关键词: 西北地区, 遥感估产, 专家系统, 人工神经网络

Abstract: With social developing,science technical improving and theory maturing,it's very important that crop estimation by remote sensing from fragment module to aggregation system.This paper introduces the basic contents and basic structure of the main crop estimation information system by remote sensing(MCEISRS) for Northwest China;describes the MCEISRS sub module function of the Resources and Environment(background data).System,Database System,GIS(Geographical Information System) spatial analysis,Model Base System,Expert System,Artificial Neural Network,and Data Transfer with INTERNET;and discusses how to improve crop estimation by remote sensing accuracy in the future.Ⅰ.Resources and environment system includes county land use data(scale 1:100 000),soil type and distribution scope,the TM interpreting keys,plant regularity and crop climate-temperate characters,districts for crop estimation,etc.Expert system can obtain acknowledge from it.Ⅱ.Database system includes all kinds database for crop estimation model produced,such as rainfall,temperature,sun days,NDVI,etc.This is an opening system.User can add data,modify data,and delete data if he(she) gets permission.Ⅲ.GIS(Geographical Information System) Spatial Analysis Module offers the crop planting area for the model.And it also can get the spatial analysis model from Model Base System.Then it feedback the data to the crop estimation model directly.Ⅳ.Model Base System introduces model type,representation methods and ways;describes the function of model base management system.Ⅴ.Expert System and Artificial Neural Network.This part introduces their concepts,basic structure and characters about ES ANN.It can deal with the non-structure problems with ES ANN,such as crop type identifying.Ⅵ.Dates Transfer with INTERNET.It must be sent out in time when you get the last data.The best way is INTERNET.You can visit the remote computer by INTERNET and use data in it.The remote user can also visit MCEISRS by INTERNET.Of course the user must have some competence,if not it will be refused.This function can protect the data from being not used by the illegal user.

Key words: Northwest China, crop yield estimation by remote sensing, expert system, artificial neural network