Flood Risk Mapping by Remote Sensing Data and Random Forest Technique

Modelling species distributions with remote sensing data: bridging disciplinary perspectives

This observation-based and explainable approach demonstrates the applicability of multi-source remote sensing for flood assessment in data-scarce regions, offering a robust scientific basis for flood management and spatial planning in monsoon-affected areas. Floods are among the most frequent and devastating natural hazards, particularly in developing countries such as Myanmar, where monsoon-driven rainfall and inadequate flood-control infrastructure exacerbate disaster impacts. This study presents a satellite-driven and interpretable framework for high-resolution flood susceptibility and risk assessment by integrating multi-source remote sensing and geospatial data with ensemble machine-learning models—Extreme Gradient Boosting (XGBoost) and Light Gradient Boosting Machine (LightGBM)—implemented on the Google Earth Engine (GEE) platform. Eleven satellite- and GIS-derived predictors were used, including the Digital Elevation Model (DEM), slope, curvature, precipitation frequency, the Normalized Difference Vegetation Index (NDVI), land-use type, and distance to rivers, to develop flood susceptibility models. The Jenks natural breaks method was applied to classify flood susceptibility into five categories across Myanmar. Both models achieved excellent predictive performance, with area under the receiver operating characteristic curve (AUC) values of 0.943 for XGBoost and 0.936 for LightGBM, effectively distinguishing flood-prone from non-prone areas. XGBoost estimated that 26.1% of Myanmar’s territory falls within medium- to high-susceptibility zones, while LightGBM yielded a similar estimate of 25.3%. High-susceptibility regions were concentrated in the Ayeyarwady Delta, Rakhine coastal plains, and the Yangon region. SHapley Additive exPlanations (SHAP) analysis identified precipitation frequency, NDVI, and DEM as dominant factors, highlighting the ability of satellite-observed environmental indicators to capture flood-relevant surface processes. To incorporate exposure, population density and nighttime-light intensity were integrated with the susceptibility results to construct a natural–social flood risk framework. This observation-based and explainable approach demonstrates the applicability of multi-source remote sensing for flood assessment in data-scarce regions, offering a robust scientific basis for flood management and spatial planning in monsoon-affected areas.

Qilian Mountains is an important water conservation area in Northwest China, which is the boundary between the first and second steps of China's topography and is sensitive to climate change. Based on the data of temperature, precipitation, normal difference vegetation index (NDVI), and digital elevation model (DEM) data, we analyzed NDVI change and its relationship with temperature and precipitation along the elevation, slope and slope aspect in the southern slope of Qilian Mountains using tendency analysis method, wavelet analysis and correlation analysis. The results showed that, from 1998 to 2017, NDVI value of the growing season presented increasing trend by a rate of 0.023·10 a-1. Changes of NDVI differed at different elevations, slopes and slope aspects. NDVI increased first and then decreased with elevation. The vegetation coverage at 2700-3700 m was good, and degraded in the area of >4700 m. NDVI reduced with the increases of slope, which showed little difference in different slope aspects but was better in sunny slope than in shade slope. NDVI of the growing season was closely related with temperature and precipitation. NDVI, temperature and precipitation in growing season all had a 14-year cycle. Vegetation at different elevations, slopes and slope aspects was differently affected by temperature and precipitation. Vegetation in areas with altitude <3700 m, >4700 m, slope <25° and each slope direction was more sensitive to precipitation.

