The purpose of the DSM2 model boundary extension is to create
a direct dynamic link between the Delta and the Stateís second longest river,
the San Joaquin. Many Delta water supply, water quality, and fishery issues are
closely linked to conditions in the San Joaquin River (SJR). Extension of the
SJR boundary will provide a tool to investigate how the Delta may respond to
different SJR management strategies.
The system domain for this project is the portion of the SJR
from near Vernalis to the Mendota Pool (see Figure 3-1). The project was divided
into two phases because of substantial gaps in bathymetry data. Phase I is that
portion of the domain from the Bear Creek confluence near Stevinson down to the
current boundary near Vernalis. Phase II is that portion of the domain from
Stevinson to Mendota Pool. In general, the SJR boundary extension work reported
herein is limited to Phase I.
3.2 Model Development
A set of USGS 7.5-minute topographic maps encompassing the
project area was used to discretize the domain into 92 reaches with 93 nodes
(Phase I & II). The locations of the nodes generally correspond to a
hierarchy of major tributaries, possible point sources of inflow and outflow, or
convenient landmarks. The geographic coordinate of each node was manually
measured from the maps using the Universal Transverse Mercator, Zone 10 (UTM)
reference system. The length of each reach was manually measured from the maps
using a digital planimeter. Three values per reach were measured then averaged.
The reaches are approximately 1-2 miles long.
Bathymetry data for the system domain were obtained from the
U.S. Army Corps of Engineers (USACE). The data were transformed from the
latitude/longitude coordinate system to the UTM coordinate system using "Corpscon,"
public domain software developed by USACE. The transformed bathymetry data and
nodal coordinates were then input into the Departmentís Cross Section
Development Program (CSDP).
CSDP was used to define the system geometry, such as channel
alignment and cross sections, for input to DSM2. The modelís river reaches
were defined by aligning centerlines to follow the thalweg (low flow channel)
that was visually located from the bathymetry data graphically displayed by CSDP.
A new function was added to CSDP that calculates the reach length from the
aligned centerlines. However, special care is necessary for this function to
give sufficient results. The thalweg can be difficult to visually extract from
the data and is highly sinuous. The placement of many short centerline segments
may be necessary to accurately define a meandering channel alignment. Many short
segments were used to describe the channels in CSDP. As a benchmark, the reach
lengths computed by CSDP were compared to the manual planimeter measurements.
The net difference between the two methods was small (approximately two feet),
with CSDP yielding the greater length.
Figure 3-1: San Joaquin River.
Irregular cross sections were developed using CSDP to
approximate the riverís existing natural shape. Every channel has at least one
representative irregular cross section and some have as many as three. Personal
engineering judgement and ability to distinguish a realistic cross section from
the data displayed at chosen locations determined the initial location of the
cross sections within a channel. In most cases the thalweg of the cross section
was well defined but the floodplain was not. Digital aerial photos were used to
reasonably approximate the shape and extent of the flood plains.
Even with the use of irregular cross sections, DSM2 still
requires the definition of two rectangular cross sections per channel segment.
These rectangular cross sections are only used if there is not at least one
irregular cross section in a given channel segment. Therefore, a homogeneous
rectangular cross section width of 500 feet was specified at the upstream and
downstream sides of each node with a linear bottom slope between nodes. The
slope was calculated using the change in channel elevation from the upstream
boundary near Stevinson to Vernalis, approximately 60 feet (msl) to 0 feet (msl),
respectively, divided by the number of reaches between those locations. A stage
of 12 feet above the bottom elevation was specified for the initial condition.
Pre-Calibration Model Runs
A mock planning study was developed for the first trial run of
the model. The purpose of this exercise was to test the planning mode input
files and new geometry for design flaws. A few select periods with hydrologic
conditions representative of dry, normal and wet scenarios were chosen. The
hydrology for the Delta and major SJR tributaries was obtained from the DWR
Planning Simulation Model (DWRSIM). Agricultural consumptive use was not readily
available for the SJR and was neglected for these preliminary simulations.
The major problem encountered in the first trial run was
channels drying up for the dry hydrologic scenario. DSM2 will not allow a
discontinuity in the flow regime and model calculations will not proceed if a
channel dries up. This error can typically be attributed to large changes in
cross sectional area or dramatic changes in bottom elevation between irregular
A systematic approach was developed to debug the geometry
design. The model was run until a channel segment dried up then the irregular
cross section(s) with that channel segment was (were) removed and the model ran
again. If no irregular cross sections are defined for a given channel segment,
then the model will default to the rectangular cross section defined for that
channel segment. This process was repeated until the model ran to a successful
completion. Approximately 40 percent of the irregular cross sections were
removed, most of them consecutive and localized to four general areas. This
consecutive and highly localized trend suggested that not all of the cross
sections removed were problematic.
The elevation of a default rectangular cross section in one
channel segment may not closely match an irregular cross section in a
neighboring channel. This requires the introduction of a continuous block of
rectangular cross sections where the elevations of the upstream and downstream
ends of this section approximate the elevations of the neighboring irregular
cross sections. Also, a problematic cross section may not cause an error in its
own channel segment but may cause an error in other channel segments in close
proximity. In some cases where a channel segment had multiple cross sections,
only one cross section was the source of error.
Based on these conclusions, a refinement process was conducted
to differentiate potentially good cross sections from the problematic ones. Each
problem area was investigated independently of the others. Irregular cross
sections were reintroduced and removed in systematic combinations until only a
minimal number of irregulars were necessary to be removed. This process reduced
the number of likely problematic cross sections to approximately 35 percent.
The next step was to determine which cross section was the
likely source of the problem for the group. Two visualization methods were
applied. The first step was to plot a family of stage to cross sectional area
relationship curves for a problematic cross section and a few upstream and
downstream of that cross section (see Figure 3-2). The other was to sequentially
plot the bottom elevations of a problematic cross section and a few neighboring
cross sections (see Figure 3-3). These tools were valuable assets to determine
which geometric attribute was most likely to be causing the problem. In all
cases, the bottom elevation transition was found to be the problem. The model
generally experienced channel drying with changes in elevation greater than 5
feet between cross sections, sometimes more or less depending on the horizontal
distance between them.
Figure 3-2: Stage-Area Relationship for Channels 619 - 624.
Figure 3-3: Bottom Elevation Transition for the Irregular Cross Sections from
Channels 619 - 624.
Some of the deep poles and shallow riffles needed to be
averaged. The bottom elevations of the corresponding rectangular cross sections
were superimposed on a "bottom elevation transition" plot such as
shown in Figure 3-3. This provided a reference baseline to a working slope since
the model ran successfully when those cross sections were used as substitutes.
The bathymetry data was revisited to determine a better location in the channel
to draw a representative cross section with a bottom elevation closer to this
baseline. In a few cases where a channel had more than one irregular cross
section, a surplus section was deleted when relocation failed. After some
iteration, the model ran to a successful completion without substitution of
rectangular cross sections.
3.4 Future Directions
In coordination with the SJRMPís Water Quality Subcommittee,
- Collect historical hydrology data and calibrate DSM2-HYDRO for the
- Collect historical water quality data and calibrate DSM2-QUAL for the
boundary extension; and
- Complete Phase II.
Author: Thomas Pate
Back to Delta Modeling Section 2000 Annual Report Table of Contents
Last revised: 2000-10-23
Comments or Questions
Webmaster email to