LISFLOOD hydrological model
LISFLOOD is a spatially distributed hydrological rainfall-runoff model that can simulate the main hydrological processes occurring in a catchment. LISFLOOD has been developed by the Joint Research Centre (JRC) of the European Commission since 1997 (De Roo et al., 2000; De Roo et al., 2001; Van der Knijff et al., 2008; Burek et al., 2013) building on earlier models such as LISEM (De Roo et al., 1996), HBV (Bergström, 1995) and WOFOST (Diepen et al., 1989; Supit et al., 1994 ) and with the specific objective of developing a tool that could be used in large and transnational catchments for a variety of applications, including: flood simulation and forecasting; water resource simulation in river basins; assessing the effects of land-use changes; assessing the effects of measures such as river regulation measures and water efficiency measures; assessing the effects of climate change.
LISFLOOD is composed of different components (sub-models) that are capable of separately simulating different hydrological processes, it includes a one-dimensional hydrodynamic channel routing model for flood waves propagation in the rivers network and components to simulate water retention and release in reservoirs and lakes. The model uses a D8 modelling scheme where each model pixel can have up to 7 contributing pixels and only one outflow pixel. Pixels connection is described through the local drain direction map (LDD), which describes the flow directions from each cell to its steepest down slope neighbour.
The model is driven by meteorological forcing data (precipitation, temperature, potential evapo-transpiration, evaporation rates for open water and bare soil surfaces). At every time step and for every grid-cell, LISFLOOD calculates a complete water balance. The list of simulated processes for each grid cell includes: interception and leaf evaporation, evapo-transpiration, evaporation from soil surface and open water, snow melt, soil freezing, surface runoff, infiltration, preferential flow, redistribution of soil moisture within the soil profile, sub-surface runoff, drainage of water to the groundwater system, groundwater storage, and groundwater base flow.
Figure 1 - LISFLOOD model structure and main components.
As precipitation occurs, liquid precipitation is partially intercepted by vegetation before reaching the ground. From leaves surface, stored water either falls on the ground or evaporates. The separation of precipitation into snow and rain is performed internally by the model based on air temperature. Snow accumulates on the ground and it's then melted using a degree day snow melting model. Snow melt is added to the water on the surface of the cell.
Surface water is available for infiltration and preferential flow. Infiltration is modelled with the widely-used Xinanjiang model (also known as VIC/ARNO; Liang, 1994; Todini, 1996), whose approach assumes that the fraction of a grid cell that is contributing to surface runoff (read: saturated) is related to the total amount of soil moisture, and that this relationship can be described through a non-linear distribution function. To prevent an unrealistic model behaviour during extreme rainfall conditions, during each time step, a fraction of the water that is available for infiltration on the pixel surface is added to the groundwater directly (i.e. without first entering the soil matrix). This is called preferential flow and it is described as a power function of the relative saturation of soil. Preferential flow becomes increasingly important as the soil gets wetter. Surface water that doesn't infiltrate into the soil or doesn't leave the surface as preferential flow becomes available for surface flow and it's routed to the nearest river channel using Kinematic routing and wide cross-section assumptions.
LISFLOOD organises soil in three layers: superficial soil, upper soil and lower soil, the two topmost soils (superficial and upper soil) forming the top soil. The modelling of water flow between the soil layers and to the upper groundwater zone is based on the simplifying assumption that the flow in the soil is entirely gravity-driven. The flow is assumed to always be in downward direction (with matric potential gradient equal to zero in Darcy's equation), at a rate that equals the hydraulic conductivity of the soil. The relationship between hydraulic conductivity and soil moisture is described by the Van Genuchten equation** and it is related to soil texture (i.e. percentage content of sand, silt and clay). An iterative procedure is used to compute water flow through the soil layers (from superficial soil to upper soil layer, from upper soil to lower soil layer and from lower soil layer to groundwater) and soil moisture in the superficial, upper and lower layer. The total soil moisture in the top layer is also computed adding soil moisture in superficial and upper layers.
Water uptake and transpiration by vegetation and direct evaporation from the soil surface are modelled in LISFLOOD as two separate processes and both of them extract water from the soil. For transpiration Supit et al. (1994) and Supit and Van Der Goot (1999) approach is used and maximum transpiration is computed as a function of the potential (reference) evapo-transpiration rate, the crop coefficient and the Leaf Area Index. The potential transpiration rate is then reduced as a function of actual moisture in the soil. The maximum evaporation from the soil surface is computed as a function of the potential evaporation rate from bare soil surface and the Leaf Area Index. The actual evaporation from the soil mainly depends on the amount of soil moisture near the soil surface: evaporation decreases as the top soil is drying.
