1. Introduction: Design prerequisites
The study of river basin hydrology is focused on the analysis of the interactions between the near-surface soil and the atmospheric boundary layer (ABL), which occur mainly through the mediation of the soil itself, the vegetation, and the turbulent and radiative energy transfers taking place on the earth’s surface, and possible feedbacks from the ABL (e.g., soil moisture–precipitation feedback mechanisms). In recent years, hydrologic research has generally evolved toward a comprehensive theory describing the water, energy, and momentum exchanges between the land surface and the atmosphere at several scales (Abbott 1992; Reggiani et al. 1999), with the purpose of creating models that can provide improved mid- and long-term hydrologic forecasts and better prediction of the impacts on the hydrologic cycle and on the earth’s ecosystems resulting from changes in land use and in the climate (Grayson and Blöschl 2000). Though inspired by this trend toward improved predictions, the initial motivations behind the development of GEOtop were practical, as its development began with attempts to predict, besides river discharges, landslide and debris flow initiation, snowpack evolution and ablation, and the water budget (especially, accurate assessment of evapotranspiration and the distribution of soil moisture) in small catchments and in complex mountain terrain.
None of the above phenomena can be currently predicted with enough accuracy and consistency to be operationally useful (Committee on Hydrologic Science 2002). It is known that such phenomena are affected by heterogeneity in soil hydraulic properties, land use, landscape, topography and geological properties, and soil moisture profile and surface–subsurface interactions, but it has not been quantitatively assessed to what extent. This cannot be explicitly obtained without a small-scale distributed model that is capable of explicitly accounting for these characteristics. Furthermore, it would require that both the water and energy balances be solved (Grayson and Blöschl 2000). Solving the energy balance means, among other things, being able to estimate surface temperature Ts, and it also requires proper treatment of radiation physics (including the effect of the view angle and shadowing in complex terrain) (Blöschl et al. 1991). However, since temperature is absolutely necessary in modeling snowpack evolution (e.g., Zanotti et al. 2004; Pomeroy et al. 2003), its simulation should be included in the design of such a model. Finally, the proper description of many of the hydrologic phenomena on hillslopes (including hillslope stability) and of water fluxes in the absence of significant slopes, such as in plains or in riparian zones, requires an adequate treatment of suction head in unsaturated soils.
Just a few distributed hydrological models have all the necessary characteristics. Since the blueprint of Freeze and Harlan (1969), many “distributed” hydrological models (Beven 2000) have been implemented to predict the lateral distribution of water, ranging from more conceptualized descriptions (e.g., Beven and Kirkby 1979; Franchini and Pacciani 1991) to fully distributed approaches, such as the Système Hydrologique Européen (SHE; Abbott et al. 1986), topographic kinematic approximation and integration model (TOPKAPI; Ciarapica and Todini 2002), THALES (Grayson et al. 1992), and others (Vertessy et al. 1993; Garrote and Bras 1995; Bronstert and Plate 1997). At the catchment scale, distributed approaches range from a three-dimensional integration of the Richards’ equations as in Paniconi and Putti (1994) or in the catchment hydrological model (CATHY; Paniconi and Wood 1993; Paniconi et al. 2003a) to a distributed description of the hydrological effects of vegetation as in the Distributed Hydrological Soil Vegetation Model (DHSVM) (Wigmosta et al. 1994). Nevertheless, the Distributed Models Intercomparison Project (DMIP) (Reed et al. 2004) provides a recent, clear example of the continued emphasis in hydrologic modeling on flood forecasting; the models involved, with a few exceptions (Liang and Xie 2001; Ivanov et al. 2004), were mostly event-oriented rainfall-runoff models, which are successful in modeling floods but do not provide enough hydrological information for forecasting other processes.
Vice versa, the land surface models (LSMs), unlike the flood-forecasting models, are usually one-dimensional (1D) vertical models that represent soil–atmosphere interactions with a great degree of complexity and accuracy, but they are usually not endowed with a detailed representation of runoff and lateral fluxes (Warrach et al. 2002), as they have been developed mainly to provide a land surface interface in support of atmospheric general circulations models (GCMs). Reviews of LSMs are presented in Garratt (1992, 1993) and Lakshmi et al. (2001). State-of-the-art LSMs, inclusive of the ability to model snowmelt, freezing soil, and multilayer canopy models are, to mention a few, the biosphere–atmosphere transfer scheme (BATS; Dickinson et al. 1986), the surface energy and water balance (SEWAB; Mengelkamp et al. 1999), the Noah LSM (Chen et al. 1996), and the National Center for Atmospheric Research (NCAR) LSM (Bonan 1996).
GEOtop is designed to be an integration of the characteristics of both LSMs and flood-forecasting models and to include a detailed treatment of topographic interaction with radiation, which is not usually available in other models. Thus GEOtop is a distributed model; it is a terrain-based model; it simulates not only the water balance but also the energy balance; and it is built for remote sensing data because it provides surface temperature and an accurate treatment of radiation and moisture in the upper soil (Troch et al. 2003).
This paper is organized as follows. Section 2 contains the description of model components, which begins with a description of the discretization of the landscape in GEOtop, followed by a description of local water and energy budgets, and concluding with a description of the processes driving the water and energy fluxes. Section 3 contains three case studies based on the 1997 Southern Great Plains Hydrology Experiment (SGP97) dataset. The case studies show the reproduction of the water and energy balance at a point, a calibration of the model on a flood event on the Little Washita basin in Oklahoma, the spatial soil moisture distribution over the basin, and the subsequent simulation of a yearlong surface discharge hydrograph with no additional calibration. Conclusions follow in section 4.
2. Description of model components
GEOtop simulates the fluxes and budgets of energy and water on a landscape defined by three-dimensional grid boxes, whose surfaces come from a digital elevation model (DEM) and whose lower boundaries are located at some specified spatially varying depth. If snow accumulates on top of the land surface, a new layer is added on top of the DEM as described in Zanotti et al. (2004). Surface boundary conditions are given by hydrometeorological measurements (rainfall, temperature, wind velocity), regionalized as described in Bertoldi (2004).
