Search

The study involves an object-based segmentation method to extract feature changes in tropical rainforest cover using Landsat image and airborne LiDAR (ALS). Disturbance event that are represents the changes are examined by the classification of multisensor data; that is a highly accurate ALS with different resolutions of multispectral Landsat image. Disturbance Index (DI) derived from Tasseled Cap Transformation, Normalized Difference Vegetation Index (NDVI), and the ALS height are the variables for object-based segmentation process. The classification is categorized into two classes; disturbed and non-disturbed forest cover using Nearest Neighbor (NN), Random Forest (RF) and Support Vector Machine (SVM). The overall accuracy ranging from 88% to 96% and kappa ranging from 0.79 to 0.91. Mcnemar’s test p-value (

Seismic damping of near-surface deposits is an important input to site-response analysis for seismic hazard assessment. In Groningen, the Netherlands, gas production from a reservoir at 3 km depth causes seismicity. Above the gas field, an 800 m thick layer of unconsolidated sediments exist, which consists of a mixture of sand, gravel, clay and peat strata. Shear waves induced at 3 km depth experience most of their anelastic attenuation in these loose sediments. A good estimate of damping is therefore crucial for modelling realistic ground-motion levels. In Groningen, we take advantage of a large network of 200 m deep vertical arrays to estimate damping from recordings of the induced events. As a first step, we apply seismic interferometry by deconvolution to estimate local transfer functions over these vertical arrays. Subsequently, two different methods are employed. The first is the 'upgoing' method, where the amplitude decay of the retrieved upgoing wave is used. The second is the 'up-down' method, where the amplitude difference between retrieved up- and downgoing waves is utilized. For the upgoing method, the amplitude of the upgoing direct wave is affected by both elastic and anelastic effects. In order to estimate the anelastic attenuation, it is necessary to remove the elastic amplification first. Despite the fact that elastic compensation could be determined quite accurately, non-physical damping values were estimated for a number of boreholes. Likely, the underlying cause was small differences in effective response functions of geophones at different depths. It was found that the up-down method is more robust. With this method, elastic propagation corrections are not needed. In addition, small differences in in situ geophone response are irrelevant because the up- and downgoing waves retrieved at the same geophone are used. For the 1-D case, we showed that for estimating the local transfer function, the complex reverberations need to be included in the interferometric process. Only when this is done, the transfer function does not contain elastic transmission loss and Q estimation can be made without knowing the soil profile in detail. Uncertainty in the estimated damping was found from the signal-to-noise ratio of the estimated transfer function. The Q profiles estimated with the up-down method were used to derive a damping model for the top 200 m of the entire Groningen field. A scaling relation was derived by comparing estimated Q profiles with low-strain damping profiles that were constructed using published models for low-strain damping linked to soil properties. This scaling relation, together with the soil-property-based damping model, allowed up-scaling of the model to each grid-cell in the Groningen field. For depths below 200 m, damping was derived from the attenuation of the microseism over Groningen. The mean damping model, over a frequency band between 2 and 20 Hz, was estimated to be 2.0 per cent (0-50 m depth), 1.3 per cent (50-100 m), 0.66 per cent (100-150 m), 0.57 per cent (150-200 m) and 0.5 per cent (200-580 m). TURNkey (Towards more Earthquake-resilient Urban Societies through a Multi-sensor-based Information System enabling Earthquake Forecasting, Early Warning and Rapid Response actions); EU [H2020 - SC5 - 2018 - 2] Published version The work of ER has partly been funded by TURNkey (Towards more Earthquake-resilient Urban Societies through a Multi-sensor-based Information System enabling Earthquake Forecasting, Early Warning and Rapid Response actions), which is a project funded by the EU within the H2020 - SC5 - 2018 - 2 call. We would like to thank Jun Matsushima and an anonymous reviewer for a detailed and critical assessment.

