from a less conservative demand forecast is likely to be relatively small Were car and bus flows higher in the Base year model, the benefits of tram would, on balance, be higher, for the following reasons: higher public transport demand would lead to higher tram patronage; higher car demand would lead to higher levels of transfer to tram; higher car demand would lead to higher congestion levels, which would lead to higher transfer levels to tram; higher congestion levels would have an adverse impact on bus speeds (except where bus lanes exist), in turn leading to greater transfer to tram (and this is already evident in the forecasts for 2011 and 2026, where the increased congestion in 2026 leads to higher tram demand and transfer levels); the benefits of tram would be enhanced because the relative journey times on bus and car would be longer (except where bus lanes exist), and therefore at a higher cost to the traveller, than currently forecast; the greater level of transfer from car from a more congested network would lead to higher decongestion benefits (and again this is evident in the modelling results to date; and detailed demand forecasting was recently undertaken for airport related movements for the EARL project. In response to the Tramline 2 Committee request to provide further information on the effect of EARL on the tram patronage and revenue, the airport demand flows have been incorporated within the transport model for tram line 2 and the results presented in a response to the Committee. The adjusted model reinforces the case that the tram network models are under estimating the case for trams. Matrix Estimation 43 The above conclusions have, in part, been recently confirmed through work to address concerns expressed about the public transport model validation. A standard model adjustment technique know as matrix estimation has been applied to adjust travel demands within the model to update the Base year public transport matrices using the LUTI count data. Matrix estimation is a technique commonly used to improve the calibration of a transport model using "Best Likelihood Fit" techniques 44 Using matrix estimation, the revised model calibration assessed against the criteria set out in the DMRB is much improved.

13 APPENDIX D Forecast benchmarking 45 The benchmarking of forecasts against schemes in operation is a valuable sense check on the forecasts being made. The STAG reports set out data showing how the forecasts for Line 1 and 2 compare to tram schemes in operation. The DfT have recently released latest patronage statistics for tram in the UK 3 and hence the opportunity has been taken to update the benchmarking (see Table 2). Comparing the performance of existing systems with the forecast for 2011 (the closest year temporally) shows that Line 1 is placed in the mid-range for all three indicators; this is shown graphically in Figure 1. TABLE 2 PATRONAGE BENCHMARKING System Route length (km) No. of stops Pax boardings (M) Pax kms (M) Pax boardings per stop (M) Pax boardings per route km (M) Pax kms per route km Manchester Metrolink Croydon Tramlink Sheffield Supertram Midland Metro Nottingham Line 1:

14 APPENDIX D FIGURE 1 PATRONAGE BENCHMARKING Manchester Metrolink Croydon Tramlink Sheffield Supertram Midland Metro Nottingham Line 1 (2011) 1 0 Boardings per stop (M) Boardings per route km (M) Pax kms per route km Manchester Metrolink Croydon Tramlink Sheffield Supertram Midland Metro Nottingham Line 2 (2011) 1 0 Boardings per stop (M) Boardings per route km (M) Pax kms per route km

15 APPENDIX D Summary 46 Significant resources have been allocated to the development of a hierarchical transport model fir Edinburgh that can simulate the complex interaction between many of the fundamental relationships between land uses and demand in congested urban areas. The model was based on the Scottish Executive s model, CSTM3, which was calibrated and validated in The model has not been subsequently adjusted using matrix estimation and is largely consistent with CSTM3. The strategic level Edinburgh model is well calibrated and validated although the detailed level model is lower than observed flows by about 10%. This model has been developed using industry standard model development procedures. 47 The model forecasts for the tram patronage and revenues has been benchmarked against observations for tram networks in Britain and these found to be comparable. Detailed analysis of the demand forecasts for Edinburgh airport indicates that the current forecasts in the model are low. The demand forecasts in the model are conservative and any model adjustments are likely to improve the case for the tram. 48 The Promoter, tie Ltd and tie s advisors consider the demand forecasts to be realistic and sufficiently robust to proceed with the scheme.

16 Annex A1

17 Annex A1

18 Annex A2

19 Annex A2

20 Information Note No 30 Annex B1 Title: Edinburgh Tram Bill Committee Subject: DAM Model Validation Ref: C32431\30&31 Version No: 2 Date: 18 August 2005 Author: Paul Leck Executive Summary This note has been prepared to collate the relevant information from the CSTM3 Highway and PT model Calibration / Validation reports, to demonstrate the implied validation of the DAM models. CSTM3 was validated for a 1997 base year the DAM models are a forecast of CSTM3 from the base year to The DAM Highway and PT models started life as a sub-area of the CSTM3 model and as such, the CSTM3 validation in the Edinburgh, Fife and Lothians area is of interest. All aspects of the original CSTM3 model including the Final Highway Assignment Model and the Final Public Transport Model were extensively audited by SIAS. The key objectives of the CSTM3 audit process were: to ensure that the model had been developed using procedures and methods consistent with Scottish Executive standards and best practice; and to establish the bounds of applicability of the model. The audit assessed the procedures used during model development and where possible made use of a systematic inspection of the input data. Emphasis was placed on an examination of the model outputs, such as traffic flows and travel times. The model outputs were also interpreted against the background of a qualitative assessment of both the quality of the input data and the reliability of the model development procedures to form a recommendation regarding the robustness of the model. The conclusions of the audit were that the Final Highway Assignment Model (FHAM) is appropriate for the assessment of inter-urban schemes all except the Renfrew Trunk Road route corridor within the model area i.e. all of the DAM modelled area. the Final Public Transport Model (FPTM) is appropriate for the assessment of public transport schemes within the Glasgow, Edinburgh and Stirling triangle, including cross-forth movements i.e. all of the DAM modelled area. In conclusion, the Edinburgh Tram Bill Committee, and its advisors can be reassured that the 1997 CSTM3 Models, which formed the base for the 2001 DAM Models were extensively audited against industry standards and are acceptable for use.

21 Information Note No 30(1) Annex B1 DAM Model Validation 1 Introduction 1.1 This note has been created to collate the relevant information from the CSTM3 Highway and PT model Calibration / Validation reports to demonstrate the implied validation of the DAM models. CSTM3 was validated for a 1997 base year the DAM models are a forecast of CSTM3 from the base year to The DAM Highway and PT models started life as a sub-area of the CSTM3 model and as such, the CSTM3 validation in the Edinburgh, Fife and Lothians area is of interest. 1.2 Also included are comments from the SIAS auditor s report of the CSTM3 Model. 1.3 The note has been produced at the request of tie, following a meeting on 8th July and subsequent work instruction of 5th August. 2 Highway Model Introduction 2.1 Validation is the process of checking how well the model compares with data independent of the calibration process and will be presented in the following ways: count data not used in calibration; 24 hour AAWDT comparison for count sites. Validation Sites 2.2 Traffic count data not used in calibration has been used for the purposes of validation. In total, 468 one-way counts were used to present the validation of the Final Highway Assignment Model (FHAM). 2.3 The GEH statistic has been used to assess the overall acceptability of the results. 2.4 Tables 2.1 to 2.3 show the Observed \ Modelled Screenline flows by time period for the screenlines in the Edinburgh area of CSTM3.

22 Information Note No 30(1) Annex B1 DAM Model Validation Table 2.1 AM Peak Observed/Modelled Screenline Flows Validation Screenlines Description Count Model M-C Absolute % GEH Count Model M-C Absolute % GEH Great King Street - Edinburgh % % 11.4 Great King Street - Edinburgh % % 1.6 Hederson Row - Edinburgh % % 4.6 Waverley Bridge - Edinburgh % % 2.6 Dundas Street - Edinburgh % % 1.2 Dundas Street - Edinburgh % % 1.8 Queensferry Street - Edinburgh % % 5.2 Hanover Street - Edinburgh % % 4.4 Hanover Street - Edinburgh % % 7.5 Hanover Street - Edinburgh % % 2.9 Meadows - Edinburgh % % 3.4 Mound - Edinburgh % % 7.5 Frederick Street - Edinburgh % % 3.3 Frederick Street - Edinburgh % % 5.1 Salamander Street - Edinburgh % % 7.4 Niddrie - Edinburgh % % 0.2 Dean Bridge - Edinburgh % % 1 Yotk Place - Edinburgh % % 6.9 Hanover Street - Edinburgh % % 2.6 Ferry Road - Edinburgh % % 5 Telford Road - Edinburgh % % 3.3 Leith Street - Edinburgh % % 9.7 Queen Street - Edinburgh % % 11.4 Queen Street - Edinburgh % % 2 Queen Street - Edinburgh % % 1.6 Queen Street - Edinburgh % % 5.2 North Charlotte St - Edinburgh % % 3 Ferry Road - Edinburgh % % 1.2 Abercromby Place - Edinburgh % % 11.6 Crewe Road - Edinburgh % % 1.7 Abercromby Place - Edinburgh % % 1.5 Melville Street - Edinburgh % % 8 Heriot Row - Edinburgh % % 1.7 Dublin Street - Edinburgh % % 3.8 Duddingston Road - Edinburgh % % 5.7 Cumberland Street - Edinburgh % % 6.8 Northumberland Street - Edinburgh % % 1 Cornwall Street - Edinburgh % % 10.7 Fettes Row - Edinburgh % % 11 Cumberland Street - Edinburgh % % 3.9 Royal Circus - Edinburgh % % 5.4 Scotland Street - Edinburgh % % 6.3 Doune Terrace - Edinburgh % % 5.5 Castle Street - Edinburgh % % 0.6 Ferry Road - Edinburgh % % 6 Sir Harry Lauder Rd - Edinburgh % % 7.8 Broughton Street - Edinburgh % % 9.9 Dean Park - Edinburgh % % 8.1 Constitution Street - Edinburgh % % 0.9 Inverleith Row - Edinburgh % % 1 Granton Road - Edinburgh % % 1.8 Great Junction Street - Edinburgh % % 10.4 Ferry Road - Edinburgh % % 1.6 Crewe Road South - Edinburgh % % 8 Crewe Road North - Edinburgh % % 10.2 West Granton Road - Edinburgh % % 1.3 West Granton Road - Edinburgh % % 0.7 Asda/Milton Road - Edinburgh % % 3.6 A702 Biggar Road - Edinburgh % % 6.2 A702 Biggar Road - Edinburgh % % 8.3 Lanark Road - Edinburgh % % 2