We present here the components of the first public version of GEOtop. Information on previous versions can be found in Bertoldi and Rigon (2004).
a. Landscape and equation discretization
GEOtop requires preprocessing of the catchment DEM for which a special set of routines has been implemented in an open source GIS, the Java Geographic Resource Analysis Support System (JGRASS; http://www.hydrologis.com). The outputs of these routines include estimation of drainage directions, slopes, curvature, the channel network structure, shadowing, and the sky view factor. Surface runoff is modeled to follow the terrain surface according to a so-called D8 topology as in Orlandini et al. (2003). Advantages and disadvantages of various methodologies for carrying out these computations are thoroughly discussed in Rodriguez-Iturbe and Rinaldo (1997) and in Orlandini et al. (2003) and are not further discussed here. Soil thickness can be defined either using field data or the model by Heimsath et al. (1997), as discussed Bertoldi et al. (2006a).
The DEM identifies also the plan view of a three-dimensional grid on which all the model’s equations are discretized. The thickness of the discretized soil layers is set up at run time; it is usually appropriate to assume that the top layer is quite thin (e.g., 5 cm thick). According to criteria that the user can specify (e.g., Prosser and Abernethy 1996), some of the grid cells can be identified as river network cells. River network cells are treated the same as hillslope cells except for the routing of surface runoff, which is modeled as described in section 2c.
b. Water and energy budgets in GEOtop
The budgets of mass and energy in GEOtop may be divided into three parts: the water budget, the energy budget, and the radiation budget. The radiation budget is presented separately from the energy budget, as it creates the link between the model and remotely sensed data.
1) Water budgets
2) Energy budget
3) Radiation budget
Remote sensors detect only limited bands of the radiation spectrum and not the whole integrated infrared (LW) and visible parts (SW) of it as in (5). The radiation budget code was separated in GEOtop from the energy budget code because future versions of the model will be able to return the radiative intensity separated into optical radiation bands where remote sensors operate.
c. Dynamic hydrology in GEOtop
Water and energy fluxes, which control the budgets described above, are computed according to gradients of energy potentials as described below. A more detailed description of the dynamics including complete equations can be found in Bertoldi and Rigon (2004). The snow module and its preliminary application to a case study are described in Zanotti et al. (2004).
1) Subsurface saturated–unsaturated flow
To model soil moisture dynamics properly in the soil layers, GEOtop utilizes the integration of Richards’ equations (Richards 1931; Freeze and Harlan 1972) in three dimensions (3D). For some of the processes estimation, simpler approaches could have been used such as the hillslope-storage Boussinesq (hsB) equation developed by Troch et al. (2003), Paniconi et al. (2003b), and Hilberts et al. (2004), which has been shown to work generally very well as compared to the 3D Richards’ equation for simulation of saturated subsurface flow dynamics and to be much more computationally efficient. However, GEOtop is intended to simulate also the suction dynamics, and, in this, solving Richards’ equation is necessary.
Runoff at the surface is generated because the precipitation intensity is greater than the soil surface infiltration capacity (infiltration excess runoff; Horton 1933) or the water table level rises above the soil surface (saturation excess runoff; Dunne 1978). Both of these possibilities are dealt with the Richards’ equation solver and are implemented according to the comprehensive analysis of boundary conditions made by Paniconi and Putti (1994). In addition, Richards’ equation allows for water to redistribute laterally and even move uphill or exfiltrate according to local suction.
2) Surface runoff
3) Channel flow
4) Turbulent fluxes
d. Numerics
GEOtop equations are discretized according to a finite-volume method, and time resolution is the smallest required for the numerical stability of all the processes described (or by the time step of the forcings), usually between 1 and 5 min. Grid cell horizontal dimensions can range from 2 to 500 m, and the number of soil layers and the maximum soil thickness is arbitrary. All the processes are solved using the same time step. GEOtop is coded in ANSI C and is compiled to run on Linux, Mac OS X, and Windows. The first public version was released in December 2005 under a General Public License (GPL), and can be downloaded from http://bedu.ing.unitn.it/GEOtopWiki.The code will be thoroughly documented with doxygen (http://www.doxygen.org). Further information about the numerics of the model can be found in Bertoldi et al. (2006b).
3. Tests of GEOtop in the Little Washita watershed
An opportunity to validate and analyze the performance of GEOtop is offered by the SGP97 dataset. Three different kinds of simulations were performed. The first set of simulations (A) uses GEOtop to reproduce the energy and water fluxes at the pixel scale to assess the physical realism present in the model. The second set of simulations (B) reproduces a flood event and soil moisture evolution and calibrates some routing parameters. The initial water content profile was obtained starting with a saturated profile and running the model several times on the meteorological forcings of the period analyzed until the spatial average surface soil moisture reached the same value as the Electrically Steered Thinned Array Radiometer (ESTAR) average. This set of simulations is further used for a simple parameter sensitivity study. The third simulation (C), finally, forecasts a yearlong record of measured discharges without any further calibration, using the parameters derived from simulations A and B. Two versions of this simulation were performed. One was performed without any additional calibration, and in the other, an adjustment of the vegetation cover fraction was made.
a. The dataset
SGP97 took place from 18 June to 17 July 1997 in central Oklahoma and is discussed in detail by Jackson et al. (1999), Mohanty et al. (2002), and Mohr et al. (2000). The basin primarily investigated was the Little Washita watershed, which has been the focus of extensive hydrologic research by the Agricultural Research Service (ARS) (Allen and Naney 1991). The modeled portion of the watershed (i.e., upstream of U.S. Geological Survey stream gauge 07327550, east of Ninnekah) covers an area of 603 km2. The topography of the region is moderately rolling with a maximum relief over the basin of less than 200 m (Fig. la). Land use is dominated by rangeland and pasture (63%), with significant areas of winter wheat and other crops concentrated in the floodplain and western portions of the watershed area. As part of the ARS Micronet, there are a total of 42 recording rain gauges distributed at a 5-km spacing over the watershed (Fig. 1b); these provided the precipitation data used in this analysis. For meteorological data other than precipitation, the three Oklahoma Mesonet stations, Acme (ACME), Apache (APAC), and Ninnekah (NINN), located inside or very near the basin were used. Moreover, at the National Oceanic and Atmospheric Administration (NOAA) (LW02) flux site (marked in Fig. 1b), measurements of soil properties, soil moisture and temperature, and surface energy fluxes are available.
