A Distributed Hydrological Model With Coupled Water and Energy Budgets

JUNE 2006 RIGON ET AL.
371
GEOtop: A Distributed Hydrological Model with Coupled Water and Energy Budgets
RICCARDO RIGON AND GIACOMO BERTOLDI
Department of Civil and Environmental Engineering, CUDAM, University of Trento, Trento, Italy
THOMAS M. OVER
Department of Geology/Geography, Eastern Illinois University, Charleston, Illinois
(Manuscript received 1 March 2005, in final form 11 August 2005)
ABSTRACT
This paper describes a new distributed hydrological model, called GEOtop. The model accommodates
very complex topography and, besides the water balance, unlike most other hydrological models, integrates
all the terms in the surface energy balance equation. GEOtop uses a discretization of the landscape based
on digital elevation data. These digital elevation data are preprocessed to allow modeling of the effect of
topography on the radiation incident on the surface, both shortwave (including shadowing) and longwave
(accounting for the sky view factor). For saturated and unsaturated subsurface flow, GEOtop makes use of
a numerical solution of the 3D Richards’ equation in order to properly model, besides the lateral flow, the
vertical structure of water content and the suction dynamics. These characteristics are deemed necessary for
consistently modeling hillslope processes, initiation of landslides, snowmelt processes, and ecohydrological
phenomena as well as discharges during floods and interstorm periods. An accurate treatment of radiation
inputs is implemented in order to be able to return surface temperature. The motivation behind the model
is to combine the strengths and overcome the weaknesses of flood forecasting and land surface models. The
former often include detailed spatial description and lateral fluxes but usually lack appropriate knowledge
of the vertical ones. The latter are focused on vertical structure and usually lack spatial structure and
prediction of lateral fluxes. Outlines of the processes simulated and the methods used to simulate them are
given. A series of applications of the model to the Little Washita basin of Oklahoma using data from the
Southern Great Plains 1997 Hydrology Experiment (SGP97) is presented. These show the model’s ability
to reproduce the pointwise energy and water balance, showing that just an elementary calibration of a few
parameters is needed for an acceptable reproduction of discharge at the outlet, for the prediction of the
spatial distribution of soil moisture content, and for the simulation of a full year’s streamflow without
additional calibration.
1. Introduction: Design prerequisites and momentum exchanges between the land surface
and the atmosphere at several scales (Abbott 1992;
The study of river basin hydrology is focused on the
Reggiani et al. 1999), with the purpose of creating mod-
analysis of the interactions between the near-surface
els that can provide improved mid- and long-term hy-
soil and the atmospheric boundary layer (ABL), which
drologic forecasts and better prediction of the impacts
occur mainly through the mediation of the soil itself,
on the hydrologic cycle and on the earth’s ecosystems
the vegetation, and the turbulent and radiative energy
resulting from changes in land use and in the climate
transfers taking place on the earth’s surface, and pos-
(Grayson and Blöschl 2000). Though inspired by this
sible feedbacks from the ABL (e.g., soil moisture–
trend toward improved predictions, the initial motiva-
precipitation feedback mechanisms). In recent years,
tions behind the development of GEOtop were practi-
hydrologic research has generally evolved toward a
cal, as its development began with attempts to predict,
comprehensive theory describing the water, energy,
besides river discharges, landslide and debris flow ini-
tiation, snowpack evolution and ablation, and the water
budget (especially, accurate assessment of evapotrans-
Corresponding author address: Riccardo Rigon, Dept. of Civil
and Environmental Engineering, CUDAM, University of Trento, piration and the distribution of soil moisture) in small
Via Mesiano, 77, Trento TN 38050, Italy. catchments and in complex mountain terrain.
E-mail: riccardo.rigon@ing.unitn.it None of the above phenomena can be currently pre-
© 2006 American Meteorological Society
Brought to you by University of Maryland, McKeldin Library | Unauthenticated | Downloaded 07/06/22 05:06 PM UTC
JHM497
372 JOURNAL OF HYDROMETEOROLOGY VOLUME 7
dicted with enough accuracy and consistency to be op- floods but do not provide enough hydrological infor-
erationally useful (Committee on Hydrologic Science mation for forecasting other processes.
2002). It is known that such phenomena are affected by Vice versa, the land surface models (LSMs), unlike
heterogeneity in soil hydraulic properties, land use, the flood-forecasting models, are usually one-
landscape, topography and geological properties, and dimensional (1D) vertical models that represent soil–
soil moisture profile and surface–subsurface interac- atmosphere interactions with a great degree of com-
tions, but it has not been quantitatively assessed to what plexity and accuracy, but they are usually not endowed
extent. This cannot be explicitly obtained without a with a detailed representation of runoff and lateral
small-scale distributed model that is capable of explic- fluxes (Warrach et al. 2002), as they have been devel-
itly accounting for these characteristics. Furthermore, it oped mainly to provide a land surface interface in sup-
would require that both the water and energy balances port of atmospheric general circulations models
be solved (Grayson and Blöschl 2000). Solving the en- (GCMs). Reviews of LSMs are presented in Garratt
ergy balance means, among other things, being able to (1992, 1993) and Lakshmi et al. (2001). State-of-the-art
estimate surface temperature Ts, and it also requires LSMs, inclusive of the ability to model snowmelt, freez-
proper treatment of radiation physics (including the ef- ing soil, and multilayer canopy models are, to mention
fect of the view angle and shadowing in complex ter- a few, the biosphere–atmosphere transfer scheme
rain) (Blöschl et al. 1991). However, since temperature (BATS; Dickinson et al. 1986), the surface energy and
is absolutely necessary in modeling snowpack evolution water balance (SEWAB; Mengelkamp et al. 1999), the
(e.g., Zanotti et al. 2004; Pomeroy et al. 2003), its simu- Noah LSM (Chen et al. 1996), and the National Center
lation should be included in the design of such a model. for Atmospheric Research (NCAR) LSM (Bonan 1996).
Finally, the proper description of many of the hydro- GEOtop is designed to be an integration of the char-
logic phenomena on hillslopes (including hill- acteristics of both LSMs and flood-forecasting models
slope stability) and of water fluxes in the absence of and to include a detailed treatment of topographic in-
significant slopes, such as in plains or in riparian zones, teraction with radiation, which is not usually available
requires an adequate treatment of suction head in un- in other models. Thus GEOtop is a distributed model;
saturated soils. it is a terrain-based model; it simulates not only the
Just a few distributed hydrological models have all water balance but also the energy balance; and it is built
the necessary characteristics. Since the blueprint of for remote sensing data because it provides surface
Freeze and Harlan (1969), many “distributed” hydro- temperature and an accurate treatment of radiation and
logical models (Beven 2000) have been implemented to moisture in the upper soil (Troch et al. 2003).
predict the lateral distribution of water, ranging from This paper is organized as follows. Section 2 contains
more conceptualized descriptions (e.g., Beven and the description of model components, which begins
Kirkby 1979; Franchini and Pacciani 1991) to fully dis- with a description of the discretization of the landscape
tributed approaches, such as the Système Hydrologique in GEOtop, followed by a description of local water
Européen (SHE; Abbott et al. 1986), topographic ki- and energy budgets, and concluding with a description
nematic approximation and integration model (TOP- of the processes driving the water and energy fluxes.
KAPI; Ciarapica and Todini 2002), THALES (Grayson Section 3 contains three case studies based on the 1997
et al. 1992), and others (Vertessy et al. 1993; Garrote Southern Great Plains Hydrology Experiment (SGP97)
and Bras 1995; Bronstert and Plate 1997). At the catch- dataset. The case studies show the reproduction of the
ment scale, distributed approaches range from a three- water and energy balance at a point, a calibration of the
dimensional integration of the Richards’ equations as in model on a flood event on the Little Washita basin in
Paniconi and Putti (1994) or in the catchment hydro- Oklahoma, the spatial soil moisture distribution over
logical model (CATHY; Paniconi and Wood 1993; Pa- the basin, and the subsequent simulation of a yearlong
niconi et al. 2003a) to a distributed description of the surface discharge hydrograph with no additional cali-
hydrological effects of vegetation as in the Distributed bration. Conclusions follow in section 4.
Hydrological Soil Vegetation Model (DHSVM) (Wig-
mosta et al. 1994). Nevertheless, the Distributed Mod- 2. Description of model components
els Intercomparison Project (DMIP) (Reed et al. 2004) GEOtop simulates the fluxes and budgets of energy
provides a recent, clear example of the continued em- and water on a landscape defined by three-dimensional
phasis in hydrologic modeling on flood forecasting; the grid boxes, whose surfaces come from a digital eleva-
models involved, with a few exceptions (Liang and Xie tion model (DEM) and whose lower boundaries are
