Sensitivity of the snowmelt runoff model to underestimates of remotely sensed snow covered area

Three methods for estimating snow covered area (SCA) from Terra MODIS data were used to derive conventional depletion curves for input to the Snowmelt Runoff Model (SRM). We compared the MOD10 binary and fractional snow cover products and a method for estimating sub-pixel snow cover using spectral mixture analysis (SMA). All three methods underestimated SCA and this contributed to underestimates in runoff modeled by SRM. The closest relationship between measured and computed runoff was achieved when SRM was run with conventional depletion curves derived from the MODIS binary snow cover product (R
²
= 0.91). Although the MODIS fractional snow cover product and SMA did not perform as well as the binary snow cover product (R
²
= 0.70 and R
²
= 0.72 respectively) we anticipate that either of these methods may be reworked to better account for forest cover in our study area and so improve SCA estimates.