b. Simulations set A: Pixel-scale simulations of the vertical profile dynamics at the NOAA (LW02) flux site
The purpose of these simulations is to test the pixel-scale behavior of GEOtop. Because of this, the lateral water fluxes have been inhibited (but infiltration is allowed). The model was initialized with the measured soil moisture and temperature profiles. Calibration was performed only on the following parameters related to the energy budget: the thermal roughness length zT (Garratt 1993), of which the turbulent transport coefficients [CH and CE in Eq. (14)] are a function; the fraction of the cell covered by canopy [sυ in Eq. (15)], which controls the rate between bare soil evaporation and transpiration; and the soil thermal conductivity Λs. Calibrated values for this location are zT = 8.6 mm, sυ = 0.9, and Λs = 0.077 J kg−1 K−1. Other soil properties have been kept equal to local field values as derived from Mohanty et al. (2002). The pixel-scale hourly evolution of energy balance components are illustrated in Fig. 2, and rainfall, surface soil moisture, and temperature variations are illustrated in Fig. 3. The model is capable of simulating infiltration related to small rainfall events and daily soil moisture evaporation cycles, as shown in Fig. 3. The model outputs show a good agreement with the local energy fluxes measured with eddy correlation systems, except for a slight overestimation of nocturnal sensible heat fluxes (Fig. 2), which perhaps reveals some deficiencies of the Kot and Song (1998) parameterization of the nocturnal stable boundary layer. There is also a systematic overestimation of the ground heat flux during early morning, resulting from the model’s energy balance closure constraints, which is also reflected by a slight overestimation of the soil temperature as shown in Fig. 3. Moreover, negative values of sensible heat flux H in the dataset could be due to the procedure chosen to process the measurements. A model structural change could be then suggested that, for instance, used a two-layer canopy model to provide a more accurate estimation of the diurnal cycle of soil temperature.
c. Simulation set B: Calibration and parameter sensitivity of discharge for the period from 26 June to 16 July 1997
The goal of this set of simulations is to determine the surface and subsurface flow parameters later used in a longer simulation, and to do some sensitivity analysis. During the simulated period the weather was quite fair, with mostly hot sunny days, during which the soil dried quickly as result of intense summer evaporation. A strong convective storm on 11 July (Fig. 1b) with more than 50 mm of rainfall in 7 h (averaged over the basin) caused a flood with a peak discharge of 60 m3 s−1. Rain patterns for this event were reconstructed by means of kriging techniques. Simulations were performed with a grid resolution of 200 m and five soil layers of increasing thickness (the first one of 5 cm). Soil depth was given by field measurements by the Oklahoma State University for Application of Remote Sensing Laboratory and ranges from 0.5 to 1.5 m. As a lower boundary condition, an impermeable surface was imposed. Heat fluxes at the bottom were regulated on the mean annual surface temperature and excursion, as explained in section 2b.
Surface soil moisture maps with a resolution of 800 m, derived from Jackson et al. (1999) from the flights of the National Aeronautics and Space Administration (NASA) P-3B aircraft fitted with the ESTAR, an L-band (1.413 GHz) passive microwave sensor, were used for model initialization and verification. The initial water content profile was obtained starting with a saturated profile and running the model several times using the meteorological forcing from 26 June to 16 July 1997 until the spatial-averaged surface soil moisture reached the same value as the ESTAR average.
Surface values of hydraulic conductivity and the van Genuchten parameters [see Eq. (8)] were obtained from soil texture data from Mohanty et al. (2002) using the Vereecken et al. (1989) pedotransfer function approach. To assign the 3D field structure of the saturated hydraulic conductivity, an exponential decay with depth was assumed according to Ks = e−fz z′, where z′ is the cell depth below the surface; the factor fz (L−1) was determined by calibration. The canopy fraction Sυ, the root depth zroot, the roughness length zT, and the albedo a are assigned according to the land-cover map (data available from Goddard Earth Sciences Data and Information Services Center; Fig. 1c). Global parameters related to the water budget were calibrated against the measured discharges: the channel celerity uc, the hydraulic diffusivity D, the surface runoff resistance coefficient Cm [all in Eq. (13)], and the anisotropy ratio αk [in Eq. (6)]. Values of the spatial average of the parameters are given in Table 1.
Comparison of the measured and calibrated discharge hydrograph is shown in Fig. 4a, and the ESTAR measured and simulated basin-average soil moisture is shown in Fig. 4b. It can be seen that it was possible to obtain a good calibration to the main flood hydrograph. Also fairly good is the uncalibrated forecasting of the average soil moisture, which implies that, on average, evapotranspiration (ET) was properly simulated. Cumulative water balance simulation results (Fig. 5) show that total evaporation ET flux dominates the water losses during this period compared to discharge Qc; in particular, only a small fraction of the 11 July storm precipitation became runoff.