&amp;lt;p&amp;gt;Groundwater resources are biodiversity hotspots, and provide crucial ecosystems services. Yet, groundwater dependent ecosystems (GDEs) are exposed to several anthropogenic threats, including climate and land use change. Tackling these threats requires improving the on-the-ground identification of GDEs at the global scale and especially in vulnerable areas such as the Mediterranean biome where water is scarce.&amp;lt;/p&amp;gt; &amp;lt;p&amp;gt;In order to identify the location of groundwater dependent vegetation (GDV) in the landscape and create a harmonized global map of GDV a novel multi-instrument and multi-scale approach was developed. The approach combines a geodata-based potential GDV zones-index (pGDVZ) together with high-resolution vegetative, hydrogeological and topographic remote sensing parameters.&amp;lt;/p&amp;gt; &amp;lt;p&amp;gt;The pGDVZ integrates global and openly available datasets to combine groundwater vegetation interaction, land use, soil characteristics and landscape wetness potential. The index is currently tested for the whole Mediterranean. Results can help to pinpoint areas of high GDVZ potential where regional high-resolution identification of GDV is necessary.&amp;lt;/p&amp;gt; &amp;lt;p&amp;gt;The regional GDV-mapping concept implements different criteria aiming at: 1) high vitality and wetness during dry periods (e.g., Enhanced Vegetation Index, Normalized Difference Vegetation Index, Normalized Difference Water Index), 2) low seasonal changes in vitality, 3) low interannual changes in vitality, 4) high topographic potential of water accumulation and low water table depth and 5) general topography (elevation, slope). Processing of different remote sensing data (e.g., Sentinel 1;2, Digital Elevation Models, MODIS) is performed using the Google Earth Engine. Botanical mapping as well as integration of several geodata is used for validation and calibration of derived GDV-likelihoods. Furthermore, the integration of vegetation plots extracted from sPlot, the global vegetation database introduces a novel methodology to train machine learning algorithm for classification and modelling of GDV.&amp;lt;/p&amp;gt; &amp;lt;p&amp;gt;After successfully testing the mapping approach at local scale in the &amp;amp;#8216;Cilento, Vallo di Diano and Alburni National Park&amp;amp;#8217; in Campania (Italy), a two-step upscaling methodology is currently designed to implement the concept on regional (county) and global (biome) scale.&amp;lt;/p&amp;gt;

The formation and expansion of glacial lakes, driven by ongoing climate warming, is a potentially hazardous phenomenon with significant impacts on glacier melt and regional hydrology. In regions like Central Asia, decades of glacier retreat and thinning have fostered the development and growth of various types of glacial lakes, including ice-marginal, moraine-dammed, and supraglacial lakes. These water bodies store meltwater and accelerate glacier mass loss and terminus retreat through heat exchange and intensified melt, ice calving, and ice-marginal destabilization. Once formed, glacial lakes can trigger positive feedback mechanisms that decouple their evolution from direct climate forcing, resulting in rapid glacier downwasting and retreat.Despite advances in remote sensing, accurately capturing the full size and spatial distribution of these lakes remains challenging. Current inventories, largely based on moderate-resolution imagery (Landsat, Sentinel-2), often overlook smaller lakes. These smaller lakes are a critical yet underappreciated component of the cryosphere and can expand rapidly, posing risks of Glacier Lake Outburst Floods (GLOFs).In this study, we present a comprehensive, high-resolution glacial lake inventory for Central Asia, derived from Planet&amp;#8217;s PlanetScope imagery. With a spatial resolution of approximately 3 m/pixel&amp;#8212;an order of magnitude finer than Landsat and Sentinel-2 &amp;#8212;PlanetScope data enables the delineation of lakes as small as tens of meters in size, overcoming the spatial limitations of previous satellite-based inventories. The study focuses on Central Asia, covering regions including Dzhungarsky Alatau, Tien Shan, Pamir-Alay, and Pamir.Our approach consists of four main steps: (i) Water Pixel Identification: Water pixels are detected from PlanetScope images using the Normalized Difference Water Index (NDWI) and Coloured Dissolved Organic Matter (CDOM), with thresholds determined by Otsu&amp;#8217;s algorithm. (ii) Lake Clustering: Detected water pixels are grouped into clusters, with glacial lakes defined as clusters exceeding 22 connected pixels (~200 m&amp;#178;). (iii) Boundary Refinement: Lake boundaries are further refined using an NDWI- and CDOM--based Otsu method. (iv) Lake Inventory Compilation: A detailed inventory is produced, including geographical coordinates, elevation, area, and NDWI statistics for each identified lake. The algorithm is implemented within the Google Earth Engine (GEE) environment, enhancing computational efficiency and minimizing the need for local data storage.The new Central Asia Glacial Lake Inventory (CAGLI) comprises approximately 14,000 water bodies, each with an area exceeding 200 m&amp;#178; with a combined area of 470 km&amp;#178; as of August-September 2023. This dataset allows for a detailed characterization of lake size distribution in the region, efficient monitoring of lake formation and growth, and analysis of the impact of surface ponds on glacier evolution. It also enhances the delineation of debris-covered glaciers, where water pond formation is more likely than on non-glacierized terrain. Importantly, the new algorithm provides practitioners in Central Asia and other glacierized mountain regions with a highly efficient tool for monitoring lake changes, supporting early warning systems and risk reduction strategies.

