MACROSCALE LAND SURFACE HYDROLOGY MODELING AT THE UNIVERSITY OF WASHINGTON
Dr. Dennis P. Lettenmaier
Professor of Civil Engineering, University of Washington, Seattle, WA
Dr. Fayez A. Abdulla
now Assistant Professor of Civil Engineering, Jordan University of Science and Technology, Irbid,
Jordan - formerly at Department of Civil Engineering, University of Washington
Dr. Xu Liang
now Postdoctoral Research Associate, Civil Engineering & Operations Research,
Princeton University, Princeton, NJ - formerly at Department of Civil Engineering,
University of Washington
Graduate student of Civil Engineering, University of Washington,Seattle, WA
Graduate student, GKSS Forschungszentrum, Institut fur Atmosphaerenphysik, Geesthacht, Germany -
currently at Department of Civil Engineering, University of Washington
Dr. Eric F. Wood
Professor of Civil Engineering & Operations Research, Princeton University,
National Science Foundation
NOAA Office of Global Programs
Western Center for Global Environmental Change
The land surface plays an important role in partitioning incoming radiation
at the land surface into latent and sensible heat, and in partitioning
precipitation into runoff and evaporation. The representation of the
land surface in numerical weather prediction and climate models must,
of necessity, be highly parameterized due to inconsistencies between the
spatial scales at which these models operate (several tens to hundreds
of km) and the length scale over which important variations in soil moisture,
or instance, occur (meters). Considerable progress has been made in the
last few years in understanding the spatial variability of water and energy
fluxes over natural catchments. Field studies have addressed the spatial
patterns of soil moisture and associated runoff production at the catchment
(tens or so of square km) scale. Remote sensing of soil moisture has
contributed to understanding of the spatial distribution of soil moisture
at larger scales. The result is a better appreciation for the role of
topography and heterogeneities in soil properties on
the distribution of soil moisture, runoff production, and evaporation. These
results have been confirmed through large scale land-atmosphere field
experiments like FIFE, the First ISLSCP Field Experiment, conducted in central
Kansas in 1987 and 1989.
The current state of understanding of land surface hydrology has not,
however, found its way into the land surface parameterizations used in
global general circulation models (GCMs) used for numerical weather prediction
and climate studies. It is our contention that the current generation
of water-energy-biospheric models that have been advanced for this purpose
are inappropriate for determining large scale energy and water fluxes
due to their inconsistency between vertical and horizontal resolution.
This lack of horizontal variability (which has lead to their characterization
as 'big-leaf' models) can result in significant biases in soil moisture
and evaporation when estimated over large areas.
To address this problem, we have developed a two-layer generalization
of the VIC (Variable Infiltration Capacity) model (Wood et al., 1992),
which in turn is a generalization of the Xanianjing model originally used
for flood forecasting and other hydrologic purposes in China (Zhao, 1977)
and elsewhere. The revised version of the model, termed VIC-2L, is described
in detail by Liang et al. (1994). The distinguishing feature of VIC-2L
is that it represents the spatial variability of infiltration, and hence
runoff production, as a spatial probability distribution
. The model, in its one-layer form, has been used to model the
land surface in GCMs by Stamm et al (1994) and Dumenil and Todini (1992).
The two-layer version of the model incorporates a nonlinear drainage (baseflow)
representation, and an explicit
representation of vegetation. Multiple land cover types,
including bare soil, are modeled using a "tile" approach (aggregates all occurrences
of a given land cover type within the area to a single equivalent subarea) .
Validation of VIC-2L for FIFE (Central U.S. grassland) and
ABRACOS (Brazilian cleared forest) is reported by Liang (1994).
Generally, the model reproduces observed latent and
sensible heat fluxes quite well. Errors in ground heat flux predictions
sometimes occur, some of which can be attributed to measurement errors.
Nonetheless, recent improvements in the model's ground heat flux algorithm
have improved its performance in this respect.
VIC-2L in PILPS
VIC-2L has been a participant in the GEWEX project PILPS, Project for
Intercomparison of Land Surface Parameterizations (Henderson-Sellers et
al., 1995). The goal of PILPS is to assess the performance of current
land surface schemes used in the current generation of climate and numerical
weather prediction models, with the ultimate aim of better predicting
soil moisture, and hence the partitioning of energy at the land surface.
In its first phase, some 30 land surface schemes were tested using prescribed
surface forcings for two hyopthetical sites: a tropical forest and a
northern mid-latitude grassland. A series of consistency tests were imposed
on each of the models, including long-term balancing of energy and water
at the land surface. Although there was no "right" answer, useful information
about the range of model performance in terms annual, seasonal, and diurnal
partitioning of net radiation into sensible, latent, and ground heat fluxes,
and partitioning of precipitation into evaporation and runoff, was obtained.
The long-term partitioning of net radiation
into latent and sensible heat by VIC-2L for both the tropical forest and
grassland sites was near the middle of the range for the models tested.
In Phase 2, the models were tested using data from an experimental agricultural
field site at Cabauw, the Netherlands. Results of a similar set of experiments
show that VIC-2L performed quite well in this environment as well, both for
the energy balance, and the water balance.
In November, 1994, a subset of fifteen PILPS participants attended a soil
moisture simulation workshop held at Macquarie University. At this workshop,
each of the models was applied to data collected during the HAPEX-MOBILHY
experiment in France. Although the annual water balance predicted
by VIC-2L matched the observations reasonably well, the upper soil layer
tended to dry more rapidly than the observations in the spring and summer.
