Meteorological Satellite Center MSC of JMA Making BRDF

Meteorological Satellite Center (MSC) of JMA Making BRDF for GSICS-DCC with RT Calculation & Voronoi-particle + other related things *YOGO Yusuke (+ many thanks to Seb, Tim, Dave and Raj) Japan Meteorological Agency / ex. Meteorological Satellite Center Made on 2020 -03 -13

Meteorological Satellite Center (MSC) of JMA VIS/NIR Inter-cal Using DCC (GSICS-DCC) 2020 -03 -13 2

Meteorological Satellite Center (MSC) of JMA GSICS-DCC • GSICS-DCC – Procedure 1. Collect DCC pixels – – – Lat, Lon ≤ SSP +/- 20 deg SZA, VZA ≤ 40 deg 10 ≤ RAA ≤ 170 deg T(11 μm) ≤ 205 K Regarding VIS (0. 6 μm) Radiance: St. Dev 3*3 px / Average 3*3 px ≤ 0. 03 Regarding TIR (11 μm) Brightness Temp. : St. Dev 3*3 px ≤ 1 K DCC ☞ Need for preparing ice cloud BRDF when applied to NIR frequency 2. Apply BRDF for normalizing toward SZA = VZA = 0 deg 3. Make radiance and reflectance PDF 4. Derive cal. coefficient by comparing mode values of PDFs radiance 2020 -03 -13 3

Meteorological Satellite Center (MSC) of JMA Particle shapes in ice clouds • VIS/NIR Mie-scattering property above cloud strongly depends on cloud particle shapes Schmitt and Heymsfield (2010) – liquid water cloud: we can assume sphere simply – Ice cloud: fractal-like complex shape • Regarding ice cloud RT calculation, we need to assume appropriate particle shape or result will be change largely • Voronoi-aggregate particle: simulates actual ice particles by using “Voronoi aggregation” Most alike Bullet-rosette Droxtal Voronoi Sphere 球形 Hexagonal Plate – Some JMA’s satellite products (e. g. OCA) will also introduce Voronoi particle 2020 -03 -13 4

Meteorological Satellite Center (MSC) of JMA Voronoi’s scattering property • Scattering phase function Voronoi shows uniform backscattering ＆ almost no peak at θ = 180 (a) Sphere (b) Hexagonal Halo (c) Voronoi scatter incident λ = 0. 64 μm, Reff = 40 μm Hayashi (2018) • Weak at θ ～ 90 • Peak at θ ～ 135 (rainbow) • Peak at θ ～ 180 2020 -03 -13 • Peak at θ ～ 20 (halo) • Peak at θ ～ 180 • Uniform backscattering • No peak at θ ～ 180 • No unique peak Voronoi has more smooth scattering property than others 5

Meteorological Satellite Center (MSC) of JMA Platnick et al. (2015): MODIS Cloud Optical Properties: User Guide for the Collection 6 Level-2 MOD 06/MYD 06 Product and Associated Level-3 Datasets https: //modis-images. gsfc. nasa. gov/_docs/C 6 MOD 06 OPUser. Guide. pdf Compare with MODIS C 6 (aggregated hexagonal column) • Similar (Uniform backscattering) MODIS C 6 Voronoi scatter incident 2020 -03 -13 6

Meteorological Satellite Center (MSC) of JMA Calculate Voronoi ice cloud top BRDF with RSTAR 7 Result 2020 -03 -13 7

Meteorological Satellite Center (MSC) of JMA Result: 0. 64 μm (= H 8/AHI B 03) • Optical thickness = 200, effective radius = 20 μm Liquid, Sphere forward scattering when large SZA Ice, Sphere Viewing Zenith: outside 80 ← center 0 → outside 80 Relative Azimuth: left 180 ↶ right 0 backscattering SZA=VZA, RAA=180 Ice, Voronoi Ice, Hexagonal Normalized reflectance (average of whole region at each SZA = 1) 2020 -03 -13 No strong backscattering 8

