Correa (2026) Runoff Potential Index (RPI): 3D modelling of surface-driven hydrological dynamics for drought resilience

Identification

• Journal: Scientific Reports
• Year: 2026
• Date: 2026-01-07
• Authors: Edgar S. Correa
• DOI: 10.1038/s41598-025-34699-5

Research Groups

• School of Engineering, Pontificia Universidad Javeriana, Bogota, Colombia
• UMR AGAP Institut, Univ. Montpellier, CIRAD, INRAE, Institut Agro, Montpellier, France
• CIRAD, UMR AGAP Institut, Montpellier, France

Short Summary

This study introduces the Runoff Potential Index (RPI), a novel divergence-based terrain metric, and integrates it with Earth observation-driven CERES-Rice crop modeling to assess drought vulnerability and optimize sowing strategies in rainfed agricultural systems. The RPI demonstrates superior sensitivity to microtopographic variations, and the combined framework provides field-specific recommendations to mitigate significant yield losses, supporting global climate adaptation.

Objective

• To introduce the Runoff Potential Index (RPI) for 3D modeling of surface-driven hydrological dynamics and integrate it with Earth observation-driven CERES-Rice crop modeling to assess drought vulnerability and provide adaptation strategies in rainfed agricultural systems.

Study Configuration

• Spatial Scale: Southern and southeastern Senegal (Casamance and Eastern Senegal regions), covering approximately 70,000 km² across 506 georeferenced locations (approximately one point per 140 km²). Topographic data at 30-meter resolution, soil data at 250-meter resolution, and climate data at approximately 50-kilometer resolution. RPI analysis aggregated at 2.5-kilometer search radius.
• Temporal Scale: 20-year simulation period (2000–2019) with daily time-step crop modeling, evaluating five sequential sowing dates at 15-day intervals (late June to late August).

Methodology and Data

• Models used:
  • Runoff Potential Index (RPI): A novel divergence-based terrain metric, formulated as (\text {RPI}(x, y) = {\nabla ^2 z} / ({|\nabla z| + \varepsilon })).
  • CERES-Rice model: A process-based crop model within the DSSAT framework.
  • Comparative analysis with Topographic Wetness Index (TWI).
• Data sources:
  • Topographic: Shuttle Radar Topography Mission (SRTM) 1 Arc-Second Global dataset (30 m spatial resolution).
  • Soil: SoilGrids global soil information system (250 m spatial resolution) for sand, silt, clay percentages, soil organic carbon, pH, bulk density, cation-exchange capacity, and volumetric water content at 15 cm, 30 cm, and 100 cm depths.
  • Climate: NASA POWER (Prediction Of Worldwide Energy Resource) database (approximately 0.5° or 50 km spatial resolution) for daily minimum/maximum temperature, solar radiation, wind speed, relative humidity, and precipitation.
  • Genetic Coefficients: Calibrated genetic coefficients for NERICA rice cultivars from previous field trial data in Senegal.
  • Computational Framework: Custom-developed Runoff Potential Index: Upland–Lowland Differentiation Toolbox (v1.0.0) in MATLAB R2024b.

Main Results

• The Runoff Potential Index (RPI) demonstrated superior analytical sensitivity and numerical stability compared to the Topographic Wetness Index (TWI), particularly in low-gradient terrains and subtle elevation changes (0.7–1.8 m), accurately detecting microtopographic variations critical for water retention.
• Terrain analysis using RPI revealed that lowland areas achieved approximately 200 kg/ha higher yields than upland areas.
• CERES-Rice simulations (2000–2019) identified optimal sowing windows, with delayed sowing (S4-S5) causing significant yield reductions exceeding 1,500 kg/ha (45–73% loss) compared to early sowing (S1-S2).
• The second week of July (S2) was identified as the optimal sowing date for 84.2% of the territory, achieving mean yields of 3090 ± 211 kg/ha.
• Annual precipitation trends showed an initial increasing phase (2000-2010: +93.3 mm/year) followed by a declining phase (2010-2019: -36.7 mm/year).
• While lowland areas provided systematic yield advantages (200–300 kg/ha), these benefits were smaller than the yield variations associated with sowing date optimization, reinforcing the primacy of temporal over spatial optimization strategies.

Contributions

• Introduction of the Runoff Potential Index (RPI), a physically-based, divergence-driven terrain metric that captures flow dynamics and maintains sensitivity in low-gradient systems where conventional indices fail.
• Development of a fully satellite-driven framework utilizing NASA POWER, SoilGrids, and SRTM data to characterize water redistribution patterns and assess drought vulnerability across regional scales without ground-based measurements.
• Systematic comparison between terrain analysis (RPI) and process-based crop model outputs (CERES-Rice), highlighting opportunities to enrich crop models with explicit representation of morphology-driven water redistribution.
• Application of the framework to drought vulnerability assessment across 506 locations and 20-year climatic scenarios, providing spatially explicit environmental predictions and field-specific sowing recommendations to prevent 45–73% yield losses, directly supporting SDG 13.1 and 13.3.

Funding

• Agropolis Fondation (CropModAdapt project, Contract No. 2201-026)
• ClimBeR initiative - France-CGIAR action plan on climate change (ICARDA Agreement No. 200303)

Citation

@article{Correa2026Runoff,
  author = {Correa, Edgar S.},
  title = {Runoff Potential Index (RPI): 3D modelling of surface-driven hydrological dynamics for drought resilience},
  journal = {Scientific Reports},
  year = {2026},
  doi = {10.1038/s41598-025-34699-5},
  url = {https://doi.org/10.1038/s41598-025-34699-5}
}