Forecasting Spring Wheat Yield Using Time Series Analysis

Techniques commonly used for wheat yield estimation employ weather data over the growing season. However, yield estimates are also required before w heat is sown - particularly by the grain exporting agencies to help them deter mine, in advance, wheat-export targets. In that case, time series techniques relying only on past y ield data can be used for yield forecasting. In this paper, a procedure for applying time series an alysis to forecast yield is described. A few relevant techniques (linear trend, quadratic trend, simple exponential smoothing, double exponential smoothing, simple moving averaging, and double moving averaging) are tested to model the average spring wheat yield series for Sa skatchewan, Canada. Using 1975-1993, 19751994, and 1975-1995 spring wheat yield data, yields were forecasted for 1994, 1995, and 1996, respectively. Based on a deterministic measure (i.e ., mean squared error), it was found that the quadratic model produced most accurate forecast during the model development periods (197593, 1975-94, and 1975-95) and model testing periods (1994, 1995, and 1996, respectively). Nonetheless, on the basis of stochastic measures (c oefficient of determination, R 2 , DurbanWatsun statistic, and coefficient of autocorrelatio n), the simple moving averaging technique was found to be the best during the model development p eriod. Further, a discussion is provided on improving the forecast by dividing the heterogeneou s cropping region of Saskatchewan into rather yield-based homogeneous regions using spatia l analysis tool of geographic information system software. Abbreviations: AR, autoregressive; ARMA, auto regressive moving average; ARIMA, auto regressive integrated moving average; MSE, mean squared error, GIS, geographic information system.