Cooperative Agreement NNX11AK35A

SEVERAL MODIS-DERIVED EARTH SYSTEM DATA RECORDS HAVE BEEN DEVELOPED TO ASSESS SNOW RESOURCES IN MOUNTAIN AREAS. HOWEVER THEY HAVE NOT BEEN RIGOROUSLY VALIDATED AND THEIR UNCERTAINTIES AND SYSTEMATIC ERRORS ARE NOT WELL CHARACTERIZED. WE PROPOSE TO VALIDATE: (I)DAILY MODIS FRACTIONAL SNOW COVER CALCULATED BY BOTH NORMALIZED DIFFERENCE SNOW INDEX (NASA STANDARD PRODUCTS MOD10A1 AND MYD10A1) AND SPECTRAL UNMIXING; (II) DAILY SNOW ALBEDO; AND (III) SPATIOTEMPORALLY SMOOTHED FRACTIONAL SNOW COVER AND ALBEDO AS A MONTHLY PRODUCT. FOR SEVERAL WET AND DRY YEARS WE WILL ASSEMBLE A SET OF PRODUCTS AND THEIR VALIDATION DATASETS. THE VALIDATION DATA WILL INCLUDE A COMPREHENSIVE (~1000) SET OF LANDSAT SCENES AUGMENTED AT SOME LOCATIONS AND TIMES BY HIGHER RESOLUTION DATA (DIGITIZED AERIAL PHOTOGRAPHS QUICKBIRD IMAGER AND DECLASSIFIED IMAGES FROM NATIONAL RECONNAISSANCE SATELLITES) FROM WHICH ACCURATE BINARY MAPS OF SNOW COVER CAN BE CREATED. WE WILL VALIDATE SNOW IN THE FORESTS WITH EXISTING LIDAR SNOW SURVEY DATA TO CHARACTERIZE SNOW DEPTH VARIABILITY AT SUB-METER SCALES INCLUDING SUB-CANOPY AREAS. WE WILL USE ALSO USE EXISTING ULTRASONIC SNOW DEPTH SENSORS TO CHARACTERIZE SNOW ACCUMULATION MELT AND DURATION IN OPEN CANOPY-EDGE AND SUB-CANOPY CONDITIONS. MODIS LOOKS ##55## FROM NADIR YIELDING (BECAUSE OF EARTH CURVATURE) A ##65## GROUND ZENITH ANGLE. DURING PERSISTENT CLOUD COVER THE ONLY VIEW OF THE GROUND FOR A WEEK OR TWO MIGHT WELL BE AT SUCH A HIGHLY OBLIQUE ANGLE. TO EVALUATE FRACTIONAL SNOW COVER IN THOSE SITUATIONS WE WILL USE LANDSAT TM OR ETM+ SCENES THAT COINCIDE WITH THE EDGES OF SNOW-COVERED MODIS IMAGES. WE WILL COMPARE TIME-SPACE SMOOTHING AND INTERPOLATION TO A SEQUENCE OF LANDSAT TM OR ETM+ SCENES OVER THE SIERRA NEVADA FROM 2000 TO PRESENT FOCUSING ESPECIALLY ON LOCATIONS AND TIMES WHEN THE SMOOTHED AND INTERPOLATED PRODUCT DIFFERS FROM THE RAW MODIS FRACTIONAL SNOW COVER. WE WILL VALIDATE BOTH THE TIME-SPACE CONTINUITY PRODUCT'S PREDICTIVE MODE (ESTIMATE SNOW-COVER ON A GIVEN DAY USING ONLY DATA PRIOR TO THAT DAY) AND ITS RETROSPECTIVE MODE (RECONSTRUCT THE HISTORY OF SNOW PROPERTIES FOR A PRIOR PERIOD.) WE HAVE ACCESS TO SNOW ALBEDO DATA AND COINCIDENT AVIRIS IMAGERY FROM FIELD SITES AT THE SENATOR BECK BASIN IN THE SAN JUAN MOUNTAINS COLORADO AND AT MAMMOTH MOUNTAIN IN THE SIERRA NEVADA CALIFORNIA. WE WILL USE THESE DATA TO VALIDATE AVIRIS-DERIVED SNOW ALBEDO AND THEN COINCIDENT AVIRIS/MODIS DATA TO VALIDATE MODIS-DERIVED ALBEDO. THE MODIS CLOUD PRODUCT CONTAINS ERRORS OF BOTH OMISSION AND COMMISSION AND ALL MODIS SNOW PRODUCTS SOMETIMES CONFUSE CLOUDS AND SNOW. THE TIME-SPACE CONTINUITY PRODUCT ATTEMPTS TO CORRECT THIS BY FORCING SNOW THAT IS TOO FINE-GRAINED AND COVERS THE WHOLE PIXEL TO BE CLASSIFIED AS CLOUD WHILE CLOUDS THAT ARE TOO COARSE-GRAINED ARE CLASSIFIED AS SNOW. WE WILL VALIDATE THIS APPROACH USING LANDSAT TM AND ETM+ WHICH CAN IDENTIFY CLOUDS BOTH BY TEXTURE AND BY SNOW/CLOUD CONTRAST IN BAND 5. SNOWMELT RUNOFF FORECASTING IN MOUNTAINOUS AREAS HAS HISTORICALLY USED EMPIRICAL MODELS FORCED BY SPARSE IN SITU MEASUREMENTS. THESE MODELS ALREADY HAVE LARGE ERRORS IN SOME YEARS AND SINCE THEY RELY ON A DATA RECORD THAT ASSUMES STATIONARITY THEY ARE THEORETICALLY ILL-SUITED FOR WATER MANAGEMENT IN A CHANGING CLIMATE. COUPLING FRACTIONAL SNOW COVER AND ALBEDO TO SNOWMELT MODELS AND RESERVOIR OPERATIONS WOULD BE SIGNIFICANTLY ADVANCED BY OUR PROPOSED INVESTIGATION WHICH WOULD VALIDATE THE PRODUCTS ANALYZE THE STRUCTURE OF ERRORS AND ADVISE USERS OF CAVEATS AND LIKELY ACCURACY. MOREOVER THE VALIDATION DATASETS WILL BE USEFUL AFTER THE INVESTIGATION IS FINISHED AS COMPREHENSIVE TEST CASES FOR FUTURE RESEARCHERS.