Multi-source Satellite Imagery Research Articles

Enhancing flood susceptibility modeling using integration of multi-source satellite imagery and multi-input convolutional neural network

Enhancing flood susceptibility modeling using integration of multi-source satellite imagery and multi-input convolutional neural network

Flood inundation monitoring using multi-source satellite imagery: a knowledge transfer strategy for heterogeneous image change detection

Flood emergency mapping is essential for flood management, often requiring near real-time extraction of large-scale flood extents by combining pre- and post-event multi-source remote sensing images. Pre-event optical imagery delineates the normal water extent, providing a benchmark for estimation of post-flood changes. Synthetic aperture radar (SAR) imagery provides rapid and accurate interpretation of flood inundation extent during the rainy period. Post-classification comparison is one of the fundamental methods for flood mapping of multi-source imagery, as the image characteristics of SAR and optical imagery are inherently different. However, the accuracy of flood mapping depends on the accuracy of water body delineation from single temporal imagery, which can lead to error propagation in flood extent determination. Change detection methods can extract flood extent directly from pre- and post-event imagery, but the models need to learn a consistent flood feature description between optical and SAR imagery from a limited flood dataset. Here, we propose an automatic method for flood inundation extent extraction using pre- and post-event multi-source imagery that requires only a small number of pseudo-change labels. The methodology adopts a pseudo-Siamese network framework as a Heterogeneous - Flood Inundation Extraction Network (H-FIENet), to detect flood extent between the pre‐ and post-event heterogeneous images. Spatially inconsistent multi-source pre- and post-event imagery was used to create a pseudo flood dataset, and this dataset was used to transfer water body feature knowledge from the pre-trained model to the flood extraction model using a cross-task knowledge transfer strategy. Pre- and post-event images from different satellite sources and different flood scenarios were used to evaluate the performance of H-FIENet. We found that: (1) The methodology produced an overall accuracy of over 0.94 for flood inundation extraction from multi-source heterogeneous pre- and post-event imagery. (2) H-FIENet can detect both expansion and recession of the flood extent using any dual-temporal imagery. Our work makes it possible to automate time-series flood monitoring from multi-source remote sensing imagery.

Combined use of multi-source satellite imagery and deep learning for automated mapping of glacial lakes in the Bhutan Himalaya

Himalayan glacial lakes have been rapidly developing and expanding in recent decades under climate change and glacier mass loss. These growing glacial lakes can produce glacial lake outburst floods (GLOFs) events with far-reaching and devastating consequences. However, the latest spatial distribution and temporal evolution of the Himalayan glacial lakes is not timely updated due to the inaccessibility of high mountain areas and the lack of an effective automated mapping method that can leverage the availability of wide-ranging remote sensing data. To frequently update glacial lake inventory in GLOF-vulnerable regions, we developed the state-of-the-art glacial lake mapping approaches based on deep learning technique and multi-source remote sensing imagery. DeepLabv3+, an advanced semantic segmentation algorithm, was trained to delineate glacial lakes with areas larger than 0.005 km2 from multi-source imagery and their derivatives, including PlanetScope red-green-blue (RGB), PlanetScope-derived Normalized Difference Water Index (NDWI), Sentinel-2 RGB, Sentinel-2-derived NDWI, Sentinel-1 Synthetic Aperture Radar (SAR), and Landsat-8 RGB images. The well-trained deep learning models achieved high mapping accuracy in the northern Bhutan test region, with the F1 score varying from 0.74 (Sentinel-1) to 0.91 (Planet-RGB) among the six types of images. We applied the well-trained models to automatically map the glacial lakes from multi-source satellite imagery. After manually cataloging the mapping results, we compiled a glacial lake inventory for the Bhutan Himalaya in 2021 that includes 2563 glacial lakes with a total area of 153.85 ± 9.33 km2. Our results demonstrated the mapping capability of deep learning on multiple satellite imagery, the key roles of PlanetScope optical images for accurate glacial lake mapping, and the essential supplementary usage of SAR images and NDWI images to complement the glacial lake inventory over Bhutan Himalaya. This study provides an advanced and transferable workflow for inventorying glacial lakes from multi-source satellite imagery, as well as provides a high-quality and comprehensive glacial lake inventory for outburst flood studies.

