ICIS:Intelligent and Label-Free crop identifying using Multi-Source time series and model transferOA
Crop cultivation is intrinsically linked to human well-being,and the timely identification of crop types is critical for agricultural monitoring and food security assessment.However,without extensive field surveys,achieving timely,large-scale,high-resolution crop identification remains challenging.This study proposes an intelligent crop identification strategy(ICIS)that integrates crop phenological development information with a model transfer strategy based on machine learning,generating 10-meter resolution crop distribution maps of major crop types from 2020 to 2023 in Northeast China(NE).ICIS utilizes time series data from Sentinel-1(S1)and Sentinel-2(S2)to extract multi-source temporal features covering the full growth cycles of maize,rice,and soybeans.A Random Forest(RF)classifier was trained on representative sample sites,and model transfer was employed to predict crop distribution across different years and regions.The experimental results show that the models achieved F1-scores above 0.92 across provinces,peaking at 0.97.For rice,the F1-scores for Jilin,Heilongjiang,and Liaoning remained stable at 0.94.Maize F1-scores consistently exceeded 0.92,and although the soybean accuracy varied spatially,it reached 0.97 in certain areas.Comparison with crop area statistics from the China Statistical Yearbook at the city level revealed that the average root mean square error(RMSE)across four years for the three crops was 8.6×10^(4)ha,with coefficients of determination(R^(2))consistently above 0.90,reaching up to 0.96.These results demonstrated the strong generalizability and robustness of the proposed ICIS Strategy for crop mapping across both temporal and spatial scales.Moreover,this study reveals the complementary strengths and key contributions of optical and Synthetic Aperture Radar(SAR)time series features in modeling crop growth dynamics.Overall,the ICIS provides an efficient and reliable technical means for the rapidly acquiring large-scale crop information.It provides robust support for agricultural monitoring and food security assessment,demonstrates significant potential for large-scale agricultural applications,and offers valuable insights and technical support for remote sensing data processing,image analysis,and machine learning-based agricultural automation and decision-making.
Su Rina;Wu Nile;Na Mula;Sicheng Wei;Ying Guo;Jiquan Zhang;Zhijun Tong;Xingpeng Liu;Chunli Zhao;Cha Ersi
School of Environment,Northeast Normal University,Changchun 130024,China Jilin Province Science and Technology Innovation Center of Agro-Meteorological Disaster Risk Assessment and Prevention,Northeast Normal University,Changchun 130024,China Key Laboratory for Vegetation Ecology,Ministry of Education,Changchun 130024,China State Environmental Protection Key Laboratory of Wetland Ecology and Vegetation Restoration,Northeast Normal University,Changchun 130024,ChinaSchool of Environment,Northeast Normal University,Changchun 130024,China Jilin Province Science and Technology Innovation Center of Agro-Meteorological Disaster Risk Assessment and Prevention,Northeast Normal University,Changchun 130024,China Key Laboratory for Vegetation Ecology,Ministry of Education,Changchun 130024,China State Environmental Protection Key Laboratory of Wetland Ecology and Vegetation Restoration,Northeast Normal University,Changchun 130024,ChinaSchool of Environment,Northeast Normal University,Changchun 130024,China Jilin Province Science and Technology Innovation Center of Agro-Meteorological Disaster Risk Assessment and Prevention,Northeast Normal University,Changchun 130024,China Key Laboratory for Vegetation Ecology,Ministry of Education,Changchun 130024,China State Environmental Protection Key Laboratory of Wetland Ecology and Vegetation Restoration,Northeast Normal University,Changchun 130024,ChinaSchool of Environment,Northeast Normal University,Changchun 130024,China Jilin Province Science and Technology Innovation Center of Agro-Meteorological Disaster Risk Assessment and Prevention,Northeast Normal University,Changchun 130024,China Key Laboratory for Vegetation Ecology,Ministry of Education,Changchun 130024,China State Environmental Protection Key Laboratory of Wetland Ecology and Vegetation Restoration,Northeast Normal University,Changchun 130024,ChinaSchool of Environment,Northeast Normal University,Changchun 130024,China Jilin Province Science and Technology Innovation Center of Agro-Meteorological Disaster Risk Assessment and Prevention,Northeast Normal University,Changchun 130024,China Key Laboratory for Vegetation Ecology,Ministry of Education,Changchun 130024,China State Environmental Protection Key Laboratory of Wetland Ecology and Vegetation Restoration,Northeast Normal University,Changchun 130024,ChinaSchool of Environment,Northeast Normal University,Changchun 130024,China Jilin Province Science and Technology Innovation Center of Agro-Meteorological Disaster Risk Assessment and Prevention,Northeast Normal University,Changchun 130024,China Key Laboratory for Vegetation Ecology,Ministry of Education,Changchun 130024,China State Environmental Protection Key Laboratory of Wetland Ecology and Vegetation Restoration,Northeast Normal University,Changchun 130024,ChinaSchool of Environment,Northeast Normal University,Changchun 130024,China Jilin Province Science and Technology Innovation Center of Agro-Meteorological Disaster Risk Assessment and Prevention,Northeast Normal University,Changchun 130024,China Key Laboratory for Vegetation Ecology,Ministry of Education,Changchun 130024,China State Environmental Protection Key Laboratory of Wetland Ecology and Vegetation Restoration,Northeast Normal University,Changchun 130024,ChinaSchool of Environment,Northeast Normal University,Changchun 130024,China Jilin Province Science and Technology Innovation Center of Agro-Meteorological Disaster Risk Assessment and Prevention,Northeast Normal University,Changchun 130024,China Key Laboratory for Vegetation Ecology,Ministry of Education,Changchun 130024,China State Environmental Protection Key Laboratory of Wetland Ecology and Vegetation Restoration,Northeast Normal University,Changchun 130024,ChinaCollege of Forestry and Grassland,Jilin Agricultural University,Changchun 130024,ChinaSchool of Environment,Northeast Normal University,Changchun 130024,China Jilin Province Science and Technology Innovation Center of Agro-Meteorological Disaster Risk Assessment and Prevention,Northeast Normal University,Changchun 130024,China Key Laboratory for Vegetation Ecology,Ministry of Education,Changchun 130024,China State Environmental Protection Key Laboratory of Wetland Ecology and Vegetation Restoration,Northeast Normal University,Changchun 130024,China
农业科技
Food securitySentinel-1/2Intelligent Crop Identifying Strategy(ICIS)Model transferTime series features
《Information Processing in Agriculture》 2026 (2)
P.283-302,20
supported by the National Natural Science Foundation of China(U21A2040)the National K&D Program of China(2022YFD2300201)the National K&D Program of China(2023YFD2301701).
评论