Atmospheric QBO and ENSO indices with high vertical resolution from GNSS RO

Transcription

1 Atmospheric QBO and ENSO indices with high vertical resolution from GNSS RO H. Wilhelmsen, F. Ladstädter, B. Scherllin-Pirscher, A.K.Steiner Wegener Center for Climate and Global Change University of Graz, Graz, Austria

2 Outline Proxy for atmospheric temperature variability based on radio occultation temperature measurements (e.g. useful for regression). Two methods, M1 and M2 M1: PCA on monthly mean temperature anomalies M2: PCA on monthly mean temperature anomalies H. Wilhelmsen 2

3 Outline Proxy for atmospheric temperature variability based on radio occultation temperature measurements (e.g. useful for regression). Two methods, M1 and M2 M1: PCA on monthly mean temperature anomalies M2: PCA on monthly mean temperature anomalies Difference? M1: The whole input field. M2: Each altitude level separately H. Wilhelmsen 3

4 Atmospheric variability in the tropics Two main patterns Quasi-Biennial Oscillation QBO - stratosphere El Niño Southern Oscillation ENSO - troposphere H. Wilhelmsen 4

5 Commonly used indices Proxies for the variability QBO: Winds ENSO: Niño 3.4 SST index Issues (e.g. when used in temperature regression analysis) An other physical variable wind and sea surface temperature Account for timelag Cannot fully describe the vertical variability structure H. Wilhelmsen 5

6 Method: PCA Principle Component Analysis Empirical Orthogonal Function analysis Extracts the main variability from the input field Decomposed into a reduced set Spatial component: EOF s. WHERE. Corresponding time components: PC s. WHEN. Most of the variability is described by the first components H. Wilhelmsen 6

7 Monthly mean temperature anomalies Monthly mean temperatures from WEGC OPSv N 30 S Resolution: 5 x 5 latitude / longitude grid, 200 m vertical resolution Time: to Include: Latitude 30 S to 30 N, altitude: 2 km to 35 km H. Wilhelmsen 7

8 Monthly mean temperature 27.5ºN H. Wilhelmsen 8

9 Monthly mean temperature 2.5ºN H. Wilhelmsen 9

10 M1 Performing the PCA on the whole input field H. Wilhelmsen 10

11 M1: EOFs (space) and PCs (time) EOFs PCs Explained variance ratio 25% 20% 13% 7% H. Wilhelmsen 11

12 M1: 30 km No longitudinal dependency H. Wilhelmsen 12

13 M1: 27 km Node H. Wilhelmsen 13

14 M1: 25 km Opposite pattern H. Wilhelmsen 14

15 M1: 22 km Node H. Wilhelmsen 15

16 M1: 20 km H. Wilhelmsen 16

17 M1: 17 km (~tropopause) H. Wilhelmsen 17

18 M1: 14 km Node again H. Wilhelmsen 18

19 M1: 11 km H. Wilhelmsen 19

20 M1: 8 km H. Wilhelmsen 20

21 M1: PCs (indices) H. Wilhelmsen 21

22 M1: Cross correlations to known indices PC1 PC2 PC3 PC H. Wilhelmsen 22

23 Phase space diagram Wind RO temperature Adapted from Wallace et. all H. Wilhelmsen 23

24 M2 Performing the PCA at each altitude level separately H. Wilhelmsen 24

25 M2: EOFs (space) and PCs (time) EOFs PCs Explained variance ratio H. Wilhelmsen 25

26 M2: 30 km No longitudinal dependency H. Wilhelmsen 26

27 M2: 27 km H. Wilhelmsen 27

28 M2: 25 km H. Wilhelmsen 28

29 M2: 22 km H. Wilhelmsen 29

30 M2: 20 km H. Wilhelmsen 30

31 M2: 17 km (~tropopause) EOF2 adds information around the tropopause H. Wilhelmsen 31

32 M2: 14 km Node again? H. Wilhelmsen 32

33 M2: 11 km H. Wilhelmsen 33

34 M2: 8 km H. Wilhelmsen 34

35 M2: PCs (indices) H. Wilhelmsen 35

36 M2: Cross correlations QBO 30 hpa H. Wilhelmsen 36

37 M2: Cross correlations QBO 50 hpa H. Wilhelmsen 37

38 M2: Cross correlations to Niño 3.4 SST index 3 month timelag ENSO above the tropopause? H. Wilhelmsen 38

39 M2: Cross correlations Solar 10.7cm index H. Wilhelmsen 39

40 Some comparisons H. Wilhelmsen 40

41 M1: Reconstruction Reconstruction using EOF1/PC1 and EOF2/PC2 Reconstruction using EOF3/PC3 and EOF4/PC4 QBO ENSO H. Wilhelmsen 41

42 M1: Residual (anomalies - reconstruction) Monthly mean temperature Anomalies (input field) Reconstruction using all 4 EOFs and PCs H. Wilhelmsen 42

43 M1: Residual (anomalies - reconstruction) Smaller scale QBO and ENSO? H. Wilhelmsen 43

44 M2: Reconstruction Reconstruction using all the EOF1s/PC1s Reconstruction using all the EOF2s/PC2s H. Wilhelmsen 44

45 M2: Residual (anomalies - reconstruction) Monthly mean temperature Anomalies (input field) Reconstruction using all EOF1s / PC1s and all EOF2s / PC2s H. Wilhelmsen 45

46 M2: Residual (anomalies - reconstruction) H. Wilhelmsen 46 Most of the variability explained by the two first PCs

47 M1: Temperature regression ( to ) Notice the bump at the tropopause H. Wilhelmsen 47

48 M2: Temperature regression ( to ) H. Wilhelmsen 48

49 Summary Derived proxies for atmospheric variabilities from GPS RO temperatures We did this with 2 methods M1: Using the full input field M2: The same as M1 but at each altitude level separately M2 describe the vertical structure better than M H. Wilhelmsen 49