Development of a fully-distributed daily hydrologic feedback model addressing vegetation, land cover, and soil water dynamics (VELAS)

Highlights

• Quantifying the impacts of vegetation-land cover on hydrologic feedbacks.

• Fully coupled water dynamics through the soil zone.

• Integration of VELAS modeling with GIS database.

• Flexibility of the VELAS model for different scale of time and space.

Summary

A simple hydrologic feedback model has been developed to simulate daily responses of hydrologic processes including interception, runoff, evapotranspiration, infiltration, and recharge under various conditions of vegetation, land cover, and soil in a fully-distributed manner. The daily soil water balance is a key element to link surface and subsurface models as it calculates infiltration and groundwater recharge by considering a time delay routing through a vadose zone down to the groundwater table. MODFLOW is adopted to simulate groundwater flow and interaction with surface water components as well. The model also can easily be localized by simple modification of soil and crop properties. The actual application of the model for a watershed in the Geum River Basin in Korea showed reliable hydrologic feedbacks between the surface and subsurface hydrologic systems.

Introduction

The hydrological phenomena in a watershed are not independent events, but rather interactive feedbacks between hydrologic processes. Comprehensive and quantitative understanding of hydrologic feedback processes is important for the effective water budget and the provision against potential risks to water resources (Winter, 1998, Guo, 2002). When rainfall occurs, five hydrologic processes including interception, runoff, evapotranspiration, infiltration, and recharge occur in the terrestrial environments considering vegetation and surface–subsurface water flow dynamics (Brooks et al., 2003, Ward and Trimble, 2004, Brutsaert, 2005, Gupta, 2010). Various techniques and models have been developed to estimate the hydrologic processes.

Interception is the first hydrologic response to the rainfall, and its amount is significant in a highly vegetated region (Muzylo et al., 2009). Hence, the significance of interception in the water budget should not be underestimated. The interception is dependent upon the ecological properties of vegetation. Leaf area index (LAI) is the most frequently used as a critical property of vegetation for the interception calculation. The Soil and Water Assessment Tool (SWAT) model (Arnold et al., 1993) and Système Hydrologique Européen (SHE) model (Abbott et al., 1986a, Abbott et al., 1986b) calculate interception as a function of LAI. The Agricultural Policy/Environmental eXtender (APEX) model (Williams et al., 2008) calculates interception as a function of LAI and weight of above-ground plant materials. The interception process is also highly dependent on the rainfall intensity (Aston, 1979, Ramírez and Senarath, 2000). Thus, both LAI and rainfall should be taken into account in the interception calculation. Merriam (1960) suggested an exponential model accounting for LAI and rainfall. Aston (1979) improved the Merriam’s model by introducing a fraction of rainfall intercepted by canopy. The Merriam’s model has demonstrated its effectiveness in estimating interception for various cases (de Jong and Jetten, 2007, Kozak et al., 2007, Ragab and Bromley, 2010).

The water left from the interception process flows over the land surface in a form of runoff. Frequently used runoff estimation methods are the rational method, the curve number (CN) method (USDA-NRCS, 2004), and the Green–Ampt (GA) method (Green and Ampt, 1911). The rational method is easy to apply and good to estimate average peak discharge, but is limited to small watershed applications. The CN method and the GA method basically take into account the effect of soil moisture. Wilcox et al. (1990) showed that the GA method yields slightly better results than the CN method, however the CN method is simpler to use. King et al. (1999) performed a comparative study between CN and GA methods on the Goodwin Creek Watershed in the United States in different time scales and concluded that both models lead similar results. The performance of the CN method has been supported by its simplicity of the method, predictability showing average trends, stability to various watersheds, and responsiveness to complex watershed characteristics such as soil type, soil water, land cover, and surface condition (Ponce and Hawkins, 1996). Many efforts have been made to improve the CN method for various cases (Baltas et al., 2007, Kim and Lee, 2008, Hawkins et al., 2010) and integrate it into various hydrologic models (Schulze, 1995, Li and Gowing, 2005, Neitsch et al., 2005, Moretti and Montanari, 2007).

