Search Results (1,380)

In mountainous regions with uneven gauge coverage, satellite precipitation errors are amplified by nonlinear rainfall–runoff processes, while conventional hourly merging often relies on a single continuous regression that inadequately handles zero inflation and intensity heterogeneity. We propose a hierarchical intensity-aware framework comprising a wet/dry gate, a frequency-matched four-class intensity router, and a shared long short-term memory (LSTM) network with class-conditional outputs. In the upper Fujiang River basin, GPM, CMORPH, ERA5-Land and topographic variables were used as predictors; models were trained on 2010–2013, with 2014 retained for temporally held-out validation across point and areal scales, spatial cross-validation and streamflow simulation. Progressive ablation shows that intensity stratification drives the main point-scale gain (Kling–Gupta efficiency (KGE), 0.22 → 0.43), while temporal modeling improves held-out catchment-event performance (KGE 0.75). Oracle diagnosis identifies intensity routing as the main remaining bottleneck, and attribution and source-ablation analyses show that data-source value varies with prediction stage and evaluation scale. Hydrologically, merged precipitation raises the overall Nash–Sutcliffe efficiency (NSE) from −0.13 (GPM) and −0.21 (CMORPH) to 0.40 and reduces absolute peak bias from 52–55% to 32%. Routing discrimination remains the principal residual limitation under the present architecture. Full article

This study investigated the linear and nonlinear effects of environmental drivers on burn severity during the 2025 Uiseong wildfire, the largest forest fire recorded in South Korea. Burn severity, quantified using the differenced Normalized Burn Ratio (dNBR), was analyzed with the Generalized Linear Model (GLM) and Generalized Additive Model (GAM). The results identified the Normalized Difference Vegetation Index (NDVI) as the dominant predictor, with approximately eightfold greater importance than any other variable. A clear threshold response showed that the probability of High-severity burns increased sharply when NDVI exceeded approximately 0.25–0.30. Coniferous forests were 3.65 times more likely to experience High-severity burns than broadleaf forests, whereas managed forests exhibited lower burn severity. Topographic and meteorological variables generally had weaker effects, although several showed nonlinear response patterns. Overall, fuel availability and continuity were the primary determinants of burn severity, while topographic and meteorological factors played secondary, modifying roles. However, the model showed limited ability to identify Low-severity areas. These findings provide a scientific basis for improving forest management and wildfire mitigation strategies. Full article

Wildfires are increasingly shaped by climate change, as rising temperatures alter fire regimes. Accurate prediction of wildfire susceptibility under warming scenarios is essential for effective risk management and land-use planning. This study develops a Geographic Information Systems (GIS)-integrated Artificial Neural Network (ANN) framework, optimized via Bayesian hyperparameter tuning, to predict wildfire susceptibility and assess its spatial response to incremental temperature increases. The proposed methodology integrates multi-source geospatial data (Sentinel-2 NDVI, SRTM terrain models, Copernicus ERA5 reanalysis, and VIIRS active fire observations) within a GIS environment, coupling event-based feature extraction with the Bayesian-optimized ANN, which was trained on climatic, topographic, vegetation, and environmental predictor variables. Model evaluation showed that smaller, balanced training datasets achieved stronger generalization than larger ones. Scenario-based analyses simulated temperature increases of +0.5 °C, +1.0 °C, +1.5 °C, and +2.0 °C, reflecting global climate model projections, with all other variables held constant. Results show that warming intensifies susceptibility in existing fire-prone regions and drives the emergence of new wildfire-prone areas with no historical fire occurrence. Relative to current conditions, newly predicted fire-prone locations increased by approximately 36%, 51%, 85%, and 118% across the respective scenarios, indicating a nonlinear susceptibility response to warming. These findings underscore the value of incorporating climate scenarios into GIS-integrated machine learning-based wildfire susceptibility modeling and offer a spatially explicit framework for proactive mitigation and climate-resilient land-use planning. Full article

