Marine

Land

DHSVM Model Calibration Datasets

Land Surface Hydrologic and Biogeochemical Modeling

Hydrologic conditions in the approximately 1200 square mile area that form the contributing watershed to Hood Canal will be simulated using a physically based, high resolution, rainfall – runoff model. The Distributed Hydrology Soil – Vegetation Model (DHSVM) was developed at the University of Washington and Princeton University. DHSVM uses GIS derived representations of elevation, soil type, soil thickness and vegetation. These representations are used in conjunction with meteorological data to simulate water and energy fluxes at and below the land surface at resolutions ranging from 30 to 200 meters. The mathematical equations describing the physical processes are described by Wigmosta et al. (1994, 2002). The model has been extensively tested for the complex terrain, vegetation types, and climate patterns found in the Pacific Northwest. DHSVM is very effective at simulating the small scale hydrologic processes necessary to produce accurate streamflow, snow accumulation, and soil moisture patterns (Bowling 2001, Burges 1998, Kenward 2000, Leung 1996, Nijssen 1996, Perkins 1996, Storck 2000, VanShaar 2002, Wigmosta 1994 and 1999).

Currently efforts are underway to expand the existing capabilities of the DHSVM modeling system to include realistic representation of groundwater movement, stream temperatures and the biogeochemical processes that contribute to the nutrient loading and eutrophication problems within Hood Canal.

Figure 1.1

1.1 Model Inputs

Application of DHSVM to the Hood Canal drainage basin involves the development of a suite of input maps describing the basin and the development of a meteorological record to drive the rainfall-runoff process. Figure 1.1 is a graphic representation of the data sets incorporated where each layer is modeled as a 150m _ 150m grid. The data sets are georeferenced to the NAD27 UTM Zone 10 coordinate system.

1.1.1 Elevations

The Digital Elevation Model (DEM) used as the basis Hood Canal Basin model was created using the 1° _ 2° 10 meter resolution DEM of western Washington created and archived by the University of Washington PRISM project. The original source data of the PRISM product are the USGS 7.5’ quads for Washington State (WAGDA, 2004). The 10 meter resolution is aggregated to a 150 meter resolution for use in DHSVM using the ArcGIS Spatial Analysis Tool package (ESRI, 2004).

Figure 1.2

1.1.2 Basin Mask and Stream Network

The watershed boundaries and stream networks are derived from the basin’s 150 meter DEM. Stream channels are assumed to form when the upstream contributing area exceeds 0.25km2. The location of the watershed boundaries and the stream channels (Figure 1.2) are verified by comparing them against the stream network and watershed GIS layers. The comparison data sources are from the King County Land and Water Resources

Figure 1.3

1.1.3 Soil Types

Soil is classified into 18 types, not all of which are present in the Hood Canal basin. The classification is based on texture as measured by the overall fraction of fine material ranging from silt to talus. This includes additional categories for organic soils, water, and bedrock outcrops.  The soil layer was initially defined using the Washington State DNR soil survey data and classified into texture categories based on the percentage of fine particulates.  However, many areas within the Hood Canal basin have no soil data; this is particularly true for the large expanses of federal land within the national park system.  To estimate the soil types in these areas we have constructed a multi-parameter regression model. The inputs variables include the surficial geology (see below), pixel slope and upstream contributing area (combined into a single soil depth index), and a course scale soil definition derived from the NRCS STATSGO soil database.  By training the regression model with areas for which the soil data does exist, we are able to build a relationship that takes advantage of the more spatial complete data sets (the geology and the STATSGO) to estimate what the missing soil classes are likely to be.  The regression model was trained and applied over all of Washington State.  Comparisons of the soil classification model output with actual DNR data shows that the soil classification model is able to predict the correct soil classification for 57% of the area for which soil data exists.  In the final soil map product, the existing DNR soil data is used for areas that have data and the output of the soil classification model is used to fill in the gaps.  

The physical parameters used to describe each soil type include: soil texture class, lateral conductivity, conductivity exponential decrease with depth, maximum infiltration, surface albedo, number of soil layers, porosity, pore size distribution, bubbling pressure, field capacity, wilting point, bulk density, vertical conductivity, thermal conductivity, and thermal capacity.

Figure 1.4

1.1.4 Vegetation

The land cover maps for the Hood Canal DHSVM implementation are based on the land use observation efforts undertakes as a art of the Hood Canal Dissolved Oxygen Program. More information is available at: http://www.hoodcanal.washington.edu/observations/landuse.jsp

Hydrologic and vegetative parameterizations for the categories are ongoing.  Each vegetation classification is described by the following parameters: impervious fraction, over story present, under story present, fractional coverage, trunk space, aerodynamic attenuation, radiation attenuation, maximum snow interception capacity, snow interception efficiency, mass release snow drip ratio, height, monthly leaf area index, maximum wind resistance, minimum wind resistance, moisture threshold, vapor pressure, albedo, number of root zones, root zone depths, over story root fraction, and under story root fraction. Different landcover scenarios will be investigated for a potential mitigating role in eutrophication and dissolved oxygen problems.

1.1.5 Geology

To more realistically simulates the hydrologic conditions in the Hood Canal basin, particularly the low-relief areas on the Kitsap peninsula, it may be necessary to extend the features of the DHSVM hydrology model to include a groundwater component.  The groundwater component to be incorporated is envisioned as a 3-dimensional, hydraulic gradient driven flow network.  The limiting factors affecting rates of flow and storage capacity will be parameterized by the prevailing geologic features.    The initial stage of this effort, it the development of the geology type classification map and parameterization to serve as model input.

Figure 1.5

The experimental groundwater layer required the development of a spatially distributed description of the basic geologic formations that form the base layer of the hydrologic model. Each type requires parameters definitions for both vertical and lateral conductivity, effective porosity, and aquifer thickness.   The basic data source has been acquired from the Washington State Department of Natural Resources Division of Geology and Earth Resources,   “1:100,000 Scale Digital Geology of Washington State.”  The DNR data is available as 51 individual ARC/INFO coverages.  The processing of these data sets into the form necessary for use the in the terrestrial modeling process has been completed.  This required that the data sets be merged and re-projected.  A complete state wide map was then reclassified from the 3416 distinct geologic units into 18 generalized categories with similar hydrologic properties.