Validation of a simple steady-state forecast of minimum nocturnal temperatures

Abstract

A two-layer steady-state resistance model is compared with routine meteorological data collected from the Western Australian wheat belt during 2000-06. Major difficulties in implementing such a model are the correct parameterization for the incoming longwave radiation and estimation of daily soil moisture, neither of which are routinely measured. These difficulties are addressed by testing parameterizations for incoming longwave radiation calibrated to local conditions and incorporating a soil-water balance model based on routine weather data. The modified model has RMSE and biases ranging from 2.4° to 3.1°C and -0.2° to 0.8°C, respectively, across the wheat belt when comparing all minimum nocturnal temperatures. The model is shown to predict frost events approximately 55% of the time and illustrates that frost damage to foliage may occur when screen temperatures are < 2°C. © 2009 American Meteorological Society.