The study of vegetation biomass inversion based on the HJ satellite data in Yellow River wetland
基于环境卫星数据的黄河湿地植被生物量反演研究

Wetland vegetation is the important component of the wetland ecological system, and its biomass is one of the major indexes to measure the primary productivity. Remote sensing technology can be used effectively to extract information of wetland vegetation, which directly reflects their growth and development. Hence the research is very significant for the environment monitoring and protection of wetland ecosystems. This paper discusses the quantitative relationship between vegetation spectrum index and dry biomass of vegetation, based on the remote sensing data from China HJ-1A satellite and synchronous field sampling data, choosing the Zhengzhou Yellow River Wetland Nature Reserve as the study area. The methods based on vegetation indexes in this study are mainly used in vegetation biomass inversion in a large area, and have conducted quite successful algorithms and models. While a single vegetation index as input factor has a high accuracy and sensitivity in fitting the vegetation biomass model, its prediction result encounter larger errors and cannot accurately reflect true information of vegetation biomass. Authors in this paper puts variety of vegetation indexes together as the comprehensive input factors, and estimates vegetation biomass in wetland by using regression analysis method. Therefore, it improves largely the precision and reliability of vegetation biomass inversion based on multiple vegetation indexes. In this paper, the biomass of vegetation is abstracted as a phenomenon, while different kinds of vegetation indexes are considered as different influencing factors. The relationships between them are conceptualized into different mathematical models. Authors inspect the precision of inversion by comparing the estimate results from the two kinds of methods after their using the SCRM (single curve regression model) biomass method and MLRM (multiple linear regression model) biomass method respectively. Results showed that the MLRM have good precision and prediction ability, which can be well applied to estimate the wetland vegetation. The model test is significant (hitting a level of 0.000), while the model correlation coefficient is 0.9791, fitting accuracy reaches 29.8 g/m2. The model prediction results has a system error of 49.9 g/m2, the determine coefficient is 0.8742, the RMS error is 67.2 g/m2, estimation of dry vegetation biomass total to 6.849199 t/hm2 in study area, with the actual biomass estimated dry biomass has a difference of 0.323 749 t/hm2, the result is 6.525 450 t/hm2, and the relative error is 4.73%, which shows an accurate estimation on the biomass of wetland vegetation by using vegetation indexes extracted from HJ-1A remote sensing data, combined with field sampling data. Wetland is one of the three large ecological systems globally, which takes a key role on the plant biodiversity protection. And it has a great environmental function and lots ecological benefits as the most productive ecological system. Method o