Representation of hydrological processes in a rice-growing watershed in central Côte d'Ivoire, using the SWAT+ model

Código

I-EBHE0015

Autores

NGUESSAN Konan Jean-Yves, Konan-Waidhet Arthur Brice, Michal Habel, Siguyama Hironobu, Dibi Brou

Tema

WG 1.12: Development & application of river basin simulators

Resumo

Adequate management of water resources in West African rice-growing catchments remains a crucial issue for ensuring food and environmental security. In this study, the flexible version of the agro-hydrological model Soil and Water Assessment Tool (SWAT), called SWAT+, was used to simulate the hydrological processes of the Lake Taabo catchment (LTC) in central Côte d'Ivoire. The LTC includes the Nanan reservoir catchment (NRC), which is used to irrigate a rice-growing scheme. The SWAT+ model includes several improvements, such as the definition of landscape units, which enable better representation of spatio-temporal dynamics. The objectives of this study were therefore twofold: (i) firstly, to assess the ability of this improved version of the SWAT model to reproduce river flows, and (ii) secondly, to represent the different components of the water balance spatially distributed over the LTC, with particular interest in the NRC. Sensitivity analysis, calibration, and validation were carried out using monthly flow data from the Bouaflé station from 1984 to 1989 for calibration and from 1991 to 1996 for validation. The results show that the SWAT+ model is particularly sensitive to eight parameters, including the new parameter CN3_SWF, which controls soil saturation and proved particularly important during calibration. The LTC SWAT+ model, calibrated and then validated, performed very satisfactorily, with a drop in performance from calibration to validation. The statistical criteria for calibration were 0.78, 0.81, -18.58, and 0.47, respectively, for NSE, R², PBIAS, and RSR. At validation, the criteria were NSE: 0.69, R²: 0.70, PBIAS: -23.9, and RSR: 0.56. Analysis of the results also reveals considerable spatial heterogeneity in water balance components on the LTC. Rainfall and evapotranspiration are higher south of the LTC, while runoff and percolation are higher north of the LTC. On the NRC, rainfall, runoff, evapotranspiration, and percolation average 950, 210, 115, and 370 mm, respectively. The LTC water balance shows an excess of water. This study demonstrates the applicability of the SWAT+ model in the NRC and could provide useful information to decision-makers about rational water management in rice-growing catchments to increase production for sustainable food and environmental security.