The sensitivity of the discharge during the major storm event on 11 July of the parameters Ksh, Ksυ, θr, θs, α, m, n, and Sυ is shown in Fig. 6. Among the calibrated parameters summarized in Table 1, the vertical hydraulic conductivity (Fig. 6a) determines the maximum infiltration capacity and greatly influences the surface runoff formation, because it controls the amount of Hortonian-type runoff, and therefore the magnitude of the peak discharge. The horizontal hydraulic conductivity, Ksh, on the contrary, primarily influences the flood tail because it determines the subsurface runoff quantity (Fig. 6b). A large value of Ksh, however, causes soil drainage that is too rapid, leading to water content values that are too low compared to the measured values. Residual water content θr, which controls the total basin storage capacity, has a greater influence on discharge than the water content value at saturation θs (Figs. 6c and 6d). The van Genuchten parameters α, n, and m [Eq. (8)] have also a significant influence on peak discharge (Figs. 6e–g). In particular, very low values of α (which means increased suction potential for the same saturation) lead to a bimodal behavior of the hydrograph. These simulations of this event are not significantly affected by the vegetation-cover fraction Sυ (Fig. 6h), which, however, plays a major role in the energy budget calibration. Other sensitivity analyses on soil thickness and river network extension are the subject of Bertoldi et al. (2006).
d. Surface soil moisture evolution in the basin
The soil moisture distribution derived from the ESTAR data by Jackson et al. (1999), available for the same period in which we performed the calibration, was compared to that produced by the top 5-cm layer of the model. As shown in Fig. 7, GEOtop predicted surface moisture patterns that appear visually similar to ESTAR estimates. The soil moisture maps produced by the model do not show the influence of land use; however, on the contrary, the soil texture (and by inference hydraulic conductivity) patterns become evident during the dry periods (3 and 16 July), as can be seen from the comparison of Fig. 7 with Fig. 1. The spatial distribution of precipitation, as pointed out by the comparison of the precipitation field of 10 July in Fig. 1 to the soil moisture map of 11 July reported in Fig. 7, strongly determines the spatial distribution of the soil surface water content, mostly in the first hour immediately following the rainfall. According to the simulations, as qualitatively shown in the 11 July map of Fig. 7, topography starts exerting its influence only some hours later, by concentrating moisture in the topographic convergence zones.
The simulated averaged surface soil moisture values are very similar to those estimated by ESTAR, as was shown in Fig. 4b. However, the frequency distribution of global measured and simulated water content are somewhat different (Fig. 8). In particular, GEOtop predicted a larger soil moisture variance than ESTAR measured (especially in the case of rain events: see 11 and 16 July in Fig. 8). To understand the reason for this discrepancy, we also ran some simulations on the same dataset (using a coarser 1000-m grid) with the Variable Infiltration Capacity (VIC) model (Liang et al. 1994), obtaining distributions, which at least show that the difference is not simply an effect of grid resolution (recall that ESTAR has a resolution of 800 m). In fact, these differences among model simulation results and ESTAR products may be caused not only by errors in the model but also in the data assimilation used to obtain the ESTAR product. In fact, in the computation of the ESTAR soil moisture product soil dielectric properties, which themselves depend on moisture content, must be assumed.
e. Simulation set C: The forecasting of the 1997 streamflow
In this simulation for the whole year of 1997, the calibrated parameters of the previous simulations have been kept constant. Only the three Oklahoma Mesonet stations (Fig. 1) had precipitation data available for the whole year; therefore the model was forced only with these three stations. The model was initialized by forcing it repeatedly with the meteorological conditions of the year (periodic forcing) until a dynamical equilibrium was achieved after various periodic cycles had passed.
The comparison between the daily measured and the GEOtop-simulated streamflow with no additional calibration (the gray band bottom) is shown in Fig. 9 for the whole year of 1997. The model preserves fairly well the total streamflow volume, but tends to overestimate peak discharges, except for the event of late May when there is a considerable underestimation of the measured flood. The underestimation for this event, however, could be caused by inadequate rainfall inputs, since the runoff ratio of that event is considerably larger that that of the others. Unfortunately, more detailed rainfall data was not available to assess this issue. The July event used for calibration is simulated slightly differently in this simulation compared with the calibration simulation because here different initial conditions and rainfall are used.
As Fig. 9b underlines, an increased difference in base flow also appeared in summer, which could be explained with a lower actual ET than was simulated. This missing evaporation in the simulation was small in terms of total volumes but presented a definite trend in summer that eventually could be strongly decreased by changing the average fractional vegetation cover from sυ = 0.78 to sυ = 0.89, in accord with Mohanty et al. (2000), who showed that bare soil evaporates more than vegetation cover in a mixed vegetation pixel. The upper limit of the gray band of Fig. 9 shows the evolution of the discharge once sυ has been raised since the beginning of the year. It emphasizes that an incorrect evapotranspiration estimation results in an incorrect average soil moisture content with possible increasing errors as time elapses (Fig. 10). Eventually the pattern of soil moisture would also depart from actual, since ET is heterogeneous in the presence of heterogeneous topography, and hydraulic conductivity presents a strong nonlinear dependence on soil moisture volumetric content.
4. Conclusions
In this paper, a new distributed hydrological model, GEOtop, is presented. This model is an attempt to create a catchment simulator that not only accounts for water redistribution in a very detailed fashion (solving the 3D Richards’ equation), but also includes a coupled energy and radiation budget. The model accommodates very complex topography and its effects on the incident radiation, both shortwave (with shadowing) and longwave (with accounting for the sky view factor).
The application of the model to the Little Washita basin leads to the following conclusions. The set of simulations A shows that the model is capable of accurately reproducing the local energy budget and the local infiltration once initialized with measured parameters and calibrated for the roughness length, the fraction of cell covered by canopies, and the heat capacity of the soil. The set of simulations B shows that GEOtop, with calibration and accurate initialization, gives a precise reproduction of a peak flow, with no more calibration than a typical rainfall-runoff model would require (i.e., adjusting the river network celerity, the surface and subsurface celerity on the hillsides). This result confirms the common belief that it is not necessary to have a detailed evapotranspiration estimation for flood forecasting. However, since the initial soil moisture conditions were obtained with a dynamical procedure, and the set of simulations C show that flood peaks can be sensitive to such conditions, more investigation is needed to assess the interplay between initialization procedures and parameter identification.
