On the use of Deep Convective Clouds DCC

On the use of Deep Convective Clouds (DCC) to calibrate visible sensors D. Doelling NASA La. RC Dan Morstad, Benjamin Scarino, Arun Gopalan GSICS/GWRG Session Beijing, China, March 5 -8, 2012 NASA Langley Research Center / Atmospheric Sciences

Background • DCC are cold, bright, stable, tropopause level, and nearlyisotropic targets located near the equator and are observed by all GEO and LEO satellites – 90% albedos at the lowest possible SZA – Predictable DCC albedos at low SZA and VZA • DCC are easily identified by a simple IR threshold. – IR imager calibration well calibrated with onboard blackbodies – Only good visible and IR co-registration is needed, good navigation is not a requirement • The collective identified DCC are analyzed for degradation, – approximately 0. 5% of all tropical pixels are identified as DCC NASA Langley Research Center / Atmospheric Sciences

The characterization of DCC for calibration • Evaluate the impact of the IR threshold temperature, spatial standard deviations, and ADM on the temporal stability and magnitude of DCC radiance – Use Aqua-MODIS • Produce regional DCC frequency maps • Produce regional maps to show impact of IR and spatial threshold to evaluate spatial and temporal variability • Compare DCC and other cloud SCIAMACHY spectra • Compare Hyperion and SCIAMACHY DCC spectra NASA Langley Research Center / Atmospheric Sciences

Baseline Aqua-MODIS 0. 65µm DCC method Parameters DCC Identification Threshold Geographical Extent 30°N to 30°S all surface types Angular Range SZA<40°, VZA<40° 11µm Temperature BT 11μm < 205°K DCC pixel Standard deviation 11µm Spatial Homogeneity Standard deviation BT 11μm <1°K based on all 9 pixels NASA Langley Research Center / Atmospheric 0. 65µm Spatial Homogeneity Standard deviation R <3%Sciences 0. 6μm

Baseline Aqua-MODIS 0. 65µm DCC method Monthly frequency PDFs Monthly PDF Rad. AC Monthly normalized trend Month # Mean Mode Kurtosis Skewness Mode Mean s 0. 68% • Apply ADM to derive DCC anisotropic corrected radiance • Compute monthly histogram • Perform temporal trend NASA Langley Research Center / Atmospheric Sciences s 0. 48%

IR Temperature Threshold for DCC Identification August 2004 Month # Mean Mode Kurtosis Skewness • 2%, 0. 6% increase in the mean, mode PDF Rad. AC with each 5° decrease • 64%, 99. 7% data reduction between 205°K and 200°K, 195°K, 190°K • Sufficient samples with IR threshold between 205°K and 195°K NASA Langley Research Center / Atmospheric Sciences

Spatial Thresholds for DCC Identification Month # Mean Mode Kurtosis Skewness s. VIS, August 2004 Month # Mean Mode Kurtosis Skewness s. IR, August 2004 • 1. 6%, 0. 4% increase in the mean, mode PDF Rad. AC with visible spatial • 0. 8%, 0. 6% increase in the mean, mode PDF Rad. AC with IR spatial NASA Langley Research Center / Atmospheric Sciences • 56% data reduction when applying both spatial thresholds

DCC ADMs No BRDF Model Hu Model CERES - Broadband MODIS based • The DCC ADM is nearly isotropic within 40° SZA and VZA • If the Aqua orbit is maintained, and DCC consistent in space and time, then viewing geometry is replicated annually and less dependency on ADM NASA Langley Research Center / Atmospheric Sciences

S/200°K DCC frequency No spatial threshold N/205°K N/200°K N/195°K S/205°K With spatial threshold 96 months possible NASA Langley Research Center / Atmospheric Sciences Months with DCC between 2002 -2011

Mean No spatial 205°K 0. 8% 1. 4% 1. 1% 1. 8% 1. 5% E) 3. 6% G) 2. 0% 1. 3% 1. 1% 3. 6% 1. 9% • For 0. 65µm the mode is more consistent in space • For 2. 12µm the mean is more consistent in space and time • 0. 65µm land not necessarily brighter than ocean, for 2. 12µm it is • For 2. 12µm land has less temporal uncertainty NASA Langley Research Center / Atmospheric Sciences 1. 9% 2. 12µm Mode 1. 5% Monthly standard deviation (%) DCC AC radiance (Wm -2 sr-1µm-1) Mean Tropics/Land 0. 65µm Mode 147 regions 1. 3%

