Virtual Mission Stage I: Implications of a Spaceborne Surface Water Mission

1
Virtual Mission Stage I Implications of a
Spaceborne Surface Water Mission Elizabeth A.
Clark1, Doug E. Alsdorf2, Paul Bates3, Matthew D.
Wilson4, Gopi Goteti5, James S. Famiglietti5,
Delwyn K. Moller6, Ernesto Rodriguez6, and Dennis
P. Lettenmaier1 1. Department of Civil and
Environmental Engineering, Box 352700, University
of Washington, Seattle, WA 98195 2. Ohio State
University Department of Geological Sciences, 275
Mendenhall Laboratory, 125 South Oval Mall,
Columbus, OH 43210 3. University of Bristol
School of Geographical Sciences, University Road,
Bristol, BS8 1SS United Kingdom 4. Department of
Geography, University of Exeter, 3.046 UEC
Tremough, Exeter, EX4 4RJ United Kingdom 5. Earth
System Science, University of California, Irvine,
3317 Croul Hall, Irvine, CA 92697 6. Jet
Propulsion Laboratory/California Institute of
Technology, MS 300-319 4800 Oak Grove Dr.,
Pasadena, CA 91109 AGU Fall Meeting, 2004San
Francisco, California
Lena River Basin
Amazon River Basin
Ohio River Basin
ABSTRACT The interannual and interseasonal
variability of the land surface water cycle
depend on the distribution of surface water in
lakes, wetlands, reservoirs, and river systems
however, measurements of hydrologic variables are
sparsely distributed, even in industrialized
nations. Moreover, the spatial extent and storage
variations of lakes, reservoirs, and wetlands are
poorly known. We are developing a Virtual Mission
to demonstrate the feasibility of observing
surface water extent and variations from a
spaceborne platform. In the first stage of the
Virtual Mission, on which we report here, surface
water area and fluxes are emulated using
simulation modeling over three continental scale
river basins, including the Ohio River, the
Amazon River and the Lena River. The Variable
Infiltration Capacity (VIC) macroscale hydrologic
model is used to simulate evapotranspiration,
soil moisture, snow accumulation and ablation,
and runoff and streamflow over each basin at
1/8, 1/2, or 100-km resolution. The runoff from
this model is routed using a linear transfer
model to provide input to a much more detailed
flow hydraulics model. The flow hydraulics model
then routes runoff through various channel and
floodplain morphologies at a 250 m spatial and 20
second temporal resolution over a 100 km by 500
km domain. This information is used to evaluate
trade-offs between spatial and temporal
resolutions of a hypothetical high resolution
spaceborne altimeter by synthetically sampling
the resultant model-predicted water surface
elevations.
Figure 7. Study domain (outlined in blue) along
Lena River. Red arrows indicate flow routing at
100 km (Su et al., 2004). Green dots show
locations of R-Arctic Net gauging locations.
Figure 6. Study domain (outlined in blue) along
Amazon River. Red arrows indicate flow routing at
1/2. Black shows stream network derived from
SRTM 30 arc-second data set. Green dots show
locations of RivDis and GRDC gauging stations.
Figure 5. Study domain (outlined in blue) along
Ohio River mainstem Red arrows indicate flow
routing at 1/8 (Maurer et al., 2002 Wood et
al., 2002). Blue shows stream network derived
from National Hydrography Data set. Green dots
show locations of USGS gauging stations in upper
Ohio River Basin.

• Due to the possible implications of freshwater
  inflows to the Arctic Ocean for global climate
  and the unique arctic surface processes, it is
  important to increase our observation of Arctic
  hydrology. The VIC model contains many features
  specific to cold-land processes
• Two-layer energy balance snow model (Storck et
  al. 1999)
• Frozen soil/permafrost algorithm (Cherkauer et
  al. 1999, 2003)
• Lakes and wetlands model (Bowling et al. 2002)
• Blowing snow algorithm (Bowling et al. 2004)