Rapid urbanization and the replacement of vegetated surfaces by impervious materials can substantially alter urban thermal environments, while water bodies, topography, and coastal setting may further modify these patterns. Understanding how these factors contribute to land surface temperature (LST), particularly across large and geographically heterogeneous cities, is important for spatially targeted heat mitigation. This study investigates the spatial variation in summer LST across Istanbul, Türkiye, using an explainable geospatial machine learning framework that integrates satellite observations with blue–green, built-environment, topographic, and coastal variables. Landsat 8/9 observations from June to August 2020–2025 were integrated with multi-source land cover, imperviousness, topographic, and coastline data within a common 500 m grid. Following predictor screening and comparison of five tree-based algorithms, XGBoost was applied to the full study area and evaluated using nested spatially blocked cross-validation. The model achieved a mean R² of 0.9531 ± 0.0083, RMSE of 1.0616 ± 0.0293 °C, and MAE of 0.7734 ± 0.0201 °C across the five outer folds of the nested spatial cross-validation. Predictive performance decreased under the larger 20 km block configuration, indicating sensitivity to the spatial scale of validation. SHapley Additive exPlanations (SHAP) identified the Normalized Difference Vegetation Index (NDVI), used here as a spectral proxy for vegetation greenness rather than as a direct measure of vegetated area, as the largest contributor to the fitted model, followed by imperviousness, water body cover, and grassland. The importance of water body cover was sensitive to coastal grid geometry and decreased substantially when low-land-fraction coastal cells were excluded. Because vegetation-related information is also used in the emissivity adjustment of the Landsat Level-2 surface temperature product, the magnitude of the NDVI attribution is interpreted as model-based rather than as an independent estimate of vegetation cooling. A sensitivity model excluding the NDVI retained substantial predictive performance (R² = 0.8063), with imperviousness becoming the leading predictor. A sensitivity analysis replacing the NDVI with independently derived tree cover percentage yielded lower model performance (R² = 0.9160). Dependence analyses revealed nonlinear relationships between the main predictors and LST. Spatial SHAP mapping further showed that although the NDVI had the largest local SHAP attribution across most of Istanbul, water-related, coastal, and topographic factors became locally important under specific geographic conditions. These findings show the value of spatially explicit explainable machine learning for identifying both city-wide and local factors associated with urban surface temperature. Full article

Vegetation dynamics in alpine ecosystems are highly sensitive to climate and environmental changes. However, the spatial heterogeneity of vegetation cover and its lagged associations with climate variability remain insufficiently understood. This study estimated fractional vegetation cover (FVC) based on MOD13Q1 NDVI data. The Sen’s slope and Mann–Kendall tests were employed to describe trends in FVC change, while pixel-based Pearson correlation analysis was used to quantify lagged associations (0–12 months) between temperature and precipitation and FVC. Furthermore, the XGBoost–SHAP framework was used to quantify nonlinear environmental associations with the spatial variation in FVC. The results showed that, from 2001 to 2024, the growing-season FVC on the Western Sichuan Plateau exhibited a significant increasing trend, with an average value of 0.7062 and a rate of 0.0018 yr⁻¹ (R² = 0.5040, p < 0.001). Areas with high and relatively high FVC values accounted for 75.7873% of the study area, indicating generally favourable vegetation conditions with substantial spatial heterogeneity. Both temperature and precipitation exhibited lagged associations with FVC, with long-term lags (7–12 months) accounting for 45.6600% and 45.0768% of the study area, respectively. After integrating topography and land-cover information, the explanatory power of the XGBoost model was enhanced, with the R² value increasing from 0.2181 in the climate-only model to 0.6393 in the full ecological model (RMSE = 0.1800, MAE = 0.1263). SHAP analysis indicated that DEM (elevation), LC_10 (grasslands), and TMEAN (Temperature_2m) made the largest contributions to the modelled spatial variation in FVC, accounting for 39.0853%, 17.3644%, and 10.3859%, respectively. These results indicate persistent associations between FVC spatial heterogeneity and topographic, land-cover, and climatic gradients. Full article

