How detail is enough? Large scale modeling limitations

Abstract:

1.Black sea Basin

Black Sea Basin (Figure 1) drains rivers of 23 European and Asian countries from an area of 2.3 million km²to the Black Sea. Major rivers draining into the Black Sea include Danube, Dnieper and Don. The greatest sources of diffuse pollution are agricultural and households not connected to sewer systems. The Basin is inhabited by a total population of around 160 million people [BSEI, 2005]. The dominant landuse in the basin is agricultural with 65% of coverage according to MODIS Land Cover [NASA, 2001]. The Black Sea Basin (BSB), suffers from ecological unsustainability and inadequate resource management leading to severe environmental, social, and economical problems.

2.Modeling tool and set up

We used the Soil and Water Assessment Tool (SWAT) [Arnold et al., 1998] to model the hydrology of the BSB coupling water quantity, water quality, and crop yield components. SWAT is a process-based, continuous-time model that operates on a daily time step to predict the impact of management practices on water, sediment and agricultural chemical yields at the catchment scale [Arnold et al., 1998]. The spatial heterogeneity of the watershed in SWAT is preserved by topographically dividing the basin into multiple sub-basins. These subbasins are further subdivided into lumped hydrologic response units (HRUs) comprising homogeneous landscape, soil, and landuse characteristic units that are not spatially identified in a given subbasin. The hydrologic cycle as simulated by the model can be separated into the land phase and the routing phase. The land phase simulates runoff and erosion processes, soil water movement, evapotranspiration, crop growth and yield, soil nutrient and carbon cycling, pesticide and bacteria degradation, and thus controls the amount of water, sediment, nutrient and pesticide loadings entering the main channel in each subbasin. A wide range of agricultural practices including tillage, fertilizer and manure application, subsurface drainage, irrigation, ponds and wetlands are accounted for in SWAT.

The routing phase controls the movement of water, sediments, nutrients, etc through the channel network to the watershed outlet. SWAT model has undergone continuous development [Gassman et al., 2007] in the last two decades. In its current version, SWAT2009 [Neitsch et al., 2011] which is used in the present study, several watershed processes can be represented by alternative methods. Table 1 gives an overview of the relevant methods used in this study. A detailed description of the model can be obtained from Neitsch et al. [2011].

The hydrological model of the BSB was calibrated and validated considering sensitivity and uncertainty analysis. River discharges, nitrate loads, and crop yields were used to calibrate the model. Employing grid technology improved calibration computation time by more than an order of magnitude. We calculated components of water resources such as river discharge, infiltration, aquifer recharge, soil moisture, and actual and potential evapotranspiration. Furthermore, available water resources were calculated at sub-basin spatial and monthly temporal levels. The aim of this study is to discuss the challenges of building a large-scale model in fine spatial and temporal detail and highlight the pitfalls.

Table 1. SWAT processes representation as used in the study.

Processes/components	Method [Neitsch et al., 2011]
Evapotranspiration	Hargreaves
Surface runoff	SCS curve number equation
Erosion	Modified universal soil loss equation
Lateral flow	Kinematic storage model
Groundwater flow	Steady-state response from shallow aquifer
Stream flow routing	Variable storage routing

3.Challenges

In highly managed watersheds, natural processes play a secondary role. Examples of managements are dams and reservoirs, water transfers, and irrigation from deep wells. On the other hand, watershed models suffer from large model uncertainties. These can be divided into: conceptual model uncertainty, input uncertainty, and parameter uncertainty. The conceptual structural uncertainty include: model uncertainties due to simplifications in the conceptual model, model uncertainties due to processes occurring in the watershed but not included in the model, and model uncertainties due to processes that are included in the model, but their occurrences in the watershed are unknown to the modeler.

We investigated the poorly simulated river discharges in Black Sea Basin hydrological model one by one using the visualization module of SWAT-CUP [Abbaspour, 2011]. This involves projection of the study area on the Microsoft BING map to identify the reasons for the inadequate simulations. Examples of these include positioning the outlets on a wrong rive (Figure 2a,b). Wetland processes are not represented in BSB hydrological model and hence bring additional structural uncertainties due to processes occurring in the watershed but not included in the model (Figure 3). Other major problems result from an outlet being positioned downstream of a reservoir. (Figure 4, 5). Other problematic situations may arise when outlets are in a highly populated or agricultural region where water management and water transfers are large. Constructions of dams for irrigation and power generation purposes as well as other water management practices such as water abstraction and diversion create major difficulties for model calibration. As management information are usually not available, proper cautions need to be taken during calibration. These include converting outlets to inlets, weighing those outlets under the influence of management less in the objective function, or removing the outlets downstream of reservoirs from the calibration process.