Meteorological Satellite Center (MSC) of JMA Result: 2. 26 μm (= H 8/AHI B 06) • Optical thickness = 200, effective radius = 20 μm Liquid, Sphere forward scattering when large SZA Ice, Sphere Viewing Zenith: outside 80 ← center 0 → outside 80 Relative Azimuth: left 180 ↶ right 0 backscattering SZA=VZA, RAA=180 Ice, Voronoi Ice, Hexagonal Normalized reflectance (average of whole region at each SZA = 1) 2020 -03 -13 No strong backscattering 9

Meteorological Satellite Center (MSC) of JMA Compare with SNPP/VIIRS empirical BRDF 2020 -03 -13 10

Meteorological Satellite Center (MSC) of JMA Compare with SNPP/VIIRS empirical BRDF: 0. 64 μm • Optical thickness = 200, effective radius = 20 μm • Compare with empirical BRDF derived from SNPP/VIIRS DCC observation (Bhatt et al. 2017) – Voronoi looks the most similar Liquid, Sphere Ice, Sphere Solar Zenith: 20. 0 (Bhatt: 22. 5) Viewing Zenith: outside 55 ← center 0 → outside 55 Relative Azimuth: left 180 ↶ right 0 strong backscattering Ice, Hexagonal SNPP/VIIRS DCC BRDF (Bhatt et al. 2017) (no observation) Ice, Voronoi Normalized reflectance (average of whole region at each SZA = 1) 2020 -03 -13 11

Meteorological Satellite Center (MSC) of JMA Compare with SNPP/VIIRS empirical BRDF: 2. 26 μm • Optical thickness = 200, effective radius = 20 μm • Compare with empirical BRDF derived from SNPP/VIIRS DCC observation (Bhatt et al. 2017) – Voronoi looks the most similar Liquid, Sphere Ice, Sphere Solar Zenith: 20. 0 (Bhatt: 22. 5) Viewing Zenith: outside 55 ← center 0 → outside 55 Relative Azimuth: left 180 ↶ right 0 strong backscattering SNPP/VIIRS DCC BRDF (Bhatt et al. 2017) (no observation) Ice, Hexagonal Ice, Voronoi Normalized reflectance (average of whole region at each SZA = 1) 2020 -03 -13 moderate backscattering 12

Meteorological Satellite Center (MSC) of JMA Results using the new BRDFs 2020 -03 -13 13

Meteorological Satellite Center (MSC) of JMA GSICS-DCC Results • • • Black: RTM Vi. Cal w/ MODIS Green: DCC w/ MODIS Blue: DCC w/ VIIRS B 01 -04: looks promising no significant impact against Hu BRDF and no BRDF, because DCC selection criteria has already chosen stable geometry B 05 -06: small bias but looks unstable, more investigation needed MODIS: multiplied x 0. 45 for putting around 1 2020 -03 -13 14

Meteorological Satellite Center (MSC) of JMA Summary • GSICS-DCC needs ice cloud BRDF ☞ try to make BRDF LUT by RTM calculation • Scattering property on clouds strongly depends on cloud particle shapes Due to that, unlike liquid water clouds (= sphere), calculation on ice clouds is difficult • Voronoi-aggregate particle has simulated complex actual ice cloud particle shape and has smooth scattering property than others (e. g. : uniform backscattering, no significant peaks) • We made BRDF LUT on top of Voronoi clouds by RTM (RSTAR 7) calculation Compared with the other shapes (sphere & hexagonal), Voronoi shows a vastly different result and it is the most corresponding to SNPP/VIIRS empirical BRDF 2020 -03 -13 15