Retraction Note: APO-AN feature selection based Glorot Init Optimal TransCNN landslide detection from multi source satellite imagery

Retraction Note: APO-AN feature selection based Glorot Init Optimal TransCNN landslide detection from multi source satellite imagery

Application of Multi-Temporal and Multisource Satellite Imagery in the Study of Irrigated Landscapes in Arid Climates

The study of ancient irrigation is crucial in the archaeological research of arid regions. It covers a wide range of topics, with the Near East being the focus for decades. However, political instability and limited data have posed challenges to these studies. The primary objective is to establish a standardised method applicable to different arid environments using the Google Earth Engine platform, considering local relief of terrain and seasonal differences in vegetation. This study integrates multispectral data from LANDSAT 5, Sentinel-2, SAR imagery from Sentinel 1, and TanDEM-X (12 m and 30 m) DSMs. Using these datasets, calculations of selected vegetation indices such as the SMTVI and NDVSI, spectral decomposition methods such as TCT and PCA, and topography-based methods such as the MSRM contribute to a comprehensive understanding of landscape irrigation. This paper investigates the influence of modern environmental conditions on the visibility of features like levees and palaeo-channels by testing different methods and parameters. This study aims to identify the most effective approach for each case study and explore the possibility of applying a consistent method across all areas. Optimal results are achieved by combining several methods, adjusting seasonal parameters, and conducting a comparative analysis of visible features.

Optimizing Malaria Control: Granular and Cost-Effective Mosquito Habitat Index in Endemic Areas Through Satellite Imagery

Malaria, classified as a tropical disease under the Sustainable Development Goals (SDGs) indicator 3.3, remains a significant global health challenge. In this study, by taking advantage of multiple spectral composite indexes of multisource satellite imagery to capture various geospatial features relevant to the suitability of marsh mosquito habitat, we introduced the Mosquito Habitat Suitability Index (MHSI) to assess potential Anopheles mosquito breeding sites in terms of the vegetation density, water bodies, environment temperature, and humidity in any particular areas. The MHSI integrates the publicly accessible granular level of the normalized difference vegetation index, water index, land surface temperature, and moisture index from cost-effective low and medium-resolution optical satellite data. We focus on West Papua Province, Indonesia, known for diverse ecological conditions and varying malaria prevalence, as a case study area. From the built index, the risk zone map is then formed with the K-Means algorithm. One key finding is the elevated risk in Fakfak Regency, demanding particular attention, as its high-risk area represents 45% of its total. This research aids localized decision-making to combat malaria's unique challenges in West Papua Province which are relevant for implementation in other regions, contributing to SDG-aligned interventions for malaria eradication by 2030.

Automated In-Season Crop-Type Data Layer Mapping Without Ground Truth for the Conterminous United States Based on Multisource Satellite Imagery

Automated In-Season Crop-Type Data Layer Mapping Without Ground Truth for the Conterminous United States Based on Multisource Satellite Imagery

Geospatial Big Data Approaches to Estimate Granular Level Poverty Distribution in East Java, Indonesia using Machine Learning and Deep Learning Regressions

One of the economic development the focus of the Indonesian government's efforts is for reducing poverty. In Indonesia, collecting poverty data uses the conventional method, the name is National Socio-Economic Survey (SUSENAS) which takes a large cost, time, and effort. To overcome these limitations, there is a need for additional data to provide more detailed poverty data. Recent studies show that the use of geospatial big data could identify poverty at a granular level, with a lower cost and faster update because of their unique and unbiased capacity to identify physical and socioeconomic phenomena. The integrated multi-source satellite imagery data such as the normalized difference vegetation index (NDVI) for detecting rural areas based on vegetation, built-up index (BUI) for identifying urban areas through building distribution, normalized difference water index (NDWI) for land cover detection, day time land surface temperature (LST) for identifying urban regions based on surface temperature, and pollutants such as carbon monoxide (CO), nitrogen dioxide (NO2), and sulfur dioxide (SO2) to evaluate economic activities based on pollution. Additionally, point of interest (POI) density and minimum POI distance are used to measure area accessibility. Therefore, the contribution of this research is to implement the utilization of geospatial big data to estimate the numbers of poverties at a granular level to the 666 sub-districts in East Java Province using machine learning and deep learning regression models. The evaluation results to estimate sub-district level poverty shows that the best model development using Support Vector Regression (SVR) in machine learning was the best root mean squared error (RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE) values of 0.365, 0.293, and 0.032 with R-squared of 0.59 and MLP in deep learning algorithm with 0.444, 0.345, and 0.039 values of RMSE, MAE, and MAPE with R2 0.52. In addition, the results of visual identification revealed that high estimates of lower poverty are typically found in urban areas with high accessibility, and these areas are not spatially deprived areas with limited accessibility.