2001; Ivanov et al. 2004), were mostly event-oriented located at some specified spatially varying depth. If
rainfall-runoff models, which are successful in modeling snow accumulates on top of the land surface, a new
JUNE 2006 RIGON ET AL. 373
layer is added on top of the DEM as described in Zan- and in the soil
otti et al. (2004). Surface boundary conditions are given ⭸␪共x, t兲
by hydrometeorological measurements (rainfall, tem- ⫹ ⵜ · qsub共x, t兲 ⫽ ⫺S共x, t兲, 共2兲
⭸t
perature, wind velocity), regionalized as described in
Bertoldi (2004). where vectors are in bold characters, x is the position
We present here the components of the first public (L), t is time (T), qsup is the runoff discharge per unit
version of GEOtop. Information on previous versions surface area [in units of L (length) T⫺1(time)], c(x) the
can be found in Bertoldi and Rigon (2004). spatially varying kinematic wave celerity (L T⫺1), ⵜ· ⫽
(⳵/⳵x ⫹ ⳵/⳵y ⫹ ⳵/⳵z) is the divergence operator (L⫺1), qL
a. Landscape and equation discretization (T⫺1) is the unit volume flow exchange with the soil
[inflow from the soil (exfiltration) if positive, outflow to
GEOtop requires preprocessing of the catchment
the soil (infiltration) if negative], ␪ is the volumetric
DEM for which a special set of routines has been imple-
water content in the soil (or snow), and ⵜ · qsub is the
mented in an open source GIS, the Java Geographic
flux divergence of water per unit volume (T⫺1), which
Resource Analysis Support System (JGRASS; http://
includes the water fluxes from adjacent cells, regulated
www.hydrologis.com). The outputs of these routines in-
by Richards’ Eq. (6) below. The term S (T⫺1) includes
clude estimation of drainage directions, slopes, curva-
the exchanges between atmosphere and soil (evapora-
ture, the channel network structure, shadowing, and the
tion and transpiration) according to Eqs. (15)–(19). The
sky view factor. Surface runoff is modeled to follow the
treatment of the boundary conditions is derived from
terrain surface according to a so-called D8 topology as
Bixio et al. (2000) and allows for surface flow to infil-
in Orlandini et al. (2003). Advantages and disadvan-
trate (when possible) and subsurface flow to return to
tages of various methodologies for carrying out these
the surface (exfiltration and return flow). When chan-
computations are thoroughly discussed in Rodriguez-
nel pixels are present, surface runoff eventually turns
Iturbe and Rinaldo (1997) and in Orlandini et al. (2003)
into channel flow, which is routed with a separate nu-
and are not further discussed here. Soil thickness can be
merical scheme as described below. Bottom boundary
defined either using field data or the model by Heim-
conditions can be defined either by a flux at the bed-
sath et al. (1997), as discussed Bertoldi et al. (2006a).
rock–soil interface or by the conductivity of the deeper
The DEM identifies also the plan view of a three-
bedrock. Except for evapotranspiration, water vapor
dimensional grid on which all the model’s equations are
fluxes are neglected in the present implementation of
discretized. The thickness of the discretized soil layers
the model.
is set up at run time; it is usually appropriate to assume
that the top layer is quite thin (e.g., 5 cm thick). Ac- 2) ENERGY BUDGET
cording to criteria that the user can specify (e.g.,
The fluxes of energy for a unit volume of water in the
Prosser and Abernethy 1996), some of the grid cells can
earth’s gravitational field are given by the sum of a
be identified as river network cells. River network cells
“thermal” part, which is a function of the net radiation
are treated the same as hillslope cells except for the
Rn, the thermal fluxes into the ground Gr, energy inputs
routing of surface runoff, which is modeled as described
deriving from precipitation P, sensible heat flux H, la-
in section 2c.
tent heat flux ␭ET (where ␭ is the latent heat of vapor-
b. Water and energy budgets in GEOtop ization per unit mass), and a fluid-mechanical part (de-
pendent on the hydraulic head). The action of the me-
The budgets of mass and energy in GEOtop may be chanical part in producing internal energy by friction is
divided into three parts: the water budget, the energy neglected in GEOtop according to the common prac-
budget, and the radiation budget. The radiation budget tice. This results in a partial decoupling of the water and
is presented separately from the energy budget, as it energy balances, which however remain strongly con-
creates the link between the model and remotely nected through the evapotranspiration term and the de-
sensed data. pendence of hydraulic conductivity on temperature.
The energy balance is computed with respect to vol-
1) WATER BUDGETS
umes where the energy content of the water and soil are
GEOtop calculates the conservation of water accord- considered together and for the same grid boxes as the
ing to discretized versions of the continuity equations soil water balance. In continuous form, the energy bal-
for liquid water on the surface ance is written as
⭸qsup共x, t兲 ⭸U共x, t兲
⫹ c共x, t兲ⵜ · qsup共x, t兲 ⫽ c共x, t兲qL共x, t兲共1兲 ⫹ ⵜ · g共x, t兲 ⫽ 0, 共3兲
⭸t ⭸t
where U (E/L3) is the internal energy density and g is are expressed through the sky view factor V(x), which
the energy flux per unit area (E L⫺2 T⫺1). In the actual indicates the sky fraction visible at a point. The rel-
numerical implementation of Eq. (3), lateral energy evance of V(x) for the radiative balance in mountain
fluxes have been neglected, as they are usually much areas has been underlined, for instance, by Blöschl et
smaller than the vertical fluxes, even though at small al. (1991). Many models neglect shadowing and the sky
spatial scales (⬃10 m) lateral energy advection can be view factor, causing large errors in the local energy
nonnegligible (Albertson and Parlange 1999). Radia- balance estimation in mountainous catchments. The al-
tion, sensible heat, and precipitation energy input are gorithms for estimating these terms are fully docu-
effective only on the surface layer, that is, ⵜ g ⫽ Rn ⫹ mented in Bertoldi and Rigon (2004). Taking into ac-
P ⫹ Gr ⫹ H ⫹ ␭ET on the surface layer and only ⵜ g ⫽ count the effects of shadowing and of the sky view fac-
Gr ⫹ ␭ET on the other soil layers. The ␭ET term in- tor, the flux of net radiation Rn is
cludes both bare soil evaporation (only for the first Rn ⫽ 关sh共x, t兲R↓SW P ⫹ V共x兲R↓SW D兴关1 ⫺ V共x兲a兴
layer) and canopy transpiration (depending from root
density of each layer), as explained later. The top layer ⫹ V共x兲␧aR↓LW ⫺ V共x兲␧s␴T s4, 共5兲
can be the snowpack, if present, and can have any thick- ⫺2 ⫺1
where R↓SW (E L T ) is the shortwave net radiation
ness. In this case, the internal energy content U is di- at the surface obtained by correcting the solar radiation
vided between energy of the ice component and energy incoming at the atmosphere top (the subscript P refers
of the water component (Zanotti et al. 2004). Heat to the direct component and D to the diffuse; note the
fluxes in the ground between the layers, diffuse component is not influenced by shadowing but
is proportional to the sky view factor), a is the short-
⭸T共x, t兲
Gr ⫽ ⫺␭s共x, t, ␪兲 , 共4兲 wave albedo, ␧a is the atmospheric emissivity, ␧s is the
⭸z longwave soil (or snow) emissivity, ␴ is the Stefan–
are a function of the soil temperature gradient accord- Boltzman constant {5.6704 10⫺8 [W(m2K4)⫺1]}, R↓LW
ing to the Fourier heat diffusion equation where the the incoming longwave radiation, calculated according
thermal conductivity ␭s(x, t) depends on the mineral to Brutsaert (1975), depending on air temperature, air
properties of the soil and on the actual (time varying) humidity, and cloud cover, and Ts is surface tempera-
water content ␪ in the grid element. Freezing soil pro- ture. The factor [1 ⫺ V(x)a] accounts for the shortwave
cesses are included as described in Bertoldi et al. radiation reflected by the surfaces surrounding the
(2006b). The boundary condition in the deepest soil point. The last two terms of Eq. (5) are multiplied by
layer is the soil temperature obtained from the analyti- V(x) to account for the longwave radiation emitted by
cal solution of the Fourier law, using an annual sinu- the surfaces surrounding the point, under the hypoth-
soidal forcing. The soil temperature in Eq. (4) is ob- esis of radiative equilibrium.
tained in the model by the integration of the full surface Remote sensors detect only limited bands of the ra-
energy budget (Best 1998), solving the system given by diation spectrum and not the whole integrated infrared
the partial differential Eqs. (3) describing the soil tem- (LW) and visible parts (SW) of it as in (5). The radia-
perature profile. tion 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 sepa-
3) RADIATION BUDGET
rated into optical radiation bands where remote sensors
An accurate calculation of the net radiation is essen- operate.
tial for a hydrological model working in complex ter-
rain (Dubayah et al. 1990). Solar radiation geometry is c. Dynamic hydrology in GEOtop
treated in GEOtop as in Iqbal (1983), and all the ra- Water and energy fluxes, which control the budgets
diative corrections needed for complex topography, in- described above, are computed according to gradients
cluding shadowing of direct solar radiation by sur- of energy potentials as described below. A more de-
rounding mountains and the effect of topography on tailed description of the dynamics including complete
diffuse radiation, are included. Shadowing is expressed equations can be found in Bertoldi and Rigon (2004).
by the dimensionless factor sh(x, t), which is 0 if the grid The snow module and its preliminary application to a
cell is in shadow (no direct radiation) and 1 if radiation case study are described in Zanotti et al. (2004).
from the sun hits the surface. The shadowing factor is