Spatial analysis of land use and cover changes: Implications of green legacy initiative on climate action in Upper Awash Basin, Ethiopia

Digital elevation models (DEMs) are fundamental data infrastructure that support a large range of applications in environmental modelling. The distribution of solar radiation, patterns of wetness and water movement, mobility of animals and evolution of soils are all intimately linked to topographic form. DEMs for the Australian continent have progressed from 1/10 th degree resolution in 1982 to 1/40 th degree in 1991, 9 th degree) in 1996 and now 1 second (about 30 m) derived from the NASA Shuttle Radar Topographic Mission (SRTM). Each improvement in resolution and accuracy enables new applications as more detail and structure are revealed. The 1 second SRTM-derived DEMs for Australia are now publicly available through the Geoscience Australia National Elevation Data Framework (NEDF) portal (http://nedf.ga.gov.au). The raw SRTM data has been processed through a series of steps to provide several products suitable for different purposes: • DEM-S is a bare-earth elevation model with stripes and voids removed, offsets due to trees treated and noise substantially reduced in flatter areas by adaptive smoothing, which is suitable for most applications where connectivity of flow is not the primary concern such as calculation of slopes and topographic position • DEM-H is a hydrologically conditioned elevation model derived from DEM-S with drainage enforcement using 1:250,000 scale stream lines, which is suitable for applications where hydrological connectivity is of primary concern such as calculation of contributing areas and catchment delineation • DEM is the pre-cursor to DEM-S before adaptive smoothing, which may be useful when the retention of detail is more important than smoothness. There is also a cleaned digital surface model (DSM) that is the precursor to DEM-S before tree offsets are removed, which is restricted to Australian government use only. Most users of DEMs need information about the shape of the landscape, not just height, so we are producing a set of standard derived products including slope, aspect, curvatures, topographic position, relief, contributing area and wetness index. These derived surfaces are key inputs for continental scale prediction of soil properties and ecological patterns. Predictions of a suite of key soil properties on a 3 second grid for the Australian continent, using the 1 second terrain layers, are being developed and will contribute to the GlobalSoilMap.net global grid of soil properties. Catchments and stream lines derived from DEM-H will form the basis for future versions of the Australian Hydrological Geospatial Fabric (http://www.bom.gov.au/water/geofabric/index.shtml). Digital elevation data continues to improve as measurement technology improves. Airborne lidar already provides high precision terrain data with vertical accuracy of 20 cm or better at 1 m horizontal resolution or better. This data is available in parts of Australia where the better accuracy justifies the high cost, such as urban areas, coastal lowlands, wetlands and floodplains. We need to develop ways of using high resolution data covering limited areas in combination with lower resolution data covering the continent. New global DEMs are being acquired and will replace SRTM in the years ahead. The TANDEM-X satellite radar system will potentially produce global elevations at 2-3 times better resolution and accuracy than SRTM, and other systems will follow. Higher resolutions bring further technical challenges, not least in data volume: each 1 second DEM product for Australia is a 40 GB dataset and finer resolutions will be correspondingly larger.

Intensifying climate change and rapid urbanization pose a severe flood risk to India's major cities. Conventional flood assessment methods, however, are often geographically limited and fail to quantify critical uncertainties, hindering effective disaster management. To address this gap, we introduce the ALERT (Advanced Learning for Enhanced Flood Risk Prediction) framework, a scalable and integrated solution for flood analysis across the chosen Indian mega cities implemented entirely within Google Earth Engine (GEE). The framework leverages multi-source remote sensing data—including Sentinel-1 SAR and Sentinel-2 optical imagery—to perform robust Land Use/Land Cover (LULC) classification using a Random Forest algorithm. Its core innovation is a Bayesian probabilistic model that integrates SAR-derived floodwater evidence with topographic wetness index (TWI) and precipitation (ERA5) data to generate per-pixel flood probability maps, explicitly quantifying prediction uncertainty. Across five diverse megacities (Chennai, Mumbai, Bengaluru, Kolkata, Gurugram), the Random Forest classifier achieved LULC accuracies exceeding ∼90%. Critically, integrating this LULC context reduced flood detection false positives by ∼10-15% compared to traditional SAR-only methods. The Bayesian probabilistic approach further improved flood mapping accuracy by ∼3-5% over deterministic thresholding, providing a more reliable assessment of inundation risk. The validated ALERT framework delivers a replicable and scalable solution for dynamic flood monitoring, offering critical insights for risk-informed urban planning and enhancing climate resilience in vulnerable urban regions globally, with demonstrated applicability across strategically selected Indian megacities. • Random Forest identified as optimal for LULC classification in five diverse Indian cities (∼90% overall accuracy). • LULC integration quantitatively shown to reduce flood detection false positives by approximately 10-15%. • Novel Bayesian probabilistic flood mapping in GEE improves accuracy by ∼3-5% over deterministic methods. • Framework explicitly quantifies uncertainty and assesses flood risk under a +10% precipitation scenario. • Delivers scalable tools for operational flood monitoring and risk-informed urban planning.