23 Information Note No 30(1) Annex B1 DAM Model Validation Table 2.2 Off Peak Observed/Modelled Screenline Flows Validation Screenlines Description Count Model M-C Absolute % GEH Count Model M-C Absolute % GEH Great King Street - Edinburgh % % 2.5 Great King Street - Edinburgh % % 1.4 Hederson Row - Edinburgh % % 1.7 Waverley Bridge - Edinburgh % % 2.4 Dundas Street - Edinburgh % % 12.8 Dundas Street - Edinburgh % % 10.8 Queensferry Street - Edinburgh % % 1.8 Hanover Street - Edinburgh % % 7.3 Hanover Street - Edinburgh % % 12 Hanover Street - Edinburgh % % 3.8 Meadows - Edinburgh % % 4.6 Mound - Edinburgh % % 12 Frederick Street - Edinburgh % % 5.3 Frederick Street - Edinburgh % % 0.9 Salamander Street - Edinburgh % % 5.2 Niddrie - Edinburgh % % 7.1 Dean Bridge - Edinburgh % % 4.9 Yotk Place - Edinburgh % % 0.7 Hanover Street - Edinburgh % % 3.8 Ferry Road - Edinburgh % % 8.3 Telford Road - Edinburgh % % 1.3 Leith Street - Edinburgh % % 3.7 Queen Street - Edinburgh % % 5.6 Queen Street - Edinburgh % % 2.1 Queen Street - Edinburgh % % 4.1 Queen Street - Edinburgh % % 0.8 North Charlotte St - Edinburgh % % 1.7 Ferry Road - Edinburgh % % 3.3 Abercromby Place - Edinburgh % % 3.7 Crewe Road - Edinburgh % % 5.2 Abercromby Place - Edinburgh % % 0.8 Melville Street - Edinburgh % % 9.9 Heriot Row - Edinburgh % % 5.9 Dublin Street - Edinburgh % % 2 Duddingston Road - Edinburgh % % 1 Cumberland Street - Edinburgh % % 4.2 Northumberland Street - Edinburgh % % 7.6 Cornwall Street - Edinburgh % % 3.6 Fettes Row - Edinburgh % % 0.7 Cumberland Street - Edinburgh % % 2 Royal Circus - Edinburgh % % 0.9 Scotland Street - Edinburgh % % 2.7 Doune Terrace - Edinburgh % % 1 Castle Street - Edinburgh % % 3.5 Ferry Road - Edinburgh % % 9.7 Sir Harry Lauder Rd - Edinburgh % % 3.3 Broughton Street - Edinburgh % % 4.2 Dean Park - Edinburgh % % 3.3 Constitution Street - Edinburgh % % 5.5 Inverleith Row - Edinburgh % % 9.6 Granton Road - Edinburgh % % 4.1 Great Junction Street - Edinburgh % % 2.9 Ferry Road - Edinburgh % % 4.5 Crewe Road South - Edinburgh % % 15.9 Crewe Road North - Edinburgh % % 13.3 West Granton Road - Edinburgh % % 4 West Granton Road - Edinburgh % % 7.3 Asda/Milton Road - Edinburgh % % 0.1 A702 Biggar Road - Edinburgh % % 2.7 A702 Biggar Road - Edinburgh % % 8.5 Lanark Road - Edinburgh % % 2.4

24 Information Note No 30(1) Annex B1 DAM Model Validation Table 2.3 PM Peak Observed/Modelled Screenline Flows Validation Screenlines Description Count Model M-C Absolute % GEH Count Model M-C Absolute % GEH Great King Street - Edinburgh % % 6.8 Great King Street - Edinburgh % % 3.2 Hederson Row - Edinburgh % % 0.4 Waverley Bridge - Edinburgh % % 3.4 Dundas Street - Edinburgh % % 1.3 Dundas Street - Edinburgh % % 2.7 Queensferry Street - Edinburgh % % 1 Hanover Street - Edinburgh % % 7.6 Hanover Street - Edinburgh % % 1 Hanover Street - Edinburgh % % 11.1 Meadows - Edinburgh % % 5 Mound - Edinburgh % % 1 Frederick Street - Edinburgh % % 2.1 Frederick Street - Edinburgh % % 6.1 Salamander Street - Edinburgh % % 1.3 Niddrie - Edinburgh % % 0.3 Dean Bridge - Edinburgh % % 10.7 Yotk Place - Edinburgh % % 4.6 Hanover Street - Edinburgh % % 5.9 Ferry Road - Edinburgh % % 14.8 Telford Road - Edinburgh % % 6.1 Leith Street - Edinburgh % % 6.3 Queen Street - Edinburgh % % 8 Queen Street - Edinburgh % % 0.4 Queen Street - Edinburgh % % 1.9 Queen Street - Edinburgh % % 0.6 North Charlotte St - Edinburgh % % 9.2 Ferry Road - Edinburgh % % 6.7 Abercromby Place - Edinburgh % % 10 Crewe Road - Edinburgh % % 1.5 Abercromby Place - Edinburgh % % 4.5 Melville Street - Edinburgh % % 5.2 Heriot Row - Edinburgh % % 3 Dublin Street - Edinburgh % % 4.2 Duddingston Road - Edinburgh % % 4.8 Cumberland Street - Edinburgh % % 9.6 Northumberland Street - Edinburgh % % 0.7 Cornwall Street - Edinburgh % % 9.8 Fettes Row - Edinburgh % % 8.9 Cumberland Street - Edinburgh % % 5.7 Royal Circus - Edinburgh % % 8.4 Scotland Street - Edinburgh % % 1.7 Doune Terrace - Edinburgh % % 9.2 Castle Street - Edinburgh % % 5 Ferry Road - Edinburgh % % 5.9 Sir Harry Lauder Rd - Edinburgh % % 8.2 Broughton Street - Edinburgh % % 6.3 Dean Park - Edinburgh % % 6.1 Constitution Street - Edinburgh % % 1 Inverleith Row - Edinburgh % % 5.8 Granton Road - Edinburgh % % 10.7 Great Junction Street - Edinburgh % % 0.8 Ferry Road - Edinburgh % % 12 Crewe Road South - Edinburgh % % 2.7 Crewe Road North - Edinburgh % % 7.6 West Granton Road - Edinburgh % % 3.3 West Granton Road - Edinburgh % % 0.5 Asda/Milton Road - Edinburgh % % 3.6 A702 Biggar Road - Edinburgh % % 3.1 A702 Biggar Road - Edinburgh % % 2.7 Lanark Road - Edinburgh % % 3.4

25 Information Note No 30(1) Annex B1 DAM Model Validation 2.5 Table 2.4 presents a summary of the validation site analysis for the entire model. Table 2.4 Validation Sites Analysis Number of Sites With GEH Value Time Period >12 AM 45% 62% 80% 90% 10% OP 54% 69% 87% 92% 8% PM 44% 61% 81% 90% 10% 2.6 The vast majority of sites exhibit a GEH statistic less than 12, there are some sites that do not (for the entire model 10% in the AM Peak, 8% in the Off Peak and 10% in the PM Peak). These sites were mainly were in rural areas and therefore not relevant to the DAM Model. 2.7 Overall the validation is acceptable and we have confidence in the model in the key areas of interest, in addition, within the urban areas of Glasgow, Edinburgh, Stirling, Perth, Dundee, Glenrothes, Kirkcaldy, Dunfermline, Kilmarnock and Dumfries the capacity restraint and generation effects of these towns are well reflected. 24 Hour AAWDT Comparison 2.8 Traffic counts supplied by then Scottish Office from ATC sites allowed a comparison of observed and modelled Annual Average Weekday Traffic (AAWDT) 24 hour totals. The modelled AAWDT equivalent values are obtained by factoring the assigned hourly values. Site specific factors were calculated giving the relationship between the assigned hours and Demand Model periods as follows: : ; the Off Peak factor is always six, as the counts are an average of the period ; : ; : 24 hours.

26 Information Note No 30(1) Annex B1 DAM Model Validation 2.9 These factors were used to multiply the modelled flows from each time period to a 24 hour total On analysis of the results of the comparison, grouped into sectors, there is good validation on the motorways and major A-roads and on the majority of other links, and only poor on links towards the periphery of the model, i.e. not applicable to the DAM model area. 3 PT Model Introduction 3.1 This section demonstrates the level of the public transport assignment s calibration and validation through a detailed analysis of: observed and modelled cordon flows; observed and modelled screenline flows; and timetabled and modelled bus journey times. PT Cordon Flows 3.2 For reporting purposes, the screenlines have been aggregated to cordons to compare AM Peak, Inter Peak and PM Peak observed and modelled flows. Tables 3.1 to 3.5 compare modelled and observed public transport flows (rail and bus combined) for each cordon and specific screenlines. Table 3.1 Edinburgh City Centre Cordon Direction Time Period Observed Passenger Flow Modelled Passenger Flow Difference % Difference Inbound AM % IP % PM % Outbound AM % IP % PM % 3.3 It is evident from Table 3.1 that, at the Edinburgh City Centre cordon, modelled passenger flows exceed observed passenger flows in each time period and in both directions. It is possible that this is the influence of the VIPS data, which represents all movements within the Edinburgh Suburban Cordon and the majority of those crossing the Edinburgh City Centre Cordon, and has simply been updated using trend information supplied by Lothian Region Transport Ltd. However, despite the discrepancies between observed and modelled flows the differences are acceptable.

27 Information Note No 30(1) Annex B1 DAM Model Validation Table 3.2 Edinburgh Suburban Cordon Direction Time Period Observed Passenger Flow Modelled Passenger Flow % Difference Difference Inbound AM % IP % PM % Outbound AM % IP % PM % 3.4 There is an acceptable level of calibration across this cordon which covers all directions into and out of Edinburgh at the edge of the urban area. Absolute differences between observed and modelled flows are small, and the percentage difference ranges from 3.8 to -0.1% inbound and 2.1 to 1.8% outbound. Table 3.3 Edinburgh Bypass Cordon Direction Time Period Observed Passenger Flow Modelled Passenger Flow % Difference Difference Inbound AM % IP % PM % Outbound AM % IP % PM %

28 Information Note No 30(1) Annex B1 DAM Model Validation 3.5 This cordon is located along the route of the Edinburgh City Bypass to the west of the city. There is a good level of calibration, with small absolute differences in observed and modelled flows. However, there is a general bias apparent in both directions and all time periods in favour of rail. In AM Outbound and PM Inbound, where the count is relatively small, due to the tidal nature of the screenline, this leads to percentage differences greater than 10. Table 3.4 Forth Bridge Screenline Direction Time Period Observed Passenger Flow Modelled Passenger Flow Difference % Difference Southbound AM % IP % PM % Northbound AM % IP % PM % 3.6 The Forth Bridge screenline shows a good degree of calibration with small absolute differences in observed and modelled passenger flows. The percentage difference in observed and modelled passenger flows ranges from between 2.7 and -2.8% southbound and 1.3 and 5.7% northbound. Table 3.5 Linlithgow Screenline Direction Time Observed Modelled Period Passenger Passenger % Flow Flow Difference Difference Eastbound AM % IP % PM % Westbound AM % IP % PM % 3.7 This screenline, consisting of rail and bus flows east of Linlithgow, shows an acceptable degree of calibration, with small absolute differences in observed and modelled passenger flows. However, due to the relatively small magnitude of count, the percentage differences are acceptable. Screenline Flows 3.8 The analysis of the modelled screenline flows make use of a summary statistic known as GEH, which is defined as: GEH = (((observed-modelled)*(observed-modelled))/(0.5*(observed+modelled))) 0.5

29 Information Note No 30(1) Annex B1 DAM Model Validation 3.9 The GEH value is designed to be more tolerant of large percentage differences at lower flows as one would not normally be concerned about a modelled flow which differed from a count by 40% if the count was only 100, but one would be if the count were The reason for introducing such a statistic is the inability of either the absolute difference or the relative difference to reflect differences over the wide range of flows contained in the Final Public Transport Model (FPTM) Given the size, complexity and magnitude of the passenger flows contained within the FPTM, we would normally expect screenline GEH values to meet the same exacting targets used to achieve a high standard of calibration of the FHAM, namely: <5 60%; <7 80%; <10 95%; and <12 100% The screenline locations used for calibration purposes are those used in the MVESTM process and they provide the basis for calibration of the model routing as well as the input for matrix estimation. As no independent passenger count data was available for model validation, section 5.3 presents the model validation using a comparison between modelled and observed bus journey times Tables 3.6 to 3.8 present tables for the AM peak, Inter peak and PM peak observed/modelled screenline flows for the DAM Model area. The GEH statistic, reported in the last column of each table and described above, has been used to assess the overall acceptability of the screenline results.

