This solution combines multi-source satellite remote sensing, ecohydrological modeling, and community science to address spring decline and water insecurity caused by afforestation and land-use changes in Nepal's Middle Hills. The integrated approach offers a pathway to scientifically informed, community-driven forest and water management.
- Satellite Data Fusion: The first core strategy involves fusing multi-temporal and multi-sensor satellite data to assess vegetation trends, hydrological changes, and potential spring recharge zones.
- Vegetation Monitoring: Time-series NDVI, EVI, and LAI are derived using Sentinel-2, Landsat, NOAA and MODIS. These indices help detect vegetation growth trends and assess forest types based on phenological signatures.
- Cloud Mitigation: Nepal’s rugged terrain and monsoon conditions create persistent cloud cover challenges. While no perfect cloud-removal technique exists, we aim to apply machine learning and established algorithms like CLAY3 or Fmask to improve data quality for vegetation and land cover analysis.
- Hydrological Metrics: ET using MODIS, Soil moisture is mapped using SMAP data, downscaled using terrain parameters such as slope and elevation. GRACE data informs groundwater trends. Satellite-based precipitation datasets are validated against DHM station data to compensate for missing or sparse in-situ observations.
- RHESSys Ecohydrological Simulation: The RHESSys model simulates the complex interactions between vegetation, soil moisture, surface and subsurface water flow, and groundwater storage. The model is run in growth mode to evaluate how forest type changes influence spring discharge.
- MODIS ET and Sentinel-1 soil moisture serve as validation inputs.
The model provides spatial outputs including groundwater depth, lateral flow, and baseflow dynamics—critical for delineating micro-watersheds and assessing recharge efficiency.
- Recharge Zone and Spring Hotspot Mapping: Topographic indices such as TWI (Topographic Wetness Index) and HAND (Height Above Nearest Drainage), derived from ASTER DEMs, are used to identify spring recharge zones.
- These zones are further validated using RHESSys outputs, satellite data layers, and available field measurements.
- Machine learning (e.g., Random Forest) and participatory mapping help cross-check locations of active and declining springs.
- The resulting maps guide protection measures and afforestation policies targeting hydrologically sensitive areas.
- Field Validation and Community Co-Design: Local participation through spring monitoring and mapping ensures the integration of indigenous knowledge with scientific analysis. Field measurements validate model predictions and support community-driven management strategies.
Requirements
Data
- Remote sensing: NDVI, EVI, LAI, Evapotranspiration (MODIS), Soil Moisture (SMAP), satellite-based precipitation, terrestrial Groundwater storage (GRACE), Land Use Land Cover, DEM (ASTER).
- In-situ: Precipitation and temperature data from DHM Nepal
Software
- Google Earth Engine (cloud computing, satellite data analysis)
- GRASS GIS and QGIS (terrain analysis, TWI, HAND)
- RHESSys (eco-hydrological modeling)
- R/Python (statistical modeling, ML integration)
- LEAF Toolbox for LAI/phenology
Physical Requirement
- Cross-validation of spring locations and groundwater depth measurements through field visits.
Priority Support Areas: To realize objectives, we seek support in the following areas:
- High-Resolution and Cloud-Free Satellite Data: Technical assistance in accessing and processing Sentinel-1 SAR, Sentinel-2, and Landsat data, and applying cloud-removal algorithms.
- Downscaling Remote Sensing Products: Assistance in refining MODIS-based NDVI, EVI, LAI, and phenological indicators using auxiliary terrain and land cover datasets.
- Integration of Hydrological and Remote Sensing Data: Guidance on synchronizing outputs from RHESSys with MODIS, SMAP, and GRACE datasets for robust cross-validation.
- Mapping Recharge and Spring Zones: Technical support in combining terrain indices with RHESSys-derived metrics to map spring recharge zones and inform land-use planning.
Outline steps for a solution
Phase 1: Satellite Analysis and Vegetation Mapping (completed)
- Use GEE to analyze NDVI, EVI, LAI, ET from NOAA, MODIS, Sentinel, and Landsat.
- Define AOI, set date ranges, apply quality filters, visualize maps, and export CSV time series.
- LULC maps are from Global Land Cover (Chinese Academy of Sciences).
- Enhanced vegetation index (EVI)
- Leaf area index (LAI)
- Normalized Difference Vegetation Index (NDVI)
- Evapotranspiration (ET)
- Pending: Improve cloud filtering using CLAY3, Fmask; downscale to higher resolution.
Phase 2: GIS and Terrain Modeling (In Progress)
- Use DEM, LULC, TWI, HAND for recharge/discharge mapping.
Analyze terrain factors (slope, curvature, aspect, valley) for moisture prediction.
Phase 3: Hydrological Simulation and Analysis (To Do)
- Run RHESSys in growth mode to simulate hydrological-vegetation dynamics.
- Validate against spring discharge logs.
- Output: groundwater depth, saturation deficit, soil moisture, flow trends, seasonal water availability.
Phase 4: Community Co-Design and Policy Translation (To Do)
- Share vulnerability maps with CFUGs and municipal planners.
- Recommend native species afforestation rather than heavily water dependent species.
- Promote thinning, selective logging, litter removal, etc. (forest management plant).
- Water availability zone mapping for settlement
Note: "The full code and comprehensive instructions for running the model are provided in this repository."
Results
- Preliminary analysis shows increasing trends in both evapotranspiration and LAI, indicating higher water consumption by vegetation and less water available for downstream use.
- Mapped recharge zones and high-risk spring areas.
- Groundwater storage trends over time and identification of water-available zones suitable for settlement planning.
- Spatial maps and hydrologic models to support forest and water governance.
- Policy briefs on forest-water tradeoffs and spring recovery.
- Community awareness on how certain forest types and species are accelerating water loss and increasing water stress for downstream communities, prompting migration.
- Technical findings paired with management strategies offer actionable insights for land-use planning and ecosystem resilience.

