TY - JOUR
T1 - Mapping spatiotemporal dynamics of maize in China from 2005 to 2017 through designing leaf moisture based indicator from Normalized Multi-band Drought Index
AU - Qiu, Bingwen
AU - Huang, Yingze
AU - Chen, Chongchen
AU - Tang, Zhenghong
AU - Zou, Fengli
N1 - Funding Information:
This work was supported by the National Natural Science Foundation of China (grant nos. 41771468 , 41771362 ) and funding from the Science Bureau of Fujian Province ( 2017I0008 , 2017L3012 ) and the scholarship under the China Scholarship Council (CSC) ( 201806655006 ). We are grateful to the anonymous reviewers for the constructive comments. Datasets (i.e. derived maize maps) are available for download as supplementary materials of this paper.
Publisher Copyright:
© 2018 Elsevier B.V.
PY - 2018/10
Y1 - 2018/10
N2 - Maize agriculture is experiencing substantial changes in the spatiotemporal pattern of planting areas in the most populous country-China. However, there is no spatially explicit and continuous information at national scale. Mapping maize at national scale is challenging due to intra-class variability of Vegetation Indices (VIs) temporal profile. This study coped with this challenge through combined utilizations of the EVI with two bands (EVI2) and Normalized Multi-band Drought Index (NMDI) time series datasets. A novel Maize mapping algorithm was proposed through Exploring Leaf moisture variation during flowering Stage (MELS). An indicator, the Ratio of Cumulative Positive slope to Negative slope (RCPN) during flowering stage, was developed based on NMDI and utilized as the unique metric for maize mapping. The capability of the MELS method was verified using the 8-day composite MODerate resolution Imaging Spectroradiometer (MODIS) datasets in China from 2005 to 2017. The derived maize map was consistent with the agricultural census data (r2 = 0.8875 in 2015) and 2020 ground truth observations (overall accuracy = 91.49%). Validation with Landsat-interpreted images in the test regions further confirmed its fairly good accuracy, with overall accuracy of 87.91% and kappa coefficient of 0.8577. We first generated annual maize maps from 2005 to 2017 in China. Maize planting areas increased continuously 100,130 km2 (by 33.20%) during the period 2005–2015 and decreased 10,424 km2 (by 2.60%) from 2015 to 2017. The increase of cropping intensity, replacement of paddy rice and other non-maize dryland crops areas accounted for 36.48%, 34.23% and 29.29% of the dramatic increased maize areas from 2005 to 2015, respectively.
AB - Maize agriculture is experiencing substantial changes in the spatiotemporal pattern of planting areas in the most populous country-China. However, there is no spatially explicit and continuous information at national scale. Mapping maize at national scale is challenging due to intra-class variability of Vegetation Indices (VIs) temporal profile. This study coped with this challenge through combined utilizations of the EVI with two bands (EVI2) and Normalized Multi-band Drought Index (NMDI) time series datasets. A novel Maize mapping algorithm was proposed through Exploring Leaf moisture variation during flowering Stage (MELS). An indicator, the Ratio of Cumulative Positive slope to Negative slope (RCPN) during flowering stage, was developed based on NMDI and utilized as the unique metric for maize mapping. The capability of the MELS method was verified using the 8-day composite MODerate resolution Imaging Spectroradiometer (MODIS) datasets in China from 2005 to 2017. The derived maize map was consistent with the agricultural census data (r2 = 0.8875 in 2015) and 2020 ground truth observations (overall accuracy = 91.49%). Validation with Landsat-interpreted images in the test regions further confirmed its fairly good accuracy, with overall accuracy of 87.91% and kappa coefficient of 0.8577. We first generated annual maize maps from 2005 to 2017 in China. Maize planting areas increased continuously 100,130 km2 (by 33.20%) during the period 2005–2015 and decreased 10,424 km2 (by 2.60%) from 2015 to 2017. The increase of cropping intensity, replacement of paddy rice and other non-maize dryland crops areas accounted for 36.48%, 34.23% and 29.29% of the dramatic increased maize areas from 2005 to 2015, respectively.
KW - Crop
KW - Intra-class variability
KW - MODIS
KW - Phenology-based algorithm
KW - Time-series analysis
UR - http://www.scopus.com/inward/record.url?scp=85051104698&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85051104698&partnerID=8YFLogxK
U2 - 10.1016/j.compag.2018.07.039
DO - 10.1016/j.compag.2018.07.039
M3 - Article
AN - SCOPUS:85051104698
SN - 0168-1699
VL - 153
SP - 82
EP - 93
JO - Computers and Electronics in Agriculture
JF - Computers and Electronics in Agriculture
ER -