Floods are occurring across the globe, and due to climate change, flood events are expected to in... more Floods are occurring across the globe, and due to climate change, flood events are expected to increase in the coming years. Current situations urge more focus on efficient monitoring of floods and detecting impacted areas. In this study, we propose two segmentation networks for flood detection on uni-temporal Sentinel-1 Synthetic Aperture Radar data. The first network is “Attentive U-Net”. It takes VV, VH, and the ratio VV/VH as input. The network uses spatial and channel-wise attention to enhance feature maps which help in learning better segmentation. “Attentive U-Net” yields 67% Intersection Over Union (IoU) on the Sen1Floods11 dataset, which is 3% better than the benchmark IoU. The second proposed network is a dual-stream “Fusion network”, where we fuse global low-resolution elevation data and permanent water masks with Sentinel-1 (VV, VH) data. Compared to the previous benchmark on the Sen1Floods11 dataset, our fusion network gave a 4.5% better IoU score. Quantitatively, the p...
2021 IEEE International Geoscience and Remote Sensing Symposium IGARSS, 2021
In recent years, wildfires have become major devastating hazards that affect both public safety a... more In recent years, wildfires have become major devastating hazards that affect both public safety and the environment. Thus, agile detection of the wildfires is desirable to suppress wildfires in the early stage. Owing to the high temporal resolution, GOES-R satellites offer capabilities to obtain images every 15 minutes enabling a near real-time monitoring of wildfires. In this research, a time-series-based deep learning framework, composed of Gated Recurrent Units (GRU), is proposed to capture the emerging of the wildfire at early stage. By feeding the embedding of the coarse satellite imagery to Deep GRU network, the active fires are segmented out from the remote sensing imagery. The preliminary results show that proposed network can detect the wildfires earlier than the state-of-the-art fire product for 2020 wildfires in California and British Columbia, at the same time provide sufficiently high accuracy on the burned areas.
The overall objective of this research is to evaluate multitemporal spaceborne SAR and optical da... more The overall objective of this research is to evaluate multitemporal spaceborne SAR and optical data for urban land cover mapping and urbanization monitoring. Multitemporal Sentinel-1A SAR and historical ERS SAR and ENVISAT ASAR data as well as HJ-1B multispectral data were acquired in Beijing, Chendgdu and Nanchang, China where rapid urbanization has taken place. KTH-SEG, a novel object-based classification method is adopted for urban land cover mapping while KTH-Pavia Urban Extractor, a robust algorithm is improved for urban extent extraction and urbanization monitoring. The research demonstrates that, for urban land cover classification, the fusion of multitemporal SAR and optical data is superior to SAR or optical data alone. The second best classification result is achieved using fusion of 4-date SAR and one HJ-1B image. The results indicate that carefully selected multitemporal SAR dataset and its fusion with optical data could produce nearly as good classification accuracy as the whole multitemporal dataset. The results also show that KTH-SEG, the edge-aware region growing and merging segmentation algorithm, is effective for classification of SAR, optical and their fusion. KTH-SEG outperforms eCognition, the commonly used commercial software, for image segmentation and classification of linear features. For Urban extent extraction, single-date and multitemporal SAR data including ERS SAR, ENVISAT ASAR and Sentinel-1A SAR achieved very promising results in all study areas using the improved KTH-Pavia Urban Extractor. The results showed that urban areas as well as small towns and villages could be well extracted using multitemporal Sentinel-1A SAR data while major urban areas could be well extracted using a single-date single-polarization SAR image. The results clearly demonstrate that multitemporal SAR data are cost- and time-effective way for monitoring spatiotemporal patterns of urbanization.
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2017
This study aims at providing a new method to efficiently analyze detailed urban ecological condit... more This study aims at providing a new method to efficiently analyze detailed urban ecological conditions at the example of Shanghai, one of the world's most densely populated megacities. The main objective is to develop a method to effectively analyze high-resolution optical satellite data for mapping of ecologically important urban space and to evaluate ecological changes through the emerging ecosystem service supply and demand concept. Two IKONOS and GeoEye-1 scenes were used to determine land use/land cover change in Shanghai's urban core from 2000 to 2009. After preprocessing, the images were segmented and classified into seven distinct urban land use/land cover classes through SVM. The classes were then transformed into ecosystem service supply and demand budgets for regulating, provisioning and cultural services, and ecological integrity based on ecosystem functions. Decreases in continuous urban fabric and industrial areas in the favor of urban green sites and high-rise areas with commercial/residential function could be observed resulting in an increase of at least 20% in service supply budgets. Main contributor to the change is the decrease in continuous urban fabric and industrial areas. The overall results and outcome of the study strengthen the suggested application of the proposed method for urban ecosystem service budget mapping with hitherto for that purpose unutilized high-resolution data. The insights and results from this study might further contribute to sustainable urban planning, prove common grounds for interurban comparisons, or aid in enhancing ecological intraurban functionality by analyzing the distribution of urban ecospace and lead to improved accessibility and proximity to ecosystem services in urban areas.
