Combination Forecast Model Of NDVI Based On Support Vector Machine Regression

The change of the vegetation growth is affected by the climate and prompts the change of climate.Simulating and predicting the change of the vegetation coverage dynamically is conducive to carry out ecological construction work.Remote sensing observation provides reliable real-time data sources for the research of the vegetation coverage change of surface.Normalized Difference Vegetation Index(NDVI)is sensitive to green vegetation.It can Reflect the growth change of vegetation effectively.In this article NDVI is chosed as the research object,to predict the growth change of vegetation.Now the research of predicting vegetation coverage is to predict the change direction in the future.The main models include Markov Model,Generalized Additive Model,etc.Among them Markov Model is used more widely.The vegetation coverage change predicting does not yet exist a short-term quantitative prediction model.In this paper,the MODIS NDVI data between 2004 and 2015 are used to make a time sequence.We make use of the temperature and precipitation data during this period as regression factor.Support vector machine regression model is adopted to establish the NDVI short-term prediction model.The process of vegetation growth response to the change of climate factor is complex.The single forecast model is very difficult to predict the vegetation coverage change accurately.For the complexity of response process,a single model is built from different angles.First,grid search method,genetic algorithm and particle swarm optimization are used to optimize model parameters respectively.Then the best parameters are used to train support vector machine respectively.The results show that the grid search method is the best parameter optimization algorithm.We build two single prediction models of NDVI from different angle using support vector machine regression model based on the grid search method.A linear combination with the two single prediction model is made and the optimal weight coefficient is calculated.The results show that the combined model can predict NDVI effectively.