Increasing the occurrence of floods, especially in cities, and the risks to human, financial, and environmental risks due to its, make flood risk zoning of great importance. The purpose of this study is to estimate the flood risk of the Maneh and Samalghan based on determining effective criteria and spatial and non-spatial data-driven models. The criteria used in this research include Modified Fournier Index, Topographic Position Index, Curve Number, Flow Accumulation, Slope, Digital elevation model, Topographic Wetness Index, Vertical Overland Flow Distance, Horizontal Overland Flow Distance, and Normalized difference vegetation index. The novelty of this study is to present new combination approaches to determine the effective criteria in flood risk zoning (Maneh and Samalghan). In this regard, the geographically weighted regression (GWR) with exponential and bi-square kernels and artificial neural network (ANN) combined with a binary particle swarm optimization algorithm (BPSO). The best value of the fitness function (1-R2) for ANN, GWR with the exponential kernel, and GWR with bi-square kernel was obtained 0.1757, 0.0461, and 0.0097, respectively, Which indicates higher compatibility of the bi-square kernel than the other models. It was also found that the criteria used have a significant effect on the rate of flooding in the study area.

Droughts are common across various climates, typically caused by prolonged decreases in rainfall. Several factors contribute to drought, including the temperature, wind speed, and relative humidity and the timing, amount, and intensity of rainfall during the growing season. This study introduces the Soil Moisture and Evapotranspiration Revealed Drought Index (SERDI), a new index that combines soil moisture and evapotranspiration (calculated using the Penman–Monteith method) to enhance drought early warning systems. To validate the SERDI, we compared it with other established indices such as the Land Surface Temperature (LST), Vegetation Health Index (VHI), Normalized Difference Vegetation Index (NDVI), and Normalized Difference Water Index (NDWI), using metrics like the R-squared (R2), root mean square error (RMSE), mean absolute percentage error (MAPE), and p-value to assess the accuracy, data variability, and forecast conditions. The results showed a low RMSE and high R2 between the SERDI and the LST and VHI, indicating a strong correlation. However, weaker correlations were observed between the SERDI and NDVI/NDWI, as shown by the lower R2 and higher RMSE values in semi-arid areas. Regions across Iran, Iraq, Syria, Jordan, and Israel experienced mostly moderate to severe drought conditions, with a few areas in Iran and Syria showing normal conditions. The SERDI’s strong correlation with the LST and moderate correlation with the VHI can be attributed to the direct influence of the soil moisture and evapotranspiration on the surface temperature and vegetation health. On the other hand, the weaker correlation with the NDVI and NDWI is due to variability in the vegetation response, irrigation practices, and regional differences. This study concludes that the SERDI is an effective tool for the detection of drought based on soil moisture and evapotranspiration.

Este estudo teve por objectivo avaliar o potencial das imagens de satélite para monitorizar a qualidade das pastagens e a humidade na camada superficial do solo no ecossistema montado. O ensaio decorreu entre 2016 e 2018 numa parcela experimental da Herdade da Mitra, em Valverde (Évora). Foram identificados 12 pixeis de amostragem do “Sentinel-2” (área de 10m × 10m, sem árvores), onde se procedeu mensalmente, durante o período de Primavera, à recolha de amostras de pastagem e à extracção dos valores dos índices “Normalized Difference Vegetation Índex” (NDVI) e “Normalized Difference Water Índex” (NDWI). Entre Janeiro de 2017 e Junho de 2018 foram também registados mensalmente os valores da humidade na camada superficial do solo (0-0,20m). Os resultados obtidos mostraram correlações significativas entre: (i) o NDVI e o NDWI; (ii) entre ambos os índices e os parâmetros de qualidade da pastagem (proteína bruta, PB, “Neutral Detergent Fiber”, NDF, humidade, HP e índice de degradação da qualidade da pastagem, IDQP); entre o NDWI e a humidade superficial do solo (HS). Este conhecimento confirma o interesse das imagens de satélite para efeitos de gestão das pastagens e do encabeçamento animal, particularmente em termos de monitorização das necessidades de suplementação alimentar dos animais no período crítico entre o final da Primavera e o Verão.

