Evaluating the seasonal and interannual variations in water balance in northern Wisconsin using a land surface model

We evaluated the performance of the Integrated BIosphere Simulator ( IBIS) land surface model in the temperate forests of northern Wisconsin ( 46 degrees N, 89 degrees W) to determine whether model formulations, driven with daily historical precipitation, temperature, relative humidity, solar radiation, and wind speed data, were capable of simulating water flow and storage within a seasonally cold climate regime. We focused concurrently on understanding seasonal and interannual variations of both the water fluxes to the atmosphere and water partitioned into surface runoff and groundwater infiltration, with special attention to the transitions from cold-dominated ( snow, ice) to warm-dominated ( rain, liquid soil moisture) hydrology. Results showed when compared with a suite of field observations IBIS simulated water and energy cycling at daily to interannual timescales with reasonable accuracy. Because of errors associated with field observations, the accuracy with which we simulated each component of the water balance is not easily quantified. By investigating the complete land surface water balance, however, we increased the likelihood that all components were being captured. The modeled monthly energy balance, annual water balance, and drainage rates were generally within 5 - 15% of the observed values. Modeled and observed soil temperatures generally differed by less than 3 degrees C and had r(2) values that were greater than 0.9. Soil moisture values were within 5 - 20%, and freeze and thaw timing was within a few days of observations. Modeled snow dynamics captured the observed snow arrival and departure ( accumulation on the surface) within a few days of observations, but overestimated the average maximum depth by 86%. Because model formulations were subjected to varying soil conditions and water phases, this evaluation exercise enhanced our understanding of northern Wisconsin's water balance and increased model credibility for applications in seasonally cold climates.