calculated for any position, x, and for any instant, t, 1) SUBSURFACE SATURATED–UNSATURATED FLOW
depending on the interplay of solar position and topog- To model soil moisture dynamics properly in the soil
raphy. The effects of topography on diffuse radiation layers, GEOtop utilizes the integration of Richards’
equations (Richards 1931; Freeze and Harlan 1972) in where ␪r is the residual water content (dimensionless),
three dimensions (3D). For some of the processes esti- and ␣(L⫺1), m, and n (both dimensionless) are param-
mation, simpler approaches could have been used such eters. Such parameters, as well as the saturated hydrau-
as the hillslope-storage Boussinesq (hsB) equation de- lic conductivity, are 3D fields in the model and many
veloped by Troch et al. (2003), Paniconi et al. (2003b), different strategies for their assignment are possible.
and Hilberts et al. (2004), which has been shown to For instance, they can be derived from the soil texture
work generally very well as compared to the 3D Rich- by means of the pedotransfer functions proposed, for
ards’ equation for simulation of saturated subsurface example, by Vereecken et al. (1989) and Leij et al.
flow dynamics and to be much more computationally (2004) or by regionalization of the appropriate field
efficient. However, GEOtop is intended to simulate measurements. The relative hydraulic conductivity is
also the suction dynamics, and, in this, solving Rich- expressed as a function of the water content as in
ards’ equation is necessary. Mualem (1976). Even though we did not pursue a com-
The numerical scheme used is new and is described in plete thermodynamic treatment of saturated–
Tamanini (2003) and in Bertoldi and Rigon (2004). The unsaturated flow as in Milly (1974), saturated hydraulic
use of Richards’ equation removes the need to make an conductivity is made dependent on soil temperature
assumption of stationary conditions in subsurface flows through viscosity, which increases by a factor 2 from 10°
[as is assumed, e.g., in Beven and Kirkby (1979)], and to 40°C, allowing the treatment of soil freezing and
therefore it is possible to describe the transients of flow seasonal variation of hydraulic characteristics that oth-
and infiltration. Richards’ equation is here written ac- erwise require multiple calibrations of soil parameters
cording to Paniconi and Putti (1994) as through the year.
The description of vertical infiltration in GEOtop
qsub ⫽ ⫺KsKr共SW兲ⵜ共␩z ⫹ ␺兲, 共6兲 takes into account the surface fraction actually covered
in water, assuming the presence of microrelief in the
⭸␺ terrain, which is parameterized by a surface roughness
␴共SW兲 ⫽ ⵜ · 关Ks共T兲Kr共SW兲ⵜ共␺ ⫹ ␩z兲兴 ⫺ S, parameter according to Smith (2002) as
⭸t
共7兲 I ⫽ I0 if d ⱖ h0, 共9兲
I ⫽ I0dⲐh0 if d ⬍ h0, 共10兲
where the dependence from space and time of the vari-
where I is the effective infiltration rate, I0 is the infil-
ables has been dropped for notational simplicity, ␴ (SW)
tration rate with no microrelief, d is the surface water
⫽ SWSs ⫹ ␾ ⳵SW/⳵␺ is the storage term (L⫺1), SW(␺) ⫽
depth, and h0 is the microrelief height.
␪/␪s is the relative water saturation (the ratio between
Runoff at the surface is generated because the pre-
the volumetric soil moisture content ␪ and the saturated
cipitation intensity is greater than the soil surface infil-
moisture content ␪s), ␾ is the porosity (dimensionless),
tration capacity (infiltration excess runoff; Horton
Ss is the aquifer specific storage coefficient (L⫺1), ␺ is
1933) or the water table level rises above the soil sur-
the pressure hydraulic head (L), ⵜ is the gradient op-
face (saturation excess runoff; Dunne 1978). Both of
erator, Ks(T ) is the saturated hydraulic conductivity,
these possibilities are dealt with the Richards’ equation
and T is the temperature (K). Here, Kr(SW) ⬅ K(SW)/
solver and are implemented according to the compre-
KS(T ) is the relative hydraulic conductivity, z is the
hensive analysis of boundary conditions made by Pani-
vertical upward coordinate, and S represents the modu-
coni and Putti (1994). In addition, Richards’ equation
lus of the source or sink volumetric flow per unit vol-
allows for water to redistribute laterally and even move
ume [the same as in Eq. (2)]. The saturated hydraulic
uphill or exfiltrate according to local suction.
conductivity Ks(T ) has in fact two components, the ver-
tical, Ks ␷(T ), and the lateral Ks h(T ) grid-averaged val- 2) SURFACE RUNOFF
ues (Kumar 2004). In the model they are considered
proportional according to an anisotropy ratio given by Runoff is routed according to a kinematic scheme
␣K ⫽ Ks h/Ks ␷, which is used as a calibration parameter. that also accounts for subgrid-scale rilling and hetero-
The relation between ␺ and ␪ [e.g., Eq. (2)] in geneity in roughness:
GEOtop is given through the van Genuchten (1980) ⭸qsup共x, t兲 ⭸qsup共x, t兲
schematization, ⫹ c共x, t兲 ⫽ c关共x, t兲兴qL, 共11兲
⭸t ⭸s
冋冉冊册
1 ␪ ⫺ ␪r 1Ⲑm 1Ⲑn c共x, t兲 ⫽ Cmd共x, t兲␥ 关ⵜz共x兲兴0.5,
␺⫽ ⫺1 , 共8兲
␣ ␪s ⫺ ␪r 共12兲
where Eq. (11) is the continuity Eq. (1) written along through the bulk turbulent heat transfer coefficient CH
the drainage direction coordinate s (L), qsup is the run- (dimensionless), the wind speed uw, and the gradient
off discharge per unit surface area (L T⫺1) along s, c(x) between the surface soil/snow temperature Ts and the
is the spatially varying kinematic wave celerity module air temperature Ta as
(L T⫺1), Cm is a resistance coefficient, [L⫺(1⫺␥) T⫺1], d
Qs ⫽ ␳cpCH uw共Ts ⫺ Ta兲, 共14兲
is the surface water depth (L), ⵜz(x) is the local slope
(dimensionless), the exponent ␥ (dimensionless) is vari- where ␳ is the density of the air, and cp is the specific
able as a function of the runoff phenomena geometry, heat of the air at constant pressure. Here, Ts is obtained
and qL (T⫺1) is the unit volume vertical inflow (posi- from solving the energy balance, and Ta is a measured
tive: exfiltration) or outflow (negative: infiltration) external forcing; CH is a bulk turbulent transfer coeffi-
from any surface cell to the soil. The qL term contains cient that accounts for the turbulent transfer processes
the term Si from Eq. (7) and allows for runoff to rein- and the different roughness length zT of the various
filtrate downhill. The exponent 0.5 in Eq. (12) is cali- surfaces through a mapping of the land use and a pa-
brated or specified according to the literature (e.g., rameterization of the stability of the ABL. In particu-
Smith 2002). lar, CH is determined using similarity theory (Garratt
1992) and is expressed according the theory of Louis
3) CHANNEL FLOW (1979), which uses the Richardson number (a function
GEOtop does not need a channel network to be of the potential temperature gradient between soil and
identified since water can move downhill according to atmosphere) as a stability parameter. Similarity func-
(12). However, channel geometry and roughness adjust tions are given by the approach of Kot and Song (1998).
themselves to maintain certain mean flow characteris- The total evapotranspiration is a sink in (7) made up
tics (e.g., Rodriguez-Iturbe et al. 1992) and thus it is of the sum of evaporation or sublimation from the soil
generally useful to separate channel flow routing from or snow surface EG, transpiration from the vegetation
the hillslope runoff routing. In the channels, surface ETC, and evaporation of the precipitation intercepted
water flow is described by the convolution of the in- by the vegetation EVC, that is,
coming discharge with the solution of the de Saint- ET ⫽ ETC ⫹ EG ⫹ EVC. 共15兲
Venant parabolic equation using a constant celerity in
the whole network as proposed in Mesa and Mifflin Every cell has a fraction covered by vegetation and a
(1986), Rinaldo et al. (1991), and D’Odorico and Rigon fraction covered by bare soil, and the vegetation frac-
(2003): tion contains an interception store, modeled as in Men-
gelkamp et al. (1999). The model uses a single-layer
Qc共t兲 ⫽ AT 冕冕 0
t Lmax
0
sW共␶,s兲
公4␲D共t ⫺ ␶兲3
exp
canopy model, and the leaf temperature is assumed to
be equal to soil temperature, as in Garratt (1992). The
green vegetation fraction is static, and there is not a
冋 ⫺
共s ⫺ uc共t ⫺ ␶兲兲2
4D共t ⫺ ␶兲
册ds d␶, 共13兲
mechanism to automatically vary the leaf area index
(LAI) to model plant development, but it may be
changed by external input to the model.
where Qc(t) is the discharge at the basin outlet (L3 The components of the total evapotranspiration are
T⫺1), AT is the basin area, W(␶, s) (T⫺1) is the inflow of computed as functions of the potential evapotranspira-
the water from the hillsides into the channel network at tion EP, which is computed as a function of the gradient
a distance s (L) from the outlet and at a time ␶ [i.e., it between saturation-specific moisture at the soil surface
is a dynamical width function (Rodriguez-Iturbe and temperature q*(Ts) and the atmospheric specific mois-
Rinaldo 1997)], uc is a mean celerity (L T⫺1), D is a ture q(Ta) (M/M) as
hydrodynamic dispersion coefficient (L2 T⫺1), and Lmax
is the maximum distance from the outlet measured ␭EP ⫽ ␭␳CE uw关q*共Ts兲 ⫺ q共Ta兲兴, 共16兲
along the network (L). where the bulk turbulent water vapor transfer coeffi-
cient CE is taken as equal (though a simplification) to
4) TURBULENT FLUXES
CH. GEOtop does not contain a special parameteriza-
Sensible H and latent ␭ET heat fluxes are determined tion of the wind profile inside the canopy cover, but it
by similarity theory (Monin and Obukhov 1954). The uses a displacement height to properly parameterize
vertical sensible heat flux H (E T⫺1 L⫺2) is expressed in vegetation roughness. Bare soil evaporation EG is re-
a form integrated between the surface and a reference lated to the water content of the first layer through the
height as a function of the atmospheric turbulence soil resistance analogy (Bonan 1996) as
ra tained starting with a saturated profile and running the

