Research on Precipitation Prediction Model Based on Extreme Learning Machine Ensemble

Precipitation is a significant index to measure the degree of drought and flood in a region,which directly reflects the local natural changes and ecological environment.It is very important to grasp the change characteristics and law of precipitation accurately for effectively reducing disaster loss and maintaining the stable development of a social economy.In order to accurately predict precipitation,a new precipitation prediction model based on extreme learning machine ensemble(ELME)is proposed.The integrated model is based on the extreme learning machine(ELM)with different kernel functions and supporting parameters,and the submodel with the minimum root mean square error(RMSE)is found to fit the test data.Due to the complex mechanism and factors affecting precipitation change,the data have strong uncertainty and significant nonlinear variation characteristics.The mean generating function(MGF)is used to generate the continuation factor matrix,and the principal component analysis technique is employed to reduce the dimension of the continuation matrix,and the effective data features are extracted.Finally,the ELME prediction model is established by using the precipitation data of Liuzhou city from 1951 to 2021 in June,July and August,and a comparative experiment is carried out by using ELM,long-term and short-term memory neural network(LSTM)and back propagation neural network based on genetic algorithm(GA-BP).The experimental results show that the prediction accuracy of the proposed method is significantly higher than that of other models,and it has high stability and reliability,which provides a reliable method for precipitation prediction.

PREDICTION OF FLOOD SEASON PRECIPITATION IN SOUTHWEST CHINA BASED ON IMPROVED PSO-PLS

In order to achieve the best predictive effect of the Partial Least Squares(PLS) regression model, Particle Swarm Optimization(PSO) algorithm is applied to automatically filter the optimal subset of a set of candidate factors of PLS regression model in this study. An improved version of the Particle Swarm Optimization-Partial Least Squares(PSO-PLS) regression model is applied to the station data of precipitation in Southwest China during flood season.Using the PSO-PLS regression method, the prediction of flood season precipitation in Southwest China has been studied. By introducing the precipitation period series of the mean generating function(MGF) extension as an alternative factor, the MGF improved PSO-PLS regression model was also built up to improve the prediction results.Randomly selected 10%, 20%, 30% of the modeling samples were used as a test trial; random cross validation was conducted on the MGF improved PSO-PLS regression model. The results show that the accuracy of PSO-PLS regression model and the MGF improved PSO-PLS regression model are better than that of the traditional PLS regression model.The training results of the three prediction models with regard to the regional and single station precipitation are considerable, whereas the forecast results indicate that the PSO-PLS regression method and the MGF improved PSO-PLS regression method are much better than the traditional PLS regression method. The MGF improved PSO-PLS regression model has the best forecast performance on precipitation anomaly during the flood season in the southwest of China among three models. The average precipitation(PS score) of 36 stations is 74.7. With the increase of the number of modeling samples, the PS score remained stable. This shows that the PSO algorithm is objective and stable. The MGF improved PSO-PLS regression prediction model is also showed to have good prediction stability and ability.

Evaluating the Capabilities of Soil Enthalpy, Soil Moisture and Soil Temperature in Predicting Seasonal Precipitation

Soil enthalpy （H） contains the combined effects of both soil moisture （w） and soil temperature （T） in the land surface hydrothermal process. In this study, the sensitivities of H to w and T are investigated using the multi-linear regression method. Results indicate that T generally makes positive contributions to H, while w exhibits different （positive or negative） impacts due to soil ice effects. For example, w negatively contributes to H if soil contains more ice; however, after soil ice melts, w exerts positive contributions. In particular, due to lower w interannual variabilities in the deep soil layer （i.e., the fifth layer）, H is more sensitive to T than to w. Moreover, to compare the potential capabilities of H, w and T in precipitation （P） prediction, the Huanghe-Huaihe Basin （HHB） and Southeast China （SEC）, with similar sensitivities of H to w and T, are selected. Analyses show that, despite similar spatial distributions of H-P and T-P correlation coefficients, the former values are always higher than the latter ones. Furthermore, H provides the most effective signals for P prediction over HHB and SEC, i.e., a significant leading correlation between May H and early summer （June） P. In summary, H, which integrates the effects of T and w as an independent variable, has greater capabilities in monitoring land surface heating and improving seasonal P prediction relative to individual land surface factors （e.g., T and w）.

A Case Study of the Improvement of Soil Moisture Initialization in IAP-PSSCA

A prediction system is employed to investigate the potential use of a soil moisture initialization scheme in seasonal precipitation prediction through a case study of severe floods in 1998. The results show that driving the model with reasonable initial soil moisture distribution is helpful for precipitation prediction, and the initialization scheme is easy to use in operational prediction.

SVD Iteration Model and Its Use in Prediction of Summer Precipitation

A new short-term climatic prediction model based on the singular value decomposition （SVD） iteration was designed with solid mathematics and strict logical reasoning. Taking predictors into prediction model, using iteration computation, and substituting the last results into the next computation, we can acquire better results with improved precision. Precipitation prediction experiments were separately done for 16 stations in North China and 30 stations in the mid-lower catchment of the Yangtze River during 1991-2000. Their average mean square errors are 0.352 and 0.312, and the results are very stable. Mean square errors of 9 yr are less than 0.5 while only that of 1 yr is more than 0.5. The mean sign correlation coefficients between forecast and observed summer precipitation during 1991-2000 are 0.575 in North China and 0.623 in the mid-lower catchment of the Yangtze River. Librations of them in North China during the 10 years are small. Only in 1996 the sign correlation coefficient is below 0.5; the others are all over 0.5. But sign correlation coefficients in the mid-lower catchment of the Yangtze River vary obviously. The lowest is only 0.3 in 1992, and the highest is 0.9 in 1998, As the distribution of the forecast precipitation anomaly field in the summer 1998 of is examined, it is known that the model captured the positive and negative anomalyies of precipitation, and also well forecasted the anomaly distributions. But the errors are obvious in quantities between the forecast and the observed precipitation anomalies. Climate characteristics of large scale meteorological elements, such as summer precipitation have obvious differences in spatial distribution. We can forecast better if we divide a big region into many subregions according to the discrepancy of climatic characteristics in the region, and predict in each subregion. The research shows that the model of SVD iteration is a very effective forecast model and has a strongly applicable value.