First validation of ML2PP V7 full mission : Temperature and altitude data

Transcription

1 MIPAS QWG #40, IFAC, Firenze, 2-4 Nov 2015 First validation of ML2PP V7 full mission : Temperature and altitude data D. Hubert, J. Granville and J.-C. Lambert Belgian Institute for Space Aeronomy (BIRA-IASB) daan.hubert@aeronomie.be Acknowledgements Ozonesonde & lidar PIs ESA, BELSPO

2 Validation methodology Data sets & selection MIPAS V6 : ML2PP 6.00/U (35034 orbits) V7 : ML2PP 7.03/W (34929 orbits) Ground-based data Ozonesonde : ground 33km, 82N 90S (85 sites) Temperature lidar : 30 70km, 80N 20S (9 sites) Data selection MIPAS screening : V6 & V7 = V6 Readme file Ground screening : as recommended in the literature Co-location criteria : (within 500 km and ±6h) & (closest SAT-GND pair) Identical MIPAS pairs : using tolerance of 0.1s on profile timestamp

3 temperature

4 δ-validation methodology Preprocessing Unit conversion Baseline system : temperature vs pressure MIPAS & O3sonde : as found in data file T lidar : altitude pressure using ERA-Interim p/t fields Vertical smoothing Ground-based profiles smoothed with co-located MIPAS vertical AK Statistics Δ = X MIPAS X ground [K] Median ( median bias ) and 68% inter-percentile ( IP68 comp ) of Δ-distribution

5 averaging kernels (or not?)

6 Effect of smoothing on median and spread of T differences AK smoothed correlative data lead to smoother vertical bias profile reduced spread in the differences

7 changes in temperature (V7.03 versus V6.00)

8 Overview T : V7-V6 changes in absolute difference to correlative data LS + MS : O3 sonde MS + US + LM : T lidar Arctic mid N Tropics mid S (V7.03-GND) (V6.00-GND) monthly & annual median Antarctic

9 Full Resolution : change in overall bias LS + MS : O3 sonde MS + US + LM : T lidar Arctic mid N Tropics mid S Antarctic V7 bias differs clearly from V6 bias sign & magnitude of V7-V6 varies with altitude and latitude Tropics positive in LS negative in US+LM High + mid latitudes ~0K at ~10hPa negative in LS+LM

10 Optimized Resolution : change in temporal dependence LS + MS : O3 sonde MS + US + LM : T lidar Arctic mid N Tropics mid S Antarctic V7 bias drifts mostly positive relative to V6 magnitude has altitude & latitude structure, not purely linear in time Large part of stratosphere about +0.3K/OR-period in Arctic + mid N, less clear elsewhere Stratopause + lower mesosphere up to +1K/OR-period in Tropics + mid N, unclear in Arctic

11 agreement with correlative data (less biased / more stable?)

12 Arctic

13 Temperature : Arctic O3 sonde 10hPa (30km) T lidar 0.2hPa (57km) 3-month moving median V6.00-GND V7.03-GND 20hPa (26km) 0.5hPa (51km) 50hPa (20km) 1hPa (46km) 100hPa (16km) 3hPa (38km) 200hPa (11km) 10hPa (30km)

14 Temperature : Arctic O3 sonde 10hPa (30km) T lidar 0.2hPa (57km) 12-month moving median V6.00-GND V7.03-GND 20hPa (26km) 50hPa (20km) 0.5hPa (51km) 1hPa (46km) Lower + middle stratosphere Full Resolution V7 bias is larger than V6 bias Optimized Resolution V7 drift is possibly larger than V6 drift 100hPa (16km) 200hPa (11km) 3hPa (38km) 10hPa (30km) Upper stratosphere Full Resolution V7 bias is larger than V6 bias Optimized Resolution unclear Other Time-dependence is non-linear e.g. Sep 2007, Aug 2009

15 Vertical structure Bias FR : V7 bias clearly larger than for V6 OR : V6 & V7 very similar Spread Very similar for V6 & V7

16 Mid Northern

17 Temperature : mid North O3 sonde 10hPa (30km) T lidar 0.2hPa (57km) 3-month moving median V6.00-GND V7.03-GND 20hPa (26km) 0.5hPa (51km) 50hPa (20km) 1hPa (46km) 100hPa (16km) 3hPa (38km) 200hPa (11km) 10hPa (30km)

