Rainfall Monitoring System Based On Satellite Remote Sensing Image

Frequent occurrences of disaster weather in china produce huge economic losses to the national economy and social life. Especially, rainstorm has become a major weather disaster. With the rapid development of the meteorological satellite technology and the wide application of meteorological operations, the use of satellite remote sensing data to monitor precipitation has became an important aspect of the application and developed rapidly, and also become an important means of monitoring rainfall. Using meteorological satellite remote sensing data has great significance for non-linearly extrapolating of cloud cluster moving forecasts and rainfall estimates, forecasting the short-term weather, monitoring the flood disaster,making early warning of geological disasters and improving the accuracy of the weather forecast.In the current forecast work, cloud cluster forecast and precipitation estimation methods they use are the linear extrapolation methods.They can produce error. In this paper, nonlinear genetic BP neural network optimization algorithm is used for the estimation of regional cloud cluster movement forecast and precipitation. In the forecasts of regional cluster movement, I adopt numerical forecast products information to create a predictive input model of cloud image,use the empirical functions orthogonal decomposition method to do the EOF expansion of the cloud map image, and take the one-dimensional time coefficient series as the output model of cloud forecasts.In the precipitation estimates, I adopt the method of Delaunay triangulation to do the grid interpolation for the output factor and it improves the prediction accuracy of the model. Through this study, the error caused by the subjective judgment and qualitative extrapolation of the forecasters about the satellite images is avoided, and I will give the objectivity and quantitative forecasts.The main contents include the following aspects in this paper:1. Real-time processing of satellite remote sensing data.I Handle the satellite data for positioning and do the relative coordinate system transformation. By calculating the image of Latitude, longitude position to achieve the position of the image.The higher accuracy of location calculation is the bedding of accuracy of the estimates of cloud cluster forecasts and regional precipitation.2. Propose a BP neural network model (GA-BP) algorithm thought based on the genetic algorithm. According to Genetic Algorithm, BP neural network algorithm and their complementary, I design (GA-BP) Genetic BP neural network algorithm, and the algorithm is applied to the model of regional monitoring of the cloud cluster forecast and precipitation estimates. Theory explains the feasibility and reliability of the algorithm.3. Design regional cloud cluster movement prediction model. In this paper, we consider the combined effects of the cloud caused by the Environment Element in the process of evolution, and adopt numerical forecast products information to create a predictive input model of cloud image.Also I use the method of the experience function orthogonal decomposition to achieve the EOF expansion of cloud map image, and take one-dimensional time sequence of coefficients as the cloud images forecasts output model. The experiments show that using the above method, the correlation coefficient of the predicted cloud cluster and the actual cloud cluster can reach85%, and the above method can better reflect the nonlinear changes in the characteristics of cloud state.4. Precipitation nonlinear estimation model is designed. In order to solve the problem of automatic weather station he automatic weather station data scattered in space and incompleteness,so in the output model,I use Delaunay triangulation method for grid interpolation.I use the combination of the correlation of the satellite brightness temperature and precipitation, global search ability of the BP neural network algorithm and the guiding search ideas of genetic algorithm to establish a BP neural network model (GA-BP)which based on genetic optimization algorithm. The algorithm improves the accuracy of forecast model in the process of estimating regional rainfall.The experiments show that the relative error of the model15%.It greatly improves the accuracy and stability of forecasts.5. The design and implementation of the precipitation monitoring system. I adopt the precipitation data of actual ground stations to analyse of the entire system for the precipitation test.