Integrating Satellite Imageries and Multiple Geospatial Big Data for Granular Mapping of Spatial Distribution of Human Development Index in East Java, Indonesia

The availability of data on the Human Development Index (HDI) is crucial as a gauge of regional performance, particularly in terms of assessing the development of human resources. In Indonesia, the collecting of HDI data usesthe conventional method, such as undirect estimation, National Socio-Economic Survey (SUSENAS), The Ministry of Religion, or inventory of sectoral data that used the large cost, time, and effort. Additional data are required to provide more detailed poverty data at a lower cost and with more recent information to overcome these limitations. According to recent studies, the quality of life for measuring HDI can be identified down to the granular level using geospatial big data. Therefore, the contribution of this research is to implement the use of geospatial big data, such as integrated multi-source satellite imagery data and Point of Interest (POI). Besides that, this study develops the relative spatial human development index in 11 km x 11 km resolution for the granular mapping of the quality of life to measure the HDI in East Java, Indonesia. The kinds of weighted sum models used in this study such as equal weight (EWS), Pearson (PCCWS), Spearman (SCCWS) correlation-based weight, and Principal Component Analysis (PCA)-based weight (PCAWS). The best RSHDI PCCWS for representing the human development index in East Java in 2022, which was determined using a weight-sum model based on Pearson correlation, has a correlation coefficient of 0.7858 (p-value = 5.078 x 10-9) and is highly correlated with official HDI data. The use of this RSHDI as a predictor variable in the estimation of HDI data shows the ideal model had an RMSE of 3.098% and an R2 of up to 61.75% using RSHDI PCCWS. According to the findings of the descriptive analysis of this map, areas with low RSHDI scores typically in some regencies areas in Madura Island and the east area of East Java with geographically depressed, while areas with high RSHDI scores typically have dense populations and have better accessibility such as urban area in Surabaya and Kota Malang. As a result, the official human development index data can be supported by the RSHDI's ability to map spatially deprive areas

Rapid Land Cover Classification Using a 36-Year Time Series of Multi-Source Remote Sensing Data

Long time series land cover classification information is the basis for scientific research on urban sprawls, vegetation change, and the carbon cycle. The rapid development of cloud computing platforms such as the Google Earth Engine (GEE) and access to multi-source satellite imagery from Landsat and Sentinel-2 enables the application of machine learning algorithms for image classification. Here, we used the random forest algorithm to quickly achieve a time series land cover classification at different scales based on the fixed land classification sample points selected from images acquired in 2022, and the year-by-year spectral differences of the sample points. The classification accuracy was enhanced by using multi-source remote sensing data, such as synthetic aperture radar (SAR) and digital elevation model (DEM) data. The results showed that: (i) the maximum difference (threshold) of the sample points without land class change, determined by counting the sample points of each band of the Landsat time series from 1986 to 2022, was 0.25; (ii) the kappa coefficient and observed accuracy of the same sensor from Landsat 8 are higher than the results of the TM and ETM+ sensor data from 2013 to 2022; and (iii) the addition of a mining land cover type increases the kappa coefficient and overall accuracy mean values of the Sentinel 2 image classification for a complex mining and forest area. Among the land classifications via multi-source remote sensing, the combined variables of Spectral band + Index + Terrain + SAR result in the highest accuracy, but the overall improvement is limited. The method proposed is applicable to remotely sensed images at different scales and the use of sensors under complex terrain conditions. The use of the GEE cloud computing platform enabled the rapid analysis of remotely sensed data to produce land cover maps with high accuracy and a long time series.