18 Temperature : mid North O3 sonde 10hPa (30km) T lidar 0.2hPa (57km) 12-month moving median V6.00-GND V7.03-GND 20hPa (26km) 50hPa (20km) 100hPa (16km) 0.5hPa (51km) 1hPa (46km) 3hPa (38km) Lower + middle stratosphere Full Resolution V7 bias is larger than V6 bias Optimized Resolution V7 may be more stable than V6 Upper stratosphere Full Resolution V7 bias is larger than V6 bias Optimized Resolution V7 may be more stable than V6 200hPa (11km) 10hPa (30km) Other Time-dependence is non-linear e.g. Jun 2006, Dec 2008

19 Vertical structure Bias FR : V7 bias mostly larger than for V6 OR : V6 & V7 very similar Spread Very similar for V6 & V7 FR : V7 less noisy in UT/LS

20 Tropics

21 Temperature : Tropics O3 sonde 10hPa (30km) T lidar 0.2hPa (57km) 3-month moving median V6.00-GND V7.03-GND 20hPa (26km) 0.5hPa (51km) 50hPa (20km) 1hPa (46km) 100hPa (16km) 3hPa (38km) 200hPa (11km) 10hPa (30km)

22 Temperature : Tropics O3 sonde 10hPa (30km) T lidar 0.2hPa (57km) 12-month moving median V6.00-GND V7.03-GND 20hPa (26km) 50hPa (20km) 100hPa (16km) 0.5hPa (51km) 1hPa (46km) 3hPa (38km) Lower + middle stratosphere Full Resolution V7 bias is mostly larger than V6 bias Optimized Resolution stability V6 & V7 is similar Upper stratosphere Full Resolution V7 bias is mostly larger than V6 bias Optimized Resolution V7 perhaps more stable 200hPa (11km) 10hPa (30km) Other Time-dependence is non-linear e.g. Jun 2007 & 2008

23 Vertical structure Bias FR : V7 bias much larger than for V6 in UT OR : V6 & V7 very similar, except in UT where V7 is less biased than V6 Spread Very similar for V6 & V7 FR : V7 less noisy in UT

24 Mid Southern

25 Temperature : mid South O3 sonde 10hPa (30km) 3-month moving median V6.00-GND V7.03-GND 20hPa (26km) 50hPa (20km) 100hPa (16km) 200hPa (11km)

26 Temperature : mid South O3 sonde 10hPa (30km) 12-month moving median V6.00-GND V7.03-GND 20hPa (26km) 50hPa (20km) Lower + middle stratosphere Full Resolution V7 bias is larger than V6 bias Optimized Resolution stability V6 & V7 is very similar Upper stratosphere + lower mesosphere No correlative instruments 100hPa (16km) 200hPa (11km) Other Time-dependence is non-linear e.g. Jan 2007 & 2009

27 Vertical structure Bias FR : V7 bias clearly larger than for V6 OR : V6 & V7 very similar Spread Very similar for V6 & V7 FR : V7 less noisy in UT/LS

28 Antarctic

29 Temperature : Antarctic O3 sonde 10hPa (30km) 3-month moving median V6.00-GND V7.03-GND 20hPa (26km) 50hPa (20km) 100hPa (16km) 200hPa (11km)

30 Temperature : Antarctic O3 sonde 10hPa (30km) 12-month moving median V6.00-GND V7.03-GND 20hPa (26km) 50hPa (20km) Lower + middle stratosphere Full Resolution V7 bias is larger than V6 bias Optimized Resolution stability V6 & V7 is very similar Upper stratosphere + lower mesosphere No correlative instruments 100hPa (16km) 200hPa (11km) Other Time-dependence is non-linear e.g. Jan 2007, Jun 2009

31 Vertical structure Bias FR : V7 bias clearly larger than for V6 OR : V6 & V7 very similar Spread Very similar for V6 & V7 OR : V6 less noisy in UT/LS

32 First look temperature* Full Resolution V7 bias is generally larger (more negative) than that of V6 Comparison spreads are similar, and somewhat better for V7 in the UT/LS Optimized Resolution No very clear distinction in quality of V6 & V7 Bias and comparison spread are very similar The V7 data is possibly slightly more stable than V6 data, but difficult to tell *All results shown are preliminary

33 Summary table V7 compared to V6 Full Resolution Optimized Resolution Bias Spread Bias Spread Drift 60N-90N Worse by 1-2K Very similar Very similar 30N-60N Worse by ~1K 30N-30S 30S-60S* Mostly worse, especially in UT Mostly worse by 0.5-1K, except in UT 60S-90S* Worse by 1-1.5K Similar, better in UTLS Similar, better in UT Similar, better in UTLS Similar, worse in UT Very similar Similar, better in UT Very similar Very similar *No T lidar comparisons, so validated only from UT to MS. Very similar Very similar Similar Very similar Very similar Possibly worse in LS+MS Possibly better Similar, possibly better in US+LM Very similar Very similar

