A simple model to predict soil moisture: Bridging Event and Continuous Hydrological (BEACH) modelling
Introduction
The soil moisture content of the root zone is a key variable that controls nearly all the hydrological processes occurring at or near the land surface. It regulates the partitioning of precipitation into infiltration, runoff, storage in the root zone and percolation into deeper ground water storage. Soil moisture also influences evapotranspiration and water availability to plants and thus affects the success of agriculture. Therefore it is considered to be an important parameter in land surface hydrology models, climate models and general circulation models at a variety of scales. Despite this importance, due to its high spatial and temporal variability, soil moisture is not routinely monitored over the long term like precipitation and discharge (Georgakakkos, 1996, Yamaguchi and Shinoda, 2002). Generally speaking, soil moisture is measured at two extreme scales (Mohanty et al., 2000). It is either observed at a scale of square centimetres (point scale) with in situ measurement methods (e.g. gravimetric, TDR, etc.) or it is observed at a scale of several square metres (pixel size) with the use of remote-sensing techniques. Neither the in situ techniques nor the remote-sensing techniques provide observations at the appropriate resolution or sampling interval and are prone to large measurement errors (Walker, 1999, Evett et al., 2002, Casper et al., 2007). For these reasons, during the last 30 years there have been various studies that have attempted to develop a method to estimate the soil moisture content over a range of scales.
In general, the methods for estimation of spatially distributed moisture content are classified into three main groups: (i) extrapolation approaches; (ii) simulation models in open loops (without feedback mechanism); and (iii) data assimilation and integration of remote-sensing observations and computational modelling.
In the first group, area average of soil moisture is estimated by extrapolating point measurements across the landscape, either with geostatistical techniques (Western and Grayson, 1998, Wang et al., 2001, Western et al., 2004) or using wetness indices based on terrain information (e.g. Beven and Kirkby, 1979, O'Loughlin, 1986, Svetlitchnyi et al., 2003, Teuling and Troch, 2005). In practice, both methods are difficult to apply. Due to the small correlation length of soil moisture variability, the application of geostatistical methods requires a large number of soil moisture observations for medium- to large-scale catchments which is unaffordable. The usefulness of the wetness indices is limited by the restrictive assumptions underlying their derivation (Grayson and Western, 1998). Moreover, the inclusion of these functions increases the complexity of the hydrological simulations (Engman and Rogowski, 1974), which may not be justifiable for a given marginal improvement in catchment prediction (Dunin and Aston, 1981).
Another method of estimating spatially distributed soil moisture that falls within the first group is the application of “time stability” (Vachaud et al., 1985). According to this concept, particular sites in the field always display the mean behaviour while others always represent extreme values (Teuling et al., 2006).
The second group includes Soil–Vegetation–Atmosphere Transfer (SVAT) models, Land Surface Model (LSM) and unsaturated zone models, which usually require the solution of a form of the Richards' equation (Hurley and Pantelis, 1985, Van Dam and Feddes, 2000, Downer and Ogden, 2004, Moran et al., 2004). Most of these models solve the Richards' equation in 1-D vertical direction and regionalisation is carried out based on land use, or soil, or topography, or two or three of these combined (Renschler et al., 2001). In a review study, Moran et al. (2004) reported that these simulation models, which are physically based, are generally of limited practical use because of the difficulties of specifying parameters and the initial and boundary conditions. Due to large uncertainties in the input parameters, the boundary conditions and the initial conditions, the physically based simulation models have a performance quality which is not much better than the results obtained with simpler conceptual models (Loague and Freeze, 1985, Chen et al., 1994, Hsu et al., 2002).
The third group, a fairly new method for estimating a spatially distributed soil moisture profile, comprises the integration of remote-sensing observations and hydrological modelling using data assimilation. For an excellent overview of this approach, see Heathman et al. (2003) and Moran et al. (2004). In this method, the profile soil moisture content is linked to the surface soil moisture content in order to combine the advantages of spatial predictability of the remote-sensing data with the continuous and depth-wise predictability of the in situ measurement tools and 1-D hydrological modelling (Kostov and Jackson, 1993, Entekhabi et al., 1994, Georgakakkos, 1996, Hymer et al., 2000, Heathman et al., 2003, Tischler et al., 2007). However, the remotely sensed soil moisture data is prone to errors introduced by soil type, landscape roughness and vegetation cover (Houser et al., 1998). Also, there is a mismatch in scale between the in situ measurements and the area estimates from remote sensing (Grayson and Western, 1998). Remote-sensing data yields the average value of the soil moisture content at the scale of a footprint that is larger than the scale of variability of soil moisture (Charpentier and Groffman, 1992, Western and Blöschl, 1999, Heathman et al., 2003, Western et al., 2004).
