Despite the widespread use of machine learning(ML)models for geospatial applications,adaptations to imbalanced multitemporal land cover(LC)datasets remain underexplored.For over two dec-ades,studies have predominantly...Despite the widespread use of machine learning(ML)models for geospatial applications,adaptations to imbalanced multitemporal land cover(LC)datasets remain underexplored.For over two dec-ades,studies have predominantly trained ML models on a single interval of LC data to model changes,with detriments of imbal-anced training datasets managed through manual manipulations.Therefore,this study proposes and implements an ML-spatial sam-ple weighting(ML-SSW)approach to leverage available multitem-poral LC data while adjusting sample influence to reflect recency of change occurrence and class-level spatial pattern measures to enable data-driven LC change modeling.Random Forest(RF),Neural Network(NN),and Extreme Gradient Boosting Machine(XGB)models are trained under the ML-SSW strategy on three study areas located in British Columbia,Canada.The RF-SSW,NN-SSW,and XGB-SSW models forecasted more realistic changes across multiple timesteps with fewer errors than baseline configurations.The presented methodology provides a step toward establishing spatialized cost-sensitive learning strategies and extending classical ML models to multitemporal LC datasets.展开更多
Shorea robusta Gaertn. f.（Sal）is one of the important tim-ber-yielding plants in India, which dominates the vegetation of Terai landscape of Uttar Pradesh state in India forming various communities based on its asso...Shorea robusta Gaertn. f.（Sal）is one of the important tim-ber-yielding plants in India, which dominates the vegetation of Terai landscape of Uttar Pradesh state in India forming various communities based on its associations. The present study deals with delineation, map-ping and characterization of various communities of Sal （Shorea robusta） forests in Terai landscape of Uttar Pradesh, India ranging across over 16 districts. Field survey and visual interpretation based forest vegetation type classification and mapping was carried out as part of the project entitled ‘Biodiversity characterization at landscape level using remote sensing and GIS’. Indian Remote Sensing-P6 （Resourcesat-1） Linear Imaging Self Scanner-III satellite data was used during the study. The total area covered by different Sal forests was found to be approximately 2256.77 km2. Sal communities were identified and characterized based on their spectral properties, physiognomy and phytosociological charac-teristics. Following nine Sal communities were identified, delineated and mapped with reasonable accuracyviz.,Chandar,Damar, dry plains, moist plains, western alluvium, western alluvium plains, mixed moist deciduous, mixed dry deciduous andSiwalik. It is evident from the area estimates that mixed moist deciduous Sal is the most dominant commu-nity in the region covering around （1613.90 km2）, other major communi-ties were found as western alluvium plains Sal （362.44 km2）, mixed dry deciduous Sal （362.44 km2） and dry plains Sal （107.71 km2）. The Terai landscape of Uttar Pradesh faces tremendous anthropogenic pressure leading to deterioration of the forests. Community level information could be used monitoring the status as well as for micro level conserva-tion and planning of the Sal forests in Terai Landscape of Uttar Pradesh.展开更多
Modelling land-use/landcover(LULC)change is vital for addressing global environmental and sustainability issues and evaluating various land system scenarios.However,existing geosimulation methodologies for global LULC...Modelling land-use/landcover(LULC)change is vital for addressing global environmental and sustainability issues and evaluating various land system scenarios.However,existing geosimulation methodologies for global LULC change fail to account for spatial distortions caused by the Earth’s curvature and do not consider multiple LULC change processes.Thus,this research study proposes an enhanced spherical geosimulation modelling approach that integrates deep learning(DL)to simulate change of multiple classes of LULC process under the shared socioeconomic pathways(SSP)at the global level.Based on the simulation results,the frontiers of urbanization,cropland expansion,and deforestation are indicated to be in developing countries particularly in Asia and Africa.The simulation outputs also reveal 42.5%-63.2%of new urban development would occur on croplands.The proposed modelling approach can serve as a valuable tool for spatial decision-making and environmental policy formulation at the global level.展开更多
基金supported by the Natural Sciences and Engineering Research Council of Canada[RGPIN-2023-04052].
