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

Senay, GabrielEvaluating the performance of remote sensingbased evapotranspiration products using fluxtower and water balance approachesSenay, Gabriel 1 ; Singh, Ramesh 2 ; Bohms, Stefanie 3 ; Verdin,James P. 11. USGS, Sioux Falls, SD, USA2. Science, ARTS, Sioux Falls, SD, USA3. SGT, Sioux Falls, SD, USAOperational estimation of actual evapotranspiration(ETa) is important for early warning applications in droughtand crop performance monitoring. The US GeologicalSurvey (USGS) developed the Simplified Surface EnergyBalance (SSEB) model, a simplified modeling approach tocalculate ETa on an operational basis for early warningapplications. The SSEB approach has been implemented forthe conterminous US and Africa to produce ETa on an 8-daybasis that is in turn used to create monthly and seasonal ETanomalies. Thermal data sets from MODIS 8-day productwere used to calculate an ET fraction which is combinedwith reference ET calculated from NOAA-produced dataassimilatedglobal weather data sets. ET anomalies arecalculated as a percent of the average ET (2000-2010) fromthe same month or season. While ET anomalies aresufficient to detect relative crop performance or detecting adrought, water budget studies require accurate estimation ofETa in terms of absolute magnitudes. We evaluated theaccuracy of SSEB based ETa with two data sources. MonthlyETa was evaluated using 18 flux tower data sets in theconterminous US using available data from 2000 through2010. In addition, Annual ETa was evaluated usingwatershed water balance data sets at the HUC8 (hydrologicunit code, level 8) scale. Annual water balance (precipitationminus runoff) for more than 1,300 HUCs were compared toHUC average ET with the assumption that net storagechange is negligible at annual time scale with little interbasinexchange for the selected HUCs. SSEB ET showedstrong correspondence with both flux tower (R2 varied from0.6 to 0.85 depending on years) with a generaloverestimation bias and HUC based water balanceevaluations (R2 > 0.9). For the water balance basedevaluation by contrast, SSEB ET showed a slightunderestimation in lower ET regions (annual ET < 600 mm).While the relationships in the two evaluation approaches arestrong, the differences in the direction of the bias needfurther investigation. However, the initial results suggestthat a simplified modeling approach for estimating ETproduces precise (high R2) and useful products that can beused both in relative applications such as droughtmonitoring but also for water budget studies at watershedscales. Calibrated and validated watershed-scale ET productsare being evaluated for the WaterSMART program of theDepartment of the Interior under which the USGS center forEarth Resources Observation and Science (EROS) isconducting a nationwide study on water availability andwater use.Seto, ShintaNecessity of Integrated Observations of SoilMoisture and Precipitation by Microwave RemoteSensingSeto, Shinta 11. Institute of Industrial Science, University of Tokyo,Tokyo, JapanBoth precipitation and soil moisture are primarycomponents of the terrestrial water cycle, and their spatialand temporal variations are large. Satellite microwave remotesensing is an indispensible tool to measure the variations ofprecipitation and soil moisture, but it is almost impossibleto measure the two components separately from each other.Information on one component is required for moreaccurate estimation of another component. Below studyshows the necessity of integration of precipitation and soilmoisture observations. For space-borne precipitation radar,surface reference technique (SRT) is available. SRT is appliedto derive path integrated attenuation (PIA) from the changein measured surface backscattering cross section (s0m)between no-rain and raining cases. To estimate PIAaccurately, the change in actual surface backscattering crosssection (s0e) caused by the variation in surface conditionsshould be considered. An analysis of Tropical RainfallMeasuring Mission (TRMM) / Precipitation Radar (PR) datashowed that s0e tends to become higher under rainfall. Asthe change of s0e is not explicitly considered, SRTunderestimates PIA and the average of PIA in many regionsis negative which is physicaly incorrect (as shown in Fig.1).The tendency is more apparently seen in sparsely vegetatedarea such as the Sahel of Africa, but not clearly seen indensely vegetated area such as the Amazonia. The increase ins0e is considered to be caused by increase in surface soilmoisture, and is called soil moisture effect. In the latestTRMM/PR standard algorithm, 0.5dB is added to PIA overland to compensate the soil moisture effect. The offset mayincrease precipitation rate at most around 7 %, however, theoffset value should be given more appropriately consideringthe characteristics of soil moisture variations.The average of PIA estimates in previous version of the TRMM/PRstandard algorithm.132