Accurate estimation of shallow groundwater depth (SGWD) is crucial for ecological security in arid oases. However, sparse monitoring wells and strong spatiotemporal heterogeneity hinder reliable monthly estimation. Traditional interpolation methods struggle to capture complex environmental drivers in anthropogenically modified regions. To address this issue, multi-source remote sensing data were integrated with machine learning models to estimate SGWD in the Weiku Oasis, Xinjiang, China. Field-measured SGWD data from March to November 2020 were collected. Based on the driving mechanisms of SGWD, six categories of variables were extracted, including optical, radar, topographic, meteorological, soil, and human activity factors. Four modeling strategies were designed by combining these variables with random forest (RF), support vector machine (SVM), and extreme gradient boosting (XGBoost) models. The optimal variable-model combination was selected for monthly SGWD mapping. SHAP was employed to interpret monthly feature importance and variable influence mechanisms. The results showed the following: (1) Among the four modeling strategies, Strategy II and Strategy IV achieved the best performance in monthly SGWD estimation, indicating that either single optical remote sensing data or the synergy of optical and radar remote sensing data can effectively characterize the spatiotemporal heterogeneity of SGWD in similar irrigated oasis regions. (2) Under different “variable-model” combinations, the RF model performed best in March, April, May, and October, whereas the SVM model performed best in June, July, August, September, and November, with XGBoost showing the weakest overall performance. Furthermore, the machine learning models outperformed Ordinary Kriging, yielding a mean R² of 0.790 and a 5.23% MAE reduction compared to −0.325. (3) Monthly SGWD maps revealed shallower depths in central and southeastern farmlands and piedmont alluvial fans, and greater depths in northern and northwestern desert margins due to insufficient groundwater recharge. (4) The dominant controls on SGWD exhibited clear seasonal variation. Salinity indicators, radar-derived wetness, and evapotranspiration prevailed in spring, followed by topographic and thermal–evaporative drivers in summer and by topography, precipitation, and soil properties in autumn. Despite these seasonal shifts, valley depth remained among the top five predictors throughout the nine months. This study provides a scientific method and technical support for dynamic monitoring and refined management of groundwater resources in similar irrigated oasis regions. Full article

The vertical redistribution of human activities and surface thermal changes poses challenges to Himalayan resilience. Using Nepal’s 77 districts, this study integrates land-use maps for 2000, 2020, and 2023 with annual MODIS land surface temperature (LST) data for 2000–2023. Trend analysis, fixed thermal thresholds, displacement correspondence tests, lag analysis, and GeoDetector were employed to examine land-use restructuring, thermal variability, and their spatial associations. Results show that: (1) Built-up land expanded from 527 to 1817 km², with its mean elevation increasing from 412 to 617 m. However, national mean LST, fixed-threshold hot-area extent, and hot-area mean elevation exhibited no significant monotonic trends. (2) Land-use and thermal displacements showed no consistent spatial correspondence or stable temporal lag, indicating regional vertical decoupling rather than delayed thermal migration. (3) Built-up expansion was positively associated with local LST trends. (4) Spatial associations between thermal heterogeneity and elevation, built-up land proportion, and population density, highlighting the spatial heterogeneity of topographic conditions and human activities in relation to regional thermal changes. These asynchronous vertical responses highlight potential development and ecological pressures in mid-elevation transition zones, suggesting a shift from predominantly horizontal spatial regulation toward elevation-sensitive, vertically adaptive planning for Himalayan resilience. Full article

High-resolution atmospheric modeling over topographically complex coastal zones requires multi-source spatial data fusion to bridge the scale gap between satellite-assimilated global reanalysis and localized boundary layer dynamics. This study presents a spatial data fusion framework coupling convection-permitting Weather Research and Forecasting (WRF) simulations (3 km grid spacing) with machine learning calibration and GIS multi-criteria decision analysis across continuous computational grids (24,871 cells). Focusing on a high-resolution seasonal case study during the peak Caribbean Low-Level Jet (CLLJ) and Trade Winds regime along the Colombian Caribbean coast, predictive algorithms were benchmarked across day-grouped and Leave-One-Station-Out (LOSO) spatial transfer protocols against classical meteorological baselines. Support Vector Regression (SVR) with an RBF kernel achieved superior residual calibration ($R^2=0.9523$, $\mathrm{RMSE}=0.4601$ m/s) under day-grouped cross-validation, and an average Leave-One-Station-Out (LOSO) spatial transfer RMSE of $1.1030$ m/s under blind geographic transfer to unmonitored stations, corresponding to a >76% RMSE reduction relative to raw, uncalibrated WRF output in that same spatial-transfer setting. Chronological forward-forecasting tests further showed that this seasonally augmented kernel model does not extrapolate reliably to calendar months absent from training ($R^2=-0.09$), so a reduced 8-variable configuration is recommended for genuine forward forecasting into unseen seasons. Scalar calibrations were transferred to horizontal wind vectors preserving simulated flow direction, and vertically scaled to 80 m hub height using WRF prognostic shear. Calibrated fields were integrated into a Boolean GIS-MCDA model incorporating wind power density, slope limits, environmental/indigenous reserves, and infrastructure proximity. The framework delineated an $8.9\%$ high-feasibility macro-corridor (2214 cells; ≈19,926 ${\mathrm{km}}^2$) in northern La Guajira, while substantially reducing uncalibrated mesoscale overestimation in the sheltered wake of the Sierra Nevada de Santa Marta. The methodology provides a robust screening-level tool for coastal renewable resource assessment. Full article