Meteorological Satellite Center (MSC) of JMA Thank you for your attention • Reference – Bhatt et al. (2017): Development of Seasonal BRDF Models to Extend the Use of Deep Convective Clouds as Invariant Targets for Satellite SWIRBand Calibration. https: //doi. org/10. 3390/rs 9101061 – 林 (2018): ひまわり８号観測バンドにおける雲放射特性の計算方法とその応用. http: //www. data. jma. go. jp/mscweb/technotes/msctechrep 63 -1. pdf – Ishimoto et al. (2012): Irregularly shaped ice aggregates in optical modeling of convectively generated ice clouds. https: //doi. org/10. 1016/j. jqsrt. 2012. 017 – Ishimoto et al. (2013 a): Optical Modeling of Irregularly Shaped Ice Particles in Convective Cirrus. https: //doi. org/10. 1063/1. 4804737 – Ishimoto et al. (2013 b): Construction of aerosol and ice particle scattering database for advanced remote sensing algorithms. https: //suzaku. eorc. jaxa. jp/GCOM/meeting/jointws 2013/program/presen/gcomc/c_17_HIshimoto. pdf – Schmitt and Heymsfield (2010): The Dimensional Characteristics of Ice Crystal Aggregates from Fractal Geometry. https: //doi. org/10. 1175/2009 JAS 3187. 1 2020 -03 -13 16

Meteorological Satellite Center (MSC) of JMA VIS/NIR vical using DCC and LUTs derived by RTM (EUM-DCC) 2020 -03 -13 17

Meteorological Satellite Center (MSC) of JMA EUM-DCC • Vicarious (not inter-) calibration method developed by EUMETSAT and Rayference • Procedure 1. Make DCC radiance LUT by RTM (EUMETSAT’s RTMOM) calculation (COT = 100, Reff = 23 (land) or 25 (ocean) μm, LUT dimensions = SZA, VZA and RAA) 2. Collect DCC pixels as an adjusted (tightened) GSICS-DCC criteria • • • Lat, Lon ≤ SSP +/- 30 deg SZA, VZA ≤ 30 deg cone angle ≥ 5 deg T(11 μm) ≤ 205 K Regarding ALL VIS/NIR (0. 4 -2. 3 μm) Reflectance: St. Dev 5*5 px ≤ 0. 02 Regarding TIR (11 μm) Brightness Temp. : St. Dev 5*5 px ≤ 0. 5 K 3. Retrieve simulated value from LUT 4. Accumulate 30 -day result Plot each pairs on scatter-plot and make linear regression • Tentatively uses 3 obs/day: 0300, 0400, 0500 UTC (H 8/AHI local noon = 0240 UTC) 2020 -03 -13 18

Meteorological Satellite Center (MSC) of JMA First result • Default: If St. Dev 5*5 px(IR BT) < 0. 50, St. Dev 5*5 px(VIS/NIR Reflectance) < 0. 02 • ~70000 DCC pixels / 30 days • AHI obs (X-axis) have too wide range 2020 -03 -13 19

Meteorological Satellite Center (MSC) of JMA First result (adjusted) • Adjusted: If St. Dev 9*9 px(IR BT) < 0. 50, St. Dev 9*9 px(VIS/NIR Reflectance) < 0. 01 • ~ 7000 DCC pixels / 30 days • B 01 -04: looks much better – B 03: slope=0. 951 seems to be a bit small • B 05 -06: more investigation needed 2020 -03 -13 20

Meteorological Satellite Center (MSC) of JMA Thank you for your attention • • JMA has implemented EUMETSAT’s DCC vicarious calibration algorithm for Himawari-8/AHI B 01 -06 If the original criteria are used obs values will have wide range and be unstable We tried to tighten the St. Dev conditions and we got much stable result, especially B 01 -04 For B 05 -06, more investigation needed? • Reference – Wagner and Govaerts (2017) Developing Deep Convective Cloud Reference Model for Vicarious Calibration https: //www. eumetsat. int/website/home/Data/Science. Activities/Science. Studies/Developing. Deep. Convective. Cloud. Reference. Modelfor. Vicarious. C alibration/index. html 2020 -03 -13 21