Combination of differentiated prediction approach and interval analysis for the prediction of weather variables under uncertainty

In this paper, a differentiated prediction model (DPM) was combined with an interval analysis approach for the prediction of weather variables under uncertainty. The DPM was used for general trend prediction, and interval analysis was used for reflecting seasonal variations and residual terms. A case study of prediction for monthly average temperature and precipitation in Wuhan, China, was provided based on 22 years of observation data. The results indicated that uncertainties existing in weather-related processes could be effectively reflected through this hybrid approach. The predicted intervals for temperature and precipitation appear to contain most of the relevant observed values. (C) 1997 Academic Press Limited