Nautical area scattering coefficient (NASC) data from small coastal acoustic surveys are strongly zero-inflated and spatially autocorrelated. We analysed 3324 integration intervals from 19 surveys in Jinhae Bay, Korea: May–September 2023 and March–September 2024–2025. Signals meeting a 38–120 kHz backscattering-strength-difference mask were treated as small pelagic fish and integrated at 38 kHz. Concurrent set-net and eDNA sampling showed a multi-species assemblage. Four predictor sets combining hydrographic, seasonal, spatial and topographic variables were compared using a two-part random forest under nested cross-validation with 1500 m spatial blocks. Zeros comprised 67.39% of observations. Hydrographic models achieved detection AUCs of 0.874–0.882 and positive-intensity log-scale R² values of 0.325–0.330, versus 0.701 and 0.062 for topography plus month. Drop-column analysis showed that surface salinity carried the unique hydrographic signal and that surface temperature was redundant with salinity and month; topographic variables added nothing once coordinates were included. Survey identity alone reproduced the full-model AUC (0.882 versus 0.885): spatial-block performance largely reflects discrimination between surveys, not local interpolation. Leave-one-survey-out and leave-one-year-out validation fell to 0.503 ± 0.122 and 0.541 ± 0.123. Detection was more predictable than intensity, but the apparent skill is largely between-survey; the model should not predict unobserved occasions or estimate abundance. Full article

Hainan Island supports abundant tropical ancient Ficus trees owing to its distinctive hydrothermal conditions and long-term human–environment interactions, yet the environmental factors associated with their distributions and responses to future climate change remain insufficiently understood. This study focused on the ancient trees of four dominant Ficus species on Hainan Island—Ficus microcarpa, F. altissima, F. benjamina, and F. virens—using systematically surveyed occurrence records and 59 candidate variables representing climate, soil, topography, anthropogenic activity, and typhoon disturbance. Two parameter-optimized MaxEnt frameworks were developed. The Environmental Association Model (EAM) incorporated climatic, soil, topographic, anthropogenic, and typhoon-related variables to identify factors associated with the current distributions of these ancient trees, whereas the Natural-Environment Projection Model (NEPM) included climatic, soil, and topographic variables to estimate potential habitat suitability under current conditions and SSP126 and SSP585 scenarios for the 2050s and 2070s; only climatic variables were updated in future projections. Under the EAM framework, population density was the key variable associated with the modeled distributions of all four ancient-tree groups, accounting for 65.3%, 46.1%, 39.6%, and 26.3% of the model contribution for ancient F. microcarpa, F. altissima, F. benjamina, and F. virens, respectively. Climatic and soil variables showed species-specific effects; topographic variables contributed mainly to ancient F. altissima and F. virens, whereas typhoon impact intensity contributed only slightly to ancient F. benjamina. Under current climatic conditions, total suitable habitat ranged from 7771.31 to 10,489.97 km². Ancient F. microcarpa had the largest total suitable area, ancient F. benjamina the smallest, and ancient F. altissima the largest highly suitable area. Highly suitable habitats were concentrated mainly in northern and northwestern Hainan, with scattered central patches. Under future scenarios, the total and highly suitable habitats of ancient F. benjamina generally expanded, particularly under SSP585. In contrast, those of ancient F. microcarpa, F. altissima, and F. virens generally contracted, with substantial losses of highly suitable habitat; ancient F. virens faced the greatest habitat-loss risk. SSP585 intensified contraction for ancient F. microcarpa and F. virens, whereas ancient F. altissima showed more complex, scenario-dependent changes. Sensitivity analyses using alternative regularization settings supported these overall trends. These findings support conservation zoning, habitat restoration, germplasm preservation, and climate-adaptive management of ancient Ficus trees on Hainan Island. Full article

