Assessment of the WRF-Hydro uncoupled hydro-meteorological model on flashy watersheds of the Grande Terre tropical island of New Caledonia (South-West Pacific)

Abstract

Study region: Six watersheds of the main island of New Caledonia. Study focus: In the context of a projected reduction of extreme precipitation events by ~ 20% by 2100 combined to large land cover variability, a distributed hydrological approach was adopted over New Caledonia. The WRF-Hydro model was implemented using a one-way WRF meteorological forcing, with rainfall being spatially interpolated from rain gauge observations using the Thiessen polygon and IDEW methods. The Dynamically Dimensioned Search algorithm was applied for calibration on three contrasted flash-floods. Two-year validation allowed the dynamics of the watersheds to be studied and compared. New hydrological insights for the region: Good results in terms of Nash-Sutcliffe efficiency were reached on all watersheds (NSE ≥ 0.6). WRF-Hydro showed consistent water budget components for two flash-floods with 90% correlation between simulated and observed runoff coefficients of all basins. Adjusted physical parameters matched known watershed morphological and hydrological features. Both methods of rainfall interpolation produced similar NSE but strongly different water distributions, with simulated soil moisture changes varying by up to 60% during flash-floods. Spatial patterns of soil moisture and flood decomposition led to improved understanding of watershed behavior and to a revision of infiltration properties for one of them. Overall, WRF-Hydro proved to be a suitable framework to better understand hydrological processes and efficiently handle prospective modifications of New Caledonia’s land cover and climate regimes.

Keywords

• New Caledonia
• Hydrological modeling
• WRF-Hydro
• DDS parameter adjustment
• Rainfall spatial interpolation
• Hydrograph separation