Initial Results on the Cross-Calibration of QuikSCAT and Oceansat-2 ScatterometersDavid G. Long

Department of Electrical and Computer EngineeringBrigham Young UniversityMay 2011

QuikSCAT vs OSCAT

ParameterQuikSCATOceanSat-2 ScatterometerSponsoring OrganizationNASA/JPLISROOperational Frequency13.402 GHz13.515 GHzPolarization (Inner/Outer)HH/VVHH/VVAntenna Diameter1 m1 mAltitude at Equator803 km720 kmOrbit Near Repeat Cycle4 days2 daysLocal time at asc/desc node6:00 a.m. at asc nodenoon at desc nodeHH 3dB footprint (Az x El)24 x 31 km26.8 x 45.1 kmVV 3 dB footprint (Az x El)26 x 36 km29.7 x 68.5 kmIncidence Angle (Inner)46 deg49 degIncidence Angle (Outer)54 deg57 degSwath Diameter (Inner)1400 km1400 kmSwath Diameter (Outer)1800 km1836 kmPointing Accuracy+/- 0.05 deg+/- 0.15 degElevation Pointing (w/ attitude error)+/- 0.05 deg+/- 0.25 degBegin / End Date201-1999/ 327-2009309-2009/ present

OSCAT & QuikSCAT Hi ResQuikSCAT and OceanSat-2 Scatterometer (OSCAT) have comparable spatial sampling and resolution

Similar image enhancement possibleCan use QuikSCAT algorithms for OSCAT

OSCATGRDQuikSCATGRDQuikSCATSIROSCATSIR

Enhanced Resolution* OSCAT ImageJD 309, 2009* Preliminary 2.225 km/pixelConventional resolution 25 km/pixel

QuikSCAT/OSCAT Image ComparisonComparison of one day BYU backscatter imagesOSCAT and QuikSCAT have nearly identical characteristicsDifferences due to azimuth and local time of daySimilar variances and meansSimilar spatial enhancement possible

OSCAT can contribute to the multi-decade scatterometer climate record of land and ice observations

OSCAT HQuikSCAT HOSCAT VQuikSCAT V

OSCAT/QuikSCAT Differences for the Land/Ice Scatterometer Climate RecordNominal incidence angle differsH: QuikSCAT=46, OSCAT=48V: QuikSCAT=54, OSCAT=56

Orbit geometry differsOSCAT has better coverage near poles (smaller holes)Time of orbit ascending node differ QuikSCAT=6:30 am OSCAT=noon Local time of measurements vary (location dependent)Orbit revisit time (Q=4 day repeat, O=2 day repeat)Azimuth angle distributions differ and varyNeed to apply azimuth angle corrections

Improved sigma-0 cross-calibration needed

Linear model for sigma-0 vs incidence angle Simplified modelASCAT Amazon Rain Forest ExampleASCAT Can also use for egg/slice incidence angle correctionASCAT(dB)(dB)

Backscatter Anisotropy Due to sastrugi and topography, some polar regions exhibit anisotropic backscatter response - Differences in azimuth geometry can be confused withclimate changes if not accounted for

Do not expect azimuth variations over the Amazon ASCAT Wilkes Land ExampleASCAT (C-Band)ASCAT (C-Band) V-pol40 inc(dB)

QuikSCAT AnisotropyQuikSCAT has fixed incidence angles but high diversity in azimuth angle observationsSimilar anisotropy observedQuikSCAT H-pol (Ku-Band)QuikSCAT V-pol (Ku-Band)QuikSCAT (Ku-Band)54 inc46 inc

QuikSCAT / ASCAT ComparisonDifferent frequencies (5.4 GHz vs 13.5 GHz) and incidence angles (40 V vs 46 H & 54 V)Consistent with dominant sastrugi scattering

OSCAT Azimuth Modulation AnalysisLocations of OSCATsigma-0 measurementswithin study regionJD 309-327, 2009OSCAT slice measurements: 13,099QuikSCAT slice measurements: 14,118OSCAT 310-311, 200925 km25 km