30 Information Note No 30(1) Annex B1 DAM Model Validation Table 3.6 AM Peak Screenline Flows Screenline Name Mode Count Assigne Diff % Diff GEH 1 Edinburgh Suburban E IB Bus % Edinburgh Suburban E OB Bus % Edinburgh Suburban E IB Rail % Edinburgh Suburban E OB Rail % Edinburgh Suburban SE IB Bus % Edinburgh Suburban SE OB Bus % Edinburgh City Centre E IB Bus % Edinburgh City Centre E OB Bus % Edinburgh City Centre SE IB Bus % Edinburgh City Centre SE OB Bus % Edinburgh City Centre S IB Bus % Edinburgh City Centre S OB Bus % Edinburgh City Centre SW IB Bus % Edinburgh City Centre SW OB Bus % Edinburgh City Centre NW IB Bus % Edinburgh City Centre NW OB Bus % Haymarket-Edinburgh CC IB Rail % Haymarket-Edinburgh CC OB Rail % Edinburgh Suburban SW IB Bus % Edinburgh Suburban SW OB Bus % Edinburgh Suburban SW IB Rail % Edinburgh Suburban SW OB Rail % Edinburgh Suburban W IB Bus % Edinburgh Suburban W OB Bus % Edinburgh Suburban W IB Rail % Edinburgh Suburban W OB Rail % Edinburgh Suburban NW IB Bus % Edinburgh Suburban NW OB Bus % Edinburgh Bypass W IB Bus % Edinburgh Bypass W OB Bus % Edinburgh Bypass W IB Rail % Edinburgh Bypass W OB Rail % Forth Bridge IB Bus % Forth Bridge OB Bus % Forth Bridge IB Rail % Forth Bridge OB Rail % Linlithgow EB Bus % Linlithgow WB Bus % Linlithgow EB Rail % Linlithgow WB Rail % 5.51

31 Information Note No 30(1) Annex B1 DAM Model Validation Table 3.7 Inter Peak Screenline Flows Screenline Name Mode Count Assigne Diff % Diff GEH 1 Edinburgh Suburban E IB Bus % Edinburgh Suburban E OB Bus % Edinburgh Suburban E IB Rail % Edinburgh Suburban E OB Rail % Edinburgh Suburban SE IB Bus % Edinburgh Suburban SE OB Bus % Edinburgh City Centre E IB Bus % Edinburgh City Centre E OB Bus % Edinburgh City Centre SE IB Bus % Edinburgh City Centre SE OB Bus % Edinburgh City Centre S IB Bus % Edinburgh City Centre S OB Bus % Edinburgh City Centre SW IB Bus % Edinburgh City Centre SW OB Bus % Edinburgh City Centre NW IB Bus % Edinburgh City Centre NW OB Bus % Haymarket-Edinburgh CC IB Rail % Haymarket-Edinburgh CC OB Rail % Edinburgh Suburban SW IB Bus % Edinburgh Suburban SW OB Bus % Edinburgh Suburban SW IB Rail % Edinburgh Suburban SW OB Rail % Edinburgh Suburban W IB Bus % Edinburgh Suburban W OB Bus % Edinburgh Suburban W IB Rail % Edinburgh Suburban W OB Rail % Edinburgh Suburban NW IB Bus % Edinburgh Suburban NW OB Bus % Edinburgh Bypass W IB Bus % Edinburgh Bypass W OB Bus % Edinburgh Bypass W IB Rail % Edinburgh Bypass W OB Rail % Forth Bridge IB Bus % Forth Bridge OB Bus % Forth Bridge IB Rail % Forth Bridge OB Rail % Linlithgow EB Bus % Linlithgow WB Bus % Linlithgow EB Rail % Linlithgow WB Rail % 2.56

32 Information Note No 30(1) Annex B1 DAM Model Validation Table 3.8 PM Peak Screenline Flows Screenline Name Mode Count Assigne Diff % Diff GEH 1 Edinburgh Suburban E IB Bus % Edinburgh Suburban E OB Bus % Edinburgh Suburban E IB Rail % Edinburgh Suburban E OB Rail % Edinburgh Suburban SE IB Bus % Edinburgh Suburban SE OB Bus % Edinburgh City Centre E IB Bus % Edinburgh City Centre E OB Bus % Edinburgh City Centre SE IB Bus % Edinburgh City Centre SE OB Bus % Edinburgh City Centre S IB Bus % Edinburgh City Centre S OB Bus % Edinburgh City Centre SW IB Bus % Edinburgh City Centre SW OB Bus % Edinburgh City Centre NW IB Bus % Edinburgh City Centre NW OB Bus % Haymarket-Edinburgh CC IB Rail % Haymarket-Edinburgh CC OB Rail % Edinburgh Suburban SW IB Bus % Edinburgh Suburban SW OB Bus % Edinburgh Suburban SW IB Rail % Edinburgh Suburban SW OB Rail % Edinburgh Suburban W IB Bus % Edinburgh Suburban W OB Bus % Edinburgh Suburban W IB Rail % Edinburgh Suburban W OB Rail % Edinburgh Suburban NW IB Bus % Edinburgh Suburban NW OB Bus % Edinburgh Bypass W IB Bus % Edinburgh Bypass W OB Bus % Edinburgh Bypass W IB Rail % Edinburgh Bypass W OB Rail % Forth Bridge IB Bus % Forth Bridge OB Bus % Forth Bridge IB Rail % Forth Bridge OB Rail % Linlithgow EB Bus % Linlithgow WB Bus % Linlithgow EB Rail % Linlithgow WB Rail % Table 3.9 summarises the screenline GEH analysis for each time period and all 78 screenlines used in model calibration for FPTM. The GEH values for the screenlines show a GEH value of less than the ideal target of 12 for all screenlines in the DAM Model area, indicating that the public transport model is well calibrated. The maximum screenline GEH value is in the AM peak, 5.30 in the Inter peak

33 Information Note No 30(1) Annex B1 DAM Model Validation and in the PM peak for CSTM3, none of which are in the DAM modelled area. Table 5.11 clearly shows that all screenline values with the exception of one in the AM Peak fall within the target levels, and the vast majority are less than 5. Table 3.9 Screenline Flow GEH Analysis % Screenline Flows lower than the GEH value Time Period <5 <7 <10 <12 Target AM peak Inter peak PM peak Table 3.10 shows that all rail screenlines and the majority of bus screenlines fall within the even more exacting GEH values of less than 3%. Table 3.10 Screenline Flow GEH Analysis by Mode Number of Screenline Flows Time Period <3% 3-5% >5 AM Peak Bus Rail Inter Peak Bus Rail PM Peak Bus Rail Audit Report 4.1 All aspects of the original CSTM3 were extensively audited by SIAS. The conclusions most relevant to the use of the CSTM3 model within the Edinburgh area are as follows: the Final Highway Assignment Model (FHAM) is appropriate for the assessment of inter-urban schemes all except the Renfrew Trunk Road route corridor within the model area i.e. all of the DAM modelled area. the FHAM s network coverage is good and the level of detail, although variable, is appropriate and acceptable given the focus of the model

34 Information Note No 30(1) Annex B1 DAM Model Validation the trip matrices for the route corridors within the Glasgow, Edinburgh, Dundee triangle have been assessed as having an acceptable level of robustness the assignment procedure used in the CSTM3 is considered appropriate for the congested and semi-congested areas of the model. That is the assignment procedure is considered appropriate for the Central Belt, an area which forms the immediate focus of the CSTM3 through the application of MVESTM, the global level of calibration to modelled hour link counts in the FHAM is good. Additionally, more detailed analysis has indicated that this good level of calibration is maintained on the Trunk Road network at the route corridor level the global validation of the FHAM to link traffic flows (at the modelled hour and 24 hour levels) is acceptable. There are good levels of validation quality on the Trunk Road network in the Edinburgh to Glasgow corridor, the Carlisle to Stranraer corridor and the Edinburgh conurbation the journey time validation comparisons meet DMRB required standards over the Glasgow, Edinburgh, Dundee triangle as a whole the CSTM3 Final Public Transport Model (FPTM) is appropriate for the assessment of public transport schemes within the Glasgow, Edinburgh and Stirling triangle, including cross-forth movements all reasonable and appropriate data sources have been utilised during model development ; overall the network coverage is good and the level of detail, although variable, is appropriate given the current focus of the model a localised anomaly has been identified in the North Berwick area, which will need addressing should the model be applied to test a scheme in this area ; the lines information (bus and rail routes) contained within the FPTM is appropriate for the immediate application and use of the model within the Glasgow, Edinburgh and Stirling triangle the fares model used in the FPTM is simple and appropriate for current conditions and the purpose of the CSTM3 ; the FPTM matrices are representative of public transport trips within the Central Belt ; the representation of public transport trips within Dundee, Perth and the major towns in Fife is extremely limited although all movements from Dundee, Perth and the major towns in Fife to/from Edinburgh or Glasgow are represented ; it is also considered that the procedures used to develop the FPTM matrices have made the best use of the available data, without rebuilding matrices using the trip record data from the donor models ;

35 Information Note No 30(1) Annex B1 DAM Model Validation the model calibration to passenger counts indicates that the FPTM has a very good fit with the area calibrated. The mode split comparisons (between bus and rail) also show a good fit ; within the Central Belt the overall level of generalised cost has the expected order of magnitude and should therefore form a suitable input into the Demand Model. However, it was noted that in some instances uncharacteristic modal choice can occur in Glasgow and Edinburgh when a transfer penalty and the relatively low walk time weighting factor (i.e. 1.6 compared to a HEN2 value of 2.0) combine to make walking over comparatively long distances a more attractive option than taking a local bus service PL

36 Annex B2 Number: 01 Public Domain: No Project Title: Consultant: Date: February 1999 Type of Model: TRIPS City Wide Model Zones: 587 Validation Coverage: Screenlines (26) and sites on screenlines Validation Performance: Screenlines (26) AM OP PM <5% 11(42%) 13(50%) 12(46%) Sites GEH <5 44% 51% 52% <10 72% 75% 73%

37 Annex B2 Number: 02 Public Domain: No Project Title: Consultant: Date: November 2000 Type of Model: Sub regional / Conurbation TRIPS AM peak Zones: 330 Validation Coverage: Calibration and Validation Counts (Screenlines too few links) Validation Performance: GEH <5 67%

38 Annex B2 Number: 03 Public Domain: No Project Title: Consultant: Date: 2002 Type of Model: SATURN developed for use in TRANSLINK appraisal Zones: 183 Validation Coverage: 13 sites Validation Performance: AM Inter peak GEH <5 10(77%) 10(77%)

