![Thresholds of Detection for Falling Snow from Satellite-Borne Active and Passive Sensors IGARSS 2011 Vancouver, Canada Gail Skofronick Jackson Benjamin Johnson Joe Munchak NASA Goddard Space Flight Center, Greenbelt, Maryland [email_address]](https://image.slidesharecdn.com/2-snow-igarss-2011-v6-110727153900-phpapp02/85/Thresholds-of-Detection-for-Falling-Snow-from-Satellite-Borne-Active-and-Passive-Sensors-1-320.jpg)
Thresholds of Detection for Falling Snow from Satellite-Borne Active and Passive Sensors
- 1. Thresholds of Detection for Falling Snow from Satellite-Borne Active and Passive Sensors IGARSS 2011 Vancouver, Canada Gail Skofronick Jackson Benjamin Johnson Joe Munchak NASA Goddard Space Flight Center, Greenbelt, Maryland [email_address]
- 2. Presentation Outline Contributions to Brightness Temperatures Falling Snow Detection Thresholds Analysis framework Active thresholds based on instrument sensitivity Passive thresholds Comparison between active and passive Future improvements Snow Field Campaign (Jan – Feb 2012) Summary
- 3. Percentages from Surface , Snow , & Water Vapor Lake Effect 2-3km tops (0.5 to 1.0 IWP) Synoptic 5-7km tops (0.5 to 1.0 IWP) Blizzard ~10km tops (0.5 to 1.0 IWP) Blizzard ~10km tops (9 to 10 IWP) “ Surface and Atmospheric Contributions to Microwave Brightness Temperatures for Falling Snow Events,” by Gail Skofronick-Jackson and Benjamin Johnson, JGR-Atmos, published Jan 2011. (a) (b) (a) (b) Macro and microphysical cloud characteristics affect TB signal These use dendrite snowflakes
- 4. Falling Snow Detection Thresholds What are the thresholds of detection in terms of IWP or IWC of falling snow? Analysis Approach: Use WRF models of Lake Effect and Synoptic snow Vertical profiles: IWC, temperature, water vapor profiles Surface: temperature, land classification, snow depth Joint active and passive computations of Z and TB Use Liu’s 2004 DDA tables for abs, scat, asymmetry, & backscatter 11 non-spherical snowflake shapes Adjust N0 to integrate Liu’s min-max DDA sizes to ensure WRF IWC is preserved
- 5. (1) Surface Emissivity Part 1 Urban crop land deciduous evergreen/mixed water Surface Temperature Vegetation Type Snow Depth WRF Simulations Courtesy of W.-K. Tao & team Lake Effect Case Synoptic Snow Case IWP (Jan 20 0400UTC) IWP (Jan 22 0600UTC)
- 6. Radar Calculations W-Band (-26dBZ) Ka-Band (12dBZ) Ku-Band (18dBZ) Thresholds of Detection for Falling Snow from Satellite-borne Active and Passive Sensors by G. Skofronick-Jackson, et al., IEEE TGRS, submit 9/11 These use 3-bullet rosette snowflakes
- 7. Reflectivities Depend on Particle Shape W-Band Ka-Band Ku-Band
- 8. Reflectivities Depend on Particle Shape W-Band Ka-Band Ku-Band Ka
- 9. Z-Thresholds Depend on Particle Shape Average IWC Detected at Surface Assumed minimum instrument Z: Ku: 18 dBZ Ka: 12 dBZ W: -15 dBZ ±One std dev of variability over 11 shapes is plotted Snowflake Shape (#) Ku-Band Ka-Band W-Band Long Hex Col. (0) 0.037 0.020 0.0020 Short Hex Col. (1) 0.037 0.020 0.0019 Block Hexag. Col. (2) 0.039 0.020 0.0020 Thick Hex Plate (3) 0.035 0.019 0.0019 Thin Hex Plate (4) 0.033 0.018 0.0022 3-Bullet Rosette (5) 0.062 0.038 0.0018 4-Bullet Rosette (6) 0.065 0.052 0.0026 5-Bullet Rosette (7) 0.062 0.047 0.0022 Six Bullet Rosette (8) 0.063 0.101 0.0023 Sector Snowflake (9) 0.077 0.049 0.0018 Dendrite Snow (10) 0.079 0.145 0.0032
- 10. Radiometer Threshold Procedure Y-Axis: TBhydr – TBclearair (with perfect surface, etc knowledge) X-Axis: IWP (max of 6 kg/m 2 ) 3-Bullet Rosette Shape: Red Line = Land surfaces, Blue line = Water Surfaces These use 3-bullet rosette snowflakes 10V 183 ± 3V 166V 89V 37V 183 ± 7V
- 11. Radiometer Thresholds Depend on Shape 89V 166V 166V 166H 183 ± 3V 183 ± 7V 166V 22 Jan
- 12. Radiometer Thresholds Depend on Snow Vertical Structure and Surface Type Channel (GHz) Total Threshold Cutoff (rounded up) (in K) From 0.05 error in emissivity From 10 o C error in surface T From 10% change in Tprofile From 10% change in RHprof 10 25 14 10 0 0 19 25 14 10 0 0 23 25 14 10 0 0 37 25 13 10 0 0 89 25 13 9 0 0 166 20 11 8 1 1 183±3 5 1 2 1 1 183±7 15 5 6 0 1
