Forest Carbon Sink Metering Algorithm Research Based On BP Neural Network

With the introduction and signing of the Kyoto Protocol, More and more countries started to increase the intensity of forest carbon sinks research, the proportion of economic development of carbon sinks in the economic market is also gradually expanding, the carbon sinks trade bring economic benefits, also changing environment, and create the corresponding environmental benefits for mankind. The formation of forest carbon stocks is the result of the role of forest photosynthesis and respiration, But also affected and constrained by the surrounding environmental factors, so, the form of forest carbon sinks is complex, there are many carbon sinks measurement algorithm, but not yet formed a unified standard, this is also the important and difficult for carbon sink metering algorithm research.This article based on the current domestic and international research status for carbon sink estimation method research, in-depth study the basic theory of several commonly used algorithms, analysis the pros and cons of various algorithms,cope of application and measurement results, summarizes several important factors which affect carbon stocks, and make them as the key factors to calculate CO2flux, learn from the idea of relaxation eddy accumulation method and chamber method, by the introduction of the knowledge of the neural network, establish the BP neural network model. Selected the research data of tropical rainforest of Xishuangbanna as the sample of the network model, due to the limited sample size, in order to make the model better reflect mapping relationship between the input and output, model training by grouping, and respectively cross validation of the model, finally get the ideal network model. The samples simulation training and validation test results show that, carbon sequestration estimates BP neural network model training times is reasonable, training error meets the accuracy requirement, and has good generalization performance, the testing results and the actual measured value are basically consistent, so, BP neural network model in the study of forest carbon sink metering algorithm has played an important role in high efficiency, low cost, small error, realize automation and intelligent of the carbon sinks measurement, this has far-reaching affect and historical significance on the development of Forest carbon sinks and sinks economic.