39 Annex B2 Number: 04 Public Domain: No Project Title: Consultant: Date: Type of Model: SATURN City Centre Model Zones: Validation Coverage: Individual sites (about 110) Validation Performance: AM Inter Peak PM GEH <5 86(80)% 94(86%) 84(81%)

40 Annex B2 Number: 05 Public Domain: No Project Title: Consultant: Date: March 2001 Type of Model: Trips urban area Zones: 180 Validation Coverage: 14 Screenlines and 148 individual sites on the screenlines Validation Performance: Screenlines (14) AM Inter < Individual Sites (148) R Slope

41 Annex B2 Number: 06 Public Domain: No Project Title: Consultant: Date: Undated but assumed 1996/97 Type of Model: SATURN sub-regional model emphasis on Thames Crossings Zones: 390 Validation Coverage: River Thames Screenline plus other screenlines plus individual sites Validation Performance: Thames Screenline Sites (14) AM Inter Peak GEH <5 7 (50%) 6 (43%) <10 10 (71%) 12 (86%) Screenlines (14) AM Inter Peak <5% 4 (29%) 4 (29%)

42 Annex B2 Number: 07 Public Domain: Yes Project Title: A3 Hindhead Study Consultant: MVA Date: June 2001 Type of Model: SATURN Regional with local emphasis at Hindhead Zones: 150 Validation Coverage: Screenline Sites (74) (most screenlines <5 sites) % flow difference criteria used due to low flows on many sites Validation Performance: AM Inter Peak PM Sunday Number of Sites inside Guidelines >95%

43 Annex B2 Number: 08 Public Domain: Yes Project Title: Central Scotland Transport Model (CSTM3A) Consultant: MVA Date: November 2001 Type of Model: TRIPS National Model Zones: 1296 Validation Coverage: Calibration Key Links (131) Links on Calibration screenlines Independent Links Validation Performance: Key Links AM Inter PM GEH <5 70% 79% 79% <10 93% 98% 95% Links on Calibration Screenlines <5 65% 78% 79% <10 94% 98% 95% Independent Sites (288) <5 39% 51% 41% <10 69% 78% 70%

44 Annex B2 Number: 09 Public Domain: Yes Project Title: West Sussex County Traffic Model Consultant: MVA Date: Type of Model: TRIPS County Model AM Peak Zones: Validation Coverage: 95 Independent sites plus 5 screenlines Validation Performance: Independent Sites GEH <5 40 (42%) <10 71 (75%) screenline sites GEH <5 (93%)

45 Annex B2 Number: 10 Public Domain: Yes Project Title: South Hampshire Rapid Transit Consultant: SDG Date: February 2002 Type of Model: SATURN Zones: 161 Validation Coverage: 12 Screenlines no mention of individual sites Validation Performance: AM All screenlines <5% Inter Peak 8 screenlines <10% PM 8 screenlines <10%

46 Annex B2 Number: 11 Public Domain: Yes Project Title: M1 Junction 19 Road Based Study Consultant: Halcrow Group Limited Date: February 2002 Type of Model: SATURN Regional Model Zones: 111 Validation Coverage: Selected screenlines in vicinity of M1 J19 plus individual sites on screenlines Validation Performance: AM Peak Hour 42 sites GEH <5 37 (88%) <10 42 (100%) Inter Peak GEH <5 39 (93%) <10 42 (100%) PM Peak GEH <5 37 (88%) <10 42 (100%)

47 Annex B2 Number: 12 Public Domain: No Project Title: Consultant: Date: October 2002 Type of Model: SATURN Free Standing City Zones: 315 Validation Coverage: Screenlines 14 Individual sites 88 (all on screenlines) Validation Performance: Screenlines AM Inter Peak <5% 7 (50%) 3 (21%) Individual Sites AM Inter Peak GEH <5 51 (58%) 45 (51%) <10 71 (81%) 72 (82%)

48 Annex B2 Number: 13 Public Domain: No Project Title: Consultant: Date: April 1999 Type of Model: SATURN Update of Mott MacDonald Work (No 14) Zones: 185 Validation Coverage: Crawley Cordon (24 sites) Validation Performance: AM Peak GEH <5 12 (50%) <10 22 (92%)

49 Annex B2 Number: 14 Public Domain: No Project Title: Consultant: Date: May 1996 Type of Model: SATURN Group of three close towns Zones: 185 Validation Coverage: 10 Screenlines for each time period 56/8 Individual sites for each screenline Validation Performance: Screenlines AM Inter PM GEH < Diff <5% Individual Sites AM Inter PM GEH < <5 37 (64%) 32 (84%) 47 (84%) <10 56 (97%) 38 (100%) 53 (95%)

50 Annex B2 Number: 15 Public Domain: No Project Title: Consultant: Date: June 2001 Type of Model: SATURN Free Standing City with some small settlements in rural area Zones: 290 Validation Coverage: Screenlines Total (14) Individual sites on screenlines (100) Validation Performance: Screenline Totals 3 (21%) <5% 3 (21%) <10% 11 (79%) <20% Individual Sites Notes GEH <5 33 (33%) <10 64 (64%) Level of Validation attributed to Grid Pattern of Central Milton Keynes

51 Annex B2 Number: 16 Public Domain: No Project Title: Consultant: Date: November 2001 Type of Model: TRIPS Small free standing market town Zones: 82 Validation Coverage: Screenline Totals Percent Differences Individual Sites (34) regression analysis on screenlines Validation Performance: Screenline Totals <1% (5%) Individual Sites Slope 1.01 ( ) Co-efficient of Correlation 0.82 (>0.95)

52 Annex B2 Number: 17 Public Domain: Yes Project Title: A453 Nottingham to M1 J24 MMS Consultant: Pell Frischmann Date: July 2002 Type of Model: Strategic Routes around M1 J24 Zones: 192 Validation Coverage: Comparisons on individual RSI sites (22) and Motorway sites (16) and Independent sites (16) Validation Performance: RSI Sites (22) AM Inter PM GEH <5 15 (68%) 19 (86%) 18 (82%) GEH <10 22 (100%) 22 (100%) 22 (100%) Motorway Sites (16) AM Inter PM GEH <5 7 (44%) 9 (56%) 7 (44%) GEH <10 10 (63%) 11 (69%) 10 (63%) Independent Sites (16) GEH <5 7 (44%) 9 (56%) 9 (56%) GEH <10 14 (88%) 13 (81%) 13 (81%) Total Sites 54

53 Annex B2 Number: 18 Public Domain: Yes Project Title: Hull Multi-Modal East West Corridor Study Consultant: Oscar Faber Date: October 2001 Type of Model: TRIPS Regional Model Zones: 164 Validation Coverage: Screenlines and Individual Sites (29) in the Screenlines - GEH and % difference Validation Performance: AM Inter PM GEH <5 24 (83%) 25 (86%) 18 (62%) <10 28 (97%) 27 (93%) 27 (93%)

54 Annex B2 Number: 19 Public Domain: Yes Project Title: East Midlands Multi Modal Study Consultant: WS Atkins Date: May 2001 Type of Model: Multi Modal Model covering M1 Zones: about 300 Validation Coverage: Individual Sites Validation Performance: AM GEH <5 69 (56%) <10 98 (79%)

55 Annex B2 Number: 20 Public Domain: No Project Title: 2002 Glasgow SATURN Model Consultant: MVA Date: October 2002 Type of Model: City Wide SATURN Model extending to adjacent suburban areas Zones: 743 Validation Coverage: 1109 individual sites Validation Performance: AM Inter PM GEH <5 70% 81% 68% <10 91% 96% 90%

56 Information Note No 32 Annex C1 Title: Edinburgh Tram Bill Committee Subject: PT Model Validation Ref: C32431\31 Version No: 1 Date: 19 October 2005 Author: Paul Leck 1 Introduction 1.1 This note has been produced to collate relevant information from PT model validation reports to consider the validation of the DAM models in context with other studies. CSTM3 was validated for a 1997 base year the DAM models are a forecast of CSTM3 from the base year to The DAM PT model started life as a sub-area of the CSTM3 model and as such, the CSTM3 PT validation in the Edinburgh, Fife and Lothians area is of interest. This note discusses PT Model Validation and compares CSTM3 validation with other PT models. Full details of the validation of CSTM3 in the DAM modelled area are given in Information Note PT Assignment Validation 2.1 Recent Guidance on PT model validation is given on the DfT Webtag website in TAG unit , Section 10 (June 2005). 2.2 The guidance suggested is given below in paragraphs Validation of the assignment should involve comparing modelled and observed: passenger flows across screenlines and cordons, usually by public transport mode and sometimes at the level of individual bus or train services; and passengers boarding and alighting in urban centres. 2.4 Across modelled screenlines, modelled flows should, in total, be within 15% of the observed values. 2.5 On individual links in the network, modelled flows should be within 25% of the counts, except where observed flows are particularly low (less than 150). 2.6 The guidance given for PT assignment validation differs from advice in DMRB for Highway Assignment Models which suggests that more stringent GEH Statistics should be used. Although PT assignment models do not have to use GEH Statistics, they are quite often quoted by practitioners in validation reports and values are compared to the Highway guidelines. DMRB guidance on Highway model validation can be found in DMRB Volume 12 Section 2 Part 1 Chapter 4 paragraphs to Note that on the DfT website in Annex E of Major scheme appraisal in local transport plans: Part 3, it does advise that it maybe be appropriate to supplement the recommended information with the GEH

57 Information Note No 32(1) Annex C1 DAM Model Validation statistic. It recognises the difficulty that the level of variation in public transport link flow data is likely to be greater than equivalent highway data due to the difficulty of accurately assessing passenger volumes. This can lead to a high margin of error within public transport survey data and could affect the degree of validation achieved. 2.7 In order to compare the PT model of the DAM Area of CSTM3 with other PT models, GEH Statistics will be examined along with the suggested criteria detailed above in 2.4 & 2.5 (where quoted in reports). 3 Comparison of Actual Model Validation Levels 3.1 The reported PT model validation statistics for a number of studies have been assembled. In addition to the DAM Modelled Area of CSTM3, there are 8 studies consisting of 6 studies by MVA and 2 by other consultants. The information for the other studies has been acquired through a report made available to MVA in the course of a project and the other as part of this exercise. 3.2 A summary of the 9 studies is presented in Table 3.1. These studies cover a range of towns and cities in the UK, and some cover more than one urban area. Where given in reports, GEH Statistics are presented as the percentage of links within percentage limits. Where it has been possible to calculate, or where values have been quoted, the recommended validation criteria given in section 2.4 & 2.5 are presented. 3.3 Taking each criteria in turn, Table 3.1 demonstrates that the validation of the DAM modelled area of CSTM3 is of a very high standard. 3.4 Examining the stringent GEH Statistics, the CSTM3 values are 100% less than 10 for the bus and rail count sites in the DAM area. The percentages fall to 79% and 94% for Rail and Bus respectively for GEH values less than 5, but these percentages are still better than a number of accepted studies. 3.5 On consideration of the recommended criteria for assessing PT models, again the DAM Modelled Area of CSTM3 performs very well. Some 98% and 100% of modelled bus and rail flows greater than 150 respectively were less than 25% from the observed value. 90% of modelled cordon counts were within 15% of the observed values. 3.6 The range of values of accepted models varies greatly and the Dam Modelled Area of CSTM3 ranks highly amongst the examples found. The examples range from a PT Model at the national level down to a study of a town centre. The levels of validation accepted depend on the size and complexity of the model.

