CO 2 Slicing Method for IASI

Transcription

1 CO 2 Slicing Method for IASI Arlindo Arriaga, Peter Schlüssel, Xavier Calbet Thomas August, Olusoji Oduleye, Tim Hultberg European Organisation for the Exploitation of Meteorological Satellites ( Slide: 1 ITSC-16, Angra dos Reis, Brazil 6-13 May 2008

2 IASI algorithms recommended methods from ISSWG Cloud detection within IFOV profile retrievals CTP retrieval CO 2 slicing method (Smith and Frey, 1990) Introduction Prototyping Processing Facility for IASI L2 products (MET division) Implement methods, optimise algorithms & coefficients Input: ECMWF (or ATOVS) co-located profiles & T skin IASI L1 radiance, surface emissivity, RTIASI-4, auxiliary file (list of channels) Output: CTP and CFR iterative profile retrieval Algorithm CO 2 channel selection Validation Slide: 2 ITSC-16, Angra dos Reis, Brazil 6-13 May 2008

3 Algorithm x k,n = R R clear k clear ref R R cloudy k cloudy ref R R clear k clear ref R R B k B ref (p ) n (p ) n P k = P(min( x k,n ) P c = M w k= 1 M w k= 1 2 k P 2 k k Algorithm CO 2 channel selection Validation Slide: 3 ITSC-16, Angra dos Reis, Brazil 6-13 May 2008

4 Algorithm implementation Non noisy radiance ( R cloudy w R B w (p suf ) ) R k,ref SNR NEδE k,ref with SNR = 3 (R clear ref R cloudy ref ) SNR 2 NEδE ref and (R clear k R cloudy k ) SNR 2 NEδE k o Search profile for inversions below 500 hpa (flag levels p i ) Flag levels p n with noisy cloud signal R ref (p n ) skip levels p n, p i Retrieve cloud top pressure P k with a single CO 2 channel Calculate effective C fk using the window channel exclude channel k if C fk outside (0,1] CTP histogram from single channel retrievals Retrieved P c as in Smith and Frey P k in most populated class If inversion exists and P c > 600 hpa use P k below inversion basis only Effective cloud fraction C f P c and window channel P c and C f delivered if C f > 10% (Other results for quality control: P k rms, ΔP k, δp c from NEδR and T error, Number of channels used) Algorithm CO 2 channel selection Validation Slide: 4 ITSC-16, Angra dos Reis, Brazil 6-13 May 2008

5 CO 2 channel selection (1) window: cm -1 ( μm) reference: cm -1 ( μm) Retrievals of scenarios (RTIASI) 41 CO 2 channels ( cm -1 ) Two global data sets (all cloud fractions, surface types, day/night, seasons) (A) single level black clouds and (B) multilevel, water/ice clouds, 6 types CTP of highest cloud (data set B) Algorithm CO 2 channel selection Validation Slide: 5 ITSC-16, Angra dos Reis, Brazil 6-13 May 2008

6 Algorithm CO 2 channel selection Validation Slide: 6 ITSC-16, Angra dos Reis, Brazil 6-13 May 2008 CO 2 channel selection (2) Retrievals with single channel k 545 IASI frequencies (645.0 to cm -1 ) Significance of channel k: Q% scenarios {k NOT used OR error critical E} Given a critical error E n identify (among 545 channels) the set M ni of channels with a pre-defined Q i and find the set S ni of scenarios left with no channels {E n < 100 hpa, S ni 1%} <> Q i = 10% => exclude 498 (A) [ 264 (B) channels ] hpa hpa hpa hpa hpa hpa hpa hpa 5.0% 5.0% excld. excld non-expl. non-expl % 10.0% 10.0% excld non-expl. non-expl %

7 Statistics of retrievals (data set B) using 41 or 281 CO 2 channels Layer (100, 400] hpa CO 2 channel selection (3) Layer (400, 800] hpa 41 chs: chs: Algorithm CO 2 channel selection Validation Slide: 7 ITSC-16, Angra dos Reis, Brazil 6-13 May 2008

8 CO 2 channel selection (4) Frequency (%) of retrievals vs. cloud cover of highest cloud (data set with multilevel water/ice clouds, six types) retrievals 41 CO 2 channels retrievals 281 CO 2 channels Algorithm CO 2 channel selection Validation Slide: 8 ITSC-16, Angra dos Reis, Brazil 6-13 May 2008

9 Validation (1a) Consistency with AVHRR images: frontal system & post frontal convection (north Atlantic) Algorithm CO 2 channel selection Validation Slide: 9 ITSC-16, Angra dos Reis, Brazil 6-13 May 2008

10 onsistency with AVHRR images: frontal system (western south Pacific) Validation (1b) Algorithm CO 2 channel selection Validation Slide: 10 ITSC-16, Angra dos Reis, Brazil 6-13 May 2008

11 Validation (1c) onsistency with AVHRR images: ost frontal squall-line and cumulus clouds organised as open cells (western north Atlantic) Algorithm CO 2 channel selection Validation Slide: 11 ITSC-16, Angra dos Reis, Brazil 6-13 May 2008

12 Validation (2) Metop sounding campaign Lindenberg Observatory (near Berlin) Jun-Aug (+Sep) overpasses (~9 AM, ~8 PM) with IASI data Co-located ECMWF forecasts (profiles, T skin ) Co-located surface emissivity (as in PPF) Cloud data: type & CFR (overpass time) + CTP Observer (Jun-Aug) and AVHRR images Cloud radar (1 min-integrated) reflectivity + Sounding CTP from radar measurements (zenith): CTP = mode of radar samples within a sampling time interval ( hpa wind) / 7 km (~ IFOV radius) Algorithm CO 2 channel selection Validation Slide: 12 ITSC-16, Angra dos Reis, Brazil 6-13 May 2008

13 Validation (3) Mean of retrieved CTP (and CFR) within IASI IFOVS up to 7 km of Lindenberg 50 km of Lindenberg if CTP in [P R -100, P R +100] Algorithm CO 2 channel selection Validation Slide: 13 ITSC-16, Angra dos Reis, Brazil 6-13 May 2008

14 Retrieved cloud top pressure error Validation (4) All 82 cases: bias = 29.4 hpa std = 49.2 hpa 72 cases (CFR > 40%): Slide: 14 ITSC-16, Angra dos Reis, Brazil 6-13 May 2008

15 Retrieved cloud cover error Validation (5) 66 cases (Jun-Aug): CFR bias = - 6.2% CFR std = 22.4% (with respect to cloud cover reported by the observer) Slide: 15 ITSC-16, Angra dos Reis, Brazil 6-13 May 2008

16 Thank you Slide: 16 ITSC-16, Angra dos Reis, Brazil 6-13 May 2008

17 Slide: 17 ITSC-16, Angra dos Reis, Brazil 6-13 May 2008

18 Slide: 18 ITSC-16, Angra dos Reis, Brazil 6-13 May 2008

19 Slide: 19 ITSC-16, Angra dos Reis, Brazil 6-13 May 2008

20 Relative frequency (%) of retrievals (10 layers) Slide: 20 ITSC-16, Angra dos Reis, Brazil 6-13 May 2008

21 Relative frequency (%) of non-retrieved cases Slide: 21 ITSC-16, Angra dos Reis, Brazil 6-13 May 2008

22 Slide: 22 ITSC-16, Angra dos Reis, Brazil 6-13 May 2008