2012 AGU Chapman Conference on Remote Sensing of the ...

oth the quantification of human water use and thedefinition of water availability. Current statistics on humanwater use are often outdated or inaccurately reportednationally, especially for groundwater. This study usesremote sensing to improve these estimates. We use NASA’sGravity Recovery and Climate Experiment (GRACE) toisolate the anthropogenic signal in water storage anomalies,which we equate to water use. Water availability hastraditionally been limited to “renewable” water, whichignores large, stored water sources that humans use. Wecompare water stress estimates derived using eitherrenewable water or the total volume of water globally. We usethe best-available data to quantify total aquifer and surfacewater volumes, as compared to groundwater recharge andsurface water runoff from land-surface models. The workpresented here should provide a more realistic image ofwater stress by explicitly quantifying groundwater, definingwater availability as total water supply, and using GRACE tomore accurately quantify water use.Rittger, Karl E.Assessment of Viewable and Canopy AdjustedSnow Cover from MODISRittger, Karl E. 1 ; Painter, Thomas H. 2 ; Raleigh, Mark S. 3 ;Lundquist, Jessica D. 3 ; Dozier, Jeff 11. Bren School of Env. Science, Univ of Calif- Santa Barbara,Santa Barbara, CA, USA2. Jet Propulsion Laboratory, California Institute ofTechnology, Pasadena, CA, USA3. Civil and Environmental Engineering, University ofWashington, Seattle, WA, USACharacterization of snow is critical for understandingEarth’s water and energy cycles. The cryosphere’s response toforcings largely determines Earth’s climate sensitivity tochanges in atmospheric composition, and one-fifth ofEarth’s population depends on snow or glaciers for waterresources. Maps of snow from MODIS have seen growing usein investigations of climate, hydrology, and glaciology, butthe lack of rigorous validation of different snow mappingmethods compromises these studies. We examine threewidely used MODIS snow products: the “binary” (i.e. snowyes/no) global snow maps, MOD10A1 binary, that relies onthe normalized difference snow index (NDSI); a regressionbasedMODIS fractional snow product, MOD10A1fractional, that relies on an empirical relationship withNDSI; and a physically-based fractional snow product,MODSCAG, that relies on spectral mixture analysis. Wecompare them to maps of snow obtained from LandsatETM+ data to evaluate viewable snow cover. The assessmentuses 172 images spanning a range of snow and vegetationconditions, including the Colorado Rocky Mountains, theUpper Rio Grande, California’s Sierra Nevada, and the NepalHimalaya. MOD10A1 binary and fractional fail to retrievesnow in the transitional periods during accumulation andmelt while MODSCAG consistently maintains its retrievalability during these periods. Fractional statistics show theRMSE for MOD10A1 fractional and MODSCAG are 0.23and 0.10 averaged over all regions. MODSCAG performs themost consistently through accumulation, mid-winter andmelt with median differences ranging from –0.16 to 0.04while differences for MOD10A1 fractional range from –0.34to 0.35. MODSCAG maintains its performance over all landcover classes and throughout a larger range of land surfaceproperties. Estimating snow cover in densely vegetated areasfrom optical remote sensing remains an outstandingproblem. We use a network of ground temperature sensorsto monitor daily snow presence in the Sierra Nevada in threesites with varying forest canopy density. For MODSCAG, weestimate snow cover under the canopy using standardmethods, adjusting with the daily fractional vegetation fromthe MODSCAG algorithm. These adjustments are superiorto static maps such as the National Land Cover Datasetbecause both fractional vegetation and fractional snow covervary as a function of MODIS view angle and season.MOD10A1 viewable snow cover is compared directly to thefield data without vegetation corrections because of itssizeable overestimates in forested areas. Characterizingviewable snow cover by spectral mixing is more accurate thanempirical methods based on the normalized difference snowindex, both for identifying where snow is and is not and forestimating the fractional snow cover within a sensor’sinstantaneous field-of-view. Estimating canopy adjustedsnow cover should rely on our best estimates of the viewablesnow cover to produce reliable and defendable results.Ascertaining the fractional value is particularly important inmountainous terrain and during spring and summer melt.Rodell, MatthewIntegrating Data from GRACE and OtherObserving Systems for Hydrological Research andApplicationsRodell, Matthew 1 ; Famiglietti, James S. 2 ; McWilliams, EricB. 3 ; Beaudoing, Hiroko K. 1, 4 ; Li, Bailing 1, 4 ; Zaitchik,Benjamin F. 5 ; Reichle, Rolf H. 6 ; Bolten, John D. 11. Hydrological Sciences Laboratory, NASA/GSFC,Greenbelt, MD, USA2. Earth System Science, UC Irvine, Irvine, CA, USA3. Atmospheric and Oceanic Science, University ofMaryland, College Park, MD, USA4. Earth System Science Interdisciplinary Center, Universityof Maryland, College Park, MD, USA5. Earth and Planetary Sciences, Johns Hopkins University,Baltimore, MD, USA6. Global Modeling and Assimilation Office, NASA/GSFC,Greenbelt, MD, USAThe Gravity Recovery and Climate Experiment (GRACE)mission provides a unique view of water cycle dynamics,enabling the only space based observations of water on andbeneath the land surface that are not limited by depth.GRACE data are immediately useful for large scaleapplications such as ice sheet ablation monitoring, but theyare even more valuable when combined with other types ofobservations, either directly or within a data assimilationsystem. Here we describe recent results of hydrological125