EG ⫽ 共1 ⫺ s␷兲EP , 共17兲
ra ⫹ rs model several times on the meteorological forcings of
the period analyzed until the spatial average surface
where s␷ (dimensionless) is the fraction of soil covered
soil moisture reached the same value as the Electrically
by the vegetation and ra (L2 T M⫺1) is the aerodynamic
Steered Thinned Array Radiometer (ESTAR) average.
resistance; ra ⫽ 1/(␳ CE uw). The soil resistance rs is a
This set of simulations is further used for a simple pa-
function of the water content of the first layer as in
rameter sensitivity study. The third simulation (C), fi-
Feddes et al. (2001). The evaporation from wet vegeta-
nally, forecasts a yearlong record of measured dis-
tion is calculated as
charges without any further calibration, using the pa-
EVC ⫽ s␷ EP␦W, 共18兲 rameters derived from simulations A and B. Two
where ␦W is the wet vegetation fraction, computed fol- versions of this simulation were performed. One was
lowing Deardorff (1978). The transpiration is calcu- performed without any additional calibration, and in
lated as the other, an adjustment of the vegetation cover frac-
n i tion was made.
兺r
f root ra
ETC ⫽ s␷EP共1 ⫺ ␦W兲 . 共19兲
i a ⫹ r ic a. The dataset
The root fraction f iroot of each soil layer is calculated SGP97 took place from 18 June to 17 July 1997 in
decreasing linearly from the surface to a maximum root central Oklahoma and is discussed in detail by Jackson
depth zroot, depending on the vegetation type. The et al. (1999), Mohanty et al. (2002), and Mohr et al.
canopy resistance r ic in soil layer i depends on solar (2000). The basin primarily investigated was the Little
radiation, vapor pressure deficit, temperature as in Best Washita watershed, which has been the focus of exten-
(1998), and on water content in the root zone as in sive hydrologic research by the Agricultural Research
Wigmosta et al. (1994). Service (ARS) (Allen and Naney 1991). The modeled
portion of the watershed (i.e., upstream of U.S. Geo-
d. Numerics
logical Survey stream gauge 07327550, east of Ninne-
GEOtop equations are discretized according to a fi- kah) covers an area of 603 km2. The topography of the
nite-volume method, and time resolution is the smallest region is moderately rolling with a maximum relief over
required for the numerical stability of all the processes the basin of less than 200 m (Fig. la). Land use is domi-
described (or by the time step of the forcings), usually nated by rangeland and pasture (63%), with significant
between 1 and 5 min. Grid cell horizontal dimensions areas of winter wheat and other crops concentrated in
can range from 2 to 500 m, and the number of soil layers the floodplain and western portions of the watershed
and the maximum soil thickness is arbitrary. All the area. As part of the ARS Micronet, there are a total of
processes are solved using the same time step. GEOtop 42 recording rain gauges distributed at a 5-km spacing
is coded in ANSI C and is compiled to run on Linux, over the watershed (Fig. 1b); these provided the pre-
Mac OS X, and Windows. The first public version was cipitation data used in this analysis. For meteorological
released in December 2005 under a General Public Li- data other than precipitation, the three Oklahoma Me-
cense (GPL), and can be downloaded from http:// sonet stations, Acme (ACME), Apache (APAC), and
bedu.ing.unitn.it/GEOtopWiki. The code will be thor- Ninnekah (NINN), located inside or very near the basin
oughly documented with doxygen (http://www. were used. Moreover, at the National Oceanic and At-
doxygen.org). Further information about the numerics mospheric Administration (NOAA) (LW02) flux site
of the model can be found in Bertoldi et al. (2006b). (marked in Fig. 1b), measurements of soil properties,
soil moisture and temperature, and surface energy
3. Tests of GEOtop in the Little Washita fluxes are available.
watershed
b. Simulations set A: Pixel-scale simulations of the
An opportunity to validate and analyze the perfor-
vertical profile dynamics at the NOAA (LW02)
mance of GEOtop is offered by the SGP97 dataset.
flux site
Three different kinds of simulations were performed.
The first set of simulations (A) uses GEOtop to repro- The purpose of these simulations is to test the pixel-
duce the energy and water fluxes at the pixel scale to scale behavior of GEOtop. Because of this, the lateral
assess the physical realism present in the model. The water fluxes have been inhibited (but infiltration is al-
second set of simulations (B) reproduces a flood event lowed). The model was initialized with the measured
and soil moisture evolution and calibrates some routing soil moisture and temperature profiles. Calibration was
parameters. The initial water content profile was ob- performed only on the following parameters related to
FIG. 1. 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).
the energy budget: the thermal roughness length zT eterization of the nocturnal stable boundary layer.
(Garratt 1993), of which the turbulent transport coeffi- There is also a systematic overestimation of the ground
cients [CH and CE in Eq. (14)] are a function; the frac- heat flux during early morning, resulting from the mod-
tion of the cell covered by canopy [s␷ in Eq. (15)], which el’s energy balance closure constraints, which is also
controls the rate between bare soil evaporation and reflected by a slight overestimation of the soil tempera-
transpiration; and the soil thermal conductivity ␭s. Cali- ture as shown in Fig. 3. Moreover, negative values of
brated values for this location are zT ⫽ 8.6 mm, s␷ ⫽ 0.9, sensible heat flux H in the dataset could be due to the
and ␭s ⫽ 0.077 J kg⫺1 K⫺1. Other soil properties have procedure chosen to process the measurements. A
been kept equal to local field values as derived from model structural change could be then suggested that,
Mohanty et al. (2002). The pixel-scale hourly evolution for instance, used a two-layer canopy model to provide
of energy balance components are illustrated in Fig. 2, a more accurate estimation of the diurnal cycle of soil
and rainfall, surface soil moisture, and temperature temperature.
variations are illustrated in Fig. 3. The model is capable
c. Simulation set B: Calibration and parameter
of simulating infiltration related to small rainfall events
sensitivity of discharge for the period from
and daily soil moisture evaporation cycles, as shown in
26 June to 16 July 1997
Fig. 3. The model outputs show a good agreement with
the local energy fluxes measured with eddy correlation The goal of this set of simulations is to determine the
systems, except for a slight overestimation of nocturnal surface and subsurface flow parameters later used in a
sensible heat fluxes (Fig. 2), which perhaps reveals longer simulation, and to do some sensitivity analysis.
some deficiencies of the Kot and Song (1998) param- During the simulated period the weather was quite fair,
FIG. 2. Comparison between modeled and observed energy balance components at the
NOAA flux site, 27 Jun to 17 Jul 1997.
with mostly hot sunny days, during which the soil dried the meteorological forcing from 26 June to 16 July 1997
quickly as result of intense summer evaporation. A until the spatial-averaged surface soil moisture reached
strong convective storm on 11 July (Fig. 1b) with more the same value as the ESTAR average.
than 50 mm of rainfall in 7 h (averaged over the basin) Surface values of hydraulic conductivity and the van
caused a flood with a peak discharge of 60 m3 s⫺1. Rain Genuchten parameters [see Eq. (8)] were obtained
patterns for this event were reconstructed by means of from soil texture data from Mohanty et al. (2002) using
kriging techniques. Simulations were performed with a the Vereecken et al. (1989) pedotransfer function ap-
grid resolution of 200 m and five soil layers of increasing proach. To assign the 3D field structure of the saturated
thickness (the first one of 5 cm). Soil depth was given by hydraulic conductivity, an exponential decay with
field measurements by the Oklahoma State University for depth was assumed according to Ks ⫽ e⫺fz z⬘, where z⬘
Application of Remote Sensing Laboratory and ranges is the cell depth below the surface; the factor fz (L⫺1)
from 0.5 to 1.5 m. As a lower boundary condition, an was determined by calibration. The canopy fraction S␷,
impermeable surface was imposed. Heat fluxes at the the root depth zroot, the roughness length zT, and the
bottom were regulated on the mean annual surface albedo a are assigned according to the land-cover map
temperature and excursion, as explained in section 2b. (data available from Goddard Earth Sciences Data and
Surface soil moisture maps with a resolution of 800 Information Services Center; Fig. 1c). Global param-
m, derived from Jackson et al. (1999) from the flights of eters related to the water budget were calibrated
the National Aeronautics and Space Administration against the measured discharges: the channel celerity
(NASA) P-3B aircraft fitted with the ESTAR, an L- uc, the hydraulic diffusivity D, the surface runoff resis-
band (1.413 GHz) passive microwave sensor, were used tance coefficient Cm [all in Eq. (13)], and the anisotropy
for model initialization and verification. The initial wa- ratio ␣k [in Eq. (6)]. Values of the spatial average of the
ter content profile was obtained starting with a satu- parameters are given in Table 1.
rated profile and running the model several times using Comparison of the measured and calibrated dis-
FIG. 3. (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.
charge hydrograph is shown in Fig. 4a, and the ESTAR The sensitivity of the discharge during the major
measured and simulated basin-average soil moisture is storm event on 11 July of the parameters Ks h, Ks ␷, ␪r,
shown in Fig. 4b. It can be seen that it was possible to ␪s, ␣, m, n, and S␷ is shown in Fig. 6. Among the cali-
obtain a good calibration to the main flood hydrograph. brated parameters summarized in Table 1, the vertical
Also fairly good is the uncalibrated forecasting of the hydraulic conductivity (Fig. 6a) determines the maxi-
average soil moisture, which implies that, on average, mum infiltration capacity and greatly influences the
evapotranspiration (ET) was properly simulated. Cu- surface runoff formation, because it controls the
mulative water balance simulation results (Fig. 5) show amount of Hortonian-type runoff, and therefore the
that total evaporation ET flux dominates the water magnitude of the peak discharge. The horizontal hy-
losses during this period compared to discharge Qc; in draulic conductivity, Ks h, on the contrary, primarily in-
particular, only a small fraction of the 11 July storm fluences the flood tail because it determines the sub-
precipitation became runoff. surface runoff quantity (Fig. 6b). A large value of Ks h,
TABLE 1. Basin-averaged values of the parameters in the application of GEOtop to the Little Washita basin.
Parameter symbol Description Units Basin-averaged value Source

