63 rd EASTERN SNOW CONFERENCE Newark, Delaware USA 2006 153 Combination of Active and Passive Microwave to Estimate Snowpack Properties in Great Lakes Area AMIR E. AZAR 1 , TARENDRA LAKHANKAR 1 , NARGES SHAHROUDI 1 , AND REZA KHANBILVARDI 1 ABSTRACT In this research we examine active and passive microwave to snow water equivalent (SWE) and to investigate the potential of combining active and passive microwaves to improve the estimation of SWE. The study area is located in Great Lakes area between the latitudes of 41N–49N and the longitudes of 87W–98W. Passive microwave are obtained from DMSP SSM/I sensors provided by NSIDC. Active microwave were obtained from different sensors: 1) RADARSAT C-Band SAR. 2) QuikSCAT Ku-band (13.4GHz) for both vertical and horizontal polarizations. The ground truth data was obtained from SNODAS data set produced by NOHRSC. An Artificial Neural Network model was defined to model various combinations of inputs to SWE. The results indicate that none of the active microwave channels produce satisfactory results. However, when combined with passive microwave, they improve the estimated SWE. INTRODUCTION Microwave remote sensing techniques have been effective for monitoring snowpack parameters (snow extend, depth, water equivalent, wet/dry state). Snow parameters are extremely important for input to hydrological models for understanding changes in climate due to global warming. Snow parameters been investigated by numerous researchers using many sensors such as SMMR and SSMI for passive microwave and SAR and QSCAT for active microwave. Space-borne microwave sensors can monitor characteristics of seasonal snow cover at high latitudes regardless of lighting conditions, time of the day, and vegetation. In passive microwave radiometer, microwave energy emitted from the ground surface is transmitted through the snow layer into the atmosphere and recorded by the sensor. Snow parameters can be extracted from remote sensing data by empirical algorithms. Hallikainen (1984) introduced his algorithm for estimating SWE using passive microwave SMMR data. The process involved the subtraction vertical polarizations of 18 and 37 GHz frequencies. The subtracted value, dT, was used to define linear relationships between dT and SWE. Chang et al. (1987) proposed using the difference between the horizontally polarized channels SMMR 37 GHz and 18 GHz to derive snow depth – brightness temperature relationship for a uniform snow field (Chang et al 1987). Goodison and Walker (1995) introduced the most widely used algorithm for North America. The algorithm was originally for Canadian prairies. It defines a linear relationship between GTV ([37V–19V]/18) and SWE. They also suggested using 37H and 37H polarization differences for identifying wet snow. Derksen et al. (2004) developed a new algorithm which derives SWE for open environments, deciduous, coniferous, and spars forest cover [SWE = F D SWE D + F C SWE C + F S SWE S + F O SWE O ]. The algorithm represents an improvement, however 1 NOAA-CREST, City University of NY, 137thst & Convent Ave. New York, NY.