58 Information Note No 32 Annex C1 Table 3.1 Summary of PT Model Validation Levels % GEH<5 %GEH<7.5 %GEH<10 Flow>150 and % diff <25% Within 15% of Observed DAM Area of CSTM3 Bus count sites 94% 98% 100% 98% Rail count sites 79% 100% 100% 100% Combined Cordons 90% PT Model 1 Bus count sites 93% 100% 100% 85% Bus boarding / alighting 90% 96% 98% 100% Rail boarding / alighting 100% 100% 100% 100% PT Model 2 Bus count sites 92% 97% 99% 90% Rail boarding / alighting 99% 100% 100% 96% LRT boarding / alighting 99% 99% 99% 95% PT Model 3 Bus count sites 72% 90% 92% 72% Bus boarding / alighting breakdown not given breakdown not given 100% breakdown not given Rail boarding / alighting 86% 97% 100% 86% Bus screenlines 92% PT Model 4 Bus count sites 46% 67% 80% 37% Rail boarding / alighting 50% 67% 83% 50% PT Model 5 Bus count sites 93% not presented 100% not presented Rail boarding / alighting 100% 100% 100% not presented PT Model 6 Bus count sites 58% 83% 94% 31% Bus boarding / alighting 78% 87% 91% n/a Bus screenlines 96% PT Model 7 Bus count sites 35% 46% 52% 19% Rail count sites 8% 31% 42% 31% Bus screenlines 22% Rail screenlines 50% PT Model 8 Screenlines not given not given not given not given 50%

59 Information Note No 32 Annex C1 4 Conclusion 4.1 The guidance for PT Models recognises the difficulty associated with this process and the criteria to assess the validation are not as rigorous as those of a Highway model. However, similar methods to assessing Highway assignment models can be applied to complement the basic requirements. 4.2 A comparison of CSTM3 validation within the DAM Modelled area with a number of other studies shows a wide range of acceptable values, showing that acceptability also takes in to account other factors such as the size and complexity of the model and its intended application. 4.3 Even in consideration of the above, the CSTM3 Dam Modelled Area performs well against all criteria including additional GEH Statistics when compared with a number of other PT Model applications. In the above comparison, CSTM3 is in the top third of the models reported here when measured against most of the criteria. It also performs well against the criteria given in the recent guidance.

60 Annex C2 Title: Edinburgh Tram Bill Committee Subject: Highway Model Validation Ref: C32431\31 Version No: 1 Date: 20 October 2005 Author: Scott Leitham 1 Introduction 1.1 In 2003, MVA produced a note for Glasgow City Council which discussed published validation criteria to be used in highway assignment modelling. It also reported the levels of validation achieved in 20 different highway models in order to provide context for the model validation which was under discussion at that time. The substantive elements of that paper are also relevant for the Edinburgh modelling and are reproduced in the remaining sections. 2 Assignment Model Validation 2.1 Guidance on model validation is included in DMRB Volume 12 Section 2 Part 1 Chapter 4 paragraphs to This guidance is recommended practice for the whole of the UK. The Republic of Ireland has also adopted the guidance along with other aspects of DMRB. 2.2 In addition to general guidance on model validation, DMRB also includes validation acceptability guidelines. These are quantitative guidelines based on comparisons of observed flows with those produced by the model. Three methods are described. 1) Comparison of Assigned Hourly Flows with observed flows percentage comparisons. 2) Comparison of Assigned Hourly Flows with observed flows GEH statistic comparisons. 3) Correlation Analysis between Assigned Hourly Flows and observed flows. 2.3 Acceptability Guidelines for methods 1 and 2 are included in Table 4.2 in DMRB, the GEH statistic being defined in paragraph Acceptability guidelines for method 3 are also included in paragraph An abstract of the relevant paragraphs of DMRB is included as Annex A to this note. 2.4 For methods 1 and 2 it is not clear whether the guidelines are such that either the percentage comparisons should be used or the GEH comparisons, or both. Most practitioners appear to use one or the other, mostly the GEH method. Figure One attached shows the comparison between the two methods. The average GEH values for the percentage comparisons varies from 7.0 down to 4.0 for flows 200 vph to 700 vph, then up to 7.8 for a flow of 2700 vph, falling thereafter for flows >2700 vph. For flows between about 400 and 1100vph the percentage comparison method is more stringent than the GEH method. 2.5 One thing to be noted in comparison methods 1 and 2 is that the unit of flow is vehicles per hour. Since the GEH statistic is not unit free, different guideline values will be relevant if units of flow are different from vehicles per hour (e.g. PCUs per hour). The units for GEH are the square root of flow,

61 Annex C2 so that for example with a pcu/vehicle ratio of about 1.2 the GEH guideline value moves from 5.0 to 5.5. Alternatively both assigned and observed flows can be divided by 1.2 prior to calculating the GEH statistic and the DMRB value of 5.0 retained as the guideline. 2.6 The correlation analysis in comparison method 3 tends to be used much less than methods 1 and 2. It is most commonly used when the validation data consists of a number of count sites spread around the model area, as opposed to count sites being organised into well defined screenlines and cordons. There are no obvious ways to compare the correlation co-efficient guidelines with the other methods. 2.7 There are a number of factors to be taken into account when considering how a real validation exercise compares with the guidelines in DMRB as follows. The guidelines are guidelines and not rigid criteria. There does not appear to be any indication in DMRB as to where these guidelines are derived from. There is a common view amongst practitioners that they are mainly relevant to small scale studies. In a real study, the validation process is almost always constrained by timescale and budget. The scale of the study is highly relevant. Small bypass type studies with straightforward networks and limited route choices will inevitably have models which are easier to calibrate and validate. The intended model application is important. There is mention in DMRB of models developed for the appraisal of a specific scheme and having the highest validation quality in the area of the study which is of direct relevance to the scheme. In contrast, many urban models are developed with much wider applications in mind and consequently the model development and validation process will be more complex and more onerous. Matrix estimation is a commonly used technique in model development. There are two main methods, MVESTM in the TRIPS package and ME2 in the SATURN package. MVESTM is a superior methodology since, as well as making use of a wider range of data sources, it allows sites to be grouped into screenlines which gives emphasis to matrix adjustments which are largely independent of flows on individual links. The SATURN based ME2 method on the other hand does not allow grouping of sites into screenlines and consequently focuses on individual sites. In practice this tends to give a better model validation when comparing observed and assigned link flows. Selective reporting may occur. It is possible to report on model validation selectively such that relatively poor comparisons are omitted. This should always be borne in mind when comparing validation performance between models. Flow levels affect validation. Within studies which model inter peak as well as peak, we would expect a better validation in the lower flow time periods. 3 Comparison of Actual Model Validation Levels 3.1 What is evident from the research we have undertaken is that, in general, acceptance criteria are seldom set by practitioners. What tends to happen is that validation achieved is simply reported and the reader is left to consider it in light of the DMRB criteria. MVA s practice has tended to be to set

62 Annex C2 criteria in advance based on the size and complexity of the model but this has not always been the case. 3.2 The reported validation statistics for a number of studies have been assembled. There are 20 studies consisting of 7 studies by MVA and 13 by other consultants. The information for the other studies has been acquired in part through reports made available to MVA in the course of separate projects, and from the public domain on the internet. 3.3 Details of each study are included in Annex B to this note. 3.4 The reported statistics have been extracted either from LMVRs or from the appropriate chapter in a more general study report. In some cases there may exist more detailed technical notes (which we have not seen) which give more information about particular features of the validation. 3.5 An overall summary of the 20 studies has been included in Table 1. This shows the reported percentage of links (or screenlines where noted) with GEH values <5 and <10, or the percentage of links meeting the percentage flow comparison guidelines. 3.6 The following summarises the key features of Table 1. The percentage of links meeting the DMRB guidelines ranges from 33% to 95% in the AM Peak, from 43% to 95% in the Inter Peak and from 52% to 95% in the PM Peak. In the 11 studies which have an Inter Peak model in addition to a morning peak, the percentage of links meeting the DMRB guidelines is higher for the Inter Peak in almost all cases. For the 17 studies which have AM Peak percentages of links meeting flow comparison guidelines, we can see that 50% of studies have less than 62% of links meeting flow comparison guidelines, and 85% of studies do not meet the DMRB guidelines. For the Inter Peak, where applicable, 50% of studies have less than 80% of links meeting flow comparison guidelines, and 67% of studies do not meet the DMRB guidelines. For the PM Peak, 50% of studies have less than 68% of links meeting the DMRB flow comparison guidelines, and 72% of studies do not meet the DMRB guidelines; The two studies employing the correlation method have co-efficients of 0.82 and 0.88, which are both lower than the DMRB guideline figure of 0.95.

63 Annex C2 4 Conclusion 4.1 It must be noted that the GEH statistic cannot be read in isolation. It is only an indicator of the model validation and says nothing about the quality of the matrix for forecasting or the area covered by the counts within the model. 4.2 An examination of the reported validation statistics for 20 studies shows that there is a wide range of percentage of links with flow comparisons meeting DMRB guidelines. In the AM Peak 85% of studies do not meet the guidelines. 4.3 The mid-point for the studies is at about 60%, that is, only one half of studies achieve more than 60% of link flow comparison meeting DMRB guidelines. Edinburgh Modelling 4.4 For CSTM3, similar validation statistics were published covering the entire study area (ie not just Edinburgh). These are shown in Table 4.1 below.

64 Annex C2 Table 4-1 CSTM Validation Sites Analysis Number of Sites With GEH Value Time Period >12 AM 45% 62% 80% 90% 10% OP 54% 69% 87% 92% 8% PM 44% 61% 81% 90% 10% 4.5 These statistics place the CSTM3 highway model well within the range of results seen in a wide range of other models ie it can be regarded as industry standard, especially when the spatial scale of the model is taken into consideration.

65 Information Note No 31 Annex D Title: Use of CEC / tie LUTI Model Subject: 2001 Detailed Assignment Model Comparison with Count Data Ref: C Version No: 2 Date: 21 October 2005 Author: Scott Leitham 1 Introduction 1.1 This Information Note has been produced to report a comparison between the 2001 Detailed Assignment Models and (i) the count data collected during the development of the TRAM model, and (ii) the bus counts undertaken in May The Detailed Assignment Models (DAM) used in the analysis of the Edinburgh tram lines are fundamentally based on the CSTM3 transport model. The DAM models were produced by taking a sub-area of CSTM3 and creating new highway and public transport assignment models. CSTM3 was calibrated and validated to a 1997 base year; the 1997 validation in the DAM model area was reported in Information Note 30. In order to create the 2001 DAM models, CSTM3 was run in forecast mode, creating 2001 demand matrices, from which the sub-area was taken for both highway and public transport. 2 Count Data LUTI Model Development Counts 2.1 Extensive counts were undertaken and used during the development phase of the TRAM model. The following survey programmes was undertaken in May 2001: Inner cordon (corresponding to the proposed charging cordon), 25 sites (Figure 2.1); Outer cordon (corresponding to the proposed charging cordon), 17 sites (Figure 2.2); North South city screenline, 9 sites (Figure 2.3); and East West city screenline, 7 sites (Figure 2.3). 2.2 These classified counts were continuous between 0700 and 1900, and recorded: Cars / / Goods (vehicles and occupants); Buses (vehicles and occupants); and Pedal cycles and taxis. 2.3 The car, LGV and bus values are of most relevance here.

