Accurate and frequent updates of surface water have been made possible by remote sensing technology. Index methods are mostly used for surface water estimation which separates the water from the background based on a threshold value. Generally, the threshold is a fixed value, but can be challenging in the case of environmental noise, such as shadow, forest, built-up areas, snow, and clouds. One such challenging scene can be found in Nepal where no such evaluation has been done. Taking that in consideration, this study evaluates the performance of the most widely used water indices: Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI), Modified NDWI (MNDWI), and Automated Water Extraction Index (AWEI) in a Landsat 8 scene of Nepal. The scene, ranging from 60 m to 8848 m, contains various types of water bodies found in Nepal with different forms of environmental noise. The evaluation was conducted based on measures from a confusion matrix derived using validation points. Comparing visually and quantitatively, not a single method was able to extract surface water in the entire scene with better accuracy. Upon selecting optimum thresholds, the overall accuracy (OA) and kappa coefficient (kappa) was improved, but not satisfactory. NDVI and NDWI showed better results for only pure water pixels, whereas MNDWI and AWEI were unable to reject snow cover and shadows. Combining NDVI with NDWI and AWEI with shadow improved the accuracy but inherited the NDWI and AWEI characteristics. Segmenting the test scene with elevations above and below 665 m, and using NDVI and NDWI for detecting water, resulted in an OA of 0.9638 and kappa of 0.8979. The accuracy can be further improved with a smaller interval of categorical characteristics in one or multiple scenes.

Water bodies are essential to humans and other forms of life. Identification of water bodies can be useful in various ways, including estimation of water availability, demarcation of flooded regions, change detection, and so on. In past decades, Landsat satellite sensors have been used for land use classification and water body identification. Due to the introduction of a New Operational Land Imager (OLI) sensor on Landsat 8 with a high spectral resolution and improved signal-to-noise ratio, the quality of imagery sensed by Landsat 8 has improved, enabling better characterization of land cover and increased data size. Therefore, it is necessary to explore the most appropriate and practical water identification methods that take advantage of the improved image quality and use the fewest inputs based on the original OLI bands. The objective of the study is to explore the potential of a J48 decision tree (JDT) in identifying water bodies using reflectance bands from Landsat 8 OLI imagery. J48 is an open-source decision tree. The test site for the study is in the Northern Han River Basin, which is located in Gangwon province, Korea. Training data with individual bands were used to develop the JDT model and later applied to the whole study area. The performance of the model was statistically analysed using the kappa statistic and area under the curve (AUC). The results were compared with five other known water identification methods using a confusion matrix and related statistics. Almost all the methods showed high accuracy, and the JDT was successfully applied to the OLI image using only four bands, where the new additional deep blue band of OLI was found to have the third highest information gain. Thus, the JDT can be a good method for water body identification based on images with improved resolution and increased size.

With over 6000 rivers and 5358 lakes, surface water is one of the most important resources in Nepal. However, the quantity and quality of Nepal’s rivers and lakes are decreasing due to human activities and climate change. Despite the advancement of remote sensing technology and the availability of open access data and tools, the monitoring and surface water extraction works has not been carried out in Nepal. Single or multiple water index methods have been applied in the extraction of surface water with satisfactory results. Extending our previous study, the authors evaluated six different machine learning algorithms: Naive Bayes (NB), recursive partitioning and regression trees (RPART), neural networks (NNET), support vector machines (SVM), random forest (RF), and gradient boosted machines (GBM) to extract surface water in Nepal. With three secondary bands, slope, NDVI and NDWI, the algorithms were evaluated for performance with the addition of extra information. As a result, all the applied machine learning algorithms, except NB and RPART, showed good performance. RF showed overall accuracy (OA) and kappa coefficient (Kappa) of 1 for the all the multiband data with the reference dataset, followed by GBM, NNET, and SVM in metrics. The performances were better in the hilly regions and flat lands, but not well in the Himalayas with ice, snow and shadows, and the addition of slope and NDWI showed improvement in the results. Adding single secondary bands is better than adding multiple in most algorithms except NNET. From current and previous studies, it is recommended to separate any study area with and without snow or low and high elevation, then apply machine learning algorithms in original Landsat data or with the addition of slopes or NDWI for better performance.

The study deals with the application of Google Earth Engine (GEE), Landsat data and ensemble-learning methods (ELMs) to map land cover (LC) change over a decade in the Kaski district of Nepal. As Nepal has experienced extensive changes due to natural and anthropogenic activities, monitoring such changes are crucial for understanding relationships and interactions between social and natural phenomena and to promote better decision-making. The main novelty lies in applying the XGBoost classifier for LC mapping over Nepal and monitoring the decadal changes of LC using ELMs. To map the LC change, a yearly cloud-free composite Landsat image was selected for the year 2010 and 2020. Combining the annual normalized difference vegetation index, normalized difference built-up index and modified normalized difference water index, with elevation and slope data from shuttle radar topography mission, supervised classification was performed using a random forest and extreme gradient boosting ELMs. Post classification change detection, validation and accuracy assessment were executed after the preparation of the LC maps. Three evaluation indices, namely overall accuracy (OA), Kappa coefficient, and F1 score from confusion matrix reports, were calculated for all the points used for validation purposes. We have obtained an OA of 0.8792 and 0.875 for RF and 0.8926 and 0.8603 for XGBoost at the 95% confidence level for 2010 and 2020 LC maps, which are better for mountainous terrain. The applied methodology could be significant in utilizing the big earth observation data and overcoming the traditional computational challenges using GEE. In addition, the quantification of changes over time would be helpful for decision-makers to understand current environmental dynamics in the study area.

Roads are important infrastructure that brings economic development in a nation by connecting different places. But, in Nepal, many roads are very vulnerable to landslides due to various reasons. Araniko Highway is one of the most landslide affected major road in Nepal lacking study in past. In this study, landslide susceptibility mapping along Dolalghat - Kodari section in Araniko Highway, Nepal, was done by integrating Relative Frequency (RF) and Predictor Rate (PR). PR was applied to the RF to quantify the prediction ability of the conditioning factors while producing Landslide Susceptibility Index (LSI). First, landslide inventory map of 314 landslides was prepared. Then, the database was divided into 70/30 ratio for the training and validating the model. After analysing thirteen landslide conditioning factors, susceptibility map produced using LSI was categorized into five classes. Finally, overall performance of the resulting map was assessed using the receiver operating characteristic curve technique. The success rate and prediction rate curve showed that the area under the curve for RF was 0.606 and 0.581 respectively. The result of this study showed a successful mapping of landslide susceptibility by integrating RF and PR.