文摘Despite the widespread use of machine learning(ML)models for geospatial applications,adaptations to imbalanced multitemporal land cover(LC)datasets remain underexplored.For over two dec-ades,studies have predominantly trained ML models on a single interval of LC data to model changes,with detriments of imbal-anced training datasets managed through manual manipulations.Therefore,this study proposes and implements an ML-spatial sam-ple weighting(ML-SSW)approach to leverage available multitem-poral LC data while adjusting sample influence to reflect recency of change occurrence and class-level spatial pattern measures to enable data-driven LC change modeling.Random Forest(RF),Neural Network(NN),and Extreme Gradient Boosting Machine(XGB)models are trained under the ML-SSW strategy on three study areas located in British Columbia,Canada.The RF-SSW,NN-SSW,and XGB-SSW models forecasted more realistic changes across multiple timesteps with fewer errors than baseline configurations.The presented methodology provides a step toward establishing spatialized cost-sensitive learning strategies and extending classical ML models to multitemporal LC datasets.
基金part of the Department of Space/Department of Biotechnology sponsored project entitled "Biodiversity Characterization at Landscape level using Remote Sensing and GIS for Uttar Pradesh state except Vindhyan Hills"
文摘Shorea robusta Gaertn. f.（Sal）is one of the important tim-ber-yielding plants in India, which dominates the vegetation of Terai landscape of Uttar Pradesh state in India forming various communities based on its associations. The present study deals with delineation, map-ping and characterization of various communities of Sal （Shorea robusta） forests in Terai landscape of Uttar Pradesh, India ranging across over 16 districts. Field survey and visual interpretation based forest vegetation type classification and mapping was carried out as part of the project entitled ‘Biodiversity characterization at landscape level using remote sensing and GIS’. Indian Remote Sensing-P6 （Resourcesat-1） Linear Imaging Self Scanner-III satellite data was used during the study. The total area covered by different Sal forests was found to be approximately 2256.77 km2. Sal communities were identified and characterized based on their spectral properties, physiognomy and phytosociological charac-teristics. Following nine Sal communities were identified, delineated and mapped with reasonable accuracyviz.,Chandar,Damar, dry plains, moist plains, western alluvium, western alluvium plains, mixed moist deciduous, mixed dry deciduous andSiwalik. It is evident from the area estimates that mixed moist deciduous Sal is the most dominant commu-nity in the region covering around （1613.90 km2）, other major communi-ties were found as western alluvium plains Sal （362.44 km2）, mixed dry deciduous Sal （362.44 km2） and dry plains Sal （107.71 km2）. The Terai landscape of Uttar Pradesh faces tremendous anthropogenic pressure leading to deterioration of the forests. Community level information could be used monitoring the status as well as for micro level conserva-tion and planning of the Sal forests in Terai Landscape of Uttar Pradesh.
基金supported by the Natural Sciences and Engineering Research Council of Canada[RGPIN-2023-04052].
文摘Modelling land-use/landcover(LULC)change is vital for addressing global environmental and sustainability issues and evaluating various land system scenarios.However,existing geosimulation methodologies for global LULC change fail to account for spatial distortions caused by the Earth’s curvature and do not consider multiple LULC change processes.Thus,this research study proposes an enhanced spherical geosimulation modelling approach that integrates deep learning(DL)to simulate change of multiple classes of LULC process under the shared socioeconomic pathways(SSP)at the global level.Based on the simulation results,the frontiers of urbanization,cropland expansion,and deforestation are indicated to be in developing countries particularly in Asia and Africa.The simulation outputs also reveal 42.5%-63.2%of new urban development would occur on croplands.The proposed modelling approach can serve as a valuable tool for spatial decision-making and environmental policy formulation at the global level.