At the basin scale, the model is capable of capturing the spatial distribution of the surface soil moisture, with no additional calibration, and it can be used as a tool to understand what the major controls on soil moisture patterns are. Simulated data show that in the Little Washita basin the main controlling factors on the soil moisture spatial distribution are soil properties during dry periods and precipitation during wet periods, confirming the finding by Mohr et al. (2000). A few hours after precipitation, the topographic redistribution of soil moisture also exerts a slight influence. A comparison of soil moisture obtained from simulations with ESTAR data brings some confirmation of the model behavior and calls for further investigations in which remotely sensed data will be assimilated concurrently (i.e., in parallel) with the model simulations.
The set of simulations C, a yearlong simulation of streamflow with no additional calibration, shows an overestimation of some flow peak discharges, giving however the proper total flow volumes. This simulation also shows that the vegetation dynamics can have, during the simulated period, a small but systematic effect on the water budget partition and that a correct estimation of evapotranspiration is essential to have a correct total discharge volume, the most important parameter being the transpiration rate. An adjustment of a controlling parameter was adequate to correct the budget partition. However, using the same parameter for the entire year (either the first or the second) resulted in increasingly incorrect average soil moisture contents with possible increasing departures from actual and causing increasing errors in base flow estimations. Results underline the importance of incorporating an adequate description of evapotranspiration processes for an accurate prediction of long-term discharge and soil moisture patterns and suggest the value of implementing a dynamic vegetation model.
A fully distributed model predicts throughout the catchment (and not only at the outlet of the basin) a larger number of observable processes than traditional lumped models. However, it also requires a corresponding increase in input data and parameters. Field measurement campaigns like the First International Satellite Land Surface Climatology Project (ISLSCP) Field Experiment (FIFE; Sellers et al. 1992), the Program for Intercomparison of Land Surface Parameterization Schemes (PILPS; Henderson-Sellers et al. 1993; Wood et al. 1999), the Hydrological Atmospheric Pilot Experiment-Modelisation du Bilan Hydrique (HAPEX-MOBILHY; Andre et al. 1988), SGP97 (Jackson et al. 1999), Tarrawarra (Western et al. 1998), Mahurangi River Variability Experiment (MARVEX; Woods et al. 2001), and other experiments, however, have already provided integrated data based on combined satellite, remote sensing, ground-based soil moisture and vegetation mapping, eddy correlation measurements of turbulent fluxes, and boundary layer profiling at a hierarchy of scales. As digital terrain and remotely sensed surface radiation data of higher resolution and better quality become increasingly available and computer performances continue to increase, physically based, distributed models such as GEOtop will become increasingly useful and more easily connected to climate and limited-area meteorological models, and they may become accepted as the standard tool for river basin hydrology.
Acknowledgments
This research has been partially supported by the Italian Ministry of University and Research (MIUR Prin 2003) and by the TIDE fifth framework and AQUATERRA sixth framework European Projects. We thank the Civil and Environmental Engineering Department of the University of Illinois at Urbana–Champaign and the Illinois Water Sciences Center of the U.S. Geological Survey for their hospitality during the visit of the second author during the course of this research. Davide Tamanini designed and wrote the code for the integration of the Richards’ equations. The Agricultural Research Service, Grazinglands Research Laboratory (ARS-GRL), El Reno, Oklahoma, provided the Little Washita precipitation data, and the ARS-GRL, in cooperation with the U.S. Geological Survey and the Oklahoma Water Resources Board, provided the Little Washita discharge data.
REFERENCES
Abbott, M. B., 1992: The theory of the hydrologic model, or: The struggle for the soul of hydrology. Advances in Theoretical Hydrology: A Tribute to James Dooge, J. P. O’Kane, Ed., Elsevier, 237–254.
Abbott, M. B., Bathurst J. C. , Cunge J. A. , Connell P. E. O. , and Rasmussen J. , 1986: An introduction to the European Hydrological System—Systeme Hydrologique European, SHE. 1: Hystory and philosophy of a physically-based distributed modelling system. J. Hydrol., 87 , 45–59.
Albertson, J., and Parlange M. B. , 1999: Surface length scales and shear stress: Implications for land–atmosphere interaction over complex terrain. Water Resour. Res., 35 , 2121–2132.
Allen, P. B., and Naney J. W. , 1991: Hydrology of the Little Washita River Watershed, Oklahoma: Data and analyses. Tech. Rep. ARS-90, U.S. Department of Agriculture, Agricultural Research Service, 74 pp.
Andre, C., and Coauthors, 1988: Evaporation over land-surfaces, first results from HAPEX-MOBILHY special observing period. Ann. Geophys., 6 , 477–492.
Bertoldi, G., and Rigon R. , 2004: GEOTOP: A hydrological balance model: Technical description and programs guide. Tech. Rep. DICA-04-001, University of Trento E-Prints, 102 pp. [Available online at http://eprints.biblio.unitn.it/archive/00000551/.].
Bertoldi, G., Rigon R. , and Over T. M. , 2006a: Impact of watershed geomorphic characteristics on the energy and water budgets. J. Hydrometeor., 7 , 389–403.
Bertoldi, G., Rigon R. , Tamanini D. , and Zanotti F. , 2006b: Geotop version 0.875: Technical description and programs guide. Tech. Rep. dica-06-001, University of Trento E-Prints, 101 pp.
Best, M. J., 1998: A model to predict surface temperatures. Bound.-Layer Meteor., 88 , 279–306.
Beven, K. J., 2000: Rainfall-Runoff Modelling: The Primer. Wiley, 372 pp.
Beven, K. J., and Kirkby M. J. , 1979: A physically-based variable contributing area model of basin hydrology. Hydrol. Sci. Bull., 24 , 1. 43–49.
Bixio, A., Orlandini S. , Paniconi C. , and Putti M. , 2000: Physically-based distributed model for coupled surface runoff and subsurface flow formulation at the catchment scale. Surface Water Systems and Hydrology and Water Research Symposium, L. Bentley et al., Ed., Vol. 2, Computational Methods in Water Resources, Balkema, 115–122.