Mean No spatial 200°K Tropics/Land 0. 6% 0. 5% 0. 6% 1. 2% 0. 9% 0. 7% 1. 6% Mode Mean 0. 9% 3. 1% 1. 6% NASA Langley Research Center / Atmospheric Sciences 1. 5% 2. 12µm 1. 4% 1. 3% Monthly standard deviation (%) DCC AC radiance (Wm -2 sr-1µm-1) 2. 9% 1. 3% 0. 65µm Mode 147 regions 1. 1%

Mean No spatial 195°K Tropics/Land 0. 5% 0. 3% 1. 0% 0. 3% 1. 4% Mode 1. 2% 1. 1% Monthly standard deviation (%) 0. 6% 1. 3% 3. 1% 1. 6% NASA Langley Research Center / Atmospheric Sciences 1. 8% 2. 12µm 2. 6% Mean DCC AC radiance (Wm -2 sr-1µm-1) 1. 1% 0. 65µm Mode 27 regions 1. 0%

Spatial 205°K Mean Tropics/Land 0. 6% 0. 5% 1. 6% 1. 5% 0. 7% 1. 9% Mode Mean 1. 1% 1. 0% 3. 7% 2. 1% NASA Langley Research Center / Atmospheric Sciences 2. 3% 2. 12µm 2. 1% 1. 4% Monthly standard deviation (%) DCC AC radiance (Wm -2 sr-1µm-1) 3. 1% 1. 1% 0. 65µm Mode 78 regions 1. 5%

Spatial 200°K Mean 0. 5% 0. 3% 1. 2% 1. 1% 0. 3% 1. 6% 1. 8% Mode 1. 2% 0. 7% 3. 4% 1. 8% NASA Langley Research Center / Atmospheric Sciences 1. 9% 2. 12µm 2. 5% 1. 0% Monthly standard deviation (%) DCC AC radiance (Wm -2 sr-1µm-1) Mean Tropics/Land 0. 65µm Mode 43 regions 1. 1%

DCC threshold characteristics • Spatial thresholds work better over land • Temperature thresholds work better over ocean NASA Langley Research Center / Atmospheric Sciences

200°K and spatial threshold • Use PDF mode statistic for <0. 65µm • Use PDF mode statistic for >1. 0µm • Land has less temporal uncertainty than over ocean NASA Langley Research Center / Atmospheric Sciences

SCIAMACHY reflectance spectra # 3253 340 16734 1799 • DCC not effected by water vapor absorption bands • Land footprints slightly brighter than ocean <0. 65µm NASA Langley Research Center / Atmospheric Sciences

SCIAMACHY reflectance standard deviation NASA Langley Research Center / Atmospheric Sciences

SCIAMACHY reflectance standard deviation (%) NASA Langley Research Center / Atmospheric Sciences

GOES-13 and Hyperion image over hurricane Rina, October 27, 2011 15: 50 GMT 4 km IR 1 km visible • Hyperion on EO-1 10: 00 AM, • 220 spectral bands from 0. 4μm to 2. 5μm at 10 nm increments • 7. 5 by 100 km spatial image NASA Langley Research Center / Atmospheric Sciences Hyperion 30 meter 0. 6 µm

Comparison of Hyperion and SCIAMACHY spectra SCIAMACHY ocean DCC Hyperion Oct 27 mean • The reflectance spectra are similar, except maybe at 1. 4µm • Hyperion reflectance based on the USGS Hyperion solar spectra NASA Langley Research Center / Atmospheric Sciences

Hyperion reflectance ratios Brightest 10% Mean Darkest 10% • The spectral reflectance shape is independent of brightness and nearly flat < 0. 9µm • Brightest/mean or darkest/mean is within 1% between 0. 4 to 0. 9µm • The standard deviation of the NASA Langley Research Center / Atmospheric Sciences Hyperion reflectances are <2%

Conclusions • DCC calibration is most successful if the monthly PDF shape is consistent from month to month – – – Must balance the most stringent threshold with sample size TWP, Africa and South America have the most DCC PDF mode best for wavelengths < 0. 7µm PDF mean best for wavelengths > 1. 0µm Spatial threshold more effective over land Possible to calibrate all wavelengths < 2. 2µm • DCC over land provides a more predictable radiance and less uncertainty than over ocean • DCC not impacted by water vapor bands between 0. 4 and 1. 6µm as with other cloud types NASA Langley Research Center / Atmospheric Sciences