u Channel celerity m s⫺1 2.0 Calibration
D Channel diffusivity m2 s⫺1 1000 Calibration
Ks h /Ks ␷ Hydraulic conductivity anisotropy ratio 50 Calibration
fz Hydraulic conductivity decay factor m⫺1 0.5 Calibration
Cm Hydraulic surface resistance m1/3 s⫺1 0.1 Calibration
␪r Residual water content 0.06 Literature
␪s Saturated water content 0.41 Literature
␣ van Genuchten parameter cm⫺1 0.035 Literature
m van Genuchten parameter 1.0 Literature
n van Genuchten parameter 0.98 Literature
␭s Thermal soil conductivity J kg⫺1 K⫺1 0.077 Calibration
S␷ Canopy fraction 0.78 Literature
zroot Root depth m 0.82 Literature
zT Roughness length m 0.021 Literature
a Albedo 0.2 Literature
FIG. 5. Cumulative water balance of the Little Washita water-

shed, 27 Jun to 17 Jul 1997, as determined by GEOtop model
simulations with parameters indicated in Table 1.
the model do not show the influence of land use; how-

ever, 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
FIG. 4. Comparison between GEOtop simulations and measure-
of precipitation, as pointed out by the comparison of
ments in the Little Washita watershed from discharge 27 Jun to 17
Jul 1997. The parameters used in the GEOtop simulations are the precipitation field of 10 July in Fig. 1 to the soil
given in Table 1. (a) Discharge at the outlet; (b) basin-averaged moisture map of 11 July reported in Fig. 7, strongly
surface soil moisture, simulated in surface soil layer (top 5 cm) determines the spatial distribution of the soil surface
compared to ESTAR estimated value. water content, mostly in the first hour immediately fol-
lowing the rainfall. According to the simulations, as
however, causes soil drainage that is too rapid, leading qualitatively shown in the 11 July map of Fig. 7, topog-
to water content values that are too low compared to raphy starts exerting its influence only some hours
the measured values. Residual water content ␪r, which later, by concentrating moisture in the topographic con-
controls the total basin storage capacity, has a greater vergence zones.
influence on discharge than the water content value at The simulated averaged surface soil moisture values
saturation ␪s (Figs. 6c and 6d). The van Genuchten pa- are very similar to those estimated by ESTAR, as was
rameters ␣, n, and m [Eq. (8)] have also a significant shown in Fig. 4b. However, the frequency distribution
influence on peak discharge (Figs. 6e–g). In particular, of global measured and simulated water content are
very low values of ␣ (which means increased suction somewhat different (Fig. 8). In particular, GEOtop pre-
potential for the same saturation) lead to a bimodal dicted a larger soil moisture variance than ESTAR
behavior of the hydrograph. These simulations of this measured (especially in the case of rain events: see 11
event are not significantly affected by the vegetation- and 16 July in Fig. 8). To understand the reason for this
cover fraction S␷ (Fig. 6h), which, however, plays a ma- discrepancy, we also ran some simulations on the same
jor role in the energy budget calibration. Other sensi- dataset (using a coarser 1000-m grid) with the Variable
tivity analyses on soil thickness and river network ex- Infiltration Capacity (VIC) model (Liang et al. 1994),
tension are the subject of Bertoldi et al. (2006). obtaining distributions, which at least show that the
difference is not simply an effect of grid resolution (re-
d. Surface soil moisture evolution in the basin call that ESTAR has a resolution of 800 m). In fact,
The soil moisture distribution derived from the these differences among model simulation results and
ESTAR data by Jackson et al. (1999), available for the ESTAR products may be caused not only by errors in
same period in which we performed the calibration, was the model but also in the data assimilation used to ob-
compared to that produced by the top 5-cm layer of tain the ESTAR product. In fact, in the computation of
the model. As shown in Fig. 7, GEOtop predicted sur- the ESTAR soil moisture product soil dielectric prop-
face moisture patterns that appear visually similar to erties, which themselves depend on moisture content,
ESTAR estimates. The soil moisture maps produced by must be assumed.
FIG. 6. Sensitivity of the flood hydrograph of 11 Jul 1997 in the Little Washita watershed, simulated with
GEOtop, to different parameters: (a) Ks ␷ , (b) Ks h, (c) ␪r, (d) ␪s, (e) ␣, (f) n, (g) m, and (h) S␷. Here Ks ␷ and
Ks h are vertical and lateral saturated hydraulic conductivity at the surface, respectively; ␪r and ␪s are residual and
saturated water content; ␣, n, and m are parameters of the van Genuchten (1980) curve; and S␷ is the averaged
fractional canopy cover. The measured value of the peak discharge was 60 m3 s⫺1.
e. Simulation set C: The forecasting of the 1997 been kept constant. Only the three Oklahoma Mesonet
streamflow stations (Fig. 1) had precipitation data available for the
whole year; therefore the model was forced only with
In this simulation for the whole year of 1997, the these three stations. The model was initialized by forc-
calibrated parameters of the previous simulations have ing it repeatedly with the meteorological conditions of
flow also appeared in summer, which could be ex-

