Agriculture · 2016 · 154 citations · 42 references
Natural SciencesEarth SciencesMeteorologyClimatologyForecastingTime Series AnalysisHydrometeorological ForecastingPrecipitationPrecipitation EstimationPrecipitation ForecastingTime Series ForecastingTime Series ModelingMonthly ForecastingRain Rate EstimationRainfall Time Series AnalysisSemi-arid ClimateStatistical Time Series Models
This study investigates how the length of recorded data affects monthly rainfall forecasting accuracy and identifies the most critical rainfall month for agricultural scheduling across climate zones. Monthly rainfall records spanning 5, 10, and 49 years from four Iranian stations were used to calibrate time‑series models, whose forecasts were then evaluated against the following year’s data. In temperate and semi‑arid climates 60 years of observations suffice for next‑year rainfall prediction, while in arid and humid climates accuracy improves with more data; overall, time‑series models are suitable for semi‑arid monthly rainfall forecasting.
This paper reports the study of the effect of the length of the recorded data used for monthly rainfall forecasting. Monthly rainfall data for three periods of 5, 10, and 49 years were collected from Kermanshah, Mashhad, Ahvaz, and Babolsar stations and used for calibration time series models. Then, the accuracy of the forecasting models was investigated by the following year’s data. The following was concluded: In temperate and semi-arid climates, 60 observation data is sufficient for the following year’s rainfall forecasting. The accuracy of the time series models increased with increasing amounts of observation data of arid and humid climates. Time series models are appropriate tools for forecasting monthly rainfall forecasting in semi-arid climates. Determining the most critical rainfall month in each climate condition for agriculture schedules is a recommended aim for future studies.
42