New Quality Assurance Technique for Meteorological Stations Based on A Physically-Based Hydrological Model

Many groups depend on climate data from federal and state supported weather station networks such as the NRCS’s SNOTEL, and NOAA’s Climate Reference Network. While data from these stations are typically of high quality, there is no statistical parameter available to the public that quantifies the bias and the magnitude of errors in a time series. A time series error analyses could possibly detect subtle data anomalies, and problems with sensors, while providing a general level of confidence in the data for the user. HYDRUS-1D was used to construct a physically-based flow model for water and thermal fluxes for a SNOTEL site in Oregon from June 1, 2008 0:00 to October 31, 2011 23:00. Soil moisture and temperature values were predicted using the model and then compared to soil data measured with Hydra Probe soil sensors placed at 5, 10, 20, 50 and 100 cm. Simulations on hourly data sets typically yield RMSE from about 0.01 to 0.05 and 1 to 3 degrees C for soil moisture and temperature respectively. Model efficiencies decrease and RMSE increase for soil moisture with the presence of the winter snow pack when using degree day based model for snow water equivalent.