The groundwater storage and transport are represented in LISFLOOD using two interconnected groundwater zones each consisting of a linear reservoir. The outflow from each zone is estimated based on the water storage and the time constant of the reservoir. The upper groundwater zone receives groundwater recharge from the lower soil and preferential flow from the cell surface. Upper zone outflow represents fast groundwater outflow and subsurface flow through macro-pores in the soil. Water can percolate from the upper to the lower groundwater zone, which represents the slow groundwater component that generates the base flow. All water that flows out of the upper- and lower groundwater zone is routed to the nearest downstream channel pixel within one time step.
Routing in the channels network (and on the surface) is performed by solving kinematic wave equations. At every time step and for each model pixel containing a channel, a 4-point implicit finite-difference solution of the kinematic wave equations is applied to compute the flow of water to the nearest downstream channel pixel. To increase accuracy and increase stability in the resolution, for the channel routing the model computation time step (6 hours for EFAS v4.0) is divided into smaller sub-steps (1 hour for all EFAS versions). Besides, to improve the representation of routing in floodplains where wave propagation gets slower as the discharge increases because of increased friction, LISFLOOD uses a double kinematic routing approach (Chow, 1988: Fröhlich, 1996) and it splits up the channel in two parts: bankful routing and over bankful routing.
LISFLOOD model can include reservoirs as particular pixels in the channel network. The inflow to each reservoir equals the channel flow upstream of the reservoir. Each reservoir has a total storage capacity and three ‘special’ relative filling levels: the conservative storage limit that represents the lower limit of reservoir water storage (reservoirs are never completely empty), the flood storage limit that represents the upper limit of normal water storage (reservoirs are never filled completely for safety reasons) and the normal storage that is the available storage capacity of the reservoir between the two filling limits, the conservative storage limit and the flood storage limit. For each reservoir, three characteristic outflows are defined: the minimum outflow that needs to be maintained at all times for e.g. ecological reasons, the non-damaging outflow that is the maximum possible outflow that will not cause problems downstream of the reservoir and the normal outflow that is the outflow being released when the reservoir is in its normal filling level (i.e. somewhere between conservative storage limit and flood storage limit). Operational rules for reservoirs are not included explicitly in LISFLOOD, but the model mimics these operational rules by using the three different filling levels (conservative storage limit, normal storage limit and flood storage limit) and the three regulated outflows (minimum outflow for ecological reasons, normal outflow, non-damaging outflow).
Figure 2 - Scheme of reservoirs filling levels and corresponding outflows in LISFLOOD model.
Lakes are also included as points in the channel network, but outflow is computed as outflow from a rectangular weir, as a function of water level and a parameter for width, gravity and weir coefficient.
A detailed description of LISFLOOD model is available at https://ec-jrc.github.io/lisflood-model/
** Water retention curve is the relationship between the water content, θ, and the soil water potential, ψ. Water potential is the potential energy of water per unit volume relative to pure water in reference conditions. Water potential quantifies the tendency of water to move from one area to another due to osmosis, gravity, mechanical pressure and matrix effects such as capillary action (which is caused by surface tension). The water retention curve is characteristic for different types of soil, and is also called the soil moisture characteristic. The shape of water retention curves can be characterized by several models, one of them known as the van Genuchten model. Based on this parametrization a prediction model for the shape of the unsaturated hydraulic conductivity - saturation - pressure relationship was developed.
REFERENCE
Bergström, S. 1995. The HBV model. In: Singh, V.P. (Ed.) Computer Models of Watershed Hydrology. Water Resources Publications, Highlands Ranch, CO., pp. 443-476.
Chow, V.T., Maidment, D.R., Mays, L.M., 1988. Applied Hydrology, McGraw-Hill, Singapore, 572 pp.
Diepen, C.A., van, Wolf, J. and Keulen, H., van, 1989. WOFOST: a simulation model of crop production. Soil Use and Management, 5: 16-24.
Fröhlich, W., 1996. Wasserstandsvorhersage mit dem Prgramm ELBA. Wasserwirtschaft Wassertechnik, ISSN: 0043-0986, Nr. 7, 1996, 34-37.
Knijff, J. and A. De Roo. 2008. LISFLOOD: distributed water balance and flood simulation model, Revised User Manual. JRC, European Commission.
Supit I., van der Goot E., National wheat yield prediction of France as affected by the prediction level, Ecological Modelling, Volume 116, Issues 2–3, 1999, Pages 203-223, ISSN 0304-3800, https://doi.org/10.1016/S0304-3800(98)00175-6.
Van Der Knijff, J. M., Younis, J. and De Roo, A. P. J.(2008)'LISFLOOD: a GIS-based distributed model for river basin scale water balance and flood simulation',International Journal of Geographical Information Science, 24:2, 189-212, DOI: 10.1080/13658810802549154