Despite the above-mentioned problems related to the measurement and prediction of the spatial and temporal distribution of soil moisture, the initial state of soil moisture is an important input parameter in modelling various hydrological processes like event-based surface runoff generation and soil erosion over a range of spatio-temporal scales. In addition, distributed physically based models such as ANSWERS, EUROSEM, KINEROS2 and LISEM have proven to be most sensitive to the initial soil moisture (De Roo, 1993, De Roo et al., 1996, Folly et al., 1999, Hantush and Kalin, 2005).
Therefore, correct representation of antecedent soil moisture condition over the catchment is of crucial importance for accurate simulation of runoff generation, soil erosion and non-point source pollution with such models (Aubert et al., 2003). In other words, the users of physically based, event-scale models are in need of a tool that provides detailed information on the spatial distribution of soil moisture at the onset of an event.
Reviewing the hydrologic literature and accounting advantages and disadvantages of the three groups of soil moisture estimation methods reveal that the simulation models (second group) have been the most frequently and readily used methods to generate soil moisture data and still remain as most appropriate tools.
One of the main goals of modellers in the area of environmental studies is to ever increase our understanding of complex natural systems and advance the development and application of models that simplify the representation of the real systems under study (Silberstein, 2006). In light of this, during the past few decades many experimental studies on infiltration and water movement in soil profiles have resulted in considerable progress in the conceptual understanding and mathematical description of soil water dynamics within the unsaturated zone. This progress has resulted in the development of various soil water dynamic models with different levels of complexity, process description, data requirement and scale of applicability. In general, existing models in hydrology are distinguished into three types, namely 1) empirical models (data-based models); 2) conceptual models; and 3) physically based models. Each type has its advantages and disadvantages. Since the level of complexity of the physically based models is generally agreed excessive for many practical problems (Marechal and Holman, 2005), in this study we develop a conceptual soil moisture model named BEACH (Bridging Event And Continuous Hydrological modelling) using simplified representations of the component processes.
In this paper, the steps involved in development, application, and evaluation of the BEACH model are demonstrated. BEACH is a spatially distributed computational model which would be a helpful tool for educators and students of environmental sciences. Model validation is assessed through a comparison of the model results with the soil moisture observation at six locations within the catchment, the discharge data at the catchment outlet as well as the inter-comparison of the model results with the BUDGET soil water balance model (Raes, 2002).
Section snippets
Model formulation
As mentioned in the introduction, in the hydrologic literature there are many simulation models of various complexities with the ability of continuously updating soil moisture content only on a 1-D soil profile. There are also some spatially semi-distributed or fully distributed physically based models. These models are complex, data-intensive, beyond the capability of the environmental scientists to adjust or modify for regional variations, and with limited exploitation of the advantages of
Site description
We have applied the BEACH model to the Catsop catchment in South Limburg, the Netherlands (50° 95′N, 5° 78′E; Fig. 3), which has an area of 0.42 km2 and is located approximately 2 km north of Maastricht–Aachen international airport. This typical agricultural catchment is in the loess district of the Netherlands (De Bakker, 1979). The loess district is characterised by an undulating topography and deep aeolian deposits of the late-Weichsel ice age. These deposits blanket older formations of
Results and discussion
A comparison of observed and simulated results for both calibration and validation periods would be the ideal method to demonstrate a model's adequacy. However, the available data set comprised only one year of soil moisture observations of acceptable quality. This hindered the validation of the BEACH model based on a separate observed data set. For integrated models, when they have to extrapolate beyond the calibration period, Jakeman et al. (2006) suggest that the evaluation criteria have to
Conclusion
The results of the application of the BEACH model to a small catchment in South Limburg in the Netherlands showed that it has an acceptable capability to provide the spatially distributed initial soil moisture content of the root zone profile which is necessary inputs for event-based surface runoff generation and soil erosion models. Our results have demonstrated that different methods of calculating the infiltration and surface runoff generation do indeed produce a high variation in the
Acknowledgements
The authors acknowledge the Ministry of Sciences & Research and Technology (MSRT) of I.R. of Iran for financial support of this study. The Water Board of Roer and Overmaas is kindly acknowledged for providing the discharge data at the outlet. Joy Burrough advised on the English.
