Application of the Soil Water Assessment Tool in a tropical agricultural catchment of the Panama Canal Watershed: Implications for its use in watershed management activities.

摘要

The Panama Canal Watershed (PCW) provides water to operate the Canal, generate hydroelectricity, and supply water provisions to the local and metropolitan populations. With a coaxed-out water budget, however, this region has little room to accommodate the possible effects of unsustainable land-use changes or of climate change, both of which threaten to alter water flows and timings. On the other hand, the water storage capacity of the canal reservoirs, necessary for water use during the dry season deficit, is compromised by sedimentation—the result of erosion and landslides on mismanaged lands. Given this context, tools must be developed to support conservation and sustainable resource use planning, watershed management activities, and risk forecasting. The Soil and Water Assessment Tool (SWAT), a physically based semi-distributed simulation watershed model, is an instrument that meets these criteria.;To assess the ability of SWAT application for use in the context of the PCW, the model was calibrated and validated for streamflow and sediment yield over a three year period (2004-2006) in the 75 km2 pilot study area of the Caño Quebrado River subbasin of the PCW, an area of burgeoning pineapple farms and with a history of cattle ranching. The model demonstrated exceptional performance for weekly average simulated streamflow and baseflow (all Nash Sutcliffe coefficients > 0.76 except for the baseflow validation period), generated little significant error, and demonstrated highly accurate predictions of annual cumulative water yield. Although SWAT was also able to simulate cumulative sediment yields with acceptable precision, the model was a poor predictor of monthly average sediment yield (calibration Nash Sutcliffe coefficient = 0.48). A qualitative and quantitative sensitivity analysis reveals that this is likely owing to the compound effects of a number of imprecise input parameters and data uncertainties, namely apropos the Modified Universal Soil Loss Equation (MUSLE) parameters for pineapple crops and pasture lands, the resolution and the reliability of the soil data used in this study, and the inability of SWAT to adequately model pineapple plant cover. Overall, this study illustrates that SWAT could potentially be a beneficial support tool for use in the PCW; however issues of data scarcity in the area will need to be resolved, including that of soil survey data, the spatial and temporal representativeness of streamflow and sediment yield field data, and estimates for MUSLE parameters. Modifications to the model framework for the groundwater and plant growth model components would also enhance prediction accuracy.