Mountainous rainfall exhibits strong spatial heterogeneity associated with topographic effects. Understanding modulating patterns and non-linear responses of rainfall to terrain factors supports mountain storm-disaster mitigation. Taking the Yanshan Mountains in Huairou (Beijing) as the study area, we divided the region by watershed divides into northern and southern zones. Elevation, slope, and sine-cosine transformed aspect served as independent variables, while annual, flood-season, maximum-daily, and maximum-3-day extreme rainfall were dependent variables. We applied Kriging interpolation, quadratic-polynomial ordinary least-squares (OLS), multiscale geographically weighted regression (MGWR), and XGBoost-SHAP to characterize terrain-rainfall statistical associations from global-linear, spatial-non-stationary, and non-linear empirical transition perspectives. Elevation and slope dominated rainfall spatial patterns, whereas aspect exerted minor local modulation. Southern rainfall generally corresponds to higher values with rising elevation, while northern rainfall presented a U-shaped non-linear elevation response. Extreme rainfall was more sensitive to slope than routine rainfall; empirical transition-response features lie at 700–800 m elevation and 20° slope. Multi-model coupling mitigated single-model limitations and provides references for extreme-precipitation understanding and geohazard risk assessment in the North China Yanshan Mountains. Full article

Biological invasions driven by climate change and human activities pose a serious threat to biodiversity and ecosystem stability in global biodiversity hotspot regions. Lantana camara is a globally recognized aggressive invasive plant that has rapidly spread in southwest China and caused significant ecological risks. To clarify its potential geographic distribution, key driving factors, and future invasion dynamics under climate change, this study applied an MaxEnt model with feature-class optimization, integrating occurrence points, bioclimatic variables, topographic data, and the human footprint index to simulate habitat suitability in southwest China for the current period (1970–2000) and the future (2050s and 2070s) under three shared socioeconomic pathways (SSP1-2.6, SSP2-4.5, SSP5-8.5). The results show that the current high-suitability area for L. camara is 1.21 × 10⁴ km², mainly distributed in central Yunnan and the Chengdu Plain. Human activity (human footprint index, 52.3%) is the dominant driver, and the mean temperature of the coldest month (28.5%) is the primary natural limiting factor. Future suitable areas exhibit a remarkable scenario-dependent nonlinear response. By the 2070s, the high-suitability area expands most sharply under SSP2-4.5 (3.27 × 10⁴ km², an increase of about 170%), revealing that moderate warming facilitates northward expansion. Under SSP5-8.5, the high-suitability area contracts compared with the 2050s, indicating that extreme high temperatures restrain optimal habitats. Spatially, L. camara displays a “retreat south, advance north” pattern under high-emission scenarios, with its distribution centroid continuously shifting northeast, reflecting climate-driven niche changes. This study verifies that the invasion risk of L. camara does not rise linearly with climate warming. Moderate warming boosts its spread, whereas extreme warming reshapes its distribution, offering a scientific basis for risk assessment, early warning, and targeted control strategies against L. camara invasion in southwest China and worldwide. Full article

Background: Administrative wildfire records can characterize operational and temporal patterns, but causal interpretation requires caution when meteorological, fuel, topographic, and accessibility variables are unavailable. This study examined factors associated with wildfire size within the jurisdiction of the Kayseri Regional Directorate of Forestry, Türkiye, during 2016–2025. Methods: A total of 364 wildfire records were analyzed. Large-fire risk (>5 ha) was modeled with Firth-penalized logistic regression because of sparse events and separation. Burned area was analyzed using a two-part model: Firth logistic regression for zero versus positive burned area, followed by a Gamma generalized linear model with a log link for positive burned area (n = 333). Seasons were treated as categorical variables using calendar definitions, and the administrative vehicle-count variable was reported descriptively because its timing could not be verified as initial dispatch. Mann–Kendall trend tests, corrected seasonal post hoc comparisons, and exploratory ROC analysis were also performed. Results: Fifty-six fires (15.4%) exceeded 5 ha, and 31 records had zero burned area. Negligence/carelessness was the recorded cause in 91.5% of incidents. In the Firth large-fire model, initial response time was not independently associated with large-fire odds (OR = 1.007, 95% CI 0.989–1.026; p = 0.452), and no modeled predictor reached statistical significance. In the positive-area Gamma component, autumn fires had greater conditional mean burned area than summer fires (mean ratio [MR] = 1.683, 95% CI 1.233–2.295; p = 0.001), whereas initial response time was not statistically significant (MR = 1.012, 95% CI 0.998–1.026; p = 0.097). Seasonal distributions differed overall (Kruskal–Wallis H = 12.759, p = 0.005; epsilon-squared = 0.027), with the Holm-corrected autumn-versus-summer comparison remaining significant (p = 0.035). Response time alone showed weak discrimination for large fires (AUC = 0.562, bootstrap 95% CI 0.489–0.636); the 18 min value was therefore retained only as a dataset-specific exploratory reference. The response-time trend was not statistically significant (tau = −0.467, p = 0.073), and excluding 2025 further attenuated it (tau = −0.333, p = 0.260). Conclusions: The corrected analyses support seasonal heterogeneity, particularly greater positive burned area in autumn, but do not support a causal or independently significant response-time effect after conservative model specification. Future wildfire-response analyses should integrate weather, fuels, terrain, access, distance, and incident conditions at detection. Full article

