PROGRESS IN GEOGRAPHY ›› 2016, Vol. 35 ›› Issue (5): 655-663.doi: 10.18306/dlkxjz.2016.05.012

• Articles • Previous Articles    

Forest cover classification based on remote sensing threshold consistent with statistics in Heilongjiang Province

Yujuan YUAN1,2(), Yunhe YIN1,*(), Erfu DAI1, Ronggao LIU3, Shaohong WU1   

  1. 1. Key Laboratory of Land Surface Pattern and Simulation, Institute of Geographic Sciences and Natural Resources Research, CAS, Beijing 100101, China
    2. University of Chinese Academy of Sciences, Beijing 100049, China
    3. State Key Laboratory of Resources and Environmental Information System, Institute of Geographic Sciences and Natural Resources Research, CAS, Beijing 100101, China
  • Accepted:2015-05-01 Online:2016-05-27 Published:2016-05-27
  • Contact: Yunhe YIN;
  • Supported by:
    National Science and Technology Support Program of China, No.2012BAC19B02;National Natural Science Foundation of China, No.41571043;Key Program of the National Natural Science Foundation of China, No.41530749


Accurately identifying spatial distribution of forest is critically important for dynamic monitoring and sustainable management of forest resources. In this article, in order to acquire a spatially explicit forest cover classification based on the national forest inventory (NFI) statistics at the provincial scale, we developed an identification method using threshold values based on forest area from NFI statistics in 1999-2003 and the Moderate Resolution Imaging Spectroradiometer (MODIS) surface reflectance data in 2000 with a spatial resolution of 500 m for Heilongjiang Province. Based on the seasonal difference of Normalized Difference Vegetation Index (NDVI) of various forest types, threshold values between different forest types in satellite data were set using the NFI statistical data as criteria. Four forest types were differentiated: evergreen needleleaf, deciduous broadleaf, deciduous needleleaf, and mixed forests. Due to the stratified random sampling method used in this study and reliable threshold identification, the accuracy assessment result shows that the spatial pattern of forest cover classifications is highly consistent with the ground reference map, with an overall classification accuracy of 78.1%. Specifically, the applied method resulted in higher classification accuracy for deciduous forests that have distinct seasonal variations of NDVI (with user accuracy above 80%). The study provides a practical method for spatially explicit forest coverage estimation, and for quantifying changes in biomass and carbon stock in the ecosystem at the regional scale based on several periods of NFI statistics and remote sensing data.

Key words: forest inventory, remote sensing, threshold, accuracy assessment, Heilongjiang Province