This problem was attributed to a) the absence of a mechanism to move
soil water from the lower to the upper layer, and b) the inability of
the model to evaporate water quickly following small summer rainfall events,
due to the damping effect resulting from depth-averaging of soil moisture
in a relatively thick surface layer. As a result, modifications to the
model, described by Liang et al. (1995) were made which
greatly improved model performance for the HAPEX-MOBILHY site.
GRID-based Version of VIC-2L for Large Rivers
Wetzel (1994) and Abdulla (1995) have implemented VIC-2L to simulate the
water and energy balances of large river basins. Essentially, the river
basin is overlain by a rectangular grid, typically of size one or one-half
degree. These grid cells are then linked to form an equivalent channel
network. The VIC-2L model is then run for each grid cell, using one of
two formulations: a) water balance only, in which case the driving variables
are precipitation and surface air temperature; b) full energy balance,
in which case the driving variables are precipitation, downward short-
and longwave radiation at the surface, and surface meteorology, including
wind, pressure, and vapor pressure deficit. Flow generated within a grid
cell is routed to the cell outlet using a unit hydrograph routing scheme.
A separate unit hydrograph is used to route flow entering and crossing
a grid cell.
The water balance scheme has been successfully applied to the Arkansas-Red,
Colorado, Missouri, Delaware, and Columbia River systems.
VIC-2L Regional Parameter Estimation
Abdulla et al. (1995) and Abdulla and Lettenmaier (1995a; b) have developed
and tested a regional parameter estimation scheme for the VIC-2L model.
Using a set of 40 catchments in the Arkansas-Red River
basin with drainage
areas in the range 100 to 10,000 km 2 for which historical streamflow
and meteorological records exist, the parameters of the VIC-2L model were
estimated using a search procedure. Regional regression equations were
then developed to relate the estimated VIC-2L parameters for the gaged
catchments to catchment characteristics derivable from digital elevation,
soils, climatalogical, and land cover data. The prediction equations
were tested using observed streamflow data for test catchments not in
the training data set, and to predict streamflow hydrographs for the entire
Arkansas-Red River system. Model performance for the humid and sub-humid
test catchments was comparable to that for the training catchments
. However, the model did not perform as well for the arid and semi-arid
test catchments . This is attributable in part to structural
problems with application of the VIC-2L model in such climates (notably
the absence of an infiltration excess runoff production mechanism). At
the scale of the entire Arkansas-Red basin, the model performed quite
well with regional parameters, especially for the more humid, eastern
part of the catchment. Finally, spatially averaged evaporative
fluxes for the entire Arkansas-Red River basin predicted by the VIC-2L
model were compared with evapotranspiration estimated over the same area
using an atmospheric water budget. The
results were shown to agree quite closely, especially during the spring
and summer months, when evaporation is highest.
Abdulla, F., D.P. Lettenmaier, E.F. Wood, and J.A. Smith, "Application
of a macroscale hydrologic model to estimate the water balance of the
Arkansas-Red River basin", in press, Journal of Geophysical Research,
Abdulla, F.A., and D.P. Lettenmaier, "Development of regional parameter
estimation equations for a land surface hydrologic model", in review,
Journal of Hydrology, 1995.
Abdulla, F.A., and D.P. Lettenmaier, "Application of a regionalized land
surface hydrologic model to a continental river", in review, Journal of
Abdulla, F.A., "Regionalization of a macroscale hydrological model", Water
Resources Series Technical Report No. 144, Department of Civil Engineering,
University of Washington, July, 1995.
Dumenil, L., and E. Todini, "A rainfall-runoff scheme for use in the Hamburg
climate model", in Advances in theoretical hydrology, A tribute to James
Dooge. P. O'Kane, ed., European Geophysical Society Series on Hydrological
Sciences 1, Elsevier, Amsterdam, 1992.
Henderson-Sellers, A., A.J. Pitman, P. Irannejad, and T.H. Chen, "The
Project for Intercomparison of Land Surface Parameterization Schemes (PILPS):
Phases 2 and 3", Bulletin of the American Meteorological Society 76(4),
489-503, Apr., 1995.
Liang, X., "A two-layer variable infiltration capacity land surface representation
for general circulation models", Water Resources Series Technical Report
No. 140, Department of Civil Engineering, University of Washington, 1994.
Liang, X., D.P. Lettenmaier, E.F. Wood, and S.J. Burges, "A simple hydrologically
based model of land surface water and energy fluxes for GCMs", Journal
of Geophysical Research 99(D7), 14,415-14,428, 1994.
Wetzel, S., "A hydrological model for predicting the effects of climate
change", B.S. Thesis, Department of Civil Engineering and Operations Research,
Princeton University, Apr., 1994.
Wood, E.F., D.P. Lettenmaier, and V.G. Zartarian, "A land surface hydrology
parameterization with subgrid variability for general circulation models",
Journal of Geophysical Research, Vol. 97, No. D3, Feb., 1992, pp. 2717-2728.
Zhao, R.J., "Flood forecasting method for humid regions of China", East
China College of Hydraulic Engineering, 1977.
Links to Related Sites
PILPS WWW home page
Dennis P. Lettenmaier
Department of Civil Engineering
University of Washington
Seattle, Washington 98195
Eric F. Wood
Department of Civil Engineering and Operations Research
Princeton, New Jersey 08544
Return to Current Research Information
Hydrology Homepage /
University of Washington /