The spatial pattern of rice paddies is an essential parameter used for studies of greenhouse gas emissions,agricultural resource management,and environmental monitoring.On large spatial scales,previous studies have us...The spatial pattern of rice paddies is an essential parameter used for studies of greenhouse gas emissions,agricultural resource management,and environmental monitoring.On large spatial scales,previous studies have usually mapped rice paddies using a single vegetation index product based on a traditional classification method,or a combined analysis of various vegetation and water indices derived from the moderate resolution imaging spectroradiometer(MODIS)satellite data.However,different indices increase the computational cost and constrain the satellite data sources,and traditional classification methods(e.g.,maximum likelihood classification)may be time-consuming and difficult to carry out over a large area like China.In this study,we designed an auto-thresholding and single vegetation index(normalized difference vegetation index(NDVI))-based procedure to estimate the spatial distribution of rice paddies in China.The MOD09Q1 product,which was available at MODIS’s highest spatial resolution(250 m),was taken as the input source.An auto-threshold function was also introduced into the change detection process to distinguish rice paddies from other croplands.Our MODIS-derived maps were validated with ground surveys and then compared with China national statistical data of rice paddy areas.The results indicated that the best classification result was achieved for plain regions,and that the accuracy declined for hilly regions,where the complex landscape could lead to an underestimation of the rice paddy area.A comparison between the modeled results and other analyses using 500-m MODIS data suggests that rice paddies may be identified routinely using a single vegetation index with finer resolution on large spatial scales.展开更多
基金financially supported by the Strategic Priority Research Program of Chinese Academy of Sciences—Climate Change:Carbon Budget and Relevant Issues(No.XDA05020200)the Startup Foundation for Introducing Talent of Nanjing University of Information Science and Technology(NUIST),China(No.2016r036)the Innovation and Entrepreneurship Training Program for College Students of Jiangsu Provincial Department of Education,China(No.2017103000165)
文摘The spatial pattern of rice paddies is an essential parameter used for studies of greenhouse gas emissions,agricultural resource management,and environmental monitoring.On large spatial scales,previous studies have usually mapped rice paddies using a single vegetation index product based on a traditional classification method,or a combined analysis of various vegetation and water indices derived from the moderate resolution imaging spectroradiometer(MODIS)satellite data.However,different indices increase the computational cost and constrain the satellite data sources,and traditional classification methods(e.g.,maximum likelihood classification)may be time-consuming and difficult to carry out over a large area like China.In this study,we designed an auto-thresholding and single vegetation index(normalized difference vegetation index(NDVI))-based procedure to estimate the spatial distribution of rice paddies in China.The MOD09Q1 product,which was available at MODIS’s highest spatial resolution(250 m),was taken as the input source.An auto-threshold function was also introduced into the change detection process to distinguish rice paddies from other croplands.Our MODIS-derived maps were validated with ground surveys and then compared with China national statistical data of rice paddy areas.The results indicated that the best classification result was achieved for plain regions,and that the accuracy declined for hilly regions,where the complex landscape could lead to an underestimation of the rice paddy area.A comparison between the modeled results and other analyses using 500-m MODIS data suggests that rice paddies may be identified routinely using a single vegetation index with finer resolution on large spatial scales.
