Quantitative Estimate Of Forest Biomass Under NDVI Covariance Inversion

To estimate the forest biomass, the105plots were selected for individual measurement, which were located in Shougang Pinewood Park of Beijing and acquired in July to Aug,2010. After combination with simultaneous SPOT5RS images, the data were used for modeling. With applying the Geostatistical Analyst model in ArcGIS9.3and SPSS18.0. Using the RS data or not, kriging prediction model was applied to analyze the above-biomass. Although two methods also carried out the prediction value and standard error, the Kriging covariance model (KCM) using remote sensing data was a better remote sensing inversion model for forest biomass, which improving prediction accuracy because of synergy between KCM and NDVI. The later method was more rapid, effective, comprehensive and precise for forest biomass prediction. The results showed:in this study field, the biomass was96.6mg/hm2which standard error was below1.426711. Innovation:(1) The research established more accurate forest biomass estimation model, which combined forest management, forestry ecology and remote sensing technology. Covariance Kriging model was more accuracy and practical than traditional methods. With identifying the relationship between VDV1and forest biomass, the method confirmed that the collaborative variation function existed between the field data (forest biomass) and NDVI. So, it was possible to use this method for forest biomass prediction. With this reliable model, the forest biomass distribution map can be draw. Above-mentioned method was used to estimate biomass by best linear minimum bias estimator, which may effectively reduce evaluated error. Although neither of two methods achieve implemented requirement, the research idea was novel.(2) This method was the base for special sampling and predicting the forest biomass. The variation of NDVI was analyzed in the study area. The spatial statistics sampling was better than random sampling, systematic sampling methods and comprehensive sampling especially. The method has its own unique insights in this study area.(3) By making full use of the features of abundant and easy to measure, remote sensing data improved the availability prediction, precision and effectiveness of forest biomass. This is very important for forest, ecology and environment science. Further research of using remote sensing to estimate biomass and statistical method has application and reference significance.