Search

Precipitation prediction is central in hydrology and water resources planning and management. This paper introduces a semi-empirical predictive model to predict monthly precipitation and compares its predictive skill with those of machine learning (ML) methods. The stochastic method presented herein estimates monthly precipitation with one-step-ahead prediction properties. The ML predictive skill of the algorithms is evaluated by predicting monthly precipitation relying on the statistical association between precipitation and environmental and topographic factors. The semi-empirical predictive model features non-negative matrix factorization (NMF) for investigating the influence of multiple predictor variables on precipitation. The semi-empirical predictive model's parameters are optimized with the hybrid genetic algorithm (GA) and Levenberg-Marquardt algorithm (LM), or GALMA, yielding a validated model with high predictive skill. The methodologies are illustrated with data from Hubei Province, China, which comprise 27 meteorological station datasets from 1988–2017. The empirical results provide valuable insights for developing semi-empirical rainfall prediction models. [ABSTRACT FROM AUTHOR]

Hubei Province Region (HPR), located in Central China, is a concentrated area of severe convective weather. Three severe convective processes occurred in HPR were selected, namely 14–15 May 2015 (Case 1), 6–7 July 2013 (Case 2), and 11–12 September 2014 (Case 3). In order to investigate the differences between the three cases, the temporal and spatial distribution characteristics of cloud–ground lightning (CG) flashes and precipitation, the distribution of radar parameters, and the evolution of cloud environment characteristics (including water vapor (VD), liquid water content (LWC), relative humidity (RH), and temperature) were compared and analyzed by using the data of lightning locator, S-band Doppler radar, ground-based microwave radiometer (MWR), and automatic weather stations (AWS) in this study. The results showed that 80% of the CG flashes had an inverse correlation with the spatial distribution of heavy rainfall, 28.6% of positive CG (+CG) flashes occurred at the center of precipitation (>30 mm), and the percentage was higher than that of negative CG (−CG) flashes (13%). Moreover, the quantity of thunderstorm cells in Case 1 was more than other cases, the peak time of +CG flashes was prior to that of total CG flashes in Case 2 and Case 3, and the time of +CG flashes' peak in Case 2 was prior to that of precipitation at about 2 h. Based on the analysis of the cloud environment, there are three main reasons for the differences of CG flashes and precipitation. Firstly, the structure of the LWC vertical profile and the height of the LWC peak are different, and high LWC makes it difficult for the collision of ice particles to generate electricity. Secondly, the differences between convective available potential energy (CAPE), precipitation, and CG flashes is caused by the sudden increase of VD from 1.5 km to 3 km, and thirdly, the production of CG flashes is very sensitive to RH at the surface layer and the total CG flashes increase as the RH increasing. [ABSTRACT FROM AUTHOR]