Spaceborne SAR for Analysis of Urban Environment and Detection of Human Settlements Project #: DN... more Spaceborne SAR for Analysis of Urban Environment and Detection of Human Settlements Project #: DNR 125-0 : A Project Report Submitted to the Swedish National Space Board
The objective of this research is to investigate multitemporal spaceborne SAR data for urbanizati... more The objective of this research is to investigate multitemporal spaceborne SAR data for urbanization monitoring in China. A generalized version of Kittler- Illingworth minimum-error thresholding algorithm, that takes into account the non-Gaussian distribution of SAR images, was tested to automatically classify the change variable derived from SAR multitemporal images into two classes, change and no change. A modified ratio operator was examined for identifying both positive and negative changes by comparing the multitemporal SAR images on a pixel-by-pixel basis. Various probability density functions such as Log normal, Generalized Gaussian, Nakagami ratio, and Weibull ratio models were tested to model the distribution of the change and no change classes. The preliminary results showed that this unsupervised change detection algorithm is very effective in detecting temporal changes in urban areas using multitemporal SAR images. The initial findings indicated that change detection accuracy varies depending on how the assumed conditional class density function fits the histograms of change and no change classes.
Assessing the Impact of Landscape Dynamics on the Terrestrial Biodiversity Using Multisensor Renm... more Assessing the Impact of Landscape Dynamics on the Terrestrial Biodiversity Using Multisensor Renmote Sensing Project #: DNR 151/05 & DNR 151/05:2 : A Project Report Submitted to the Swedish National Space Board
2008 International Workshop on Earth Observation and Remote Sensing Applications, 2008
... 2) During 1982-2003, the dynamic evolution process ... it is important to identify the purpos... more ... 2) During 1982-2003, the dynamic evolution process ... it is important to identify the purpose of this article is just to compare the best situation of vegetation cover in Mongolia ... evolved like this or that, the future study should focus on the driving forces study by coupling regional land ...
Cartographica: The International Journal for Geographic Information and Geovisualization, 2011
This article proposes a novel framework for online visualization of 3D city models. CityGML is us... more This article proposes a novel framework for online visualization of 3D city models. CityGML is used to represent the city models, based on which 3D scenes in X3D are generated, then dynamically updated to the user side with AJAX and visualized in WebGL-supported browsers with X3DOM. The experimental results show that the proposed framework can easily be implemented using widely supported major browsers and can efficiently support online visualization of 3D city models in small areas. For the visualization of large volumes of data, generalization methods and multiple-representation data structure should be studied in future research.
Floods are occurring across the globe, and due to climate change, flood events are expected to in... more Floods are occurring across the globe, and due to climate change, flood events are expected to increase in the coming years. Current situations urge more focus on efficient monitoring of floods and detecting impacted areas. In this study, we propose two segmentation networks for flood detection on uni-temporal Sentinel-1 Synthetic Aperture Radar data. The first network is “Attentive U-Net”. It takes VV, VH, and the ratio VV/VH as input. The network uses spatial and channel-wise attention to enhance feature maps which help in learning better segmentation. “Attentive U-Net” yields 67% Intersection Over Union (IoU) on the Sen1Floods11 dataset, which is 3% better than the benchmark IoU. The second proposed network is a dual-stream “Fusion network”, where we fuse global low-resolution elevation data and permanent water masks with Sentinel-1 (VV, VH) data. Compared to the previous benchmark on the Sen1Floods11 dataset, our fusion network gave a 4.5% better IoU score. Quantitatively, the p...
2021 IEEE International Geoscience and Remote Sensing Symposium IGARSS, 2021
In recent years, wildfires have become major devastating hazards that affect both public safety a... more In recent years, wildfires have become major devastating hazards that affect both public safety and the environment. Thus, agile detection of the wildfires is desirable to suppress wildfires in the early stage. Owing to the high temporal resolution, GOES-R satellites offer capabilities to obtain images every 15 minutes enabling a near real-time monitoring of wildfires. In this research, a time-series-based deep learning framework, composed of Gated Recurrent Units (GRU), is proposed to capture the emerging of the wildfire at early stage. By feeding the embedding of the coarse satellite imagery to Deep GRU network, the active fires are segmented out from the remote sensing imagery. The preliminary results show that proposed network can detect the wildfires earlier than the state-of-the-art fire product for 2020 wildfires in California and British Columbia, at the same time provide sufficiently high accuracy on the burned areas.