- 13. Radiometer Thresholds Depend on Snow Vertical Structure and Surface Type Channel (GHz) Total Threshold Cutoff Average Detected IWP Lake Effect over Land Detected IWP Lake Effect over Lakes V-pol Detected IWP Lake Effect over Lakes H-pol Detected IWP Synoptic over Land Detected IWP Synoptic over Lakes V-pol Detected IWP Synoptic over Lakes H-pol 10 25 19 25 23 25 3.2 na na 37 25 1.2 2.0 1.1 89 25 0.4 0.5 1.5 0.5 0.6 0.8 166 20 0.2 0.2 0.2 0.3 0.3 0.3 183±3 5 1.8 na 1.1 1.1 na 183±7 15 0.4 0.4 na 0.6 0.6 na
- 14. Active Versus Passive Snow Detection Thresholds of Detection for Falling Snow from Satellite-borne Active and Passive Sensors by G. Skofronick-Jackson, et al., IEEE TGRS, submit 9/11 Active Avg. Surface IWC Detected: Ku Ka W Units 0.08 0.07 0.004 g m -3 Simple falling snow conversion (melted snow rate) 1.01 0.93 0.027 mm hr -1 Passive over land Avg. Columnar IWP Detected: 89 166 183±3 183±7 Units Land V-Pol Lake Effect 0.43 0.16 1.85 0.37 kg m -2 Land V-Pol Synoptic 0.53 0.26 1.10 0.63 kg m -2 Simple IWC conversion (correct assumption????) Lake Effect (3 km clouds) 0.14 0.05 0.62 0.12 g m -3 Synoptic (6 km clouds) 0.09 0.04 0.18 0.11 g m -3 Simple falling snow conversion (melted snow rate) Lake Effect (3 km clouds) 1.97 0.61 11.19 1.65 mm hr -1 Synoptic (6 km clouds) 1.11 0.47 2.64 1.36 mm hr -1 Thresholds for passive could be improved with additional information
- 15. RGB Composite AMSU-B Emissivity Map Three Color Emissivity Map by Joe Munchak 89 GHz (red), 150 GHz (green), 183 GHz (blue) Darker colors indicate lower emissivities (more reflective) Missing data (black).
- 16. GCPEx Snowfall Campaign (Near Toronto, Canada Jan.-Feb. 2012) GV Science Radiometer/DPR Snowfall measurement sensitivities to snow type, rate, surface and tropospheric characteristics Physics of snowfall in the column and relation to extinction characteristics Model databases for forward modeling and retrieval development. Approach: DFIR instrument clusters (account for measurement uncertainty, mitigate wind, complimentary physics) centered around X/W/Ka-KU/MRR radars and a ground-staring radiometer at CARE site. Clusters located under C-band/D3R multi-freq/dual-pol radar umbrella; D3R V-point with W and X-bands or cover clusters in scanning/RHI/spectral sampling modes. Overfly in-situ aircraft in coordination with DC-8 (APR-2 and CoSMIR radiometer); Pre and post land surface radiative measurements by Ka-Ku and radiometers. O (60 km) O (10 km) 7-8 km 0.4-0.8 km Ht. King City C-band Dual-pol DFIR Clusters x Georgian Bay CARE D3R PSD: 2DVD, Parsivel, POSS,SVI Radar: Ka/Ku,X,W(2),MRR SWER: Pluvio, Hot Plate SWE/Depth L-Band + -sensor (Land/Snow) 10-89 GHz Radiometer Aircraft: DC-8, Citation x
- 17. Today’s Messages Falling snow retrievals are complex Challenges being addressed: non-spherical particles surface emissivity (2) Thresholds of Detection Theoretical thresholds of detection are promising Differences between active and passive detection thresholds Thresholds for passive could be improved with additional constraints (3) What matters? IWP, cloud thickness, surface underneath, snow particle shapes and PSD limits, and more (4) The GCPEx Field Campaign in 2012 will provide data to help address challenges and finalize algorithms.
- 18. Questions? Questions? IEEE Geoscience and Remote Sensing Society Administrative Committee (AdCom) Member Voting is open All GRSS members can vote for new AdCom members Please vote this week at the GRSS booth or online by Sept. 16, 2011
Editor's Notes
- #5: At 340GHz and higher, spheres were used instead of Liu non-spheres
- #11: A 1kg/m^2 threshold of detection for say 89 GHz means that if you distribute that 1kg/m^2 over a 5km cloud thickness (and if I did my math correctly) this means that one would need a surface LIQUID equivalent snow rate of ~3mm/hr (Hence the focus on the blizzard like events in the literature for passive snow events). For 0.5kg/m^2 the liquid equivalent is: 1.25mm/hr
- #12: A 1kg/m^2 threshold of detection for say 89 GHz means that if you distribute that 1kg/m^2 over a 5km cloud thickness (and if I did my math correctly) this means that one would need a surface LIQUID equivalent snow rate of ~3mm/hr (Hence the focus on the blizzard like events in the literature for passive snow events). For 0.5kg/m^2 the liquid equivalent is: 1.25mm/hr
- #16: dark=low emissivity, in this case from snow cover from blizzards in December 2006), why the oceans are blue (89=red,150=green,183=blue + emissivity increases with frequency = blue oceans), and why there is missing data (cloud cover or too much water vapor for all channels to "see" surface).