Timely detection of flooding is paramount for saving lives as well as evaluating levels of damage. Floods generally occur under specific weather conditions, such as excessive precipitation, which makes the presence of clouds very likely. For this reason, radar-based sensors are most suitable for near-real-time flood mapping. The public dataset Sen1Floods11 recently released by the Cloud to Street is one example of ongoing beneficial initiatives to employ deep learning for flood detection with synthetic aperture radar. The present study used this dataset to improve flood detection using well-known segmentation architectures, such as SegNet and UNet, as networks. In addition, this study provided a deeper understanding of which set of polarized band combination is more suitable for distinguishing permanent water, as well as flooded areas from the SAR image. The overall performance of the models with various kinds of labels and a combination of bands to detect all surface water areas were also assessed. Finally, the trained models were tested on a completely different location at Kerala, India, during the 2018 flood for verifying their performance in the real-world situation of a flood event outside of the given test set in the dataset. The results prove that trained models can be used as off-the-shelf models to achieve an intersection over union (IoU) as high as 0.88 in comparison with optical images. The omission and commission error were less than 6%. However, the most important result is that the processing time for the whole satellite image was less than 1 min. This will help significantly for providing analysis and near-real-time flood mapping services to first responder organizations during flooding disasters.

The brown planthopper Nilaparvata lugens (BPH) is one of the most harmful insect pests in rice paddy fields, which causes considerable yield loss and consequent economic problems, particularly in the central plain of Thailand. Accurate and timely forecasting of pest population incidence would support farmers in planning effective mitigation. In this study, artificial neural network (ANN), random forest (RF) and classic linear multiple regression (MLR) analyses were applied and compared to forecast the BPH population using weather and host-plant phenology factors during the crop dry season from 2006 to 2016 in the central plain of Thailand. Data from satellite earth observation was used to monitor crop phenology factors affecting BPH population density. An ANN model with integrated ground-based meteorological variables and satellite-derived host plant variables was more accurate for short-term forecasting of the peak abundance of BPH when compared with RF and MLR, according to a reasonably validating dataset (RMSE of natural log-transformed (ln) BPH light trap catches = 1.686, 1.737, and 2.015, respectively). This finding indicates that the utilization of ground meteorological observations, satellite-derived NDVI time series, and ANN have the potential to predict BPH population density in support of integrated pest management programs. We expect the results from this study can be applied in conjunction with the satellite-based rice monitoring system developed by the Geo-Informatic and Space Technology Development Agency of Thailand (GISTDA) to support an effective pest early warning system.

Coastal zones are considered to be highly vulnerable to the effects of climate change, such as erosion, flooding, and storms, including sea level rise (SLR). The effects of rising sea levels endanger several nations, including Indonesia, and it potentially affects the coastal population and natural environment. Quantification is needed to determine the degree of vulnerability experienced by a coast since measuring vulnerability is a fundamental phase towards effective risk reduction. Therefore, the main objective of this research is to identify how vulnerable the coastal zone of Bali Province by develop a Coastal Vulnerability Index (CVI) of areas exposed to the sea-level rise on regional scales using remote sensing and Geographic Information System (GIS) approaches. This study was conducted in Bali Province, Indonesia, which has a beach length of ~640 km, and six parameters were considered in the creation to measure the degree of coastal vulnerability by CVI: geomorphology, shoreline change rate, coastal elevation, sea-level change rate, tidal range, and significant wave height. The different vulnerability parameters were assigned ranks ranging from 1 to 5, with 1 indicating the lowest and 5 indicating the highest vulnerabilities. The study revealed that about 138 km (22%) of the mapped shoreline is classified as being at very high vulnerability and 164 km (26%) of shoreline is at high vulnerability. Of remaining shoreline, 168 km (26%) and 169 km (26%) are at moderate and low risk of coastal vulnerability, respectively. This study outcomes can provide an updated vulnerability map and valuable information for the Bali Province coast, aimed at increasing awareness among decision-makers and related stakeholders for development in mitigation and adaptation strategies. Additionally, the result may be utilized as basic data to build and implement appropriate coastal zone management.

Online social media crowdsourced photos contain a vast amount of visual information about the physical properties and characteristics of the earth’s surface. Flickr is an important online social media platform for users seeking this information. Each day, users generate crowdsourced geotagged digital imagery containing an immense amount of information. In this paper, geotagged Flickr images are used for automatic extraction of low-level land use/land cover (LULC) features. The proposed method uses a naive Bayes classifier with color, shape, and color index descriptors. The classified images are mapped using a majority filtering approach. The classifier performance in overall accuracy, kappa coefficient, precision, recall, and f-measure was 87.94%, 82.89%, 88.20%, 87.90%, and 88%, respectively. Labeled-crowdsourced images were filtered into a spatial tile of a 30 m × 30 m resolution using the majority voting method to reduce geolocation uncertainty from the crowdsourced data. These tile datasets were used as training and validation samples to classify Landsat TM5 images. The supervised maximum likelihood method was used for the LULC classification. The results show that the geotagged Flickr images can classify LULC types with reasonable accuracy and that the proposed approach improves LULC classification efficiency if a sufficient spatial distribution of crowdsourced data exists.

Earthquakes are one of the most devastating types of natural disasters, and happen with little to no warning. This study combined Landsat-8 and interferometric ALOS-2 coherence data without training area techniques by classifying the remote sensing ratios of specific features for damage assessment. Waterbodies and highly vegetated areas were extracted by the modified normalized difference water index (MNDWI) and normalized difference vegetation index (NDVI), respectively, from after-earthquake images in order to improve the accuracy of damage maps. Urban areas were classified from pre-event interferometric coherence data. The affected areas from the earthquake were detected with the normalized difference (ND) between the pre- and co-event interferometric coherence. The results presented three damage types; namely, damage to buildings caused by ground motion, liquefaction, and landslides. The overall accuracy (94%) of the confusion matrix was excellent. Results for urban areas were divided into three damage levels (e.g., none–slight, slight–heavy, heavy–destructive) at a high (90%) overall accuracy level. Moreover, data on buildings damaged by liquefaction and landslides were in good agreement with field survey information. Overall, this study illustrates an effective damage assessment mapping approach that can support post-earthquake management activities for future events, especially in areas where geographical data are sparse.