66 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data 2.4 Note that the locations of the inbound and outbound cordons were based on the proposed congestion charging cordon. The Edinburgh cordons reported in the CSTM3 Public Transport manual are not the same, so direct comparisons between the two cannot be made. May 2003 Bus Counts 2.5 An SDG Memorandum of 8 th April 2003 details the rationale and methodology behind the May 2003 public transport surveys which were undertaken by Babtie. A programme of bus vehicles counts and estimates of passenger numbers (made from the roadside) was undertaken on a screenline basis for each of the three corridors. 2.6 These count data were compared with the 2001 model firstly by Faber Maunsell (in a File Note dated 24 June 2003), then by MVA (Modelling and Appraisal Working Group Note 12, 8 October 2003). 3 Comparison of Modelled versus Observed data, 2001 LUTI Counts 3.1 Table 3.1 below summarises the results for each of the cordons and screenlines the comparison is between car / lgv (vehicles) and bus passengers. In each case, the observed count values and modelled values are given together with the % difference between the two and the GEH values.

67 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data Table 3.1 Summary of modelled versus observed data Cars / AM Peak ( ) Off Peak (ave ) PM Peak ( ) Count Modelled % Diff GEH Count Modelled % Diff GEH Count Modelled % Diff GEH Inner Cordon Inbound 14,097 12,116-14% 17 8,912 6,997-21% 21 10,551 9,493-10% 11 Outbound 9,935 7,441-25% 27 8,414 5,779-31% 31 12,334 12,883 4% 5 Outer Cordon Inbound 19,644 17,578-11% 15 10,451 9,486-9% 10 13,228 12,260-7% 9 Outbound 12,192 10,085-17% 20 10,720 8,757-18% 20 18,709 17,156-8% 12 NS Screenline Eastbound 4,305 4,170-3% 2 3,600 3,430-5% 3 4,221 3,960-6% 4 Westbound 4,419 4,191-5% 3 3,282 3,268 0% 0 3,717 5,155 39% 22 EW Screenline Northbound 7,147 5,285-26% 24 4,642 3,210-31% 23 5,408 5,688 5% 4 Southbound 5,101 4,432-13% 10 4,529 3,194-29% 21 6,401 6,402 0% 0 Bus Passengers AM Peak ( ) Off Peak (ave ) PM Peak ( ) Count Modelled % Diff GEH Count Modelled % Diff GEH Count Modelled % Diff GEH Inner Cordon Inbound 21,965 19,410-12% 18 7,067 6,619-6% 5 6,538 7,812 19% 15 Outbound 7,940 7,503-6% 5 7,123 6,216-13% 11 18,804 16,677-11% 16 Outer Cordon Inbound 5,597 6,311 13% 9 2,391 1,395-42% 23 2,448 1,993-19% 10 Outbound 2,784 2,532-9% 5 2,101 1,244-41% 21 6,368 4,954-22% 19 NS Screenline Eastbound 3,747 6,181 65% 35 3,300 3,007-9% 5 6,716 4,681-30% 27 Westbound 5,917 4,636-22% 18 2,287 2,571 12% 6 4,066 5,716 41% 24 EW Screenline Northbound 8,136 9,355 15% 13 2,929 3,381 15% 8 2,591 4,241 64% 28 Southbound 4,206 4,459 6% 4 2,973 3,408 15% 8 11,288 8,215-27% Looking at the results on the inner cordon first, it can be seen that in general, the modelled car flows are lower than the observed. The comparison is best in the PM peak and worst in the off-peak. There is a similar pattern with bus passenger flows, with the modelled values typically around 10% below the observed values. 3.3 The comparison on the outer cordon shows that all modelled car flows are less than the observed counts, by 7-18%. The bus passenger flows show a greater variation, with off peak modelled values being particularly low compared to the counts. 3.4 Across the North-South screenline, modelled and observed car flows are very close, except westbound in the PM peak, where modelled values are 39% higher than observed. The comparison for bus passengers is less good, with large variations in the AM and PM peak. On the East-West screenline, car modelled values are low in the AM and off peak, but compare well in the PM peak. 3.5 The remaining tables show the detailed comparison for each of the cordons / screenlines, for each time period. These tables include the GEH value. This is colour coded to green to reflect a GEH value of <= 5, amber values represent a GEH >5 & <10, and red values represent GEH > 10.

68 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data Table 3.2 Inner Cordon AM Peak Hour

69 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data CARS / LGVS OUTBOUND VEHICLES (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) Broughton Road % -77% -92% Bruntsfield Place , % -45% -58% Clerk Street , ,228-33% -2% 0% Comely Bank Road % -47% -61% Dalry Road ,062-29% 24% 78% Dundee Street % -14% -73% East Fettes Avenue % n/a n/a 19.8 n/a n/a Gilmore Place % -9% -69% Holyrood Road % -38% 178% Hope Park Crescent % 0% -44% Inverleith Row % -23% 28% Leith Walk , ,240-63% -2% 6% London Road % 13% 7% Melville Drive % -67% -72% Queensferry Road % -23% -1% Ravelston Terrace % n/a n/a 8.6 n/a n/a Regent Road % 20% 702% St Leonards Street % -36% 33% West Coates % -19% 30% West Approach Road , % -68% 35% TOTAL 9, ,940 7, ,503-25% -16% -6% CARS / LGVS AM Counts AM Modelled % difference GEH Values INBOUND VEHICLES (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) Broughton Road % -73% -22% Bruntsfield Place , ,525-28% -29% 19% Clerk Street , ,674 16% -19% -2% Comely Bank Road % -40% -28% Dalry Road , ,492 53% 7% -38% Dundee Street % 17% -45% East Fettes Avenue % -100% -100% Gilmore Place % 0% -65% Holyrood Road % 340% 753% Hope Park Crescent % -11% -67% Inverleith Row % -43% -14% Leith Walk , ,078-20% -29% -19% London Road , ,683-22% -47% -40% Melville Drive % -71% -76% Queensferry Road % -68% -43% Ravelston Terrace % -100% -100% Regent Road % -100% -100% St Leonards Street 1, % -40% -39% West Coates , ,475-12% -9% 55% West Approach Road 1, , ,281-7% -37% 415% TOTAL 14, ,965 12, ,410-14% -31% -12%

70 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data

71 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data Table 3.3 Inner Cordon Off Peak (average hour)

72 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data CARS / LGVS OUTBOUND VEHICLES (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) Broughton Road % -38% -64% Bruntsfield Place % -18% -39% Clerk Street , ,201-34% -3% 4% Comely Bank Road % -18% -43% Dalry Road % 4% -11% Dundee Street % 4% -72% East Fettes Avenue % -100% -100% Gilmore Place % -35% -56% Holyrood Road % -72% -18% Hope Park Crescent % -55% -35% Inverleith Row % -12% 7% Leith Walk % -1% 0% London Road % -2% -9% Melville Drive % -79% -38% Queensferry Road % -51% -53% Ravelston Terrace % -100% -100% Regent Road % -79% -74% St Leonards Street % -12% -25% West Coates % -17% 7% West Approach Road % -48% 382% TOTAL 8, ,123 5, ,216-31% -19% -13% CARS / LGVS OP Counts OP Modelled % difference GEH Values INBOUND VEHICLES (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) Broughton Road % -11% -30% Bruntsfield Place % -6% 42% Clerk Street , ,178 19% 28% 17% Comely Bank Road % -2% -21% Dalry Road % -13% -22% Dundee Street % -15% -66% East Fettes Avenue % -100% -100% Gilmore Place % -14% -63% Holyrood Road % 51% 70% Hope Park Crescent % -26% -52% Inverleith Row % -25% 36% Leith Walk , ,000-40% -12% -26% London Road , % -2% -26% Melville Drive % -73% -28% Queensferry Road % -45% -35% Ravelston Terrace % -100% -100% Regent Road % -95% -98% St Leonards Street % -35% 31% West Coates % -8% 31% West Approach Road % -48% 141% TOTAL 8, ,067 6, ,619-21% -14% -6%

73 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data

74 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data Table 3.4 Inner Cordon PM Peak hour

75 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data CARS / LGVS OUTBOUND VEHICLES (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) Broughton Road % -25% -81% Bruntsfield Place , ,337 16% -14% -20% Clerk Street , ,428-16% -26% 37% Comely Bank Road % 13% 74% Dalry Road , ,248-14% -7% -2% Dundee Street % -18% -82% East Fettes Avenue % -100% -100% Gilmore Place % -29% -75% Holyrood Road % -46% 159% Hope Park Crescent , % -38% -51% Inverleith Row % -9% 16% Leith Walk , ,936-41% 2% 6% London Road , ,159-8% -1% -33% Melville Drive % -60% -31% Queensferry Road ,075 1, % -35% -73% Ravelston Terrace % n/a n/a 0.6 n/a n/a Regent Road % -75% -60% St Leonards Street 1, % 11% -33% West Coates , ,481 10% -18% -14% West Approach Road 1, , % -57% 447% TOTAL 12, ,804 12, ,677 4% -20% -11% CARS / LGVS PM Counts PM Modelled % difference GEH Values INBOUND VEHICLES (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) Broughton Road % 0% -57% Bruntsfield Place % 8% 78% Clerk Street ,460 45% 11% 46% Comely Bank Road % 0% 96% Dalry Road % -7% 8% Dundee Street % -7% -9% East Fettes Avenue % -67% -98% Gilmore Place % -9% -55% Holyrood Road % 92% 320% Hope Park Crescent % -56% -88% Inverleith Row % -20% 29% Leith Walk ,308-20% 10% 45% London Road % -3% -19% Melville Drive % -75% -1% Queensferry Road % -45% 152% Ravelston Terrace % n/a n/a 14.4 n/a n/a Regent Road % -75% -45% St Leonards Street % 43% 59% West Coates % -9% 11% West Approach Road 1, % -53% 63% TOTAL 10, ,538 9, ,812-10% -10% 19%

