Study On The Regularity Of Drought And Water-logging And The BP Model Prediction In Changchun City

As an important component of the nature resource, climate is vital to human survival.Each change of the climate will have an effect on nature eco-system and social economysystem. The effect will be multi-scale and multi-level. Drought and water-logging is animportant task of climate research. It has theoretical and practical significance to manyaspects, such as agriculture and hazard prevention and so on. Based on the historical droughtand water-logging grade and the modern precipitation data of Changchun station, this papermakes an interpolation of the precipitation with the GM (1, 1). Then the paper analyzes theregularity of drought and water-logging from 1825 to 2000 with the help of wavelet analysisand Markov chain. Based on the data we have, this paper constructs a back-propagationmodel to predict the precipitation of Changchun city.Based on the data analysis, this paper makes a good use of the GM (1, 1) to interpolatethe precipitation data. The drought and water-logging analysis indicates that there are twophases from 1825a to 2000a, first water-logging and then drought. The transition year is 1918and the lasting time of drought is approximately equal to water-logging. The wavelet analysisshows that the water-logging period is 60a and the drought period is 50a. The drought periodcenter is near 1985a and the center wavelet index is +3 which indicates that the drought statewill be serious in the future. The Markov chain is adaptive to the prediction of the droughtand water-logging grade in Changchun city. After trial and error, the BP model is built topredict the precipitation. Though the artificial neural network has some weak points inprediction of precipitation, it will be a powerful research tool in the future with thedevelopment of artificial neural network (ANN).