Detection of hydrothermal alteration zones (HAZs) associated with porphyry copper systems using remote sensing imagery is a crucial stage for discovering high potential zone of ore mineralization. Statistical model-based clustering methods have great potential for automatic and accurate detection of hydrothermal alteration minerals using hyperspectral remote sensing imagery. In this research, the Dirichlet Process based on Stick-Breaking (DPSB) model-based clustering algorithm was implemented to hyperion remote sensing imagery to discriminate HAZs associated with the Kuh-Panj porphyry copper deposit, south, Iran. The DPSB clustering algorithm was implemented and subsequently compared with the k-means algorithm, CLARA clustering, hierarchical clustering, Gaussian finite mixture model (GFMM), Gaussian model for high-dimensional (GMHD) and spectral clustering as well as spectral angle mapping (SAM). Results derived from the DPSB model-based clustering algorithm show 88.6% accuracy in distinguishing propylitic, argillic, advanced argillic, propylitic-argillic and phyllic alteration zones. The DPSB algorithm can be broadly implemented to hyperspectral remote sensing imagery for detecting alteration zones associated with porphyry systems.

High spatial resolution information on urban air pollution levels is unavailable in many areas globally, partially due to the high input data needs of existing estimation approaches. We introduced a computer vision method to estimate annual means for air pollution levels from street-level images. We used annual mean estimates of NO2 and PM2.5 concentrations from locally calibrated models as labels from London, New York, and Vancouver to allow for compilation of a sufficiently large dataset (similar to 250 k images for each city). Our experimental setup is designed to quantify intra- and intercity transferability of image-based model estimates. Performances were high and comparable to traditional land-use regression (LUR) and dispersion models when training and testing images from the same city (R-2 values between 0.51 and 0.95 when validated on data from ground monitoring stations). Similar to LUR models, transferability of models between cities in different geographies is more difficult. Specifically, transferability between the three cities (London, New York, and Vancouver), which have similar pollution source profiles, was moderately successful (R-2 values between zero and 0.67). Comparatively, performances when transferring models trained on cities with very different source profiles, such as Accra in Ghana and Hong Kong, were lower (R-2 between zero and 0.21). This suggests a need for local calibration, using additional measurement data from cities that share similar source profiles., Remote Sensing, 14 (14), ISSN:2072-4292

This paper proposes a cloud-based mangrove monitoring framework that uses Google Collaboratory and Google Earth Engine to classify mangroves in Southeast Asia (SEA) using satellite remote sensing imagery (SRSI). Three multi-class classification convolutional neural network (CNN) models were generated, showing F1-score values as high as 0.9 in only six epochs of training. Mangrove forests are tropical and subtropical environments that provide essential ecosystem services to local biota and coastal communities and are considered the most efficient vegetative carbon stock globally. Despite their importance, mangrove forest cover continues to decline worldwide, especially in SEA. Scientists have produced monitoring tools based on SRSI and CNNs to identify deforestation hotspots and drive targeted interventions. Nevertheless, although CNNs excel in distinguishing between different landcover types, their greatest limitation remains the need for significant computing power to operate. This may not always be feasible, especially in developing countries. The proposed framework is believed to provide a robust, low-cost, cloud-based, near-real-time monitoring tool that could serve governments, environmental agencies, and researchers, to help map mangroves in SEA.

Governments worldwide have rapidly deployed non-pharmaceutical interventions (NPIs) to mitigate the COVID- 19 pandemic. However, the effect of these individual NPI measures across space and time has yet to be sufficiently assessed, especially with the increase of policy fatigue and the urge for NPI relaxation in the vaccination era. Using the decay ratio in the suppression of COVID-19 infections and multi-source big data, we investigated the changing performance of different NPIs across waves from global and regional levels (in 133 countries) to national and subnational (in the United States of America [USA]) scales before the implementation of mass vaccination. The synergistic effectiveness of all NPIs for reducing COVID-19 infections declined along waves, from 95.4% in the first wave to 56.0% in the third wave recently at the global level and similarly from 83.3% to 58.7% at the USA national level, while it had fluctuating performance across waves on regional and subnational scales. Regardless of geographical scale, gathering restrictions and facial coverings played significant roles in epidemic mitigation before the vaccine rollout. Our findings have important implications for continued tailoring and implementation of NPI strategies, together with vaccination, to mitigate future COVID-19 waves, caused by new variants, and other emerging respiratory infectious diseases. Published version