plained 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
s␷ ⫽ 0.78 to s␷ ⫽ 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 evolu-
tion of the discharge once s␷ 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 to-
pography, and hydraulic conductivity presents a strong
nonlinear dependence on soil moisture volumetric con-
tent.
4. Conclusions
In this paper, a new distributed hydrological model,
GEOtop, is presented. This model is an attempt to cre-
ate 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
FIG. 7. Volumetric surface soil water content of the Little energy and radiation budget. The model accommodates
Washita watershed. (left) Estimated using the ESTAR remote very complex topography and its effects on the incident
sensor. (right) Simulated using the GEOtop model. For (top) 3, radiation, both shortwave (with shadowing) and long-
(middle) 11, and (bottom) 16 July 1997. wave (with accounting for the sky view factor).
The application of the model to the Little Washita
the year (periodic forcing) until a dynamical equilib- basin leads to the following conclusions. The set of
rium was achieved after various periodic cycles had simulations A shows that the model is capable of accu-
passed. rately reproducing the local energy budget and the local
The comparison between the daily measured and the infiltration once initialized with measured parameters
GEOtop-simulated streamflow with no additional cali- and calibrated for the roughness length, the fraction of
bration (the gray band bottom) is shown in Fig. 9 for cell covered by canopies, and the heat capacity of the
the whole year of 1997. The model preserves fairly well soil. The set of simulations B shows that GEOtop, with
the total streamflow volume, but tends to overestimate calibration and accurate initialization, gives a precise
peak discharges, except for the event of late May when reproduction of a peak flow, with no more calibration
there is a considerable underestimation of the mea- than a typical rainfall-runoff model would require (i.e.,
sured flood. The underestimation for this event, how- adjusting the river network celerity, the surface and
ever, could be caused by inadequate rainfall inputs, subsurface celerity on the hillsides). This result con-
since the runoff ratio of that event is considerably firms the common belief that it is not necessary to have
larger that that of the others. Unfortunately, more de- a detailed evapotranspiration estimation for flood fore-
tailed rainfall data was not available to assess this issue. casting. However, since the initial soil moisture condi-
The July event used for calibration is simulated slightly tions were obtained with a dynamical procedure, and
differently in this simulation compared with the calibra- the set of simulations C show that flood peaks can be
tion simulation because here different initial conditions sensitive to such conditions, more investigation is
and rainfall are used. needed to assess the interplay between initialization
As Fig. 9b underlines, an increased difference in base procedures and parameter identification.
FIG. 8. 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.
At the basin scale, the model is capable of capturing Washita basin the main controlling factors on the soil
the spatial distribution of the surface soil moisture, with moisture spatial distribution are soil properties during
no additional calibration, and it can be used as a tool to dry periods and precipitation during wet periods, con-
understand what the major controls on soil moisture firming the finding by Mohr et al. (2000). A few hours
patterns are. Simulated data show that in the Little after precipitation, the topographic redistribution of
FIG. 9. 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 s␷ ⫽ 0.89 (with less bare soil evaporation and more transpiration); the lower
limit of the gray band is the simulated discharge with s␷ ⫽ 0.78.
FIG. 10. Comparison between simulations in the Little Washita watershed for the whole
1997 year with different averaged canopy fraction S␷ : (a) evapotranspiration, (b) total soil
column saturation. The black lines refer to the simulation with s␷ ⫽ 0.89, the gray one with
s␷ ⫽ 0.78.
soil moisture also exerts a slight influence. A compari- Satellite Land Surface Climatology Project (ISLSCP)
son of soil moisture obtained from simulations with Field Experiment (FIFE; Sellers et al. 1992), the Pro-
ESTAR data brings some confirmation of the model gram for Intercomparison of Land Surface Parameter-
behavior and calls for further investigations in which ization Schemes (PILPS; Henderson-Sellers et al.
remotely sensed data will be assimilated concurrently 1993; Wood et al. 1999), the Hydrological Atmospheric
(i.e., in parallel) with the model simulations. Pilot Experiment-Modelisation du Bilan Hydrique
The set of simulations C, a yearlong simulation of (HAPEX-MOBILHY; Andre et al. 1988), SGP97 (Jack-
streamflow with no additional calibration, shows an son et al. 1999), Tarrawarra (Western et al. 1998),
overestimation of some flow peak discharges, giving Mahurangi River Variability Experiment (MARVEX;
however the proper total flow volumes. This simulation Woods et al. 2001), and other experiments, however, have
also shows that the vegetation dynamics can have, dur- already provided integrated data based on combined sat-
ing the simulated period, a small but systematic effect ellite, remote sensing, ground-based soil moisture and
on the water budget partition and that a correct esti- vegetation mapping, eddy correlation measurements of
mation of evapotranspiration is essential to have a cor- turbulent fluxes, and boundary layer profiling at a hierar-
rect total discharge volume, the most important param- chy of scales. As digital terrain and remotely sensed sur-
eter being the transpiration rate. An adjustment of a face radiation data of higher resolution and better quality
controlling parameter was adequate to correct the bud- become increasingly available and computer perfor-
get partition. However, using the same parameter for mances continue to increase, physically based, distributed
the entire year (either the first or the second) resulted models such as GEOtop will become increasingly useful
in increasingly incorrect average soil moisture contents and more easily connected to climate and limited-area
with possible increasing departures from actual and meteorological models, and they may become accepted as
causing increasing errors in base flow estimations. Re- the standard tool for river basin hydrology.
sults underline the importance of incorporating an ad-
equate description of evapotranspiration processes for Acknowledgments. This research has been partially
an accurate prediction of long-term discharge and soil supported by the Italian Ministry of University and Re-
moisture patterns and suggest the value of implement- search (MIUR Prin 2003) and by the TIDE fifth frame-
ing a dynamic vegetation model. work and AQUATERRA sixth framework European
A fully distributed model predicts throughout the Projects. We thank the Civil and Environmental Engi-
catchment (and not only at the outlet of the basin) a neering Department of the University of Illinois at Ur-
larger number of observable processes than traditional bana–Champaign and the Illinois Water Sciences Cen-
lumped models. However, it also requires a corre- ter of the U.S. Geological Survey for their hospitality
sponding increase in input data and parameters. Field during the visit of the second author during the course
measurement campaigns like the First International of this research. Davide Tamanini designed and wrote
