Daily Data Assimilation of a Hydrologic Model Using the Ensemble Kalman Filter

TL;DR

This paper develops a data assimilation framework using the Ensemble Kalman Filter to improve hydrologic model runoff forecasts by integrating Snow Water Equivalent and runoff measurements, showing significant improvements in wet years.

Contribution

It introduces a novel joint state-parameter assimilation approach for daily hydrologic modeling, enhancing runoff prediction accuracy in mountainous regions.

Findings

Assimilating SWE daily updates the modeled SWE but does not improve runoff output.

Augmenting state with parameters and runoff data significantly reduces runoff RMSE, up to 60%.

The method shows consistent improvements across multiple years, especially in wet conditions.

Abstract

Accurate runoff forecasting is crucial for reservoir operators as it allows optimized water management, flood control and hydropower generation. Land surface models in mountainous regions depend on climatic inputs such as precipitation, temperature and solar radiation to model the water and energy dynamics and produce runoff as output. With the rapid development of cheap electronics applied in various systems, such as Wireless Sensor Networks (WSNs), satellite and airborne technologies, the prospect of practically measuring spatial Snow Water Equivalent in a dense temporal scale is increasing. We present a framework for updating the Precipitation Runoff Modeling System (PRMS) with Snow Water Equivalent (SWE) maps and runoff measurements on a daily timescale based on the Ensemble Kalman Filter (ENKF). Results show that by assimilating SWE daily, the modeled SWE gets updated accordingly, however no improvement is observed at the runoff model output. Instead, a deterioration consistently occurs. Augmenting the state space with model parameters and runoff model output allows for filter update with previous day measured runoff using the joint state-parameter method, and showed a considerable improvement in the daily runoff output of up to 60% reduction in RMSE for the wet water year 2011 relative to the no assimilation scenario, and improvement of up to 28% compared to a naive autoregressive AR(1) filter. Additional simulation years showed consistent improvement compared to no assimilation, but varied relative to the previous day autoregressive forecast during the dry year 2014.