Editorial: Remote sensing and data science for mapping climate change impacts in mountainous regions

Mountain environments occupy a paradoxical position in the Earth system. They are among the most ecologically rich yet environmentally fragile landscapes on Earth. As regional water storage, biodiversity hotspots, and early indicators of climatic change, they demand regular monitoring. Their heterogeneous topography, climatic severity, and geographic remoteness pose a limitation for field data Research Topic. The fragmented ground-based observation networks create a persistent data gap that weakens the capacity to understand and respond to change.

Remote sensing techniques, with their synoptic view and frequency of observation, offer the cost-effective solution to fill these data gaps. This Research Topic presents six contributions that applied remote sensing and machine learning to mountain environments. Thes contributions present methodological trends that extend beyond any single study. First, nearly all depend on a hybrid design that couples multi-source Earth Observation data, optical, thermal, and increasingly SAR, with machine learning or deep learning architectures (Random Forest, XGBoost, ANN, DLNN, SVR, and ensemble combinations thereof), rather than either data stream alone. Second, several papers explicitly incorporate interpretability (SHAP-based attribution in Li et al.'s forecasting work) or comparative model benchmarking (Paramanik et al.'s four-architecture ensemble), a sign that the field is moving from raw predictive accuracy toward accountable, decision-ready outputs. Third, these studies highlight the value of combining optical and microwave sensors to overcome cloud cover in monsoon, a solution applied effectively across flood, vegetation, and limnology research.

Beyond their methodological contributions, these studies help us understand how climate change is already changing mountain regions. The groundwater and flood studies show that both water shortages and floods are getting worse at the same time, not as separate problems, but because the same causes (changing rainfall and melting glaciers) drive both. The vegetation studies show that plant health is harder to judge than it looks: an area can look “green” and healthy on satellite images while its ecosystem is actually declining, which means simple greenness measurements can be misleading. The air quality and lake/river studies show that these changes also affect human health and water quality, not just plants and hydrology, problems in one part of the system spread to others. Together, this shows that mountain hazards are connected, not separate, and adaptation plans need to treat them that way instead of tackling each one on its own.

Hydroclimatic Vulnerability and Atmospheric Forecasting: Kunwar et al. offer a systematic synthesis of the impact of climate change on groundwater. Their review highlights the inherent data scarcity and coarse spatial modeling prevent current climate frameworks from accurately representing complex aquifer recharge dynamics under altered precipitation and glacial retreat scenarios. Data scarcity is a major constraint to developing reliable adaptation strategies in mountain hydrogeological systems. Li et al. shift the focus toward predictive modelling by developing a spatially explicit forecasting system for extreme climate events in complex terrains. The fusion of multi-source remote sensing indicators like Normalized Difference Vegetation Index (NDVI), Land Surface Temperature (LST) and albedo, with ground-based meteorological records, processed through an XGBoost (eXtreme Gradient Boosting) model with SHAP-based interpretability, is methodologically sound. Their ability to predict indicators like maximum daily temperature and single-day precipitation anomalies with high spatial precision represents a kind of early-warning infrastructure for exposed mountain regions. Paramanik et al. address a different kind of hazard, projecting flood susceptibility for the eastern Himalayan Kalimpong district out to the year 2100. Combining CMIP6 climate projections across three Shared Socioeconomic Pathways with four machine learning architectures (ANN, DLNN, XGBoost and a deep-boosting hybrid) they show that rainfall, slope and land use are the dominant predictors of flood risk and that susceptibility expands sharply as emissions rise from moderate to extreme scenarios, underscoring the value of model ensembles for climate adaptation planning in steep, monsoon-dominated terrain.

Vegetation Dynamics under Compound Environmental Stress: Two studies examine the responses of terrestrial vegetation to climatic and atmospheric pressures. Mehmood et al. deploys a dual-index framework, kernel NDVI and the Vegetation Health Index (VHI), both derived from remote sensing sources and using a Random Forest algorithm to disentangle the two phenomena that can appear superficially similar in spectral data. The transient greening that follows flood inundation and the persistent, physiologically meaningful decline driven by sustained moisture deficit and thermal stress. This distinction is important for land-use policy. An apparent increase in vegetation greenness is not evidence of ecosystem recovery if it is moisture-driven and followed by accelerated degradation. Li et al. examines the spatial and temporal relationship between fine particulate matter and fractional vegetation cover across altitudinal and urban-rural gradients. They find a consistently negative correlation between vegetation density and pollution concentration, suggesting that mountainous landscapes function as natural ecological refugia and regulating regional air quality. This has both monitoring and policy implications, particularly as land-use pressures encroach on vegetated buffer zones at mid-elevations.

Limnological monitoring serves as an environmental bellwether that rapidly responds to perturbations in catchments, precipitation regime, sediment flux, nutrient loading and temperature. These changes offer a valuable window into the cumulative effects of upstream land-use changes, agricultural runoff, and climate-driven monsoon inflows. Joshi et al. address this through an ensemble machine learning architecture combining Support Vector Regression and Random Forest, applied to long-term Chlorophyll-a dynamics in Phewa Lake, Nepal. Their methodological contribution lies in contrasting optical (Sentinel-2) and microwave (Sentinel-1) data. Cloud cover during the monsoon season limits optical retrievals, while the SAR-based approach can overcome this limitation. The resulting framework is scalable and its extension to other monsoon-dominated mountain lakes represents a concrete research priority.

Future research in mountain environments should prioritize foundation models combined with multi-sensor integration, fusing satellite, drone, and LiDAR data, to address persistent data scarcity and better capture fine-scale terrain variability in steep, complex landscapes. A second priority could be translating these frameworks into near-real-time monitoring systems capable of delivering timely early warnings for floods and other climate-related hazards. Finally, integrating these physical monitoring systems with socioeconomic vulnerability data will be essential to ensure that environmental forecasts translate into equitable, community-centered policy responses. Despite ongoing limitations in satellite and climate models’ ability to resolve fine-scale mountain processes, these three priorities, model-sensor fusion, real-time delivery, and vulnerability integration, offer a clear roadmap for advancing the field.

We thank to the contributing authors for their rigorous and challenging work, and the reviewers for their critical engagement. This Research Topic serves as a record of current capability foundation and a provocation for the next-generation of inquiry into the world’s most climatically vulnerable mountain landscapes.

Statements

Summary

Keywords

data science, Himalaya, machine learning, mountain, remote sensing

Citation

Mishra B, Kumar M and Stroppiana D (2026) Editorial: Remote sensing and data science for mapping climate change impacts in mountainous regions. Front. Environ. Sci. 14:1940158. doi: 10.3389/fenvs.2026.1940158

Received

16 July 2026

Revised

12 August 2026

Accepted

13 August 2026

Published

07 September 2026

Volume

14 - 2026

Edited and reviewed by

Sawaid Abbas, University of the Punjab, Pakistan

Updates

Check for updates

Copyright

© 2026 Mishra, Kumar and Stroppiana.

This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

*Correspondence: Bhogendra Mishra, bmishra@sciencehub.org.np

†

ORCID: Bhogendra Mishra, orcid.org/0000-0002-8998-8160; Manoj Kumar, orcid.org/0000-0003-2360-5451; Daniela Stroppiana, orcid.org/0000-0002-5619-4305

Disclaimer

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.