Distribution and mapping of temperate savanna in the sandy lands of eastern China  

作　　者：Xiaoya LI Feng WANG Tao DUAN Kaijie YANG Bin YANG Chunmei WANG Xin TIAN Qi LU 

机构地区：[1]Institute of Ecological Conservation and Restoration,Institute of Desertification Studies,Chinese Academy of Forestry,Beijing,100091,China [2]Institute of Desertification,Institute of Desertification Studies,Chinese Academy of Forestry,Beijing,100091,China [3]Institute of Great Green Wall,Bayan Nur,015200,China [4]Satellite Application Centre for Ecology and Environment,MEE,Beijing,100094,China [5]Institute of Microelectronics,Chinese Academy of Sciences,Beijing,100029,China [6]Aerospace Information Research Institute,Chinese Academy of Sciences,Beijing,100094,China [7]Institute of Forest Resource Information Techniques,Chinese Academy of Forestry,Beijing,100091,China

出　　处：《Science China Earth Sciences》2025年第1期128-141,共14页中国科学(地球科学英文版)

基　　金：supported by the Chinese Academy of Forestry Science Fund(Grant Nos.CAFYBB2020QD002,CAFYBB2021MC002,CAFYBB2023ZA009);the National Natural Science Foundation of China(Grant No.32171875);the National Key Research and Development Program of China(Grant No.2023YFF1304103);the Open Bidding for Selecting the Best Candidates Project supported by National Forestry and Grassland Administration(Grant No.202401)。

摘　　要：Savannas are distributed globally across various regions.They perform a significant socioeconomic role in regulating net primary productivity and the terrestrial carbon cycle.Owing to the high spatial heterogeneity induced by the mixed distribution of trees and grasses,current global land-cover classification maps often classify savannas as forests or grasslands Temperate savannas in China cover a vast area,yet their spatial patterns and extent remain poorly understood.This study developed a novel remote sensing temperate savannas mapping method by integrating very high-resolution(VHR)unmanned aerial vehicle(UAV)and GF-6 satellite imagery based on the vegetation structure characteristics of temperate savannas using the FAO definition of savannas.Training and validation datasets were collected from VHR UAV imagery covering the main temperate savannah types with a ground area of 1818 ha.The fractional woody vegetation cover(FWVC)and fractional herbaceous vegetation cover(FHVC)in temperate savannas were estimated by integrating Classification and Regression Tree and Random Forest algorithms.The current distribution of temperate savannas in the Horqin and Otindag Sandy Lands of Eastern China was mapped to a unit area of 0.5 ha.The root-mean-square error(RMSE)values for the FWVC and FHVC models were 10.69%and 11.67%,and the R^(2)values were 0.66 and 0.63,respectively.The assessment indicated that the new temperate savanna map had high identification accuracy,with a Kappa coefficient of 0.70,overall accuracy of 0.88,and average accuracy of 0.85.Temperate savannas in the sandy lands of eastern China were primarily distributed in the northern Horqin Sandy Land the central and eastern regions of the Otindag Sandy Land,and the transition zone from the southern slopes of the Greater Khingan Mountains to the plains,covering a total area of 3.27×10^(6)ha.The main savannah types were elm,oak/maple,and shrub savannas.Elm savannas were widely distributed in the Otindag Sandy Land and the eastern and northern Horqin Sandy Land a

关 键 词：Temperate savanna UAV imagery GF-6 Fractional woody vegetation cover Fractional herbaceous vegetation cover Machine learning 

分 类 号：S812[农业科学—草业科学] P285[农业科学—畜牧学]