the code for the integration of the Richards’ equations. and soil moisture dynamics for hillslopes and microcatch-
The Agricultural Research Service, Grazinglands Re- ments. J. Hydrol., 198, 177–195.
Brutsaert, W., 1975: On a derivable formula for long-wave radia-
search Laboratory (ARS-GRL), El Reno, Oklahoma,
tion from clear skies. Water Resour. Res., 11, 742–744.
provided the Little Washita precipitation data, and the Chen, F., and Coauthors, 1996: Modeling of land surface evapo-
ARS-GRL, in cooperation with the U.S. Geological ration by four schemes and comparison with FIFE observa-
Survey and the Oklahoma Water Resources Board, tions. J. Geophys. Res., 101 (D3), 7251–7268.
provided the Little Washita discharge data. Ciarapica, L., and E. Todini, 2002: TOPKAPI: A model for the
representation of the rainfall-runoff process at different
scales. Hydrol. Processes, 16, 207–229.
REFERENCES Committee on Hydrologic Science, 2002: Report of a Workshop
on Predictability and Limits-to-Prediction in Hydrological
Abbott, M. B., 1992: The theory of the hydrologic model, or: The
Systems. Water Sciences and Technology Board, National
struggle for the soul of hydrology. Advances in Theoretical
Research Council, National Academy Press, 118 pp.
Hydrology: A Tribute to James Dooge, J. P. O’Kane, Ed.,
Deardorff, J. W., 1978: Efficient prediction of ground surface tem-
Elsevier, 237–254.
perature and moisture with inclusion of a layer of vegetation.
——, J. C. Bathurst, J. A. Cunge, P. E. O. Connell, and J. Ras-
J. Geophys. Res., 83 (C4), 1889–1903.
mussen, 1986: An introduction to the European Hydrological
Dickinson, R. E., A. Henderson-Sellers, P. J. Kennedy, and M.
System—Systeme Hydrologique European, SHE. 1: Hystory
Wilson, 1986: Biosphere Atmosphere Transfer Scheme
and philosophy of a physically-based distributed modelling
(BATS) for the NCAR Community Climate Model. Tech.
system. J. Hydrol., 87, 45–59.
Note NCAR/TN⫺275⫹STR, 72 pp.
Albertson, J., and M. B. Parlange, 1999: Surface length scales and
D’Odorico, P., and R. Rigon, 2003: Hillslope and channel contri-
shear stress: Implications for land–atmosphere interaction
butions to the hydrologic response. Water Resour. Res., 39,
over complex terrain. Water Resour. Res., 35, 2121–2132.
1–9.
Allen, P. B., and J. W. Naney, 1991: Hydrology of the Little Dubayah, A., J. Dozier, and F. W. Davis, 1990: Topographic dis-
Washita River Watershed, Oklahoma: Data and analyses. tribution of clear-sky radiation over the Konza Prairie, Kan-
Tech. Rep. ARS-90, U.S. Department of Agriculture, Agri- sas. Water Resour. Res., 26, 679–690.
cultural Research Service, 74 pp.
Dunne, T., 1978: Field studies of hillslope flow processes. Hill-
Andre, C., and Coauthors, 1988: Evaporation over land-surfaces, slope Hydrology, M. J. Kirkby, Ed., Wiley-Interscience, 227–
first results from HAPEX-MOBILHY special observing pe- 293.
riod. Ann. Geophys., 6, 477–492. Feddes, R. A., and Coauthors, 2001: Modeling root water uptake
Bertoldi, G., and R. Rigon, 2004: GEOTOP: A hydrological bal- in hydrological and climate models. Bull. Amer. Meteor. Soc.,
ance model: Technical description and programs guide. Tech. 82, 2797–2809.
Rep. DICA-04-001, University of Trento E-Prints, 102 pp. Franchini, M., and M. Pacciani, 1991: Comparative analysis of
[Available online at http://eprints.biblio.unitn.it/archive/ several conceptual rainfall-runoff models. J. Hydrol., 122,
00000551/.] 161–219.
——, ——, and T. M. Over, 2006a: Impact of watershed geomor- Freeze, R. A., and R. L. Harlan, 1969: Blueprint for a physically-
phic characteristics on the energy and water budgets. J. Hy- based, digitally simulated hydrologic response model. J. Hy-
drometeor., 7, 389–403. drol., 9, 237–258.
——, ——, D. Tamanini, and F. Zanotti, 2006b: Geotop version ——, and ——, 1972: Role of subsurface flow in generating sur-
0.875: Technical description and programs guide. Tech. Rep. face runoff. 2. Upstream source areas. Water Resour. Res., 8,
dica-06-001, University of Trento E-Prints, 101 pp. 1272–1283.
Best, M. J., 1998: A model to predict surface temperatures. Garratt, J. R., 1992: The Atmospheric Boundary Layer. Cam-
Bound.-Layer Meteor., 88, 279–306. bridge University Press, 316 pp.
Beven, K. J., 2000: Rainfall-Runoff Modelling: The Primer. Wiley, ——, 1993: Sensitivity of climate simulations to land surface and
372 pp. atmospheric boundary-layer treatments—A review. J. Cli-
——, and M. J. Kirkby, 1979: A physically-based variable contrib- mate, 6, 419–449.
uting area model of basin hydrology. Hydrol. Sci. Bull., 24 Garrote, L., and R. Bras, 1995: A distributed model for real-time
(1), 43–49. flood forecasting using digital elevation models. J. Hydrol.,
Bixio, A., S. Orlandini, C. Paniconi, and M. Putti, 2000: Physically- 167, 279–306.
based distributed model for coupled surface runoff and sub- Grayson, R., and G. Blöschl, 2000: Spatial Patterns in Catchment
surface flow formulation at the catchment scale. Surface Wa- Hydrology, Observations and Modelling. Cambridge Univer-
ter Systems and Hydrology and Water Research Symposium, sity Press, 404 pp.
L. Bentley et al., Ed., Vol. 2, Computational Methods in Wa- ——, I. Moore, and T. McMahon, 1992: Physically based hydro-
ter Resources, Balkema, 115–122. logic modelling. 1. A terrain-based model for investigative
Blöschl, G., R. Kirnbauer, and D. Gutknecht, 1991: Distributed purposes. Water Resour. Res., 28, 2639–2658.
snowmelt simulations in an alpine catchment. 1. Model evalu- Heimsath, M. A., W. E. Dietrich, K. Nishiizumi, and R. Finkel,
ation on the basis of snow cover patterns. Water Resour. Res., 1997: The soil production function and landscape equilib-
27, 3171–3179. rium. Nature, 388, 358–361.
Bonan, G., 1996: A land surface model for ecological, hydrologi- Henderson-Sellers, A., Z. Yang, and R. Dickinson, 1993: The
cal, and atmospheric studies: Technical description and user’s Project for Intercomparison of Land-surface Parameteriza-
guide. Tech. Note NCAR/TN-417⫹STR, 150 pp. tion Schemes. Bull. Amer. Meteor. Soc., 74, 1335–1349.
Bronstert, A., and E. Plate, 1997: Modelling of runoff generation Hilberts, A., E. van Loon, P. A. Troch, and C. Paniconi, 2004: The
hillslope-storage Boussinesq model for non-constant bedrock mixing in the ground layer of the atmosphere. Trans. Geo-
slope. J. Hydrol., 291, 160–173. phys. Inst. Akad., 151, 163–187.
Horton, R. E., 1933: The role of infiltration in the hydrological Mualem, Y., 1976: A new model for predicting the hydraulic con-
cycle. Eos, Trans. Amer. Geophys. Union, 14, 446–460. ductivity of unsaturated porous media. Water Resour. Res.,
Iqbal, M., 1983: An Introduction to Solar Radiation. Academic 12, 513–522.
Press, 390 pp. Orlandini, S., G. Moretti, M. Franchini, B. Aldighieri, and B.
Ivanov, V. Y., E. R. Vivoni, R. L. Bras, and D. Entekhabi, 2004: Testa, 2003: Path-based methods for the determination of
Preserving high-resolution surface and rainfall data in opera- nondipersive drainage directions in grid-based digital eleva-
tional-scale basin hydrology: A fully-distributed physically- tions models. Water Resour. Res., 39, 1144, doi:10.1029/
based approach. J. Hydrol., 298, 80–111. 2002WR001639.
Paniconi, C., and E. F. Wood, 1993: A detailed model for simu-
Jackson, T. J., D. M. LeVine, A. Hsu, A. Oldak, P. Starks, C.
lation of catchment scale subsurface hydrologic processes.
Swift, J. Isham, and M. Haken, 1999: Soil moisture mapping
Water Resour. Res., 29, 1601–1620.
at regional scales using microwave radiometry: The Southern
Great Plains Hydrology Experiment. IEEE Trans. Geosci. ——, and M. Putti, 1994: A comparison of Picard and Newton