Blöschl, G., Kirnbauer R. , and Gutknecht D. , 1991: Distributed snowmelt simulations in an alpine catchment. 1. Model evaluation on the basis of snow cover patterns. Water Resour. Res., 27 , 3171–3179.
Bonan, G., 1996: A land surface model for ecological, hydrological, and atmospheric studies: Technical description and user’s guide. Tech. Note NCAR/TN-417+STR, 150 pp.
Bronstert, A., and Plate E. , 1997: Modelling of runoff generation and soil moisture dynamics for hillslopes and microcatchments. J. Hydrol., 198 , 177–195.
Brutsaert, W., 1975: On a derivable formula for long-wave radiation from clear skies. Water Resour. Res., 11 , 742–744.
Chen, F., and Coauthors, 1996: Modeling of land surface evaporation by four schemes and comparison with FIFE observations. J. Geophys. Res., 101 , D3. 7251–7268.
Ciarapica, L., and Todini E. , 2002: TOPKAPI: A model for the representation of the rainfall-runoff process at different scales. Hydrol. Processes, 16 , 207–229.
Committee on Hydrologic Science, 2002: Report of a Workshop on Predictability and Limits-to-Prediction in Hydrological Systems. Water Sciences and Technology Board, National Research Council, National Academy Press, 118 pp.
Deardorff, J. W., 1978: Efficient prediction of ground surface temperature and moisture with inclusion of a layer of vegetation. J. Geophys. Res., 83 , C4. 1889–1903.
Dickinson, R. E., Henderson-Sellers A. , Kennedy P. J. , and Wilson M. , 1986: Biosphere Atmosphere Transfer Scheme (BATS) for the NCAR Community Climate Model. Tech. Note NCAR/TN−275+STR, 72 pp.
D’Odorico, P., and Rigon R. , 2003: Hillslope and channel contributions to the hydrologic response. Water Resour. Res., 39 , 1–9.
Dubayah, A., Dozier J. , and Davis F. W. , 1990: Topographic distribution of clear-sky radiation over the Konza Prairie, Kansas. Water Resour. Res., 26 , 679–690.
Dunne, T., 1978: Field studies of hillslope flow processes. Hillslope Hydrology, M. J. Kirkby, Ed., Wiley-Interscience, 227–293.
Feddes, R. A., and Coauthors, 2001: Modeling root water uptake in hydrological and climate models. Bull. Amer. Meteor. Soc., 82 , 2797–2809.
Franchini, M., and Pacciani M. , 1991: Comparative analysis of several conceptual rainfall-runoff models. J. Hydrol., 122 , 161–219.
Freeze, R. A., and Harlan R. L. , 1969: Blueprint for a physically-based, digitally simulated hydrologic response model. J. Hydrol., 9 , 237–258.
Freeze, R. A., and Harlan R. L. , 1972: Role of subsurface flow in generating surface runoff. 2. Upstream source areas. Water Resour. Res., 8 , 1272–1283.
Garratt, J. R., 1992: The Atmospheric Boundary Layer. Cambridge University Press, 316 pp.
Garratt, J. R., 1993: Sensitivity of climate simulations to land surface and atmospheric boundary-layer treatments—A review. J. Climate, 6 , 419–449.
Garrote, L., and Bras R. , 1995: A distributed model for real-time flood forecasting using digital elevation models. J. Hydrol., 167 , 279–306.
Grayson, R., and Blöschl G. , 2000: Spatial Patterns in Catchment Hydrology, Observations and Modelling. Cambridge University Press, 404 pp.
Grayson, R. B., Moore I. , and McMahon T. , 1992: Physically based hydrologic modelling. 1. A terrain-based model for investigative purposes. Water Resour. Res., 28 , 2639–2658.
Heimsath, M. A., Dietrich W. E. , Nishiizumi K. , and Finkel R. , 1997: The soil production function and landscape equilibrium. Nature, 388 , 358–361.
Henderson-Sellers, A., Yang Z. , and Dickinson R. , 1993: The Project for Intercomparison of Land-surface Parameterization Schemes. Bull. Amer. Meteor. Soc., 74 , 1335–1349.
Hilberts, A., van Loon E. , Troch P. A. , and Paniconi C. , 2004: The hillslope-storage Boussinesq model for non-constant bedrock slope. J. Hydrol., 291 , 160–173.
Horton, R. E., 1933: The role of infiltration in the hydrological cycle. Eos, Trans. Amer. Geophys. Union, 14 , 446–460.
Iqbal, M., 1983: An Introduction to Solar Radiation. Academic Press, 390 pp.
Ivanov, V. Y., Vivoni E. R. , Bras R. L. , and Entekhabi D. , 2004: Preserving high-resolution surface and rainfall data in operational-scale basin hydrology: A fully-distributed physically-based approach. J. Hydrol., 298 , 80–111.
Jackson, T. J., LeVine D. M. , Hsu A. , Oldak A. , Starks P. , Swift C. , Isham J. , and Haken M. , 1999: Soil moisture mapping at regional scales using microwave radiometry: The Southern Great Plains Hydrology Experiment. IEEE Trans. Geosci. Remote Sens., 37 , 2136–2151.
Kot, S. C., and Song Y. , 1998: An improvement of the Louis scheme for the surface layer in an atmospheric modelling system. Bound.-Layer Meteor., 88 , 239–254.
Kumar, P., 2004: Layer averaged Richards’ equation with lateral flow. Adv. Water Resour., 27 , 521–531.
Lakshmi, V., Albertson J. , and Schaake J. , 2001: Land Surface Hydrology, Meteorology, and Climate: Observations and Modeling. Amer. Geophys. Union, 246 pp.
Leij, F., Romano N. , Palladino M. , Schaap M. , and Coppola A. , 2004: Topographical attributes to predict soil hydraulic properties along a hillslope transect. Water Resour. Res., 40 .W02407, doi:10.1029/2002WR001641.
