S 5 L 2 PP Verification Workshop October
S 5 L 2 PP Verification Workshop, October 2018 S 5 SO 2 definition : update N. Theys, C. Lerot, I. De Smedt, H. Yu, J. Vlietinck, J. van Gent, M. Van Roozendael (BIRA-IASB), J. Pukite, S. Warnach (MPIC), P. Hedelt (DLR) Sinabung, Indonesia
Overview
S 5 SO 2 prototype SO 2 layer height requirement: uncertainty < 1 km (breakthrough) to 2 km (threshold) for SO 2 VCD> 25 DU
Algorithm concept Joint SO 2 LH and VCD retrieval: iterative SO 2 optical depth fitting (see PM 4 slides) Fitting window: 310. 5 -326 nm LHi+1 = LHi + β (LH 0 = 7. 5 km) VCDi+1=VCDi + α (VCD 0=VCDdoas (LH 0)) Convergence: (1)&(2) | β |<0. 25 km (1) | α | < 5%. VCDi (2) Max. iter. : 10
Application to real/synthetic spectra • SO 2 LH v 1: development of scientific algorithm well advanced (Matlab). • It includes: ü reading routines (of L 1, L 2 AUX data) ü wavelength calib from QDOAS (optimized wvl grid + optimized SFPs) ü handling of SO 2 OD LUT (memory mapping + spline interpolation) ü LER retrieval ü Iterative spectral fitting module ü writing routine (variables as agreed with EUMETSAT) in he 5 format (placeholders for flags and errors) • Look-up-tables: convolution with slit functions SO 2 OD: 9 dimensions (wvl, SZA, VZA, RAA, TO 3, LER alb, LER height, SO 2 VC, SO 2 LH) 38. 5 106 spectra = 15. 1 Gb (uint 16) Rad @340 nm: 5 dimensions (SZA, VZA, RAA, LER alb, LER height) 87 Kb
Tests on BIRA synthetic spectra Retrieved SO 2 Layer Height (km) Ratio retrieved/true SO 2 Vertical Column SO 2 VCD (DU) O 3 VCD: 385 DU, SZA: 30°, Albedo: 5%
Tests on MPIC synthetic spectra • • • Use S 4 TDS v 2 simulated spectra Wvl range : 308 -345 nm, high resolution-sampling of Kurucz Post-processing convolution and resampling : representative of S 5 ISRF Spectral noise: no + yes Geometry: 2 locations, varying SZA (hourly and seasonally) • Benkowski (VZA : 55°) & [Helsinki (VZA : 71°)] Measurements : hourly measurements of 5 days in June and Dezember each no LER, BRFs are used SO 2 profile: • Volcanic eruption (2 days with 3 different loads varying daily/semidaily): • SO 2 VCD: 1 E 18 (med erupt. ) & 1 E 19 (high erupt. ) • Hourly changing plume height (4 -5, 6 -7 & 8 -9 km) Aerosols (volcanic cases): • All spectra calculated without and with aerosols • Sulfate Aerosol & volcanic ash for all 3 volcanic days • Aerosol in same height as SO 2 plume
Tests on MPIC synthetic spectra BL scenarios, E, no aerosols, no clouds, Benkowski (VZA: 55°), summer. threshold reqt. breakthrough reqt.
Tests on MPIC synthetic spectra BL scenarios, E, no aerosols, no clouds, Benkowski (VZA: 55°), summer. Differences in XS data used for generating LUT and radiance spectra 50% reqt. 30% reqt.
Tests on MPIC synthetic spectra BL scenarios, E, no aerosols, no clouds, Helsinki (VZA: 71°), summer.
Tests on MPIC synthetic spectra BL scenarios, S, sulfate aerosols, no clouds, Benkowski (VZA: 55°), summer.
Tests on MPIC synthetic spectra BL scenarios, S, sulfate aerosols, no clouds, Benkowski (VZA: 55°), summer.
Tests on MPIC synthetic spectra BL scenarios, V, volcanic ash, no clouds, Benkowski (VZA: 55°), summer.
Tests on MPIC synthetic spectra BL scenarios, V, volcanic ash, no clouds, Benkowski (VZA: 55°), summer.
Tests on MPIC synthetic spectra Sulfate aerosols Volcanic ash Difference in optical depth
Application to TROPOMI data • Transfer OMI TROPOMI almost ‘as is’ • 1 st test in 310. 5 -326 nm : failed (probably because of unconsolidated Irr in short UV) 312 -326 nm • Input needed : ü S 5 P_SO 2_L 2 ü S 5 P_L 1_BD 3 (Rad + Irr) • Preliminary results!
Application to TROPOMI data 2017 Ambae degassing
Application to TROPOMI data 2017 Ambae degassing Courtesy of L. Clarisse (ULB)
Application to TROPOMI data 2018 Sinabung eruption 13: 30 local time 10: 30 local time 19 Februari 2018 CALIOP track S 5 P passed several hours after the start of the eruption
Application to TROPOMI data
Application to TROPOMI data
Application to TROPOMI data Courtesy of L. Clarisse (ULB)
Application to TROPOMI data TROPOMI ALH (experimental) Courtesy of M. de Graaf and J. de Laat (KNMI)
Application to TROPOMI data Comparison to CALIOP data tropopause volcano TROPOMI
Application to TROPOMI data 2018 Fuego eruption S 5 P passed ~1 h after the start of the eruption AAI SO 2
Application to TROPOMI data
Conclusions (synthetic data) BL volc scenarios, E, no aerosols, no clouds. • Reasonable results even at high VZA, withing reqts • Not perfect because of differences in xs input data -> it would be helpful to regenerate a few TDS files. BL volc scenarios, S, sulphuric aerosols, no clouds. • SO 2 LH are often within reqts • SO 2 VCD strong underestimation (limitation in LER approach) BL volc scenarios, V, volcanic ash, no clouds. • SO 2 LH and VCD completely off. -> it would be helpful to regenerate a few TDS files with aerosol/cloud layer below SO 2 layer for further testing LER approach. -> it would be helpful to regenerate a few TDS files with intermediate AODs (e. g. 0. 25, 0. 5, 1). -> a new approach is needed for enhanced aerosols scenes
Conclusions (TROPOMI) • TROPOMI SO 2 LH algorithm: (very) preliminary results already show interesting features. Comparisons to other estimates (TROPOMI ALH, IASI, MLS, CALIOP) are reasonable as long as AAIs are not too high. • Future work: Ø Ø Ø Ø Fine tuning (e. g. extension of LUT for higher LER height) Consolidated solar measurements and wavelength calib. Treatment of aerosols Impact of polarization => impact on LH <0. 5 km Validation/verification Flags and error budget Optimization of processing time: currently 2. 5 x too slow compared to requirements (~ 130 min for 1/10 of S 5 orbit).
- Slides: 28