References (71)
- et al.
Sequential assimilation of soil moisture and streamflow data in a conceptual rainfall–runoff model
J. Hydrol.
(2003) - et al.
Evaluation of the EUROSEM model using data from the Catsop watershed, the Netherlands
Catena
(1999) - et al.
Towards areal estimation of soil water content from point measurements: time and space stability of mean response
J. Hydrol.
(1998) - et al.
Assimilation of surface soil moisture to estimate profile soil water balance content
J. Hydrol.
(2003) - et al.
Ten iterative steps in development and evaluation of environmental models
Environ. Model. Softw.
(2006) - et al.
Development and application of a soil classification-based conceptual catchment-scale hydrological model
J. Hydrol.
(2005) A linear root water uptake model
J. Hydrol.
(1988)- et al.
Towards integrating GIS and catchment models
Environ. Model. Softw.
(2000) - et al.
Ten steps applied to development and evaluation of process-based biogeochemical models of estuaries
Environ. Model. Softw.
(2008) Hydrological models are so good, do we still need data?
Environ. Model. Softw.
(2006)
Spatial distribution of soil moisture content within catchments and its modelling on the basis of topographic data
J. Hydrol.
A GIS framework for surface-layer soil moisture estimation combining satellite radar measurements and land surface modeling with soil physical property estimation
Environ. Model. Softw.
Numerical simulation of infiltration, evaporation and shallow ground water levels with Richards' equation. Simulation of field-scale water flow and bromide transport in a cracked clay soil
J. Hydrol.
Water balance modelling in Bowen, Queensland, and the ten iterative steps in model development and evaluation
Environ. Model. Softw.
On the spatial scaling of soil moisture
J. Hydrol.
Spatial correlation of soil moisture in small catchments and its relationship to dominant spatial hydrological processes
J. Hydrol.
Crop evapotranspiration: guidelines for computing crop water requirements
A physically based variable contributing area model of basin hydrology
Hydrol. Sci. Bull.
Simple model for daily evaporation from fallow tilled soil under spring conditions in a temperate climate
Netherlands J. Agri. Sci.
Fuzzy logic-based rainfall–runoff modelling using soil moisture measurements to represent system state
Hydrol. Sci. J.
Soil moisture variability within remote sensing pixels
J. Geophys. Res.
Spatial averaging of unsaturated flow equations under infiltration conditions over areally heterogeneous fields 2. Numerical simulations
Water Resour. Res.
Major Soils and Soil Regions in the Netherlands
LISEM: a single event physically-based hydrologic and soil erosion model for drainage Basins. II: sensitivity analysis, validation and application
Hydrol. Proc.
Appropriate vertical discretization of Richards' equation for two-dimensional watershed-scale modeling
Hydrol. Proc.
Spatial variability in the water balance of an experimental catchment
Aust. J. Soil Res.
A partial area model for storm flow synthesis
Water Resour. Res.
Solving the inverse problem for soil moisture and temperature profiles by sequential assimilation of multifrequency remotely sensed observations
IEEE Trans. Geosci. Remote Sens.
A GIS-based variable source area hydrology model
Hydrol. Proc.
Soil moisture theories and observations
J. Hydrol.
Uncertainty and sensitivity analysis of runoff and sediment yield in a small agricultural watershed with KINEROS2
Hydrol. Sci. J.
Intercomparison of flow and transport models applied to vertical drainage in cropped lysimeters
Vadose Zone J.
Integration of soil moisture remote sensing and hydrologic modelling using data assimilation
Water Resour. Res.
Cited by (80)
Sand fixation and human activities on the Qinghai-Tibet Plateau for ecological conservation and sustainable development
2024, Science of the Total EnvironmentImpact of a large-scale replacement of maize by soybean on water deficit in Europe
2023, Agricultural and Forest MeteorologyChanges in flood-associated rainfall losses under climate change
2023, Journal of HydrologyBeyond prediction: An integrated post-hoc approach to interpret complex model in hydrometeorology
2023, Environmental Modelling and SoftwareA sensitivity analysis of a FAO-56 dual crop coefficient-based model under various field conditions
2023, Environmental Modelling and Software