Liang, X., and Xie Z. , 2001: A new surface runoff parametrization with subgrid-scale soil hetereogeneity for land surface models. Adv. Water Resour., 24 , 1173–1193.
Liang, X., Lettenmaier D. P. , Wood E. F. , and Burges S. J. , 1994: A simple hydrologically based model of land surface water and energy fluxes for general circulation models. J. Geophys. Res., 99D , 14415–14428.
Louis, J. F., 1979: A parametric model of vertical eddy fluxes in the atmosphere. Bound.-Layer Meteor., 17 , 187–202.
Mengelkamp, H., Warrach K. , and Raschke E. , 1999: SEWAB— A parametrization of the surface energy and water balance for atmospheric and hydrologic models. Adv. Water Resour., 23 , 165–175.
Mesa, O. J., and Mifflin E. R. , 1986: On the relative role of hillslope and network geometry in hydrologic response. Scale Problems in Hydrology, V. K. Gupta, I. Rodriguez-Iturbe and E. Wood, Eds., D. Reidel, 181–190.
Milly, P. D., 1974: Moisture and heat transport in hysteretic, inhomegeneous porous media: A matric head-based formulation and a numerical model. Water Resour. Res., 10 , 514–520.
Mohanty, B. P., Famiglietti J. , and Skaggs T. , 2000: Evolution of soil moisture spatial structure in a mixed-vegetation pixel during the SGP97 Hydrology Experiment. Water Resour. Res., 36 , 3675–3686.
Mohanty, B. P., Shouse P. J. , Miller D. A. , and van Genuchten M. T. , 2002: Soil property database: Southern Great Plains 1997 Hydrology Experiment. Water Resour. Res., 38 .1047, doi:10.1029/2000WR000076.
Mohr, K. I., Famiglietti J. S. , Boone A. , and Starks P. , 2000: Modeling soil moisture and surface flux variability with an untuned land surface scheme: A case study from the Southern Great Plains 1997 hydrology experiment. J. Hydrometeor., 1 , 154–169.
Monin, A. S., and Obukhov A. M. , 1954: Basic laws of turbulent mixing in the ground layer of the atmosphere. Trans. Geophys. Inst. Akad., 151 , 163–187.
Mualem, Y., 1976: A new model for predicting the hydraulic conductivity of unsaturated porous media. Water Resour. Res., 12 , 513–522.
Orlandini, S., Moretti G. , Franchini M. , Aldighieri B. , and Testa B. , 2003: Path-based methods for the determination of nondipersive drainage directions in grid-based digital elevations models. Water Resour. Res., 39 .1144, doi:10.1029/2002WR001639.
Paniconi, C., and Wood E. F. , 1993: A detailed model for simulation of catchment scale subsurface hydrologic processes. Water Resour. Res., 29 , 1601–1620.
Paniconi, C., and Putti M. , 1994: A comparison of Picard and Newton iteration in the numerical simulation of multidimensional variably saturated flow problems. Water Resour. Res., 30 , 3357–3374.
Paniconi, C., Marrocu M. , Putti M. , and Verbunt M. , 2003a: Newtonian nudging for a Richards equation-based distributed hydrological model. Adv. Water Resour., 26 , 161–178.
Paniconi, C., Troch P. A. , van Loon E. , and Hilberts A. G. J. , 2003b: Hillslope-storage Boussinesq model for subsurface flow and variable source areas along complex hillslopes: 2. Intercomparison with a three-dimensional Richards equation model. Water Resour. Res., 39 .1317, doi:10.1029/2002WR001728.
Pomeroy, J. W., Toth B. , Granger R. J. , Hedstrom N. R. , and Essery R. L. H. , 2003: Variation in surface energetics during snowmelt in a subarctic mountain catchment. J. Hydrometeor., 4 , 702–719.
Prosser, I. P., and Abernethy B. , 1996: Predicting the topographic limit to a gully network using a digital terrain model and process thresholds. Water Resour. Res., 32 , 2289–2298.
Reed, S., Koren V. , Smith M. , Zhang Z. , Moreda F. , and Seo D. , and DMIP participants, 2004: Overall Distributed Model Intercomparison Project results. J. Hydrol., 298 , 27–60.
Reggiani, P., Hassanizadeh S. , Sivapalan M. , and Gray W. , 1999: A unifying framework for watershed thermodynamics: Constitutive relationships. Adv. Water Resour., 23 , 15–39.
Richards, L. A., 1931: Capillary conduction of liquids in porous mediums. Physics, 1 , 318–333.
Rinaldo, A., Marani A. , and Rigon R. , 1991: Geomorphological dispersion. Water Resour. Res., 27 , 513–525.
Rodriguez-Iturbe, I., and Rinaldo A. , 1997: Fractal River Networks: Chance and Self-Organization. Cambridge University Press, 547 pp.
Rodriguez-Iturbe, I., Rinaldo A. , Rigon R. , Bras R. L. , Marani A. , and Ijjasz-Vasquez E. , 1992: Energy dissipation, runoff production, and the three-dimensional structures of river basins. Water Resour. Res., 28 , 1095–1103.
Sellers, P. J., Hall F. G. , Asrar G. , Strebel D. E. , and Murphy R. E. , 1992: An overview of the First International Satellite Land Surface Climatology Project (ISLSCP) Field Experiment (FIFE). J. Geophys. Res., 97 , 345–373.
Smith, R. E., Ed. 2002: Infiltration Theory for Hydrologic Applications.Water Resources Monogr., Vol. 15, Amer. Geophys. Union, 212 pp.
Tamanini, D., 2003: Una nuova modellazione dei flussi nell’insaturo nel modello idrologico distribuito GEOTOP. Tesi di laurea (Masters thesis), Corso di laurea in Ingegneria per l’Ambiente e il Territorio, University of Trento, 149 pp.
