Preliminary Research on Grassland Fine-classification Based on MODIS

35th International Symposium on Remote Sensing of Environment

地球科学;生态环境科学

Hu, Z.W.^1,2,3 ; Zhang, S.^1,2,3 ; Yu, X.Y.^1,2,3 ; Wang, X.S.^1,2,3

College of Resources Environment and Tourism, Capital Normal University, Beijing 100048, China^1

Key Lab of Resources Environment and GIS, Beijing 100048, China^2

Key Lab of Integrated Disaster Assessment and Risk Governance, Ministry of Civil Affairs, Beijing 100048, China^3

关键词: Classification accuracy;    Grassland classifications;    Grassland ecosystems;    Grassland resource;    Grassland types;    Monitoring methods;    Spectral feature;    Vegetation index;   

Others  :  https://iopscience.iop.org/article/10.1088/1755-1315/17/1/012079/pdf
DOI  :  10.1088/1755-1315/17/1/012079

学科分类：环境科学（综合）

来源: IOP

【 摘 要 】

Grassland ecosystem is important for climatic regulation, maintaining the soil and water. Research on the grassland monitoring method could provide effective reference for grassland resource investigation. In this study, we used the vegetation index method for grassland classification. There are several types of climate in China. Therefore, we need to use China's Main Climate Zone Maps and divide the study region into four climate zones. Based on grassland classification system of the first nation-wide grass resource survey in China, we established a new grassland classification system which is only suitable for this research. We used MODIS images as the basic data resources, and use the expert classifier method to perform grassland classification. Based on the 1:1,000,000 Grassland Resource Map of China, we obtained the basic distribution of all the grassland types and selected 20 samples evenly distributed in each type, then used NDVI/EVI product to summarize different spectral features of different grassland types. Finally, we introduced other classification auxiliary data, such as elevation, accumulate temperature (AT), humidity index (HI) and rainfall. China's nation-wide grassland classification map is resulted by merging the grassland in different climate zone. The overall classification accuracy is 60.4%. The result indicated that expert classifier is proper for national wide grassland classification, but the classification accuracy need to be improved.