76 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data

77 OUTBOUND CARS / LGVS Annex D Information Note No 31(1) A199, Edinburgh Road % -62% 40% A1, Musselburgh Bypass % -64% -33% Newcraighall Road % 30% 55% A6106, The Wisp % -67% n/a A7, 2001 Old Dalkeith Detailed Road Assignment 458Model 18 Comparison with Count 13 Data % -28% 39% A772, Gilmerton Road % -57% -42% Lasswade Road % -50% -25% A701, Straton Road % -36% -85% Lang Loan % -100% -100% A702, Biggar Road % -64% -90% Dreghorn Link % -100% -100% A70, Lanark Road % 8% -66% Baberton Mains View % -100% n/a n/a A71, Calder Road 1, % -27% -12% A8, Glasgow Road 2, , % -55% -48% South Gyle Broadway % 40% 1088% Gogar Roundabout n/a n/a n/a n/a n/a n/a A90, Cramond Bridge 1, , % -25% -11% TOTAL 12, ,784 10, ,532-17% -39% -9% Table 3.5 Outer Cordon AM Peak hour VEHICLES (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) CARS / LGVS AM Counts AM Modelled % difference GEH Values INBOUND VEHICLES (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) A199, Edinburgh Road 1, % -74% -37% A1, Musselburgh Bypass 1, , % -80% -92% Newcraighall Road % 8% 211% A6106, The Wisp % -75% 61% A7, Old Dalkeith Road 1, % 8% 109% A772, Gilmerton Road % -50% 82% Lasswade Road % -60% 152% A701, Straton Road 1, , % -58% -27% Lang Loan % -100% n/a n/a A702, Biggar Road 1, , % -69% -96% Dreghorn Link % -100% -100% A70, Lanark Road , % -50% -50% Baberton Mains View % -100% -100% A71, Calder Road 2, , % -36% -50% A8, Glasgow Road 1, , ,178 66% -50% 124% South Gyle Broadway 2, , % 200% 1550% Gogar Roundabout % -18% 0% A90, Cramond Bridge 1, , % -52% -30% TOTAL 19, ,597 17, ,311-11% -48% 13%

78 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data

79 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data Table 3.6 Outer Cordon Off Peak (average hour)

80 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data CARS / LGVS OUTBOUND VEHICLES (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) A199, Edinburgh Road % -63% -55% A1, Musselburgh Bypass 1, % 4% -3% Newcraighall Road % 58% 12% A6106, The Wisp % -75% -81% A7, Old Dalkeith Road % -1% 21% A772, Gilmerton Road % -56% -32% Lasswade Road % -42% 72% A701, Straton Road % -32% -73% Lang Loan % n/a n/a 8.2 n/a n/a A702, Biggar Road % -58% -85% Dreghorn Link % -100% -100% A70, Lanark Road % -12% -51% Baberton Mains View % -100% -100% n/a A71, Calder Road 1, % -27% -61% A8, Glasgow Road 1, , % -47% -38% South Gyle Broadway 1, , % 56% 20% Gogar Roundabout n/a n/a n/a n/a n/a n/a A90, Cramond Bridge 1, , % -31% -45% TOTAL 10, ,101 8, ,244-18% -31% -41% CARS / LGVS Off Peak Counts Off Peak Modelled % difference GEH Values INBOUND VEHICLES (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) A199, Edinburgh Road % -65% -42% A1, Musselburgh Bypass % -33% -45% Newcraighall Road % -1% 94% A6106, The Wisp % -57% -73% A7, Old Dalkeith Road % 58% 86% A772, Gilmerton Road % -59% -31% Lasswade Road % -22% -9% A701, Straton Road % -24% -69% Lang Loan % -100% n/a n/a A702, Biggar Road % -56% -96% Dreghorn Link % -100% -100% A70, Lanark Road % 1% -78% Baberton Mains View % -100% -100% A71, Calder Road 1, % -25% -70% A8, Glasgow Road 1, , % -50% -58% South Gyle Broadway , % 53% 53% Gogar Roundabout % 7% 141% A90, Cramond Bridge 1, , % -26% -58% TOTAL 10, ,391 9, ,395-9% -30% -42%

81 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data

82 PM Counts PM Modelled % difference GEH Values OUTBOUND CARS / LGVS Annex D Information Note No 31(1) A199, Edinburgh Road , % -62% -42% A1, Musselburgh Bypass 1, , % -62% -89% Newcraighall Road 1, % 27% -27% A6106, The Wisp % -75% n/a A7, 2001 Old Dalkeith Detailed Road Assignment 867 Model 16 Comparison with Count 13 Data % -19% 44% A772, Gilmerton Road % -70% 23% Lasswade Road % -44% 133% A701, Straton Road % -35% -43% Lang Loan % n/a n/a 12.7 n/a n/a A702, Biggar Road 1, , % -57% -95% Dreghorn Link % -100% -100% A70, Lanark Road , % -11% -68% Baberton Mains View % -100% n/a n/a A71, Calder Road 1, , % -18% -10% A8, Glasgow Road 2, ,036 2, % -38% -9% South Gyle Broadway 2, , % -24% -51% Gogar Roundabout n/a n/a n/a n/a n/a n/a A90, Cramond Bridge 2, , % 0% -27% TOTAL 18, ,368 17, ,954-8% -37% -22% Table 3.7 Outer cordon, PM peak hour VEHICLES (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) CARS / LGVS INBOUND VEHICLES (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) A199, Edinburgh Road % -62% -49% A1, Musselburgh Bypass 1, , % -40% 2421% Newcraighall Road % 0% 399% A6106, The Wisp % 0% -88% A7, Old Dalkeith Road % 0% 44% A772, Gilmerton Road % -60% -8% Lasswade Road % 0% n/a A701, Straton Road % -18% -74% Lang Loan % -100% n/a n/a A702, Biggar Road 1, % -25% n/a Dreghorn Link % -100% -100% A70, Lanark Road % 27% -50% Baberton Mains View % n/a n/a 10.2 n/a n/a A71, Calder Road 1, , % -8% -40% A8, Glasgow Road 1, , % -57% 20% South Gyle Broadway % 1100% 458% Gogar Roundabout % -10% 85% A90, Cramond Bridge 1, , % -42% -84% TOTAL 13, ,448 12, ,993-7% -30% -19%

83 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data Table 3.8 North South Screenline AM peak hour CARS / LGVS EASTBOUND VEHICLES (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) Henderson Row % 0% -91% Howe Street % 33% -51% Heriot Row % -100% n/a n/a Queen Street 1, , % -100% n/a n/a George Street % -20% -10% Princes Street , , % -12% 177% Johnston Terrace % 14% -64% Grassmarket % -82% -100% Laurieston Place % -16% -92% TOTAL 4, ,747 4, ,181-3% -17% 65% CARS / LGVS AM Counts AM Modelled % difference GEH Values WESTBOUND VEHICLES (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) Henderson Row % 0% -47% Howe Street % -17% -20% Heriot Row % -100% n/a n/a Queen Street % -100% n/a n/a George Street % -13% -35% Princes Street , ,806-51% -26% -12% Johnston Terrace % 50% -40% Grassmarket % -67% -91% Laurieston Place % -43% -58% TOTAL 4, ,917 4, ,636-5% -26% -22% Table 3.9 North-South Screenline, Off Peak (average hour)

84 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data CARS / LGVS EASTBOUND VEHICLES (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) Henderson Row % -11% -24% Howe Street % 26% -35% Heriot Row % -100% n/a n/a Queen Street 1, , % -100% -100% George Street % -19% -13% Princes Street , , % -8% 20% Johnston Terrace % -57% -63% Grassmarket % -77% -95% Laurieston Place % -28% -70% TOTAL 3, ,300 3, ,007-5% -20% -9% CARS / LGVS OP Counts OP Modelled % difference GEH Values WESTBOUND VEHICLES (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) Henderson Row % 9% -8% Howe Street % -13% -65% Heriot Row % -100% n/a n/a Queen Street % -100% -100% George Street % -10% 6% Princes Street , ,015-21% -6% 62% Johnston Terrace % 33% 3% Grassmarket % -69% -86% Laurieston Place % -58% -59% TOTAL 3, ,287 3, ,571 0% -18% 12%

85 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data Table 3.10 North-South Screenline, PM peak hour CARS / LGVS PM Counts PM Modelled % difference GEH Values EASTBOUND VEHICLES (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) Henderson Row % 0% -59% Howe Street % 50% -9% Heriot Row % n/a n/a 8.5 n/a n/a Queen Street 1, , % -100% -100% George Street % -29% -44% Princes Street , , % -20% -27% Johnston Terrace % -13% 121% Grassmarket % -50% -98% Laurieston Place % 0% -71% TOTAL 4, ,716 3, ,681-6% -20% -30% CARS / LGVS WESTBOUND VEHICLES (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) Henderson Row % 0% 47% Howe Street % 50% 36% Heriot Row % -100% -100% Queen Street % -100% -100% George Street % 30% -47% Princes Street , ,939 24% -8% 112% Johnston Terrace % 17% -14% Grassmarket % -57% -89% Laurieston Place % -29% -85% TOTAL 3, ,066 5, ,716 39% -7% 41% Table 3.11 East West Screenline, AM peak hour

86 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data CARS / LGVS NORTHBOUND VEHICLES (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) A199 Seafield Road 1, , % -100% -100% A1 at Meadowbank Stadium , ,560-62% -29% -24% North Bridge , ,896-17% -24% 69% Lothian Road 1, , ,337-38% -13% 62% Balgreen Road % -67% -79% Broomhouse Road % -18% 200% South Gyle Access 1, , % -56% -78% TOTAL 7, ,136 5, ,355-26% -30% 15% CARS / LGVS AM Counts AM Modelled % difference GEH Values SOUTHBOUND VEHICLES (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) A199 Seafield Road % -100% n/a n/a A1 at Meadowbank Stadium % 20% 85% North Bridge , ,955-21% -3% 31% Lothian Road 1, , ,292-21% -44% -28% Balgreen Road % -80% -21% Broomhouse Road 1, % 13% -15% South Gyle Access % -47% 49% TOTAL 5, ,206 4, ,459-13% -22% 6%

87 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data Table 3.12 East West Screenline, Off Peak average hour CARS / LGVS NORTHBOUND VEHICLES (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) A199 Seafield Road % -100% -100% A1 at Meadowbank Stadium % 1% 49% North Bridge , ,215-8% 2% 14% Lothian Road , ,129-47% -45% 12% Balgreen Road % -83% -95% Broomhouse Road % 36% 79% South Gyle Access % -19% -54% TOTAL 4, ,929 3, ,381-31% -21% 15% CARS / LGVS Off Peak Counts Off Peak Modelled % difference GEH Values SOUTHBOUND VEHICLES (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) A199 Seafield Road % -100% -100% A1 at Meadowbank Stadium % 9% 53% North Bridge , ,348-36% -6% 24% Lothian Road ,061-7% -30% 8% Balgreen Road % -75% -88% Broomhouse Road % 23% 109% South Gyle Access % -4% -86% TOTAL 4, ,973 3, ,408-29% -15% 15% Table 3.13 East West Screenline, PM peak hour

88 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data CARS / LGVS NORTHBOUND VEHICLES (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) A199 Seafield Road % -100% -100% A1 at Meadowbank Stadium % 12% 117% North Bridge , ,041 46% 8% 62% Lothian Road 1, ,174-27% -20% 65% Balgreen Road , % -50% n/a Broomhouse Road 1, % 18% 50% South Gyle Access , % -23% -48% TOTAL 5, ,591 5, ,241 5% -6% 64% CARS / LGVS PM Counts PM Modelled % difference GEH Values SOUTHBOUND VEHICLES (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) (vehicles) A199 Seafield Road 1, , % -100% n/a n/a A1 at Meadowbank Stadium , ,290-3% 18% -20% North Bridge , ,291-3% -19% -16% Lothian Road 1, ,442 1, ,136-15% -49% -38% Balgreen Road % -50% -79% Broomhouse Road % 67% 92% South Gyle Access 1, , % -16% -84% TOTAL 6, ,288 6, ,215 0% -19% -27% Comparison of Modelled versus Observed data, 2003 Bus Counts 4.1 Tables 4.1 to 4.3 below give the comparison between the 2003 bus counts and the 2001 base modelled values. The comparison is similar to that reported in MVA MAWG Note 12. The differences can be accounted for by the change in bus services, Note 12 used 2003 bus services, the comparison here used 2001 bus services. 4.2 Note that the 2001 demand matrices used here precede the Line 3 matrix estimation exercise.