The overall objective of this research is to evaluate multitemporal spaceborne SAR and optical da... more The overall objective of this research is to evaluate multitemporal spaceborne SAR and optical data for urban land cover mapping and urbanization monitoring. Multitemporal Sentinel-1A SAR and historical ERS SAR and ENVISAT ASAR data as well as HJ-1B multispectral data were acquired in Beijing, Chendgdu and Nanchang, China where rapid urbanization has taken place. KTH-SEG, a novel object-based classification method is adopted for urban land cover mapping while KTH-Pavia Urban Extractor, a robust algorithm is improved for urban extent extraction and urbanization monitoring. The research demonstrates that, for urban land cover classification, the fusion of multitemporal SAR and optical data is superior to SAR or optical data alone. The second best classification result is achieved using fusion of 4-date SAR and one HJ-1B image. The results indicate that carefully selected multitemporal SAR dataset and its fusion with optical data could produce nearly as good classification accuracy as the whole multitemporal dataset. The results also show that KTH-SEG, the edge-aware region growing and merging segmentation algorithm, is effective for classification of SAR, optical and their fusion. KTH-SEG outperforms eCognition, the commonly used commercial software, for image segmentation and classification of linear features. For Urban extent extraction, single-date and multitemporal SAR data including ERS SAR, ENVISAT ASAR and Sentinel-1A SAR achieved very promising results in all study areas using the improved KTH-Pavia Urban Extractor. The results showed that urban areas as well as small towns and villages could be well extracted using multitemporal Sentinel-1A SAR data while major urban areas could be well extracted using a single-date single-polarization SAR image. The results clearly demonstrate that multitemporal SAR data are cost- and time-effective way for monitoring spatiotemporal patterns of urbanization.
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2017
This study aims at providing a new method to efficiently analyze detailed urban ecological condit... more This study aims at providing a new method to efficiently analyze detailed urban ecological conditions at the example of Shanghai, one of the world's most densely populated megacities. The main objective is to develop a method to effectively analyze high-resolution optical satellite data for mapping of ecologically important urban space and to evaluate ecological changes through the emerging ecosystem service supply and demand concept. Two IKONOS and GeoEye-1 scenes were used to determine land use/land cover change in Shanghai's urban core from 2000 to 2009. After preprocessing, the images were segmented and classified into seven distinct urban land use/land cover classes through SVM. The classes were then transformed into ecosystem service supply and demand budgets for regulating, provisioning and cultural services, and ecological integrity based on ecosystem functions. Decreases in continuous urban fabric and industrial areas in the favor of urban green sites and high-rise areas with commercial/residential function could be observed resulting in an increase of at least 20% in service supply budgets. Main contributor to the change is the decrease in continuous urban fabric and industrial areas. The overall results and outcome of the study strengthen the suggested application of the proposed method for urban ecosystem service budget mapping with hitherto for that purpose unutilized high-resolution data. The insights and results from this study might further contribute to sustainable urban planning, prove common grounds for interurban comparisons, or aid in enhancing ecological intraurban functionality by analyzing the distribution of urban ecospace and lead to improved accessibility and proximity to ecosystem services in urban areas.
Spaceborne SAR for Analysis of Urban Environment and Detection of Human Settlements Project #: DN... more Spaceborne SAR for Analysis of Urban Environment and Detection of Human Settlements Project #: DNR 125-0 : A Project Report Submitted to the Swedish National Space Board
The objective of this research is to investigate multitemporal spaceborne SAR data for urbanizati... more The objective of this research is to investigate multitemporal spaceborne SAR data for urbanization monitoring in China. A generalized version of Kittler- Illingworth minimum-error thresholding algorithm, that takes into account the non-Gaussian distribution of SAR images, was tested to automatically classify the change variable derived from SAR multitemporal images into two classes, change and no change. A modified ratio operator was examined for identifying both positive and negative changes by comparing the multitemporal SAR images on a pixel-by-pixel basis. Various probability density functions such as Log normal, Generalized Gaussian, Nakagami ratio, and Weibull ratio models were tested to model the distribution of the change and no change classes. The preliminary results showed that this unsupervised change detection algorithm is very effective in detecting temporal changes in urban areas using multitemporal SAR images. The initial findings indicated that change detection accuracy varies depending on how the assumed conditional class density function fits the histograms of change and no change classes.
Assessing the Impact of Landscape Dynamics on the Terrestrial Biodiversity Using Multisensor Renm... more Assessing the Impact of Landscape Dynamics on the Terrestrial Biodiversity Using Multisensor Renmote Sensing Project #: DNR 151/05 & DNR 151/05:2 : A Project Report Submitted to the Swedish National Space Board
2008 International Workshop on Earth Observation and Remote Sensing Applications, 2008
... 2) During 1982-2003, the dynamic evolution process ... it is important to identify the purpos... more ... 2) During 1982-2003, the dynamic evolution process ... it is important to identify the purpose of this article is just to compare the best situation of vegetation cover in Mongolia ... evolved like this or that, the future study should focus on the driving forces study by coupling regional land ...
Cartographica: The International Journal for Geographic Information and Geovisualization, 2011
This article proposes a novel framework for online visualization of 3D city models. CityGML is us... more This article proposes a novel framework for online visualization of 3D city models. CityGML is used to represent the city models, based on which 3D scenes in X3D are generated, then dynamically updated to the user side with AJAX and visualized in WebGL-supported browsers with X3DOM. The experimental results show that the proposed framework can easily be implemented using widely supported major browsers and can efficiently support online visualization of 3D city models in small areas. For the visualization of large volumes of data, generalization methods and multiple-representation data structure should be studied in future research.
Uploads
Papers by Yifang Ban