Research And Application Of The Forecast Model Of Summer Precipitation In The Middle And Lower Reaches Of The Yangtze River Based On Data Mining Method

In order to provide the short-term climate prediction of summer precipitation in the middle and down-stream reaches of the Yangtze River, we use the National Climate Center National 160 stations in 1951-2014 years of monthly precipitation data, NCEP/NCAR from 1951 to 2014 atmospheric monthly mean reanalysis data, National Climate Center of 74 ring flow index and NOAA reorganize all sorts of index data, using correlation analysis, synthetic analysis, regression analysis and other methods of the change trend of summer precipitation in the middle and lower reaches of the Yangtze River, the effect of key factors on the precipitation in summer, use the data mining methods (C4.5 algorithm, CART algorithm) and early spring early winter and summer at the same time influence factor to summer rainfall prediction model is established. The main results are as follows:(1)In recent 63 years, the precipitation in the middle and lower reaches of the Yangtze River has been rising in the summer, and the precipitation in the middle and lower reaches of the Yangtze River has changed in 1979. In the middle and lower reaches of the Yangtze River during the summer rainy year, the most precipitation areas in Jiangxi, Anhui, Zhejiang provinces at the junction of the three provinces. Less rain in the middle and lower reaches of the Yangtze River in summer,the least precipitation area is mainly in the north of Yueyang, near Hunan and junction of Hubei province and Jiangxi province.(2)Using C4.5 algorithm in data mining and early spring influencing factors to establish the forecast model of summer precipitation in the middle and lower reaches of the Yangtze river. The accuracy of precipitation is 91.67% and 85.71% respectively. By using the CART algorithm and the early winter influence factors and the corresponding summer influence factors, the prediction model of summer precipitation was established. The establishment of early winter rainfall is too much and the factor is less model prediction accuracy was 70.59%,83.33%, the prediction accuracy of the two model established by the same period is 82.35% and 75% respectively.(3) Based on the methods of data mining in the middle and lower reaches of the Yangtze River in summer precipitation prediction model has higher prediction accuracy, model construction with synoptic significance, discriminant rules on the basis of big data theory, widely considered related factor, multiple diagnostic factor and a lot of index factor of permutation and combination, intelligent choice key factor, easy to objective, automation implementation, provides new ideas and methods for short-term climate prediction of flood season precipitation in the Yangtze River Basin.(4) During the spring of Pacific subtropical high intensity index, the northern hemisphere subtropical high intensity index strengthening, strengthening of the Western Pacific subtropical high in summer, center west, make for transporting the southwest warm and moisture flow in the Yangtze River downstream area, Baikal in southern anticyclonic promoted the cold air southward, for the middle and lower reaches of the Yangtze River region provides warm air and cold air, is conducive to the summer precipitation increased; and vice versa is not conducive to the formation of precipitation. It also shows that using summer precipitation prediction model is more feasible and factors to establish the physical meaning.