Hydrologic model calibration using remotely-sensed data in Alaska

Watershed models in Alaska are critical for understanding snow- and glacier-dominated hydrologic processes in a changing climate. The highly diverse, frozen, and remote landscapes in Alaska present a host of new challenges for broad-scale hydrologic model development, including a notable absence of field-measured streamflow data. Without this commonly-used data for model calibration, alternative methods need to be developed in order to model hydrologic processes Alaska. Calibration methods that use remotely-sensed data in multi-objective, step-wise procedures were developed in this study. A calibration method using snow covered area (SCA) measured by NASA's Moderate Resolution Imaging Spectroradiometer (MODIS) was developed for a daily, deterministic, physically-based watershed model. The model selected for this study was the U.S. Geological Survey's Precipitation Runoff Modeling System (PRMS), and focused on a 62,500-sqkm test basin in southeastern Alaska from 2000 to 2014. Watershed model calibration to SCA ensures model processes adequately estimate transitional model states during snowmelt, a dominant hydrologic process in the basin. Gridded SCA data measuring fractional coverage were spatially aggregated to hydrologic response units within the basin. Model results were aggregated to eleven subbasins for calibration, comparison, and evaluation. Calibration of the subbasin watersheds to intermediate snowmelt process states, such as daily SCA, will likely also improve model estimates of solar radiation, potential evapotranspiration, annual water balance, and components of daily runoff. In Alaska where snow and glacier-fed systems are abundant, observed data are often scarce; therefore the development of calibration methods with remotely-sensed data are critical for improvement of watershed models which can then be used to estimate response to climate change. This study provides a method using remotely-sensed snow cover data to overcome field-measured data scarcity towards the development of a broad-scale PRMS model in Alaska.