H104-04
An assessment of the ability of the SWAT model to simulate land-use change impacts in tropical montane areas of Central Veracruz, Mexico

Thursday, 10 December 2020: 19:09
Virtual
Sergio Miguel Lopez Ramirez1, Alex S Mayer2, Leonardo Saenz2, Lyssette E Munoz Villers3, Friso Holwerda4, Nathaniel Thomas Looker5 and Z. Carter Berry6, (1)Michigan Technological University, Houghton, MI, United States, (2)Michigan Technological University, Civil and Environmental Engineering, Houghton, MI, United States, (3)Universidad Nacional Autonoma de Mexico, Mexico City, Mexico, (4)Universidad Nacional Autonoma de Mexico, Centro de Ciencias de la Atmósfera, Mexico City, Mexico, (5)Montana State University, Bozeman, MN, United States, (6)Chapman University, Orange, CA, United States
Abstract:
Contrasting a physically-based ecohydrological model against data and expert knowledge of the dominant hydrologic processes in catchments dominated by tropical montane cloud forest (TMCF) offers the opportunity to advance our understanding of land-use impacts on the hydrologic cycle. This study evaluates an improved version of the Soil and Water Assessment Tool model for the Tropics (SWAT-T, hereafter SWAT) to simulate hydrologic partitioning in areas influenced by TMCF and anthropogenic land covers. We evaluated the effectiveness of SWAT in capturing the dominant processes in a watershed influenced by TMCF ( 42 km2). Specifically, we assessed the performance of the potential evapotranspiration (PET) methods available in SWAT in these environments; and evaluated the capacity of the SWAT model to simulate streamflow in contrasting land covers (mature and secondary TMCF, shade coffee, and pasture).

Meteorological and hydrological measurements combined with extensive soil and vegetation data were integrated into the SWAT model. PET was calculated using the three methods provided by SWAT: (i) Penman-Monteith (PM), (ii) Hargreaves (HA), and (iii) Priestly-Taylor (PT). A dense network of rainfall gauges yielded an accurate spatial representation of precipitation (P). Plant growth was manually calibrated based on LAI measurements combined with MODIS time series, and the canopy water storage capacity was adjusted with field data. Finally, a temporally distributed Global Sensitivity Analysis and calibration of daily streamflow at the watershed outlet were conducted, including the use of signature metrics for five parts of the hydrograph at the catchment scale.

Satisfactory model performance SWAT was achieved with respect to predicting daily streamflow at the outlet of the catchment (R2 > 0.7, NS >0.5, PBIAS < 5%). The model accurately represented the observed baseflow/streamflow ratio; however, simultaneously capturing the extreme parts of the hydrograph proved more challenging, specifically for very low flow conditions. It was found that the model underestimated the rainfall interception in the forests. The HA and PT methods provided more realistic PET estimates than the PM method. The model needs improvements in the simulation of canopy evaporation to better capture ET in forests.

<< Previous Abstract | Next Abstract >>