Landscape Pattern of Sloping Garden Erosion Based on CSLE and Multi-Source Satellite Imagery in Tropical Xishuangbanna, Southwest China

Inappropriate soil management accelerates soil erosion and thus poses a serious threat to food security and biodiversity. Due to poor data availability and fragmented terrain, the landscape pattern of garden erosion in tropical Xishuangbanna is not clear. In this study, by integrating multi-source satellite imagery, field investigation and visual interpretation, we realized high-resolution mapping of gardens and soil conservation measures at the landscape scale. The Chinese Soil Loss Equation (CSLE) model was then performed to estimate the garden erosion rates and to identify critical erosion-prone areas; the landscape pattern of soil erosion was further discussed. Results showed the following: (1) For the three major plantations, teas have the largest degree of fragmentation and orchards suffer the highest soil erosion rate, while rubbers show the largest patch area, aggregation degree and soil erosion ratio. (2) The average garden erosion rate is 1595.08 t·km−2a−1, resulting in an annual soil loss of 9.73 × 106 t. Soil erosion is more susceptible to elevation and vegetation cover rather than the slope gradient. Meanwhile, irreversible erosion rates only occur in gardens with fraction vegetation coverage (FVC) lower than 30%, and they contribute 68.19% of total soil loss with the smallest land portion, indicating that new plantations are suffering serious erosion problems. (3) Garden patches with high erosion intensity grades and aggregation indexes should be recognized as priorities for centralized treatment. For elevations near 1900 m and lowlands (<950 m), the decrease in the fractal dimension index of erosion-prone areas indicates that patches are more regular and aggregated, suggesting a more optimistic conservation situation.

Multi-Factors Synthetically Contribute to Ulva prolifera Outbreaks in the South Yellow Sea of China

In recent years, the frequent outbreaks of Ulva prolifera in the South Yellow Sea have become the largest-scale green tide disasters in the world. The causes of its outbreaks have garnered widespread attention, particularly the coupled effects of multiple factors. Leveraging the Google Earth Engine (GEE) platform, this study conducted a long-term investigation of the Yellow Sea green tide disaster and the factors using multi-source satellite imagery. Finally, the combined effects of natural environmental changes and human activities on Ulva prolifera were analyzed by redundancy analysis (RDA) and variation partitioning analysis (VPA). The results indicate: (1) Since 2018, the scale of Ulva prolifera in the South Yellow Sea has shown a distinct “biennial” trend. (2) Regarding environmental factors, SST, PAR, precipitation, and windspeed have certain effects on the growth of Ulva prolifera. However, they cannot be considered as determining factors for the outbreak of Ulva prolifera (0.002 < R2 < 0.14). Regarding anthropogenic factors, the recycle time of Pyropia yezoensis culture rafts has a relatively minor influence on the extent of Ulva prolifera. There exists a certain positive correlation (R2 = 0.45) between the human footprint index (HFI) of Jiangsu Province and the annual variation in Ulva prolifera area in the South Yellow Sea. (3) The combined effects of multiple factors influence green tide outbreaks. The Ulvatotal explanatory power of SST, PAR, precipitation, windspeed, HFI, and the recycle time of Pyropia yezoensis culture rafts for the annual variation in the Ulva prolifera area is 31.8%, with these factors interweaving and mutually influencing each other. This study offers important insights into quantifying the driving forces behind Ulva prolifera in the South Yellow Sea, providing valuable information for a deeper comprehension of the complexity of marine ecosystems and sustainable management.

