Research On Rice Diseases And Pests Comprehensive Management System Based On GRASS

Rice is one of the most important crops in China. the economic losses caused by rice diseaseand pest annually is so astonishing every year, How to improve the comprehensive managementefficiency and forecast accuracy of rice disease and pest in rice production and droppingeconomic losses are very important. At present, the management of agricultural disease andpest in China is through labor paper record or database management which lack of spacemanagement and analysis ability. Some researchers had used advanced Geographic InformationSystem (GIS) to study rice disease and pest management systems. However, these systems werebased on GIS commercial softwares which had high cost and it was difficult to be used widely.In this paper, an open source GIS softerware——Geographic Resources Analysis SupportSystem (GRASS) and open source database were used in researching GIS of rice disease andpest. A rice disease and pest geographic information database was built. A comprehensivemanagement system of rice disease and pest was developed based on GRASS, which couldmanage the data of rice disease and pest and had the basic space analysis. The open sourcestatistics software R was used to establish negative binomial regression model and zero inflationnegative binomial regression models to forcast rice disease and pest. The main results are asfollows:(1) At present, few people use and research GRASS software. The Chinese references are alsovery limited. Through reading a large number of foreign references, this thesis introducedGRASS platform，function module, the data format and how to install and configurationthe software in detail. It provided a reference for some application researches based onGRASS.(2) A GIS database of rice disease and pest was built based on open-source databases(PostgreSQL and PostGIS), it could save and manage history data of rice disease and pest,meteorological data and map data. It improved the efficiency of data management, avoidedthe data loss and poor sharing of paper record.(3) In forcasting model of rice disease and pest, all the subsets were used to regression analysisfor screening meteorological factors. It avoided some shortcomings of stepwise regressionanalysis and other methods in factor screening process. Factor selection could eliminatesome factors which are irrelevant or low significant. It reduced computer burden, andimproved operation efficiency. (4) The occurrence data of rice disease and pest is a kind of count data.Two models of negativeNegative Binomial Regression Model (NBRM) and Zero Inflation Negative BinomialRegression Models (ZINBRM) were used to research the frocast of Cnaphalocrocismedinalis (Guenee). The result showed that NBRM and ZINBRM both are suitable to fitthe relation between occurrence data of Cnaphalocrocis medinalis and meteorologicalfactors and they could realize the accurate prediction of Cnaphalocrocis medinalis. Itprovided a new idea for other plant disease and pest by NBRM and ZINBRM.(5) A GIS system of rice disease and pest was developed based on GRASS. This system hadsome functions of data management, GIS spatial analysis, statistical analysis, GPS, dataquery and so on. This system could be extended to the management of disease and pest inthe other crops.