Globally, the frequency, intensity, and damage rate of floods have increased extensively in the last few decades. The present study proposes a spatial-based multi-criteria approach integrating mitigation capacity to map the degree of flood risk. Flood risk is calculated considering 14 relevant factors under three risk components. Thematic layers were ranked through fuzzy membership functions, and risk components were prepared using fuzzy overlay operation. To make a comparison, first, we calculated risk with mitigation capacity and later without it. Results exhibit, without mitigation capacity around 67% of the study area is exposed to high flood risk, while after integrating mitigation capacity flood risks decreased to 62%. Moreover, validation evaluation suggests risk map integrating mitigation capacity is more accurate (82%) than the risk map without it (79%). Therefore, the developed integrated risk mapping approach is more efficient to assess flood risk, and can be implemented by policymakers in similar environments.

Geohazard risk is high in Arab countries due to ineffective disaster preparedness measures, mismanagement, lack of public awareness, inadequate funding and lack of stakeholder support. One such country is Egypt, which is hit by floods every year that cost lives and bring the economy to a standstill. Moreover, not much has been done to map flood-prone areas. In this paper, flood susceptibility modelling was evaluated in the Ras Gharib region of Egypt using two effective techniques machine learning technique-MLT (Support Vector Machine (SVM)) and deep learning method-DL (Convolutional Neural Networks (CNN)). Thirteen flood related factors and flood inventory layer were prepared to construct these models. Validation was performed with 30% of the flood locations where receiver operating characteristic (ROC) curves showed that the deep learning technique (CNN) gave a prediction accuracy of 86.5% (high performance), while the MLTs (SVM) gave 71.6% (medium performance). The results show that CNN provides 17% better than SVM which indicates a powerful and accurate model in flood susceptibility mapping. Results were confirmed using the Astro Digital images shortly after the 2016 flood, in which the CNN model provides a good agreement.

Most previous studies of landslide susceptibility mapping (LSM) have not contemplated spatial heterogeneity and the commonly used models for LSM are aspatial, which could reduce model performance. Therefore, aiming to evaluate the applicability of spatial algorithms to predict landslide susceptibility, the performance of geographical random forest (GRF) was evaluated, in comparison to random forest (RF) and extreme gradient boosting (XGBoost). Based on the results, GRF presented the better performance (AUC = 0.876), followed by RF (AUC = 0.748) and XGBoost (AUC = 0.745). GRF also provided the most suitable susceptibility map. While RF and XGBoost presented almost 50% of the study area as susceptible, the GRF presented more concentrated susceptibility areas spatially, with a reasonable area for moderate (15.55%), high (8.73%) and very-high (2.59%) susceptibility classes. Finally, it can be inferred that spatial assessment may improve model performance, and that spatial models have a great potential for LSM.

In this study, the multi-criteria decision-making method was used to estimate the weights of several input factors such as slope, curvature, elevation, proximity to road, road density, proximity to land use, land use density, proximity to water bodies, river density, rail density, distance from rail, groundwater variation, lithology with amplification factors, peak ground acceleration (PGA) variation, and population density. An integrated analytic hierarchy process (AHP) and a probabilistic neural network (PNN) were applied for the Earthquake vulnerability assessment (EVA). The PNN model successfully explored the relationship between variables and weights obtained from the AHP approach. Validation results indicate that 92.5% accuracy was attained by the PNN model. According to the results, 24.26%, 15.26%, and 20.58% of the area fall under very-high, high, and moderate vulnerability category, respectively. The EVA map illustrates that high to very-high impact could be observed in coastal Odisha and few districts in the Mahanadi Graven.

Floods are considered one of the most destructive natural hydro-meteorological disasters across the world. The present study attempts to assess flood susceptibility of the Brahmaputra floodplain of Bangladesh using Deep Boost, Deep Learning Neural Network, and Artificial Neural Network. Primarily, flood inventory maps were prepared from fieldworks and satellite image classification. Consequently, the flood locations were segregated into 70% training and 30% validation samples randomly for running the models and validating the models, respectively. The complete procedure is designed to be considered 12 flood conditioning criteria under four relevant components. The efficiency assessment of DLNN, DB, and ANN models using validation data through the area under the curve (AUC) reveals that DB demonstrates higher accuracy (AUC=.917) than DLNN (AUC =.901) and ANN (AUC=.895) approaches. Therefore, proposed susceptibility mapping approaches are efficient in assessing flood susceptibility accurately and can be implemented in other flood-affected regions.

