Improvements to the accuracy of modelled soil water content from the Second Generation Prairie Agrometeorological Model
The direct measurement of soil water on a regional scale is often not practical due to large instrumental and labour requirements. Alternatively, soil water estimates can be derived using models. The Second Generation Prairie Agrometeorological Model (PAMII) models soil water, crop development and evapotranspiration (ET) in order to derive an estimate of crop water use. The objective of this study was to validate, and if necessary modify, the soil water component of PAMII using weather and soil water data collected from several spring wheat trials in Saskatchewan and Manitoba during the 2003 though 2006 growing seasons. Comparison of modelled and measured soil water values yielded a RMSE of 62 mm. For most site-years, PAMII overestimated soil water during the second half of the growing season, which was caused by an increase in modelled canopy resistance (rc) before the crop experienced water stress. The rc function was thus modified so that rc would not increase until the soil water content was < 0.5 of plant available water. Overall this modification reduced the RMSE from 62 to 56 mm. In addition, modelled soil water was underestimated during periods that experienced consecutive days of precipitation. This was because the model stopped infiltration when the top-zone reached saturation. When modified to allow infiltration to continue independent of the top-zone’s water content, the RMSE was further reduced to 53 mm. Overall, both modifications reduced the RMSE of modelled soil water by 9 mm, and this reduction was highly significant (P < 0.01). Key words: Prairie Agrometeorological Model (PAMII), soil water modelling, evapotranspiration, Canadian prairies