The suitability of using Moderate Resolution Imaging Spectroradiometer (MODIS) images for surface soil moisture estimation to investigate the importance of soil moisture in different applications, such as agriculture, hydrology, meteorology and natural disaster management, is evaluated in this study. Soil moisture field measurements and MODIS images of relevant dates have been acquired. Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI) and Normalized Difference Water Index (NDWI) are calculated from MODIS images. In addition, MODIS Land Surface Temperature (LST) data (MOD11A1) are used in this analysis. Four different soil moisture estimation models, which are based on NDVI–LST, EVI–LST, NDVI–LST–NDWI and EVI–LST–NDWI, are developed and their accuracies are assessed. Statistical analysis shows that replacing EVI with NDVI in the model that is based on LST and NDVI increases the accuracy of soil moisture estimation. Accuracy evaluation of soil moisture estimation using check points shows that the model based on LST, EVI and NDWI values gives a higher accuracy than that based on LST and EVI values. It is concluded that the model based on the three indices is a suitable model to estimate soil moisture through MODIS imagery.

In recent decades, the phenomenon of drought has become a hazard with increasing frequency, with multiple societal and environmental impacts. One of these impacts concerns water resources and their availability for various uses. Numerous indices have been used over time to quantify the severity of drought and to assess its effects on socio-economic activities and environmental components. Among the spatial indices, the most numerous belong to remote sensing, being easy to use to analyse drought consequences especially on landcover and vegetation. However, regarding the hydrological drought, the indices used are mainly calculated based on hydroclimatic data, without taking into account spatial variables, such as the topographical, geological or pedological characteristics. The aim for this study is to compute a hydrological drought index which integrates several drought control variables, using both GIS and remote sensing techniques in order to map the susceptibility to hydrological drought within the Teleorman watershed.Located in the central-southern part of Romania, the Teleorman River has a length of 169 km and a catchment area of 1.427 km2. The most part of &amp;#160;the catchment overlaps the central sector of the Romanian Plain, an important agricultural area, highly sensitive to water deficit. According to K&amp;#246;ppen-Geiger classification, the analyzed catchment has a humid continental climate with hot summers (Dfa), meaning that the drought could occur in the basin.A series of free data and information sources has been accessed in order to compute the hydrological drought index, such as: Worldclim, Landsat Archive, Geological Map of Romania, Pedological Map of Romania, Shuttle Radar Topography Mission (SRTM), Topographical Map of Romania. The following parameters were derived from these sources: Topograhic Wetness Index (TWI); Drainage Density (resulted from hydrographic network); Normalized Difference Drough Index (NDDI), resulted from ratio between Normalized Difference Vegetation Index (NDVI) and Normalized Difference Water Index (NDWI); Temperature Condition Index (TCI), extracted from Land Surface Temperature (LST); Aridity Index (AI) computed as a ratio between Precipitation and Potential Evapotranspiration (PET); De Martonne Index (based on ratio between Precipitation and Air temperature); Lithology and Soil Texture. Because some of the parameters had a different spatial resolution, the regridding method was used to bring the database to a resolution of 30 meters. Analytic hierarchy process method (AHP) was used to determine the influence of each factor and for the bonitation process. Based on the total obtained score, 5 classes from to lowest to highest hydrological drought susceptibility resulted. Finally, the Weighted Overlay and Raster Calculator tools from ArcGisPro software were used to map the index.The resulted map allows the identification in the studied watershed of areas the most susceptible to hydroclimatic drought allowing the focus in these areas of appropriate actions to improve drought risk management. GIS and Remote sensing proved to be useful tools in spatial analysis of drought based on a composite index integrating several drought control factors. In the future, we intend to improve the method by considering other variables controlling the hydrological drought, such as the streamflow and the groundwater depth.

Spatio-temporal flood frequency dynamics in Kaziranga National Park using multi-year Sentinel-1 SAR and vegetation indices

Normalized Difference Vegetation Index (NDVI) and Normalized Difference Water Index (NDWI) were compared for monitoring live fuel moisture in a shrubland ecosystem. Both indices were calculated from 500 m spatial resolution Moderate Resolution Imaging Spectroradiometer (MODIS) reflectance data covering a 33‐month period from 2000 to 2002. Both NDVI and NDWI were positively correlated with live fuel moisture measured by the Los Angeles County Fire Department (LACFD). NDVI had R 2 values ranging between 0.25 to 0.60, while NDWI had significantly higher R 2 values, varying between 0.39 and 0.80. Water absorption measures, such as NDWI, may prove more appropriate for monitoring live fuel moisture than measures of chlorophyll absorption such as NDVI.