Slice Sigma-0 vs Azimuth Angle

Azimuth Corrected Slice Sigma-0 vs Azimuth Angle

Comparison of OSCAT and QuikSCAT modulation for the study regionOSCAT azimuth modulation does not match QuikSCAT azimuth modulationImproved processing is expected to resolve thisQuikSCATOSCATH-polV-pol

OSCAT Slice Sigma-0 vs Incidence Angle(narrow incidence angle range)

Azimuth Corrected Slice Sigma-0 vs Incidence Angle

Comparison of Sigma-0 Distributions in the Antarctic Study RegionV-pol bias 0.0 dBH-pol bias 0.7 dB

Comparison of Azimuth Corrected Sigma-0 Distributions in the Antarctic Test Region

V-pol bias 0.2 dBH-pol bias 0.6 dB

Amazon Study RegionSelect region that both QuikSCAT and OSCAT sigma-0 fall within narrow range

Rain forest is a good calibration target (anisotropic), but exhibits spatial inhomogeneitySelect homogenous regionTime-of-day variationSigma-0 varies with time of day as moisture moves up/down in canopySeveral tenths of a dB effect OSCAT and QuikSCAT observe at different local timesNo azimuth variation expectedDifferent incidence anglesSmall mean differences

Egg Sigma-0 vs Incidence Angle

Incidence-Corrected Egg Sigma-0 vs Incidence Angle

Egg Sigma-0 vs Azimuth Angle

Azimuth-Corrected Egg Sigma-0 vs Azimuth Angle

Comparison of Egg Sigma-0 distribution in Amazon Study RegionV-pol bias 0.25 dBH-pol bias 0.05 dB

Comparison of Corrected Sigma-0 Distribution in Amazon Study Region V-pol bias 0.25 dBH-pol bias 0 dB

OSCAT Local Time of Day AnalysisTime in minutes from start

OSCAT Local Time of Day AnalysisArcticAntarcticEqui-latitude strips used for measurement extraction in LTD analysis superimposed upon OSCAT gridded sigma-0 images of the polar regions

Comparison of Northern Hemisphere Local Time of Day ObservationsScatterplot of LTD vs UTC in the Northern Hemisphere for different longitude bins (a) OSCAT, (b) Seawinds, (c) QuikSCATOSCATSeaWindsQuikSCATLTD (hours) = UTC + Local_Longitude / 15

Comparison of Southern Hemisphere Local Time of Day ObservationsScatterplot of LTD vs UTC in the Southern Hemisphere for different longitude bins (a) OSCAT, (b) Seawinds, (c) QuikSCATOSCATSeaWindsQuikSCATLTD (hours) = UTC + Local_Longitude / 15

Diagram of LTD Divisions for Four ScatterometersLTD (hrs) = UTC + Local_Longitude / 1524 hours24 hours

ConclusionQuikSCAT and OSCAT sensors very similarCalibrated OSCAT products will be similar to QuikSCAT productsValidated QuikSCAT land/ice SCP products Daily Antarctic iceberg products (operational)Daily sea ice extent and mapping (operational, widely distributed)Daily FY/MY ice classification (relatively new)Can be averaged to longer time scalesPost wind mission (PWM) QuikSCAT data supports OSCAT calibration PWM QuikSCAT Coverage is too limited for less than monthly maps, aliasing an issue for ice movement for monthly maps

The linear (in dB) model relating sigma-0 to incidence angle is only an approximation. Several days of data over a uniform region of the Amazon is analyzed and the measurements plotted to validate the model for this region. On the left, the raw data an is linear (in dB) fit is shown. After removing the incidence angle dependence the residual backscatter is essentially constant with incidence angle.Over most land and ice surfaces, there is little azimuthal modulation (directional dependence) of sigma-0. However, as illustrated here, in certain regions where the surface has coherent features such as snow dunes (sastrugi), azimuth variation of sigma-0 is observed. As a result, small changes in the azimuth observation geometry can result in different mean sigma-0 values. Because the orbit, which defines the azimuth angles each location on the Earth is observed at, drifts over time and season, if the azimuth variation is not accounted for, changes in sigma-0 due to geometry can be confused with long-term changes. An update to the SCP products will incorporate corrections for this effect.You are going to talk about certain things here.