34 altitude

35 Validation methodology Preprocessing Unit conversion Baseline system : altitude vs pressure MIPAS : scan_information_mds.tangent_altitude_los + ( ECMWF scale!) [pt_retrieval_mds.h_corr 0]*1d-3 O3 sonde : as found in data file Vertical smoothing Ground-based profiles regridded to fixed pressure grid Statistics Δ = X MIPAS X ground [m] Median ( median bias ) and 68% inter-percentile ( IP68 comp ) of Δ-distribution

36 changes in altitude (V7.03 versus V6.00)

37 Overview altitude : V7-V6 changes in absolute difference to sonde data Arctic mid N Tropics mid S Antarctic (V7.03-GND) (V6.00-GND) monthly & annual median

38 Overview altitude : V7-V6 changes in absolute difference to sonde data Arctic mid N Tropics mid S Antarctic V7 bias differs clearly from V6 bias magnitude of V7-V6 varies with latitude and time Full Resolution : large spread Optimized Resolution : long-term negative drift

39 agreement with correlative data (less biased / more stable?)

40 MIPAS altitude versus sonde 3-month moving median V6.00-GND V7.03-GND

41 MIPAS altitude versus sonde 12-month moving median V6.00-GND V7.03-GND

42 Observations 12-month moving median V6.00-GND V7.03-GND 90N 60S Full Resolution V7 bias is smaller than V6 bias Optimized Resolution V7 is less stable than V6 60S 90S Full Resolution V7 is more stable than V6 Optimized Resolution V6 & V7 very similar

43 Vertical structure

44 Vertical structure

45 First look altitude* Full Resolution V7 bias is much smaller than V6 bias, except in Antarctica V7 comparison spread is smaller as well Optimized Resolution V7 data quality is clearly worse than that of V6, except perhaps in Antarctica V7 has larger overall bias, spread and long-term drift *All results shown are preliminary

46 backup

47 Lower + middle stratosphere

48 Middle + upper stratosphere & lower mesosphere

49 Overview : change in absolute bias to correlative data LS + MS : O3 sonde MS + US + LM : T lidar Arctic mid N Tropics (V7.03-GND) (V6.00-GND) Monthly & annual median mid S Antarctic

50 Overview : change in absolute bias to correlative data LS + MS : O3 sonde MS + US + LM : T lidar Arctic mid N Tropics mid S (V7.03-GND) (V6.00-GND) (V7.01-GND) (V6.00-GND) Common subset : DDS orbits Antarctic

51 Structure of ex-ante uncertainty

52 altitude

53 AI_L2_36.06: Check if in the V7 delta validation the corrected altitude or the ECMWF correction has been used The QWG #36 presentation used corrected altitude, i.e. anchor = lowest engineering altitude z_corr = scan_information_mds.tangent_altitude_los + [pt_retrieval_mds.h_corr 0]*1d-3 Median difference z_corr z_sonde (collocated) QWG #36 conclusions V7 altitude mostly ~200m lower than V5/V6, and ~100m larger variability Strong temporal dependence (FR/OR, season, year) for all MIPAS versions, but most pronounced for V7 Some latitudinal dependence

54 What s the difference between altitude corrections for V7? Here, we compare to ECMWF altitude correction (layer thicknesses are same for z_corr and z_ecmwf) z_ecmwf = pt_retrieval_mds.ecmwf_corr_alt Median difference z_corr z_ecmwf (V7 TDS overpasses at 6 stations)

55 Very clear temporal structure between MIPAS-ECMWF anchors Clear structure at various temporal scales (similar to those seen in MIPAS-sonde comparisons) Full Resolution MIPAS anchor is ~ m higher than ECMWF anchor Seasonal cycle very pronounced, amplitude >700m Optimal Resolution MIPAS anchor is ~ m lower than ECMWF anchor, except in Antarctic Seasonal cycle clear southward of 30S, amplitude ~300m Long-term drift of about -50m/year (bump during ?)

56 The ECMWF anchored vertical scale is likely less biased 1. MIPAS corrected altitude versus co-located ozonesonde 2. MIPAS corrected altitude versus ECMWF corrected altitude Conclusion Difference MIPAS & ECMWF anchor of same sign and similar magnitude as difference between MIPAS & sonde altitude ECMWF altitude scale likely less biased than MIPAS altitude scale 3. Summary table But this needs to be verified with actual MIPAS- GND comparisons V7 altitude bias [m] 60N-90N 30N-60N 30N-30S 30S-60S 60S-90S MIPAS-sonde MIPAS-ECMWF ~-200 ~ ~-100 ~0