Infusion of gap-filling radar, snow data assimilation and airborne lidar for improved water supply forecasting in the inter-mountain western U.S

Large gaps in observations of precipitation and snowpack conditions in remote mountain watersheds can result in strong biases in total snowmelt-driven runoff amount, and, equally important for water management, the timing of runoff. The southern Rocky Mountains in the western U.S. suffer from such observational deficiencies leaving water supply forecasters and water managers alike with a lack of situational awareness on the state of snowpack conditions and on how winter and spring time precipitation are distributed in time, space and precipitation phase. These deficiencies are hypothesized to have a significant adverse impacts on estimates of snowpack melt-out rates and on water supply forecasts. We present a series of findings from a coordinated observational-data assimiliation-modeling study from the Upper Rio Grande River basin whose aim was to quanitfy the impact enhanced precipitation, meteorological and snowpack measurements on the simulation and prediction of snowmelt driven streamflow. Measurements from a gap-filling, polarimetric radar (NOXP) and multiple in-situ meteorological and snowpack measurement stations were assimilated into the WRF-Hydro modeling framework to provide continuous analyses of snowpack and streamflow conditions. Model performance from simulations and hindcasts using enhanced observational analyses were compared against uncalibrated WRF-Hydro model results. Airborne lidar estimates of snowpack conditions from the NASA Airborne Snow Observatory during mid-April and mid-May were also used as additional independent initialization and validation datasets for various model experiments. Our results suggest that correcting the operationally-available forcing data with the experimental observations led to improvements in the seasonal accumulation and ablation of mountain snowpack and ultimately led to improvement in model simulated streamflow as compared with streamflow observations. While late season preciptiation during April and May clearly influenced the skill of the initial April 1 seasonal water supply forecasts, it is evident that the enhanced observations provided significant value in monitoring the rate and timing of melt-out and in latter water supply hindcasts made in May and June.