Theoretical Study Of Area Estimation In CFI Based On Remote Sensing Large Plot

Continuous Forest Inventory(CFI)is not only the main way to collect forest information and master the dynamics change of forest in China,but also the most important part of forest monitoring system.In recent years,based on RS and GIS the government put forward strategies in order to build a new forest monitoring system,which brings considerable timeliness and comprehensiveness.This paper is based on the study on annual forest area monitoring methods using theoretical derivation method.In this paper the methods proposed are based on three information sources: Covering the entire computerized remote sensing classification results,interpretation large plot data by systematic sampling and RS ground test plot in interpretation large plot.The principle of estimation consists of deeply mining information from different information sources;maximize the estimation efficiency.The main idea is to establish connection between RS classification data and interpretation large plot data,deeply mine the information of RS classification data.To establish the connection between interpretation large plot data and RS ground test plot data,and deeply mining information of interpretation large plot data.Thus,two categories methods are proposed to the annual forest area monitoring.The first one category is based on single land category,that is to say,estimate only one land category in a single time.The estimation method consists of regression relationship between different kinds of data,and perhaps there are some errors which should be modified.There are three methods in the first category.The second one category is based on probability transfer matrix.This kind of method may estimate all the land categories at the same time by the way of building probability transfer matrix between different kinds of data.Because of all the land categories are estimating in the same time,it is cooperative to area results.And the second category also has three methods.In this research,continuous variable is used to measure plot area,rather than 0-1 variable which is used in the traditional CFI.In large plot survey,the continuous variable is used naturally,and to explain the superiority of continuous variable there is a computer sampling simulation about continuous variable and discrete variable,the theoretical derivation is given at the same time.The result shows that continuous variable is unbiased and has high efficiency.One of the discrete variable method based on one point in the plot is unbiased too,but its efficient is lower.Besides,the method based on dominant land category is biased,in some situation it seriously biased.It proved that continuous variable used in large plot survey is theoretically reasonable.Due to the restrictions of the conditions,this research has not been tested by the real production data.All the methods proposed in this research are all theoretical research,the feasibility of these methods are not tested by real data.Perhaps there are some problems in these methods,and the further studies are needed.