Evapotranspiration is a water transporting process from soil and vegetation into atmosphere. The Penman–Monteith (PM) equation is the most widely accepted evapotranspiration method. The PM equation, derived originally from the Penman equation (Penman, 1948), is a combined expression of energy, aerodynamics, and plant canopy resistance (Monteith, 1965). While the PM equation includes vegetation properties, other evapotranspiration equations such as the Hargreaves equation (Hargreaves et al., 1985) and the Priestley–Taylor equation (Priestley and Taylor, 1972) do not include them. Many hydrological models adopt the PM equation to estimate evapotranspiration from the vegetated surface (Aydin et al., 2005, Chen et al., 2005, Neitsch et al., 2005, Batelaan and de Smedt, 2007, Williams et al., 2008). In the Food and Agricultural Organization (FAO) irrigation and drainage paper 56, Allen et al. (1998) introduced the reference crop evapotranspiration (ET₀) for a hypothetical crop surface based-on the PM equation to provide consistent estimation of evapotranspiration in any regional and climate conditions. The FAO-PM method calculates the potential evapotranspiration (PET) for crop surface by multiplying ET₀ and a crop coefficient (K_C) representing a specific crop type. The benefits of using the FAO-PM method are its consistency of application to various crops and vegetation, simplified parameter requirement, and ease of use. The FAO-PM method has been adopted in many other evapotranspiration studies (Eilers et al., 2007, de Silva and Rushton, 2008, Sheikh et al., 2009).

The runoff, evapotranspiration, and recharge processes are fully coupled by the soil water balance in the shallow subsurface. Many studies have revealed the relationship between soil water contents and runoff under various climate conditions (Castillo et al., 2003, Scipal et al., 2005, Penna et al., 2011). The evapotranspiration from a vegetated or bare soil cover consumes both residual water and infiltrated water in the root zone. Hence, soil water content, commonly represented as soil moisture deficit (SMD) in regards of the soil water stress against evapotranspiration, is a major controlling factor for the evapotranspiration process (Bonsu, 1997, Clark, 2002, Allen et al., 2005, Li et al., 2007). The amount of recharge is usually calculated as a residual amount of water from the soil water balance in a watershed model (Sophocleous, 2004). In the SHE model and the Agricultural Catchments Research Unit (ACRU) model (Schulze, 1995), a soil profile is assumed to have two layers, in which an exceeding amount of water over the maximum holding capacity of the layer after budgeting becomes the amount of recharge. The use of a two-layer water balance has advantages of model simplification and ease of application. The APEX model and the SWAT model adopt a similar concept, but the soil profile is assumed to have multiple layers. However, such layer-based water balance methods are only suitable for a shallow aquifer system. The SWAT model uses an exponential decay weighting function to calculate time delay of recharge through a vadose zone in the soil profile.

The overview of hydrologic processes occurring in the land environments shows that the spatial and temporal heterogeneity of vegetation and land cover is an important factor affecting water balance in a watershed. While most of the crop land is covered with vegetation during the growing season, bare soil is dominant during the non-growing season. Anthropogenic activities such as an expansion of urban area increase the area of impervious surface and reduce vegetated or bare soil areas. Cruise et al. (2010) also showed that a large land transition from forest to agricultural land caused a decrease of stream discharge in the southeastern United States for the last 20 years. Zhang and Schilling (2006) investigated the effect of two different land covers, grass and bare soils, on the hydrologic responses in the Walnut Creek watershed in Iowa, US, to find out that reduced groundwater recharge was observed on the grass covered area due to the lower soil moisture by the evapotranspiration process. Hence, the vegetation-land cover dynamics should be carefully considered in the soil–water balance calculation.

It is inevitable that the integrated watershed models require an enormous amount of input data and extensive understanding of each hydrologic process to assure the accuracy of simulation results. A rapid process of hydrological modeling is also important in terms of cost and efficiency, but sometimes huge data requirements of modeling process obstruct a practical use of the model. The Water and Energy Transfer between Soil, Plants, and Atmosphere under quasi Steady State (WetSpass) model by Batelaan and de Smedt (2007) is a good example of the model capability of hydrologic simulation using readily available data. Although the WetSpass model excludes in-depth soil water balance and only considers wet and dry seasons, its optimized implementation of hydrologic processes and simple data requirement still assure model accuracy and a rapid process of simulation (Dams et al., 2008, Tilahun and Merkel, 2009, Dujardin et al., 2011).

A new model referred to as the VEgetation-LAnd cover-Soil water dynamics model (VELAS) is introduced in the present paper. VELAS calculates interception, runoff, evapotranspiration, soil water variation, and recharge under various conditions of vegetation, land cover, and soil in a fully-distributed manner. A daily response of soil water balance depending on the vegetation-land cover dynamics is a key feature of the VELAS model. The application to a watershed in the Geum River Basin in Korea demonstrates that the required input dataset of the model are commonly obtainable and simple, but the modeling results show accurate responses of hydrologic processes. The model is validated with the observed stream flow data and groundwater elevation data to show the VELAS capability of a fully synchronized simulation of hydrologic processes.