Evaluation of WGEN for generating long term weather data for crop simulations

Abstract

We evaluated the ability of the WGEN model to generate long term weather series in situations where the actual historic weather data record is only 3–10 years long. The series generated were used to simulate yield of irrigated and rainfed chickpea. To do this, four 100-year samples of weather data were generated for Tabriz, Iran. The WGEN parameters used to generate data were obtained from daily actual weather data of 3 (W3), 5 (W5), 7 (W7), and 10 (W10) recent years. The actual and generated weather series were each used as input to a chickpea crop model under irrigated and rainfed conditions at three planting dates. Results showed that the generated data are very similar to the actual data used for parameter estimation for all base periods tested. In comparison of the generated data and the historic data the means and the distributions of weather data variables differed significantly. However, with increasing the number of years used for parameter estimation of WGEN from 3 to 10, percent of significant differences were 38, 26, 17 and 13% for W3–W10, respectively. When generated weather data were evaluated as input to a chickpea crop model, simulated yields obtained using generated data were significantly different from that obtained using actual data in 50 and 8% of cases under irrigated and rainfed conditions, respectively. To generate data similar to long term historic data, a longer base period (>10 years) would be required for parameter estimation. However, when it is required that the generated data represent recent history rather than a long term period, the WGEN can be used as a reliable source of weather data even if it’s required parameters are obtained from only 3–10 years of actual historical weather data.