89 AM Counts AM Modelled % difference OP Counts OP Modelled % difference PM Counts PM Modelled % difference Line 1 Area Information Note No 31 Annex D Table 4.1 Comparison with 2003 bus counts, Line 1 area Bus vehicles Passengers Bus vehicles Passengers Bus vehicles Passengers Bus vehicles Passengers Bus vehicles Passengers Bus vehicles Passengers Bus vehicles Passengers Bus vehicles Passengers Bus vehicles Passengers Inbound - West Coates (A8 Inner) 62 2, ,326-3% 25% 52 1, % -26% % -7% Outbound - West Coates (A8 Inner) % 98% % -4% 57 2, ,321 16% -6% Inbound - West Coates (A8 Inner) Airport Bus Outbound - West Coates (A8 Inner) Airport Bus Inbound - West Coates (A8 Inner) 62 2, ,538 6% 33% 52 1, % -10% ,156 8% 16% Outbound - West Coates (A8 Inner) ,135 3% 133% % 22% 57 2, ,556 26% 3% Inbound - Queensferry Road 42 1, % -53% % -30% % 46% Outbound - Queensferry Road % -23% % -36% 28 1, % -49% Inbound - Orchard Brae % -38% % -49% % -16% Outbound - Orchard Brae % -26% % -10% % -27% Inbound - Comely Bank Road % -46% % -50% % -16% Outbound - Comely Bank Road % -61% % -48% % 7% Inbound - Inverleith Road % 3% % 23% % 39% Outbound - Inverleith Road % 198% % 53% % 56% Inbound - Broughton Road % -21% % 33% % 89% Outbound - Broughton Road % -82% % -23% % -38% Inbound - Leith Walk 62 2, ,078-18% -20% ,000-9% 5% ,308 13% 35% Outbound - Leith Walk 58 1, ,240-14% -5% 55 1, % -16% 49 2, ,936 0% -29% Inbound - London Road 53 3, ,665 21% -23% 53 1, % -47% 42 1, % -56% Outbound - London Road % 27% 52 1, % -38% 63 3, ,140 14% -43% Inbound - Regent Road % -100% % -99% % -67% Outbound - Regent Road % 260% % -79% % -29% Inbound Ferry Road % -28% % -41% % 58% Outbound Ferry Road % -43% % -53% % -52% Inbound Starbank Road % 166% % 1% % 62% Outbound Starbank Road % 235% % 103% % 12% Inbound , ,878-4% -16% 266 5, ,970 8% -26% 237 5, ,022 14% -2% Outbound 256 4, ,991 10% 15% 246 4, ,638 19% -17% , ,063 13% -25% TOTAL , ,869 2% -8% 512 9, ,608 13% -22% , ,085 14% -18%

90 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data Table 4.2 Comparison with 2003 bus counts, Line 2 area

91 AM Counts AM Modelled % difference OP Counts OP Modelled % difference PM Counts PM Modelled % difference Line 2 Area Bus vehicles Passengers Bus vehicles Passengers Bus vehicles Passengers Bus vehicles Passengers Bus vehicles Passengers Bus vehicles Passengers Bus vehicles Passengers Bus vehicles Passengers Bus vehicles Passengers Inbound - West Coates (A8 Inner) 62 2, ,326-3% 25% 52 1, % -26% % -7% Information Outbound - West Coates Note (A8 Inner) No 31(1) % 98% Annex 586 D 3% -4% 57 2, ,321 16% -6% Inbound - West Coates (A8 Inner) - Airport Bus Outbound - West Coates (A8 Inner) - Airport Bus Inbound - West Coates (A8 Inner) 62 2, ,538 6% 33% 52 1, % -10% ,156 8% 16% Outbound - West Coates (A8 Inner) ,135 3% 133% % 22% 57 2, ,556 26% 3% Inbound - Dalry Road (A70) Outer 46 2, ,492 7% -46% % -44% 41 1, % -36% 2001 Outbound Detailed - Dalry Road (A70) Assignment Outer Model Comparison with Count 56 Data 1,062 44% 6% % -16% 44 2, ,248 25% -6% Inbound - Western Approach Road, Fountain Park Stop inner ,281 20% 183% % 123% % 26% Outbound - Western Approach, Fountain Park Stop inner % -3% % 125% % 168% Inbound - Dundee Street inner % -64% % -65% % -42% Outbound - Dundee Street inner % -50% % -46% % -69% Inbound - Dalry 132 6, ,518 7% 1% 121 2, ,979 5% -20% 116 3, ,742 16% -11% Outbound - Dalry 125 2, ,900 12% 25% 118 1, ,950 15% 11% 126 5, ,905 21% 4% TOTAL - Dalry 257 8, ,418 9% 7% 239 4, ,930 10% -7% 242 8, ,647 19% -1% Inbound - Glasgow Road (A8) Outer 38 1, ,178-42% -9% % -43% % 40% Outbound - Glasgow Road (A8) Outer % 38% % -58% 34 1, % -37% Inbound - Glasgow Road (A8) Outer - Airport Bus Outbound - Glasgow Road (A8) Outer - Airport Bus Inbound - Glasgow Road (A8) Outer 38 1, ,420-26% 10% % 15% % 142% Outbound - Glasgow Road (A8) Outer % 101% % -12% 34 1, ,205 6% -19% Inbound - South Gyle Broadway (Outer) % 19% % -75% % -15% Outbound - South Gyle Broadway (Outer) 30 1, % -83% % -73% % -91% Inbound - Calder Road (A71) Outer % -44% % -71% % -66% Outbound - Calder Road (A71) Outer % 17% % -66% % -3% Inbound - Ring Road 89 2, ,914-28% -6% % -38% 79 1, ,099-18% -20% Outbound - Ring Road 81 1, ,370-21% -27% 87 1, % -48% 83 2, ,740-8% -32% TOTAL - Ring Road 170 3, ,283-25% -16% 174 1, ,058-33% -44% 162 3, ,839-13% -28% Inbound - A8 East of Newbridge Interchange ,277-29% 223% % 24% % 53% Outbound - A8 East of Newbridge Interchange % 436% % -8% % 141% TOTAL - A8 East of Newbridge Interchange ,529 4% 246% % 11% ,380 17% 107% TOTAL - Inbound 238 8, ,709-9% 9% 221 3, ,687-10% -22% 211 4, ,242-1% -10% TOTAL - Outbound 214 4, ,522 2% 6% 215 2, ,592-6% -11% 223 8, ,624 14% 0% TOTAL , ,231-4% 8% 436 6, ,279-8% -17% , ,866 7% -4%

92 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data

93 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data Table 4.3 Comparison with 2003 bus counts, Line 3 area

94 AM Counts AM Modelled % difference OP Counts OP Modelled % difference PM Counts PM Modelled % difference Line 3 Area Bus vehicles Passengers Bus vehicles Passengers Bus vehicles Passengers Bus vehicles Passengers Bus vehicles Passengers Bus vehicles Passengers Bus vehicles Passengers Bus vehicles Passengers Bus vehicles Passengers Inbound - St. Leonard's St % -13% % 51% % 20% Information Outbound - St. Leonard's Note Street No 31(1) % 45% Annex 114 D 60% -12% % -27% Inbound - Clerk Street 49 3, ,674 31% 34% 53 1, ,178 33% -5% 52 1, ,460 17% -10% Outbound - Clerk St ,228 54% 60% ,201 53% 90% 47 1, ,428 43% 113% Inbound - Buccleuch Street % -84% % -38% % -49% Outbound - Buccleuch Street % -15% % -11% % -75% Inbound - L3 excl. screenlines 61 3, ,883 33% 25% 67 1, ,391 35% -4% 66 1, ,591 20% -10% 2001 Outbound Detailed - L3 excl. screenlines Assignment Model Comparison with Count 80 Data 1,394 54% 53% ,418 51% 61% 58 2, ,668 47% 72% TOTAL - L3 excl. screenlines 113 4, ,276 42% 30% 122 2, ,808 42% 20% 124 3, ,259 32% 35% Inbound - Peffermill Road (Just West of Bridge) % 117% % 118% % 191% Outbound - Peffermill Road (Just West of Bridge) % 255% % 127% % 72% Inbound - Old Dalkeith Road (By Bridge End) % 11% % -52% % -67% Outbound - Old Dalkeith Road (By Bridge Road) % -57% % -37% % 26% Inbound - Gilmerton Road (At North End of Schools) ,500-11% 78% % 79% % 78% Outbound - Gilmerton Road (At North End of Schools) % 28% % 74% ,051-17% 75% Inbound - Liberton Brae % 16% % -32% % -45% Outbound - Liberton Brae % -23% % -15% % -40% Inbound - Kirk Brae % 103% % -28% % -2% Outbound - Kirk Brae % -38% % 7% % 173% Inbound - Inner screenline 86 3, ,827-6% 51% 79 1, ,430 28% 16% 76 1, ,242 9% -2% Outbound - Inner screenline 62 1, ,090 29% 5% 77 1, ,440 24% 27% 87 2, ,470 0% 30% TOTAL - Inner screenline 148 4, ,917 9% 40% 155 2, ,870 26% 21% 163 3, ,712 4% 20% Inbound - Burdiehouse Road % -38% % -68% % -63% Outbound - Burdiehouse Road % -67% % -64% % -63% Inbound - Gilmerton Road % 95% % -42% % 6% Outbound - Gilmerton Road % -68% % -31% % 30% Inbound - Dalkeith Road % 122% % 47% % 36% Outbound - Dalkeith Road % 4% % 14% % 57% Inbound - Newcraighall Road % 191% % 16% % 128% Outbound - Newcraighall Road % 17% % 38% % 58% Inbound - Milton Road % 69% % 58% % 181% Outbound - Milton Road % 120% % 36% % 9% Inbound - Outer screenline 72 1, ,771-15% 52% % 1% % 49% Outbound - Outer screenline % -2% % -5% 80 2, ,441-14% 1% TOTAL - Outer screenline 132 2, ,447-5% 37% 121 1, ,467 29% -2% 144 3, ,363-7% 11% Inbound 219 8, ,481 2% 40% 205 3, ,581 32% 4% 206 3, ,756 10% 3% Outbound 174 2, ,159 29% 20% 193 2, ,565 32% 29% 225 7, ,579 7% 33% TOTAL , ,640 14% 35% 399 6, ,146 32% 15% , ,334 9% 23%

95 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data

96 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data Figure 4.1 Outer cordon count locations

97 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data

98 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data Figure 4.2 Inner cordon count locations

99 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data

100 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data Figure 4.3 North south and east west screenline count locations

101 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data

102 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data Figure 4.4. May 2003 Bus Count Locations

103 Information Note No 31(1) Annex D 2001 Detailed Assignment Model Comparison with Count Data