RETRACTED ARTICLE: APO-AN feature selection based Glorot Init Optimal TransCNN landslide detection from multi source satellite imagery

RETRACTED ARTICLE: APO-AN feature selection based Glorot Init Optimal TransCNN landslide detection from multi source satellite imagery

Attention Mechanism-Combined LSTM for Grain Yield Prediction in China Using Multi-Source Satellite Imagery

Grain yield prediction affects policy making in various aspects such as agricultural production planning, food security assurance, and adjustment of foreign trade. Accurately predicting grain yield is of great significance in ensuring global food security. This paper is based on the MODIS remote sensing image data products from 2010 to 2020, and adds band information such as vegetation index and temperature to form composite remote sensing data as a dataset. Aiming at the lack of models for large-scale forecasting and the need for human intervention in traditional models, this paper proposes a grain production estimation model based on deep learning. First, image cropping and yield mapping techniques are used to process the data to generate training samples. Then the channel and spatial attention mechanism (convolutional block attention module, CBAM) is added to extract spatial information in different remote sensing bands to improve the efficiency of the model. Long short-term memory (LSTM) neural networks are added to obtain feature information in the time dimension. Finally, a national-scale grain yield prediction model is constructed. After the study, it was found that the LSTM model using a combination of multi-source satellite images and an attention mechanism can effectively predict grain yield in China. Furthermore, the proposed model was tested on data from 2018 to 2020 showing an average R2 of 0.940 and an average RMSE of 80,020 tons, indicating that it can predict Chinese grain yield better. The model proposed in this paper extracts grain yield information directly from the composite remote sensing data, and solves the problem of small-scale research and imprecise yield prediction in an end-to-end manner.

Emerging Issues in Mapping Urban Impervious Surfaces Using High-Resolution Remote Sensing Images

Urban impervious surface (UIS) is a key parameter in climate change, environmental change, and sustainability. UIS extraction has been evolving rapidly in the past decades. However, high-resolution impervious surface mapping is a long-term need. There is an urgent requirement for impervious surface mapping from high-resolution remote sensing imagery. In this paper, we compare current extraction methods in terms of extraction units and extraction models and summarize their strengths and limitations. We discuss the challenges in impervious surface estimation from high spatial resolution remote sensing imagery in terms of selection of spatial resolution, spectral band, and extraction method. The uncertainties caused by clouds and snow, shadows, and vegetation occlusion are also analyzed. Automated sample labeling and remote sensing domain knowledge are the main directions in impervious surface extraction using deep learning methods. We should also focus on using continuous time series of high-resolution imagery and multi-source satellite imagery for dynamic monitoring of impervious surfaces.

The role of harmonized Landsat Sentinel-2 (HLS) products to reveal multiple trajectories and determinants of cropland abandonment in subtropical mountainous areas

Cropland abandonment is a widespread land-change process globally, which can stem from the accelerated outmigration of the population from rural to urban areas, socio-economic and political changes, catastrophes, and other trigger events. Clouds limit the utility of optical satellite data to monitor cropland abandonment in highly fragmented mountain agricultural landscapes of tropical and subtropical regions, including the south of China. Taking Nanjing County of China as an example, we developed a novel approach by utilizing multisource satellite (Landsat and Sentinel-2) imagery to map multiple trajectories of cropland abandonment (transitioning from cropland to grassland, shrubs and forest) in subtropical mountainous landscapes. Then, we employed a redundancy analysis (RDA) to identify the spatial association of cropland abandonment considering agricultural productivity, physiography, locational characteristics and economic factors. Results indicate the great suitability of harmonized Landsat 8 and Sentinel-2 images to distinguish multiple trajectories of cropland abandonment in subtropical mountainous areas. Our framework of mapping cropland abandonment resulted in good producer's (78.2%) and user's (81.3%) accuracies. The statistical analysis showed 31.85% of croplands cultivated in 2000 were abandoned by 2018, and more than a quarter of townships experienced cropland abandonment with high abandoned rates (>38%). Cropland abandonment mainly occurred in relatively unfavorable areas for agricultural production, for instance with a slope above 6°. Slope and the proximity to the nearest settlement explained 65.4% and 8.1% of the variation of cropland abandonment at the township level, respectively. The developed approaches on both mapping cropland abandonment and modeling determinants can be highly relevant to monitor multiple trajectories of cropland abandonment and ascribe their determinants not only in mountainous China but also elsewhere and thus promote the formulation of land-use policies that aim to steer cropland abandonment.