Assessing the extent and level of building damages is crucial to support post-earthquake rescue and relief activities. There is a large body of literature proposing novel frameworks for automating earthquake-induced building damage mapping using high-resolution remote sensing images. Yet, its deployment in real-world scenarios is largely limited to the manual interpretation of images. Although manual interpretation is costly and labor-intensive, it is preferred over automatic and semi-automatic building damage mapping frameworks such as machine learning and deep learning because of its reliability. Therefore, this review paper explores various automatic and semi-automatic building damage mapping techniques with a quest to understand the pros and cons of different methodologies to narrow the gap between research and practice. Further, the research gaps and opportunities are identified for the future development of real-world scenarios earthquake-induced building damage mapping. This review can serve as a guideline for researchers, decision-makers, and practitioners in the emergency management service domain.

Drought, a natural and very complex climatic hazard, causes impacts on natural and socio-economic environments. This study aims to produce the drought vulnerability map (DVM) considering novel ensemble machine learning algorithms (MLAs) in Jharkhand, India. Forty, drought vulnerability determining factors under the categories of exposure, sensitivity, and adaptive capacity were used. Then, four machine learning and four novel ensemble approaches of particle swarm optimized (PSO) algorithms, named random forest (RF), PSO-RF, multi-layer perceptron (MLP), PSO-MLP, support vector regression (SVM), PSO-MLP, Bagging, and PSO-Bagging, were established for DVMs. The receiver operating characteristic curve (ROC), mean-absolute-error (MAE), root-mean-square-error (RMSE), precision, and K-index were utilized for judging the performance of novel ensemble MLAs. The obtained results show that the PSO-RF had the highest performance with an AUC of 0.874, followed by RF, PSO-MLP, PSO-Bagging, Bagging, MLP, PSO-SVM and SVM, respectively. Produced DVMs would be helpful for policy intervention to minimize drought vulnerability.

Dengue Fever (DF) is a common vector-borne disease with catastrophic health implications. DF prediction modelling is a challenging task, although technologies such as Geographical Information Systems (GIS) and spatial statistics have improved our understanding of dengue dynamics. In this paper, we create a robust data analysis model to (i) provide a better understanding of confirmed dengue fever cases despite missing data, (ii) obtain better insights into risk factors associated with confirmed cases, and (iii) by means of machine learning, create clusters of patients with comparable characteristics. The last was accomplished with a self-organizing feature map (SOFM) and the density-based spatial clustering of applications with noise (DBSCAN). The approaches used to classify confirmed cases were: Decision Tree, k-nearest neighbours, Random Forest, AdaBoost, Support Vector Classification (SVC), CatBoost, and Naive Bayes. The CatBoost classifier achieved the best accuracy for the analysis of confirmed cases. Spatial analysis was conducted using the ordinary least square (OLS) and geographically weighted regression (GWR) models to identify high-risk areas. SOM can group patients with similar features into clusters, then DBSCAN detects and retrieves six clusters from this data. The clustering of confirmed cases increases CatBoost’s modelling prediction accuracy and reveals complex factors that influence prediction accuracy. Because confirmed cases in each cluster have different features, CatBoost is applied to each cluster individually to improve the prediction accuracy. Variable values in each cluster are analysed to clarify the confirmed cases of a specific subset of DF incidents. Overall, OLS outperforms GWR when identifying hotspot areas. The proposed novel, data-driven and machine-learning-based strategy facilitates the understanding and identification of patterns associated with confirmed DF cases. The study's findings can be utilized to cluster historical patient data into groups or subgroups sharing similar variables. Using identifiable patient clusters rather than raw history data improves the model accuracy provided by CatBoost.

