Calibration of Distributed Hydrology-Soils-Vegetation Model to Lake Whatcom Watershed, Washington State

1
Calibration of Distributed Hydrology-Soils-Vegetat
ion Model to Lake Whatcom Watershed, Washington
State
Katherine Callahan and Robert Mitchell, Western
Washington University, 516 High St. Bellingham,
WA 98225 Pascal Storck, 3 Tier Environmental
Forecast Group, 2825 Eastlake Ave. East, Seattle,
WA 98102
130-5 Abst. No. 66494
INTRODUCTION
METHODS
PRELIMINARY RESULTS
Lake Whatcom Watershed
Meteorologic Inputs
Comparison of Predicted and Recorded Discharge
Lake Whatcom watershed is located in the North
Cascades foothills in northwestern Washington
(Figure 1). The watershed was chosen for this
study because the lake provides the drinking
water supply for approximately 86,000 people and
water managers are concerned about sustaining the
lake as a long term drinking water source. The
watershed characteristics are described below.
Although we are in initial phase of model
calibration, we are satisfied with the
preliminary results. The model is capturing the
timing of peaks reasonably well, but it is over
estimating the volumes. The simulated flow over
estimates the gauge volume by approximately 25
for Austin Creek (Figure 8) and by about 64 for
Smith Creek (Figure 9). We believe the over
estimates are attributed to one or more of the
following
Climate data were taken from the Northshore
station (Figure 1). DHSVM requires the following
climate data

• Differences between the actual gauge location
  along the creeks and the location where the model
  is predicting discharge.
• Inadequate soil data. The soil thickness and
  permeability will influence the magnitudes of the
  peaks. We have not altered predicted soil
  thickness values in the basins or sufficiently
  quantified values for the bedrock in the
  watershed (fractured sandstone).
• Unsatisfactory precipitation lapse rate
  predictions. Point precipitation is distributed
  through the watershed via algorithms in DHSVM. We
  have not fully explored all lapse rate
  variability options.
• Inaccurate solar radiation inputs may be
  influencing transpiration and soil storage. We
  have not performed simulations using the aspect
  grid which models shortwave radiation based on
  topographic aspect and slope variability.

• Precipitation (m)
• Wind Speed (m/s)
• Relative Humidity ()
• Longwave Radiation (estimated from station data)
  (W/m2)
• Shortwave Radiation (W/m2)
• Temperature (oC)

• Watershed area 146 km2
• Lake area 21 km2
• Elevation ranges from 93 meters at the lake to
  1024 meters at the highest point
• Urban area covers 9 km2
• Forested areas cover 117 km2
• Nine stream gauges are located in the watershed
• Two climate stations exist within the watershed

Figure 4. Daily Precipitation and Temperature
Data From the Northshore Climate Station near
Smith Creek Subbasin
Input Grids
Research Objective

• DHSVM requires multiple input grids to
  characterize the watershed.
• 30 meter DEM provides a 30m pixel size for model
  calculations (Figure 1)
• Stream Network (generated by Arc/Info AML)
  (Figure 1)
• CONUS soil types (Miller and White, 1998) (Figure
  5)
• USGS National Land Cover Data Set (Figure 6)
• Soil Depth (generated by Arc/Info AML) (Figure 7)

Our goal is to quantify the surface water runoff
from the watershed into the lake under varying
climatic conditions using the Distributed
Hydrology-Soils-Vegetation Model (DHSVM). The
watershed is suitable for modeling purposes
because it contains

• Data logging stream gauges on three perennial
  streams
• Two weather stations that collect solar
  radiation, temperature, humidity, and
  precipitation data
• Gauges that log hourly lake levels and hydraulic
  inputs and outputs for the lake

Figure 8. Comparison of Austin Creek Simulated
Discharge to Recorded Flow
Figure 1. DEM of Lake Whatcom Watershed with
Stream Network, and Two Primary Subbasins
Figure 5. CONUS Soil Types
Distributed Hydrology-Soils-Vegetation (DHSVM)
Model
DHSVM is a physically based, spatially
distributed hydrology model that simulates
watershed hydrology through a multilayer grid
system at the pixel level. Each pixel in the grid
is assigned various characteristics such as soil
type, vegetation type, elevation, slope, depth to
water table (Figure 2). This model was developed
at the University of Washington specifically for
watersheds in the Puget Sound and Cascade
Mountains (Wigmosta et al., 1994).
Figure 9. Comparison of Smith Creek Simulated
Discharge to Recorded Flow
FUTURE WORK
Our preliminary calibration results are
encouraging. We are confident that we will
accurately calibrate DHSVM by refining the basin
characteristics and meteorological inputs. Once
calibrated the model will be used to explore
surface runoff scenarios in the watershed such as
the influence of logging and increased urban
development. We are also interested in
quantifying groundwater inputs into the lake
using DHSVM.
Figure 6. Land Cover Data from USGS
Figure 7. Soil Depths generated by AML Script
Figure 2. DHSVM Model Representation of the 1-D
Vertical Water Balance
Model Simulation
References Miller, D.A. and R.A. White, 1998 A
Conterminous United States Multi-Layer Soil
Characteristics Data Set for Regional Climate
and Hydrology Modeling. Earth Interactions, 2.
Available on-line at http//EarthInteractions.org
Storck, P., L. Bowling, P. Wetherbee, and D.
Lettenmair, 1998. Application of a GIS-based
distributive hydrology model for prediction of
forest harvest effects on peak stream flow in the
Pacific Northwest. Hydrological Processes vol.
12, pp 889- 904. Vogelmann, J.E., S.M. Howard, L.
Yang, C.R. Larson, B.K. Wylie, N. Van Driel,
2001. Completion of the 1990s National Land
Cover Data Set for the Conterminous United
States from Landsat Thematic Mapper Data and
Ancillary Data Sources, Photogrammetric
Engineering and Remote Sensing, 67650-652.
Wigmosta, M., L. Vail, and D. Lettenmair, 1994.
A distributed hydrology-vegetation model for
complex terrain. Water Resources Research, vol.
30 no. 6, pp 1665-1679. Acknowledgements I
would like to thank my thesis committee members
Dr. Robert Mitchell, Dr. Doug Clark, Dr. David
Wallin, and Mr. Steve Walker. In addition, I
would to thank WWU and the Institute for
Watershed Studies for their assistance in funding
this research and Jay Chennault for his continued
assistance with modeling and GIS.
Calibration means modifying the basin attributes
and/or meteorological data to satisfactorily
match the discharge predicted by the model to the
actual discharge recorded at the gauge. The
model is being calibrated to a time series of
river-discharge data collected from the Smith
Creek and Austin Creek (Figure 1).
DHSVM performs an energy and water mass balance
on each pixel. Then all the pixels are linked
through a subsurface transport method Darcys
Law determines downward movement and flow is
exchanged between pixels based on topography
(Figure 3) (Storck, et al., 1994).

• Climate data covers a period from January 1, 2001
  and ends July 31, 2003
• Initial model state of the watershed was
  determined using the entire climate input-time
  series
• Initial calibration simulation covered the time
  frame from October 1, 2001 to March 31, 2002
• Stream segments that terminate near the stream
  gauges were selected for output during model
  simulations
• It takes approximately 24 hours for DHSVM to
  perform one calibration simulation on a SUN E450

Figure 3. DHSVM Surface and Subsurface Flow
Routing and Runoff Generation