Editorial: Advanced Geospatial Data Analytics for Environmental Sustainability: Current Practices and Future Prospects

Manish Pandey¹Varun Narayan Mishra^2*Maya Kumari²Saeid Janizadeh³Romulus Costache⁴

• ¹Marwadi University, Rajkot, India
• ²Amity University, Noida, India
• ³University of Hawai'i at Manoa Hawaii Research Center for Futures Studies, Honolulu, United States
• ⁴Universitatea Transilvania din Brasov, Brașov, Romania

Environmental sustainability involves utilizing natural resources without compromising future needs and requires balancing ecological, social, and economic goals (Hariram et al., 2023).Growing global concern stems from intensified human-environment interactions and accelerating climatic pressures linked to urbanization and industrialization (Purvis et al., 2019;Terra dos Santos et al., 2023). Shifts in precipitation, increased evapotranspiration, and extreme weather events highlight the urgency for research supporting climate-resilient societies.Advances in geospatial technologies have transformed environmental monitoring and decisionmaking. Remote sensing, GIS, LiDAR, and UAVs enable the acquisition and analysis of large earth observation datasets to assess spatiotemporal changes in natural systems (Lee & Kang, 2015). These tools now support modeling, visualization, and prediction of climate-driven processes, offering improved projections of environmental phenomena and socio-economic change.The rapid growth of earth observation data, combined with artificial intelligence (AI), is expanding opportunities for advanced environmental analytics. Big geospatial data supports real-time monitoring, interactive visualization, and improved decision support. Its integration into environmental sustainability research enables the identification of spatial patterns, drivers of change, and system vulnerabilities. Applications span land and water resource management, climate action planning, agriculture, and public health. Despite these advances, challenges remain. Data heterogeneity, imbalance, uncertainty, and computational constraints undermine model generalization and interoperability (Lee & Kang, 2015). Ethical and privacy concerns are increasing as high-resolution and real-time sensing expand. The computational footprint of large AI models and the opaque "black-box" nature of machine learning introduce additional risks. Innovations such as large language models, explainable AI, semantic modeling, advanced data fusion, synthetic data generation, metaverse environments, and knowledge graphs offer potential solutions for improving transparency, scalability, and responsible data use. This special issue highlights how integrated geospatial and computational methods can strengthen environmental sustainability, reflecting a shift toward data-driven, interdisciplinary, and anticipatory approaches. The compiled contributions encourage policymakers, researchers, and practitioners to leverage emerging geospatial intelligence for sustainable development. Earth observation (EO) forms the foundation of environmental monitoring and informed decision-making. Current applications integrate satellite remote sensing, GIS, and predictive modeling to support biodiversity conservation, land-use planning, hazard mitigation, and sustainable resource management (Abebe et al., 2022;M et al., 2023;Tesfaye et al., 2024).Three major categories dominate current practice: monitoring via EO, predictive modeling for risk management, and biodiversity and ecosystem assessment. Monitoring land use and land cover (LULC) changes is one of the most established applications. Forest loss, biomass dynamics, and canopy structure can be quantified using multispectral and radar datasets such as Landsat, Sentinel, and LiDAR (Chang et al., 2021;E. Nyland et al., 2018). Built-up indices such as NDBI support urbanization assessment, while NDVI and EVI are widely used for agriculture monitoring and yield prediction. EO also supports water and soil resource monitoring, for example, through MODIS-or Landsat-derived evapotranspiration products and hyperspectral datasets for soil nutrient mapping (Huang et al., 2025).Predictive modeling extends monitoring by integrating EO-derived variables with machine learning and statistical models to assess environmental risks. Flood, drought, wildfire, landslide, and avalanche susceptibility analyses increasingly rely on DEMs, climate data, and remote sensing-based indices (Ikematsu et al., 2024;Padilha et al., 2024). Techniques such as Random Forest, Support Vector Machines, and AHP support the identification of vulnerable regions and enhance early-warning capacities. EO technologies are also central to biodiversity and ecosystem conservation. Habitat suitability models using EO-based variables help identify priority conservation areas (Bilal, 2025;Chan et al., 2024;Gomes & Cardoso, 2025;Zhou et al., 2024). Wildlife monitoring benefits from multi-sensor integration, including satellite tracking, UAV observations, and camera traps (Rojas et al., 2024). Tools such as InVEST enable mapping of ecosystem services, including carbon storage, hydrological regulation, and urban cooling, while also informing conservation strategies addressing habitat fragmentation, invasive species, and poaching pressures (Jha et al., 2025;Reddy et al., 2024). This special issue, Advanced Geospatial Data Analytics for Environmental Sustainability: Current Practices and Future Prospects, includes seven research papers demonstrating cuttingedge applications of geospatial technologies, remote sensing, and advanced computational intelligence-including machine learning and fog computing -to address pressing environmental challenges such as resource sustainability, hazard prediction, ecological health, and infrastructure modernization. Based on nature of the manuscripts, we have divided the contributions into four groups/subsections as follows: The first thematic group highlights the use of Geospatial Artificial Intelligence (GeoAI) to improve environmental prediction, particularly for water-related systems. Elmotawakkil et al. (2025) advanced groundwater recharge modeling in arid environments using a tabular deep learning approach in the Feija Basin, Morocco. Their method integrated ten remote sensing-derived conditioning factors, including elevation, slope, and rainfall, with five AI models-TabNet, TabTransformer, Multilayer Perceptron (MLP), CatBoost, and AdaBoost. TabNet performed best, achieving 97.8% accuracy and an AUC of 0.99, followed by TabTransformer (97.6%), demonstrating the strong potential of deep learning for groundwater recharge prediction-critical for designing artificial recharge and rainwater harvesting practices. Similarly, Lakra et al. (2024) evaluated machine learning models for soil moisture estimation across wheat fields using Sentinel-1 SAR backscatter and in situ observations. Random Forest (RF) performed best with an RMSE of 7.06% and R² of 0.61 (correlation: 0.8), followed by the CNN model (RMSE: 8.44%; R²: 0.43; correlation: 0.67).The study demonstrates the viability of ML-SAR fusion in soil moisture mapping to support agricultural water management. This section focuses on remote sensing-based assessment of wildfire danger and harmful algal blooms. analysis, the study confirmed NDCI as a more reliable bloom indicator, with a strong linear relationship between NDCI and in situ phycocyanin concentrations (R² = 0.893). The approach offers a reliable framework for rapid harmful algal bloom detection and lake management. This theme includes studies employing geospatial tools and multi-criteria analysis for waste management planning and habitat sustainability evaluation. Aguiñaga-Vallejo et al. (2025) developed a decision-support framework for optimal solid waste disposal site selection in Guayas, Ecuador, integrating the Gravity Center of Waste Production (GCWP) with AHP-GIS analysis. Water resource protection emerged as the highest-priority factor, and three inter-municipal partnerships were proposed to reduce operational costs while improving environmental compliance. Xu et al. (2025) analyzed habitat quality dynamics in Guilin City, China, between 2001 and 2022 using the InVEST model and Geographically Weighted Regression. Average habitat quality measured 0.59, with 47.98% categorized as good to excellent, but overall trends declined. Spatial autocorrelation (Moran's I > 0.8) showed strong clustering, while elevation was identified as the most influential driver. Dual-factor interactions contributed more strongly to habitat variation than individual variables, emphasizing the complex nature of ecological degradation.