The predictability of autumn soil moisture levels on the Canadian prairies

This paper examines time and space patterns of autumn soil moisture for the south-eastern Canadian Prairie provinces and looks for potential teleconne ctions between these patterns and remote forcing. A unique 35-year dataset of annually-measured (in the autumn) soil moisture at 35 sites was subjecte d to principal component analysis. The dataset comprised gravimetric soil w ater contents at depths of 15, 30, 46 and 76 cm. For the composite of all s oil depths, the first three principal components explained 34, 10 and 8% re spectively, of the total variance. The first component, reflecting moisture changes over the entire study area, is the only one likely to be related t o large-scale remote forcing and to be potentially predictable. The remaini ng components, related to moisture gradients within the study area, are pro bably either random fluctuations or responses to smaller-scale, less-remote forcing, as well as being less amenable to prediction. Analysis of the tim e-series amplitudes of the first mode showed significant correlations with several teleconnection indices; these were the North Pacific sea surface te mperature (NPSST), the East Pacific pattern, and several indices of Arctic temperature anomalies. The highest correlations were found with the NPSST i ndex, with early summer NPSST anomalies appearing mainly to influence autum n soil moisture, particularly the deeper layers. Although the magnitudes of the correlations were low, the results of this study contribute towards a better understanding of soil moisture variations and their potential predic tors on the eastern Canadian Prairies. This may be useful for predictions o f the succeeding year's crop and forage yields, as well as spring runoff an d summer streamflow. Copyright (C) 1999 Royal Meteorological Society.