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

esolution lidar-derived DEM was compared to the NED 10m resolution DEM, the streams delineated with the 10 mlidar data were significantly better than those modeled withthe 10 m NED data, showing that significant improvementin accuracy can be achieved with no increase in data storage.When topographic index was modeled with multipleresolutions of lidar-derived DEMs, the spatial and statisticaldistributions were both very different, with finer resolutionDEMs not accurately modeling areas of high TI.Additionally, depending on the flow accumulationalgorithm used, there were differences in the change instatistical resolution with response to initial DEMresolution.Brutsaert, WilfriedSome Indirect Estimates of Changes in HydrologicConditions During the Past Century INVITEDBrutsaert, Wilfried 11. Civil & Environmental Eng, Cornell Univ, Ithaca, NY,USAThe water budget of a natural river basin can beformulated as P - Q - E = dS/dt, where P is the precipitationrate, Q the net surface outflow rate per unit area, E theevaporation rate and S the water stored per unit area in thebasin. The variables P and Q can be and have been measureddirectly and many long-term data sets are available for basinsall over the world, with which their evolution over time canbe studied in great detail. The construction of reliable longterm data sets for E and S for climate change purposes ismore challenging; indeed, the direct and routinemeasurement of these variables is still very difficult, so thatin practice to gain information on past trends they mustinvariably be estimated by indirect methods. In the case of E,attempts have been made to estimate past trends oflandscape evaporation from available pan evaporationrecords. Because pan and landscape evaporation areintrinsically different especially under drying conditions,there is still no unanimity regarding the interpretation ofthese studies. In the case of S, it is generally agreed thatterrestrial storage in a basin is related to the baseflows or dryweather river flows from the basin; this has allowed todocument the evolution of groundwater storage in manylarge basins under widely different climate conditions fromavailable streamflow records. While these approaches toestimate E and S involve obvious challenges, they can beovercome.Castro, Lina M.Assessment of TRMM Multi-satellite PrecipitationAnalysis (TMPA) in the South of Chile’s AndesMountainsCastro, Lina M. 1 ; Miranda, Marcelo 2 ; Fernandez, Bonifacio 11. Department of Hydraulic and EnvironmentalEngineering, Pontificia Universidad Catolica de Chile,Santiago de Chile, Chile2. Department of Forest Science, Pontificia UniversidadCatólica de Chile, Santiago De Chile, ChilePrecipitation is the most crucial variable in applicationof hydrological models because it provides most of themoisture input for hydrologic processes over land.Hydrological models require accurate rainfall data at thehighest possible resolution for streamflow predictions.Nevertheless, due to the high variability in space and time ofprecipitation it is necessary to have a dense rain gaugesnetwork to achieve high accuracy. The gauge network inChile is sparse or nonexistent, especially in the AndesMountains where the most Chilean rivers are born.Although the rain gauges have the advantage of a hightemporal resolution, the scarcity and difficulty of getting thedata in real time limit their application into hydrologicalmodels for simulation and forecasting in real time. This gapcan be solved using data from space-borne sensors. In recentyears, several satellite-based, near global, high-resolutionprecipitation estimates have become available withincreasing temporal and spatial resolution. Rainfallestimates from space-borne sensors offer a valuable source ofinformation for capturing the rainfall and for understandingof terrestrial rainfall behavior over some regions that areungauged like some parts of the Andes Mountains. Thepresent work aims to assess the rainfall estimations ofTropical Rainfall Measurement Mission (TRMM) MultisatellitePrecipitation Analysis (TMPA) in a region in theSouth of Chile over the Andes Mountains. The assessmentbetween TMPA product and gauge measurements was madeusing statistical error (Bias, Root Mean Square Error, andCorrelation Coefficient) and detection measurements (FalseAlarm Ratio - FAR and Frequency Bias Index - FBI). TheTMPA product represents 50% of total rainfall in almost allground truth stations, with the highest bias values in winterseason. When the TMPA estimates show high FAR and FBIvalues, it means that satellite has overestimated the numberof rain events with highest FAR and FBI values in the dryseason. Bias, RMSE, FAR and FBI show a spatial patternwhich increases with elevation because of the orographiceffect in the rainfall distribution and intermittentoccurrence. The temporal aggregation improves Bias, RMSEand Coefficient of Correlation values. For a monthly timescale the coefficient of correlation and bias reach values 0.95and 28% respectively. Improvements on a monthly scale mayarise from the TMPA processing algorithm which usesmonthly histograms to calibrate the satellite data. TMPAproducts like other ones (STAR or CMORH) have coarsespatial resolution (between 4km – 27 km) and represent asnapshot at a given time. The spatial and temporal44