GeoAI Framework for Hydrological Resilience in Intermontane Basins: Integrating AI and NbS Solutions

An India–Japan research collaboration under the JST LOTUS Programme (ASPIRE type). The project combines geospatial AI with Nature-based Solutions (NbS) to move from hazard maps to actionable, nature-based urban resilience.

The team: Bruno Adriano (Japan-side principal investigator, IRIDeS, Tohoku University), Mahua Mukherjee (India-side principal investigator, IIT Roorkee) and Kshitiz Agarwal (invited researcher, IIT Roorkee).

The project adapts one spatial-planning framework to two cities while respecting the distinct hazards of each.

• Dehradun, India — a Himalayan basin city facing monsoon-driven flooding and dry-season water scarcity, with growing pressure on drainage corridors and open land. Here NbS can act as flood buffers, support groundwater recharge and restore river corridors.
• Sendai, Japan — a coastal city exposed to tsunamis, whose reconstruction after the 2011 Great East Japan Earthquake offers lessons for resilient spatial planning: tsunami resilience, safe open space and recovery. Sendai is studied as an exploratory case.

Shared research question: where is non-built-up land available that can support feasible, effective Nature-based Solutions?

The framework moves from multi-source geospatial data to planning decisions that are ready for intervention.

• Data inputs — remote sensing, digital elevation models (DEMs), drainage networks, urban layers and a catalog of NbS types.
• Hazard and growth — urban growth projected to 2050, and flood and tsunami scenario modeling.
• Cross-context AI — transfer learning adapts models trained on data-rich Japanese datasets to the data-scarce Himalayan setting.
• Spatial optimization — candidate sites for NbS and blue-green infrastructure (BGI) are ranked with multi-criteria decision-making (MCDM) and feasibility criteria.
• Planning outputs — opportunity maps, intervention options and a suitability matrix.

The framework is tested in Dehradun (flooding and water stress) and Sendai (tsunami risk) to support resilience planning.

Research objectives

• Urban growth and opportunity mapping — model land-use change to 2050 and locate non-built-up “opportunity windows” where NbS could be placed.
• Dynamic hazard forecasting — simulate monsoon flooding and tsunami inundation using geospatial data and AI-assisted methods.
• Spatial optimization for NbS and BGI — use MCDM to rank feasible sites by runoff reduction, groundwater recharge potential and technical suitability.

Ten-month research stay at IRIDeS (February–December 2027)

• Months 1–4: urban growth and hazard mapping for Dehradun with multi-source satellite data and transfer learning.
• Months 5–6: an NbS optimization matrix built with objective statistical weighting.
• Months 7–10: spatial simulations for NbS siting, an NbS roadmap, exploratory modeling of the Sendai datasets, and manuscripts.

Expected outputs: opportunity maps of sites for possible intervention, an NbS/BGI priority matrix with feasibility-informed site rankings, and an urban resilience framework supported by research articles.