Remote Sens., 37, 2136–2151. iteration in the numerical simulation of multidimensional
variably saturated flow problems. Water Resour. Res., 30,
Kot, S. C., and Y. Song, 1998: An improvement of the Louis
3357–3374.
scheme for the surface layer in an atmospheric modelling
——, M. Marrocu, M. Putti, and M. Verbunt, 2003a: Newtonian
system. Bound.-Layer Meteor., 88, 239–254.
nudging for a Richards equation-based distributed hydrologi-
Kumar, P., 2004: Layer averaged Richards’ equation with lateral cal model. Adv. Water Resour., 26, 161–178.
flow. Adv. Water Resour., 27, 521–531.
——, P. A. Troch, E. van Loon, and A. G. J. Hilberts, 2003b:
Lakshmi, V., J. Albertson, and J. Schaake, Eds., 2001: Land Sur- Hillslope-storage Boussinesq model for subsurface flow and
face Hydrology, Meteorology, and Climate: Observations and variable source areas along complex hillslopes: 2. Intercom-
Modeling. Amer. Geophys. Union, 246 pp. parison with a three-dimensional Richards equation model.
Leij, F., N. Romano, M. Palladino, M. Schaap, and A. Coppola, Water Resour. Res., 39, 1317, doi:10.1029/2002WR001728.
2004: Topographical attributes to predict soil hydraulic prop- Pomeroy, J. W., B. Toth, R. J. Granger, N. R. Hedstrom, and
erties along a hillslope transect. Water Resour. Res., 40, R. L. H. Essery, 2003: Variation in surface energetics during
W02407, doi:10.1029/2002WR001641. snowmelt in a subarctic mountain catchment. J. Hydro-
Liang, X., and Z. Xie, 2001: A new surface runoff parametrization meteor., 4, 702–719.
with subgrid-scale soil hetereogeneity for land surface mod- Prosser, I. P., and B. Abernethy, 1996: Predicting the topographic
els. Adv. Water Resour., 24, 1173–1193. limit to a gully network using a digital terrain model and
——, D. P. Lettenmaier, E. F. Wood, and S. J. Burges, 1994: A process thresholds. Water Resour. Res., 32, 2289–2298.
simple hydrologically based model of land surface water and Reed, S., V. Koren, M. Smith, Z. Zhang, F. Moreda, and D. Seo,
energy fluxes for general circulation models. J. Geophys. and DMIP participants, 2004: Overall Distributed Model In-
Res., 99D, 14 415–14 428. tercomparison Project results. J. Hydrol., 298, 27–60.
Louis, J. F., 1979: A parametric model of vertical eddy fluxes in Reggiani, P., S. Hassanizadeh, M. Sivapalan, and W. Gray, 1999:
the atmosphere. Bound.-Layer Meteor., 17, 187–202. A unifying framework for watershed thermodynamics: Con-
Mengelkamp, H., K. Warrach, and E. Raschke, 1999: SEWAB— stitutive relationships. Adv. Water Resour., 23, 15–39.
A parametrization of the surface energy and water balance Richards, L. A., 1931: Capillary conduction of liquids in porous
for atmospheric and hydrologic models. Adv. Water Resour., mediums. Physics, 1, 318–333.
23, 165–175. Rinaldo, A., A. Marani, and R. Rigon, 1991: Geomorphological
dispersion. Water Resour. Res., 27, 513–525.
Mesa, O. J., and E. R. Mifflin, 1986: On the relative role of hills-
Rodriguez-Iturbe, I., and A. Rinaldo, 1997: Fractal River Net-
lope and network geometry in hydrologic response. Scale
works: Chance and Self-Organization. Cambridge University
Problems in Hydrology, V. K. Gupta, I. Rodriguez-Iturbe
Press, 547 pp.
and E. Wood, Eds., D. Reidel, 181–190.
——, ——, R. Rigon, R. L. Bras, A. Marani, and E. Ijjasz-Vasquez,
Milly, P. D., 1974: Moisture and heat transport in hysteretic, in-
1992: Energy dissipation, runoff production, and the three-
homegeneous porous media: A matric head-based formula-
dimensional structures of river basins. Water Resour. Res., 28,
tion and a numerical model. Water Resour. Res., 10, 514–520.
1095–1103.
Mohanty, B. P., J. Famiglietti, and T. Skaggs, 2000: Evolution of Sellers, P. J., F. G. Hall, G. Asrar, D. E. Strebel, and R. E. Mur-
soil moisture spatial structure in a mixed-vegetation pixel phy, 1992: An overview of the First International Satellite
during the SGP97 Hydrology Experiment. Water Resour. Land Surface Climatology Project (ISLSCP) Field Experi-
Res., 36, 3675–3686. ment (FIFE). J. Geophys. Res., 97, 345–373.
——, P. J. Shouse, D. A. Miller, and M. T. van Genuchten, 2002: Smith, R. E., Ed., 2002: Infiltration Theory for Hydrologic Appli-
Soil property database: Southern Great Plains 1997 Hydrol- cations. Water Resources Monogr., Vol. 15, Amer. Geophys.
ogy Experiment. Water Resour. Res., 38, 1047, doi:10.1029/ Union, 212 pp.
2000WR000076. Tamanini, D., 2003: Una nuova modellazione dei flussi
Mohr, K. I., J. S. Famiglietti, A. Boone, and P. Starks, 2000: Mod- nell’insaturo nel modello idrologico distribuito GEOTOP.
eling soil moisture and surface flux variability with an un- Tesi di laurea (Masters thesis), Corso di laurea in Ingegneria
tuned land surface scheme: A case study from the Southern per l’Ambiente e il Territorio, University of Trento, 149 pp.
Great Plains 1997 hydrology experiment. J. Hydrometeor., 1, Troch, P. A., C. Paniconi, and E. van Loon, 2003: Hillslope-
154–169. storage Boussinesq model for subsurface flow and variable
Monin, A. S., and A. M. Obukhov, 1954: Basic laws of turbulent source areas along complex hillslopes: 1. Formulation and
characteristic response. Water Resour. Res., 39, 1316, and hydrological flux measurements. Water Resour. Res., 34,
doi:10.1029/2002WR001728. 2765–2768.
van Genuchten, M. T., 1980: A closed-form equation for predict- Wigmosta, M. S., L. Vail, and D. Lettenmaier, 1994: A distributed
ing the hydraulic conductivity of unsaturated soils. Soil Sci. hydrology–vegetation model for complex terrain. Water Re-
Soc. Amer. J., 44, 892–898. sour. Res., 30, 1665–1679.
Vereecken, H., J. Maes, J. Feyen, and P. Darius, 1989: Estimating Wood, E., and Coauthors, 1999: The Project for Intercomparison
of Land-Surface Parameterization Schemes (PILPS) Phase-
the soil moisture retention characteristic from texture, bulk
density and carbon content. Soil Sci., 148, 389–403. 2(c) Red–Arkansas river basin experiment: 1. Experiment
description and summary intercomparisons. Global Planet.
Vertessy, R., T. Hatton, P. O’Shaughnessy, and M. Jayasuriya,
Change, 19, 115–136.
1993: Predicting water yield from a mountain ash forest using
Woods, R. A., and Coauthors, 2001: Experimental design and ini-
a terrain analysis based catchment model. J. Hydrol., 150,
tial results from the Mahurangi River variability experiment:
665–700.
MARVEX. Observations and Modelling of Land Surface
Warrach, K., M. Stieglitz, H. Mengelkamp, and E. Raschke, 2002: Hydrological Processes, V. Lakshmi, J. Albertson, and J.
Advantages of a topographically controlled runoff simulation Schaake, Eds., Water Science and Application 3, Amer. Geo-
in a SVAT model. J. Hydrometeor., 3, 131–148. phys. Union, 201–213.
Western, A. W., R. B. Grayson, and T. Green, 1998: The Tar- Zanotti, F., S. Endrizzi, G. Bertoldi, and R. Rigon, 2004: The
rawarra data set: Soil moisture patterns, soil characteristics geotop snow module. Hydrol. Processes, 18, 3667–3679.

A Distributed Hydrological Model With Coupled Water and Energy Budgets

Uploaded by

Copyright:

Available Formats

A Distributed Hydrological Model With Coupled Water and Energy Budgets

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

A Distributed Hydrological Model With Coupled Water and Energy Budgets

Uploaded by

Copyright:

Available Formats

JUNE 2006 RIGON ET AL.

(Manuscript received 1 March 2005, in final form 11 August 2005)

© 2006 American Meteorological Society

ra tained starting with a saturated profile and running the

Parameter symbol Description Units Basin-averaged value Source

FIG. 5. Cumulative water balance of the Little Washita water-

the model do not show the influence of land use; how-

flow also appeared in summer, which could be ex-

You might also like