Climate change,natural disasters,pollution,and fast urbanization have made environmental security a more serious international issue.Timely,accurate,and multi-dimensional information is essential in the effective moni...Climate change,natural disasters,pollution,and fast urbanization have made environmental security a more serious international issue.Timely,accurate,and multi-dimensional information is essential in the effective monitoring and management of such complex challenges in the environment.The Earth Observation(EO)systems,including optical sensors,radar sensors,Light Detection and Ranging(LiDAR)sensors,thermal sensors,Unmanned Aerial Vehicle(UAV)sensors,and in-situ sensors,offer a good coverage of space and time,as well as provide useful information on land,water,and atmospheric processes.But the shortcomings or weaknesses of individual sensors,such as their vulnerability to weather conditions,spectral or spatial resolution,and gaps in time,can tend to limit their ability to provide a complete picture of the environment.One of the solutions has been multi-sensor fusion,which combines heterogeneous data and makes it more accurate,robust,and interpretable.This systematic review analyzes the latest methods of multi-sensor fusion,which are machine learning,deep learning,probabilistic models,and hybrid approaches,in terms of methodological principles,preprocessing needs,and computational frameworks.Applications in environmental security are highlighted,which include monitoring natural disasters,monitoring of climate and ecosystem,pollution monitoring,monitoring of land use change,and early warning systems.The review also covers evaluation measures,validation plans,and uncertainty measures,where a strict measure of evaluation is vital to making actionable decisions.Lastly,emerging issues,e.g.,data heterogeneity,computational needs,sensor interoperability,and prospects in the future,e.g.,AI-based adaptive fusion,UAVs and Internet of Things(IoT)integration,and scalable cloud-based systems,are discussed.The synthesis has highlighted the transformational capability of multi-sensor EO in terms of improving the environment in the context of environmental security and sustainable management.展开更多
文摘Climate change,natural disasters,pollution,and fast urbanization have made environmental security a more serious international issue.Timely,accurate,and multi-dimensional information is essential in the effective monitoring and management of such complex challenges in the environment.The Earth Observation(EO)systems,including optical sensors,radar sensors,Light Detection and Ranging(LiDAR)sensors,thermal sensors,Unmanned Aerial Vehicle(UAV)sensors,and in-situ sensors,offer a good coverage of space and time,as well as provide useful information on land,water,and atmospheric processes.But the shortcomings or weaknesses of individual sensors,such as their vulnerability to weather conditions,spectral or spatial resolution,and gaps in time,can tend to limit their ability to provide a complete picture of the environment.One of the solutions has been multi-sensor fusion,which combines heterogeneous data and makes it more accurate,robust,and interpretable.This systematic review analyzes the latest methods of multi-sensor fusion,which are machine learning,deep learning,probabilistic models,and hybrid approaches,in terms of methodological principles,preprocessing needs,and computational frameworks.Applications in environmental security are highlighted,which include monitoring natural disasters,monitoring of climate and ecosystem,pollution monitoring,monitoring of land use change,and early warning systems.The review also covers evaluation measures,validation plans,and uncertainty measures,where a strict measure of evaluation is vital to making actionable decisions.Lastly,emerging issues,e.g.,data heterogeneity,computational needs,sensor interoperability,and prospects in the future,e.g.,AI-based adaptive fusion,UAVs and Internet of Things(IoT)integration,and scalable cloud-based systems,are discussed.The synthesis has highlighted the transformational capability of multi-sensor EO in terms of improving the environment in the context of environmental security and sustainable management.