Analysis of ground movement rates along the coastline and upper sections of the Ventnor landslide complex was carried out utilizing Persistent Scatterer Interferometric Synthetic Aperture Radar methods using Sentinel-1 SAR data from 2015 to 2019 (four years). Results were compared with rainfall data, historical ground investigation records and monitoring surveys carried out at Ventnor to relate observations to geology, geomorphology and rainfall. Decomposition of InSAR viewing geometries to vertical and horizontal aligned well with previous ground-based studies. Subsidence of −9.8 mm a−1 at the Lowtherville Graben and heave of +8.5 mm a−1 along the coastline south of Ventnor Park were observed. Decomposition to east-west geometry results showed an eastward displacement of approximately 12.4 mm a−1 along the coastline south of Ventnor Park, and a westward displacement of −3.7 mm a−1 throughout built up sections of Ventnor town, indicating the landslide was displacing more in an eastern direction than vertically. The cause of this movement was investigated by using publicly available intrusive boreholes paired with Persistent Scatterer Interferometry, and a new ground model spanning east-west parallel to the coastline was presented. No evidence of significant ground movement was observed along heavily protected sections of the coastline, suggesting coastal defences comprised of concrete aprons and rip rap appear to be an effective coastal management/landslide stabilisation tool when compared to rip rap alone. The mechanism of this increased stability is likely due to the combination of toe weighting and reduced toe erosion. A lag of approximately 13–20 days was observed between high rainfall events and subsequent peaks in ground displacement, which was shorter than a 29 day lag observed in a previous study. Similar observations of prolonged rainfall resulting in prolonged displacements were also observed. The study demonstrates the capabilities of the PSI methodology in identifying the same ground movements that conventional methods provide. By providing detailed analysis of ground deformation of the Ventnor landslide, we demonstrate small ground movements, validated with existing ground movement surveys. Similar methodology can be applied to coastal landslides in urban environments worldwide, providing a relatively cheap and rapid resource for coastal landslide monitoring.

Parameter identification for wildfire forecasting models often relies on case-by-case tuning or posterior diagnosis/analysis, which can be computationally expensive due to the complexity of the forward prediction model. In this paper, we introduce an efficient parameter flexible fire prediction algorithm based on machine learning and reduced order modelling techniques. Using a training dataset generated by physics-based fire simulations, the method forecasts burned area at different time steps with a low computational cost. We then address the bottleneck of efficient parameter estimation by developing a novel inverse approach relying on data assimilation techniques (latent assimilation) in the reduced order space. The forward and the inverse modellings are tested on two recent large wildfire events in California. Satellite observations are used to validate the forward prediction approach and identify the model parameters. By combining these forward and inverse approaches, the system manages to integrate real-time observations for parameter adjustment, leading to more accurate future predictions.

We present an evaluation of Aerosol Optical Depth (AOD) retrievals from the Moderate Resolution Imaging Spectroradiometer (MODIS) over Australia covering the period 2001–2020. We focus on retrievals from the Deep Blue (DB) and Multi-Angle Implementation of Atmospheric Correction (MAIAC) algorithms, showing how these compare to one another in time and space. We further employ speciated AOD estimates from Copernicus Atmospheric Monitoring Service (CAMS) reanalyses to help diagnose aerosol types and hence sources. Considering Australia as a whole, monthly mean AODs show similar temporal behaviour, with a well-defined seasonal peak in the Austral summer. However, excepting periods of intense biomass burning activity, MAIAC values are systematically higher than their DB counterparts by, on average, 50%. Decomposing into seasonal maps, the patterns of behaviour show distinct differences, with DB showing a larger dynamic range in AOD, with markedly higher AODs (ΔAOD∼0.1) in northern and southeastern regions during Austral winter and summer. This is counter-balanced by typically smaller DB values across the Australian interior. Site level comparisons with all available level 2 AOD data from Australian Aerosol Robotic Network (AERONET) sites operational during the study period show that MAIAC tends to marginally outperform DB in terms of correlation (RMAIAC = 0.71, RDB = 0.65) and root-mean-square error (RMSEMAIAC = 0.065, RMSEDB = 0.072). To probe this behaviour further, we classify the sites according to the predominant surface type within a 25 km radius. This analysis shows that MAIAC’s advantage is retained across all surface types for R and all but one for RMSE. For this surface type (Bare, comprising just 1.2% of Australia) the performance of both algorithms is relatively poor, (RMAIAC = 0.403, RDB = 0.332).