Unlike the Ohio River, riverine floodplains and
wetlands are common in the Amazon River Basin.
Recent studies (Richey et al., 2002) suggest that
CO2 outgassing from large tropical rivers, like
the Amazon, might play a significant role in the
global carbon cycle. Since the amount of
outgassing depends on the area of inundation,
spaceborne observations have implications for
global carbon cycling.
As is the case with most American rivers in
temperate, humid climate regions, the Ohio River
is a regulated river with a small floodplain.
Twenty locks and dams have been constructed along
the Ohio River main stem for the purposes of
flood control and navigation (USACE, 2004).
However, for the purposes of the VM stage I, we
do not model these controls.
The primary goal of the Virtual Mission (VM) is
to examine the feasibility of collecting
spaceborne measurements of surface water extent
and elevation and to evaluate potential uses for
this information, including the derivation of
discharge values and the study of surface storage
dynamics. Three test basins were selected for the
initial phase of the Virtual Mission Ohio River
Basin, Amazon River Basin, and Lena River Basin.
Each basin
Simulated Surface Area and Depth Results
represents a different climate region and
hydrologic environment.
Figure 1. Virtual Mission Stage I experimental
design. The hydrologic model component is
described here. Bates and Wilson (this meeting
section H22C-06) discuss the hydraulics model
component. Rodriguez and Moller (this meeting
section H22C-08) describe the NASA/JPL Instrument
Simulator. Goteti et al. (this meeting section
H22C-07) discuss preliminary results regarding
spatial and temporal resolution tradeoffs and
calculation of discharge.
Water depth and extent on Jan. 21, 1995. (High
flows).
Water depth and extent on April 4, 1995. (Low
flows).
Water depth and extent on Jan. 21, 1995. This
view corresponds to red box at left.
Water depth and extent on April 4, 1995. This
view corresponds to red box at left.
These figures show example images produced from
LISFLOOD-FP using dynamic inflows from VIC. These
images are in accord with the 1 arc-second
resolution elevation data from SRTM (shown in
grayscale).
Hydrologic Simulation
Initial river discharge inputs to the VM are
simulated using the Variable Infiltration
Capacity (VIC) macroscale hydrologic model (Liang
et al., 1994). This model accounts for sub-grid
scale variability in soil, vegetation,
precipitation, and topography and treats sub-grid
hydrologic variability statistically over coarse
resolution grid cells (Figure 1). The vertical
dimension of each cell is divided into three soil
layers, in which the soil moisture storage
capacity is treated as a spatial probability
distribution. Baseflow from deepest soil layer is
produced according to a nonlinear baseflow
formulation. Evapotranspiration, surface runoff
and baseflow are calculated for each vegetation
class, and the weighted sum of each, based on
vegetation fraction, is assigned to the grid
cell. Once generated, surface runoff and baseflow
are routed to simulate streamflow using the model
developed by Lohmann et al. (1998 Figure 2).
Complete
In Progress
Simulated Streamflow as Input to LISFLOOD-FP
VIC and LISFLOOD-FP work at different spatial and
temporal scales. In this study, VIC is run at a
resolution of 1/8 (55 km) to 1000 km over
3-hourly to daily time steps, while LISFLOOD-FP
operates at a resolution of 270 m over 20-second
time steps. In addition, the resolution of the
satellite instrument is likely to vary between 2
m x 60 m and 2 m x 10 m over several day-long
repeat cycles (Rodriguez, 2004). Because the
temporal resolution of most hydrologic data sets
is much longer than 20-seconds, LISFLOOD-FP is
designed to linearly interpolate discharge data
from longer to shorter time steps. Output from
the VIC model routing is divided between the
stream segments within a given grid cell on the
basis of stream length. We are experimenting with
using drainage areas derived from finer
resolution DEMs to determine the proportions of
flow to each stream.
Gauging stations vary globally in density,
temporal continuity, and quality. In order to
dynamically simulate the surface extent of a
river as might be viewed from space, the VM
employs a 2-D hydraulics model, LISFLOOD-FP
(Bates and De Roo, 2000). This model requires
hydrologic inputs at several locations along the
channel reach. Even in the Ohio River Basin,
where stream gauges are relatively common, many
of the smaller tributaries are either ungauged or
gauged too far upstream from the main channel.
However, because VIC simulates runoff across a
mesh grid, it is able to produce inflow values
for all tributaries and along the main reaches.
The record produced by VIC is also free of data
gaps in time.
Figure 4. Colored boxes show individual VIC grid
cells. Runoff from all of these cells is routed
to the orange cell. This cell contains three
outflow points (green), so the flow assigned to
each point is the flow entering the green cell
weighted by the length of river upstream of each
point.
Figure 1. Schematic representation of VIC
hydrologic model.
Figure 3. Black crosses show location of USGS
gauging stations in the Ohio River study domain.
Red dots show the desired locations of flow
inputs.
Some Additional Features for Phase II
Site Selection Criteria

• Domain size ideally 500 km by 100 km
• Channel width 50 to 200 m
• Relatively straight mainstream reaches
• Minimal regulation to allow for natural flooding
  in tributaries
• Floodplain connected to channel for most of
  domain
• Preferably located near headwaters such that
  tributaries lie at a variety of angles to account
  for non-perpendicular orbital crossings
• Upstream and downstream ends of reach coincide
  with gauging stations
• Not parallel with orbital crossings (anticipated
  angle of inclination near 65 from equator)

• 3 NEW study domains representing the following
  regions boreal, alpine and desert.
• Application of a simple reservoir model to
  evaluate the implications for water management,
  especially across international boundaries.
• Implementation of VIC Lakes and Wetlands
  algorithm to estimate storage changes, which can
  be sampled in a similar fashion as simulated
  stream flow.
• Calculations of typical biogeochemical fluxes
  across simulated surface water areas to estimate
  potential impacts that these satellite
  measurements could have on our knowledge of the
  global carbon cycle.
• Simulations of swaths that would be generated
  from sub-orbital platforms, geostationary
  satellites, and multiple satellites.
• Development of a framework for the assimilation
  of derived stream flow into hydrologic models.

CONCLUDING REMARKS

• The Virtual Mission is designed to demonstrate
  the feasibility of a spaceborne surface water
  observation system. It employs a hydrologic model
  and a hydraulics model to discharge and surface
  extent that can then be perturbed to represent
  variable sampling conditions.
• By testing over a wide range of hydrologic and
  climate regions, we hope to demonstrate the broad
  scope of issues that could be addressed by
  providing global satellite observations of
  surface water extent and elevation.
• Further investigations are underway to create a
  recommendation for future surface hydrology
  satellite missions that would account for
  tradeoffs in spatial and temporal data collection
  resolution.

Figure 2. Schematic representation of routing
model.
Authors would like to thank Ed Maurer at UC Santa
Clara, Andrew Wood, Fengge Su, and Jennifer Adam
at University of Washington.