Flood susceptibility mapping is essential for effective flood mitigation and land management, particularly in rapidly urbanizing watersheds. This study proposes a multi-criteria framework for improving flood susceptibility modeling through conditioning factor selection. Nineteen topographic, hydrological, environmental, and land-use factors were used to develop four models: Frequency Ratio (FR), Shannon Entropy (SE), Certainty Factor (CF), and VIKOR. A flood inventory of 230 historical flood locations was compiled, with 160 locations (70%) used for model development and 70 (30%) reserved for independent validation. Model performance was evaluated using Receiver Operating Characteristic (ROC) analysis based on Success Rate and Prediction Rate datasets. A factor selection framework integrating ROC-AUC, Information Value (IV), Pearson correlation, and Variance Inflation Factor (VIF) was then applied to identify informative and non-redundant variables. The framework reduced the conditioning factors from 19 to 9. All optimized models showed improved predictive performance. The FR model showed the greatest improvement, with Success Rate AUC increasing from 58.45% to 90.41% and Prediction Rate AUC from 58.10% to 89.88%. The CF model achieved the highest overall performance after optimization, while SE and VIKOR also improved substantially. The proposed framework improves prediction while reducing input-data complexity. Full article

Flooding and drought alternate in many tropical floodplains, but how vegetation responds to both hazards over multiple years remains poorly quantified. We examined vegetation–flood associations in the Chi River Basin, Northeast Thailand, drawing on Sentinel-2 imagery from 2020, 2023, and 2024, flood records from 2017, 2018, 2021, and 2022, and machine learning methods across 230,911 point-based spatial units. Nine forms of asymmetry emerged from the analysis. The most striking was recovery asymmetry: areas that flooded at least twice between 2017 and 2022 gained vegetation during 2020–2023 (mean NDVI change = +0.0270), whereas areas with little or no flood exposure lost vegetation (mean change = −0.0430). This gap was statistically significant (Cohen’s d = 0.5586, p < 0.001) and suggests that repeated flooding may buffer vegetation against subsequent drought. Agricultural land declined less (−0.0316) than forest (−0.0937; ANOVA F = 1717.7, p < 0.001). Baseline vegetation condition, measured as NDVI in 2020, contributed 32.9% to model importance, more than twice the contribution of elevation (14.5%). Spectral indices together accounted for 76.6% of importance, compared with 23.5% for topographic variables. The 2023 El Niño year produced the largest difference between high-flood and low-flood areas (+0.0364); because only one year per ENSO phase was available, we treat this as a case-based comparison rather than a general ENSO response. Threshold analysis identified two distinct values: an operational cut-off at NDVI = 0.05 (overall accuracy 82.67%) and an ecological transition around 0.25–0.30. Spatial clustering was weak but significant (Moran’s I = 0.2179, p < 0.001). Spatial block cross-validation gave lower accuracy (0.597) than random cross-validation (0.627), pointing to spatial autocorrelation in the data. High-flood areas had 1.67 times the vulnerability index of low-flood areas (0.4306 vs. 0.2573). These patterns support differentiated management: elevation-based zoning, warning systems calibrated to local flood regimes, and focused interventions at hotspots. Full article