Troch, P. A., Paniconi C. , and van Loon E. , 2003: Hillslope-storage Boussinesq model for subsurface flow and variable source areas along complex hillslopes: 1. Formulation and characteristic response. Water Resour. Res., 39 .1316, doi:10.1029/2002WR001728.
van Genuchten, M. T., 1980: A closed-form equation for predicting the hydraulic conductivity of unsaturated soils. Soil Sci. Soc. Amer. J., 44 , 892–898.
Vereecken, H., Maes J. , Feyen J. , and Darius P. , 1989: Estimating the soil moisture retention characteristic from texture, bulk density and carbon content. Soil Sci., 148 , 389–403.
Vertessy, R., Hatton T. , O’Shaughnessy P. , and Jayasuriya M. , 1993: Predicting water yield from a mountain ash forest using a terrain analysis based catchment model. J. Hydrol., 150 , 665–700.
Warrach, K., Stieglitz M. , Mengelkamp H. , and Raschke E. , 2002: Advantages of a topographically controlled runoff simulation in a SVAT model. J. Hydrometeor., 3 , 131–148.
Western, A. W., Grayson R. B. , and Green T. , 1998: The Tarrawarra data set: Soil moisture patterns, soil characteristics and hydrological flux measurements. Water Resour. Res., 34 , 2765–2768.
Wigmosta, M. S., Vail L. , and Lettenmaier D. , 1994: A distributed hydrology–vegetation model for complex terrain. Water Resour. Res., 30 , 1665–1679.
Wood, E., and Coauthors, 1999: The Project for Intercomparison of Land-Surface Parameterization Schemes (PILPS) Phase-2(c) Red–Arkansas river basin experiment: 1. Experiment description and summary intercomparisons. Global Planet. Change, 19 , 115–136.
Woods, R. A., and Coauthors, 2001: Experimental design and initial results from the Mahurangi River variability experiment: MARVEX. Observations and Modelling of Land Surface Hydrological Processes, V. Lakshmi, J. Albertson, and J. Schaake, Eds., Water Science and Application 3, Amer. Geophys. Union, 201–213.
Zanotti, F., Endrizzi S. , Bertoldi G. , and Rigon R. , 2004: The geotop snow module. Hydrol. Processes, 18 , 3667–3679.
Spatial properties of the Little Washita watershed. (upper left) Topography, (upper right) precipitation on 10 Jul 1997 with the location of the meteorological stations used in GEOtop simulations, (bottom right) soil texture (data from Mohanty et al. 2002), and (bottom left) land cover (data from Goddard Earth Sciences Data and Information Services Center).
Citation: Journal of Hydrometeorology 7, 3; 10.1175/JHM497.1
Comparison between modeled and observed energy balance components at the NOAA flux site, 27 Jun to 17 Jul 1997.
Citation: Journal of Hydrometeorology 7, 3; 10.1175/JHM497.1
(top) Observed rainfall and comparison between modeled and observed (middle) surface soil moisture and (bottom) temperature at the NOAA flux site, 27 Jun to 17 Jul 1997.
Citation: Journal of Hydrometeorology 7, 3; 10.1175/JHM497.1
Comparison between GEOtop simulations and measurements in the Little Washita watershed from discharge 27 Jun to 17 Jul 1997. The parameters used in the GEOtop simulations are given in Table 1. (a) Discharge at the outlet; (b) basin-averaged surface soil moisture, simulated in surface soil layer (top 5 cm) compared to ESTAR estimated value.
Citation: Journal of Hydrometeorology 7, 3; 10.1175/JHM497.1
Cumulative water balance of the Little Washita watershed, 27 Jun to 17 Jul 1997, as determined by GEOtop model simulations with parameters indicated in Table 1.
Citation: Journal of Hydrometeorology 7, 3; 10.1175/JHM497.1
Sensitivity of the flood hydrograph of 11 Jul 1997 in the Little Washita watershed, simulated with GEOtop, to different parameters: (a) Ks υ, (b) Ks h, (c) θr, (d) θs, (e) α, (f) n, (g) m, and (h) Sυ. Here Ks υ and Ks h are vertical and lateral saturated hydraulic conductivity at the surface, respectively; θr and θs are residual and saturated water content; α, n, and m are parameters of the van Genuchten (1980) curve; and Sυ is the averaged fractional canopy cover. The measured value of the peak discharge was 60 m3 s−1.
Citation: Journal of Hydrometeorology 7, 3; 10.1175/JHM497.1
Volumetric surface soil water content of the Little Washita watershed. (left) Estimated using the ESTAR remote sensor. (right) Simulated using the GEOtop model. For (top) 3, (middle) 11, and (bottom) 16 July 1997.
Citation: Journal of Hydrometeorology 7, 3; 10.1175/JHM497.1
Volumetric surface (top 5 cm) soil water content frequency distributions, comparison between ESTAR data, GEOtop model, and VIC model. (upper left) 29 Jun, (upper right) 3 Jul, (bottom right) 11 Jul, and (bottom left) 16 Jul 1997.
Citation: Journal of Hydrometeorology 7, 3; 10.1175/JHM497.1
Comparison between measured discharge and the GEOtop simulated discharge in the Little Washita watershed for the whole 1997 year: (a) linear scale; (b) logarithmic scale. The black line is the measured discharge; the upper limit of the gray band is the simulated discharge with sυ = 0.89 (with less bare soil evaporation and more transpiration); the lower limit of the gray band is the simulated discharge with sυ = 0.78.
Citation: Journal of Hydrometeorology 7, 3; 10.1175/JHM497.1
Comparison between simulations in the Little Washita watershed for the whole 1997 year with different averaged canopy fraction Sυ: (a) evapotranspiration, (b) total soil column saturation. The black lines refer to the simulation with sυ = 0.89, the gray one with sυ = 0.78.
Citation: Journal of Hydrometeorology 7, 3; 10.1175/JHM497.1
Basin-averaged values of the parameters in the application of GEOtop to the Little Washita basin.