Abstract:

Based on precipitation data from observations and predictions from the ECMWF (European Centre for Medium-Range Weather Forecasts) model in Fujian and surrounding areas from April to June in 2014-2018, the main rain belts from the observations and all of the rain belts from predictions were determined using an object identification method, in which a rainfall threshold, an appropriate characteristic spatial scale, and a close degree were set.Through pairing objects with the observations, the threshold of matching degree was selected to determine the predicted main rain belts.The SAL (Structure, Amplitude, and Location) spatial verification method was improved and accounted for the biases associated with the intensity, position, and morphology of the predicting main rain belts.The results show that, compared with the observations, the maximum amount in precipitation for the main rain belts prediction by the model is smaller, but the threshold one is larger.The position of main rain belts lies more westward and southward in April, but more westward and northward in May and June.The position of the main rain belts predictions by the model lags behind the observations and tends to be located in the upstream of the moving direction of the observations.The main rain belts have a stripe shape.The morphology of the predicted main rain belts is narrower and longer, and this tendency intensifies with the forecast lead time.The predicted axial angle of main rain belts is slightly smaller and tends to oriented north-eastwards.This work provides guidance for the bias correction of the main rain belts forecasting by the model.

Key words: Rain belt property, Object identification, Adjacent degree threshold