A Multi-Sensor Satellite Approach to Characterize the Volcanic Deposits Emitted during Etna’s Lava Fountaining: The 2020–2022 Study Case

Between December 2020 and February 2022, the South East Crater of Etna has been the source of numerous eruptions, mostly characterized by the emission of lava fountains, pyroclastic material and short-lasting lava flows. Here we estimate the volume and distribution of the lava deposits by elaborating multi-source satellite imagery. SEVIRI data have been elaborated using CL-HOTSAT to estimate the lava volume emitted during each event and calculate the cumulative volume; Pléiades and WorldView-1 data have been used to derive Digital Surface Models, whose differences provide thickness distributions and hence volumes of the volcanic deposits. We find a good agreement, with the total average lava volume obtained by SEVIRI reaching 73.2 × 106 m3 and the one from optical data amounting to 67.7 × 106 m3. This proves the robustness of both techniques and the accuracy of the volume estimates, which provide important information on the lava flooding history and evolution of the volcano.

Retracted: Extraction of Long Time-Series Vegetation Indices from Combined Multisource Satellite Imagery.

[This retracts the article DOI: 10.1155/2022/3901372.].

Data Fusion for Satellite-Derived Earth Surface: The 2021 Topographic Map of Etna Volcano

We present a new automatic procedure for updating digital topographic data from multi-source satellite imagery, which consists in the production of Digital Surface Models (DSMs) from high resolution optical satellite images, followed by a context-aware fusion that exploits the complementary characteristics of the multi-source DSMs. The fused DSM minimizes blunders and artifacts due to occlusions (e.g., the presence of clouds, snow or ash plumes) in the source images, resulting in improved accuracy and quality versus those that are not merged. The procedure has been tested to produce the 2021 digital topography of Mt Etna, whose summit area is constantly changing and shows the new peak of 3347 m on the north rim of the South East Crater. We also employ the 2021 DSM to measure the volcanic deposits emplaced in the last five years, finding about 120 million cubic meters, with a yearly average volume of about 24 million cubic meters in agreement with the large eruptive rates registered at Mt Etna since the nineteen seventies. The flexibility and modularity of the presented procedure make it easily exportable to other environmental contexts, allowing for a fast and frequent reconstruction of topographic surfaces even in extreme environments.

Damage detection on antarctic ice shelves using the normalised radon transform

Areas of structural damage mechanically weaken Antarctic ice shelves. This potentially preconditions ice shelves for disintegration and enhanced grounding line retreat. The development of damage and its feedback on marine ice sheet dynamics has been identified as key to future ice shelf stability and sea level contributions from Antarctica. However, it is one of the least understood processes that impact ice shelf instability since quantifying damage efficiently and accurately is a challenging task. Challenges relate to the complex surface of Antarctica, variations in viewing-illumination geometry, snow or cloud cover and variable signal-to-noise levels in satellite imagery. Therefore, automated damage assessment approaches require careful pre- and post-processing, lacking the option to be applied to wider spatiotemporal domains. Simultaneously, studies that use manual mapping are usually limited due to the effort required for extensive mapping, which either results in a limited spatial domain or the use of low resolution data.This study proposes the NormalisEd Radon transform Damage detection (NeRD) method to detect damage features and their orientations from multi-source satellite imagery. NeRD performs robust, high resolution, large-scale damage assessments. NeRD is applied to the ice shelves in the Amundsen Sea Embayment (ASE) and validated with both manually labelled and existing fracture maps. Validation shows that NeRD detects damage with high recall and provides an accurate physical representation of multi-scale damage features and their orientation. Sensitivity analyses indicate NeRD is robust to different resolution parameter settings. NeRD consistently detects damage for different data sources ranging from optical Landsat 7/8 and Sentinel-2 optical to Synthetic Aperture Radar Sentinel-1 data. Therefore, NeRD paves the way for synergistic multi-source damage detection that overcomes remaining limitations from individual sources. Results show varying damage patterns on the ice shelves in the ASE area in austral summer 2020–2021, with most damage located on the Pine Island, Crosson and Thwaites ice shelves. We show a damage increase on the Pine Island ice shelf between 2013–2019, and display advection and rotation of crevasses. The detected damage orientation can provide insight in the type of crevasse opening mode and the development of damage over time. The damage maps produced with NeRD can help evaluate ice sheet models or machine learning approaches, improving our understanding of damage evolution.