Satellite Based Measurements of Global Surface Fresh Waters

1
Satellite Based Measurements of Global Surface
Fresh Waters
Funded by the Terrestrial Hydrology Program at
NASA Jared Entin, Program Manager
Doug Alsdorf NASA SWWG Chair alsdorf_at_geog.ucla.ed
u
www.swa.com/hydrawg/
2
Outline

• Important Hydrologic Science Questions
• Why Satellite Based Observations Are Required to
  Answer These Questions
• Present and Future Technologies
• What Needs to be Done

3
Science Questions
Alsdorf, D. and D. Lettenmaier, Science,
1485-1488, 2003. Alsdorf, D., D. Lettenmaier, C.
Vörösmarty, the NASA Surface Water Working
Group, EOS Transactions AGU, 269-276, 2003.
4
Water Energy Fluxes in Global Water Cycle
From Land Cover Land Use Change Missions (e.g.,
LandSat, MODIS, etc.)
From Precipitation (GPM, TRMM), Clouds
(CloudSat), and Soil Moisture Missions (HYDROS,
SMOS)

• Global Needs
• Surface water area for evaporation direct
  precipitation
• DS and Q

From Soil Moisture Mission (e.g., SMOS, HYDROS)
DS Qout Qin (P-E)
5
The Difficulty of In-Situ Measurements
Gauges are designed for in-channel hydraulics yet
are incapable of measuring the diffusive flow
conditions and related storage changes in these
photos of the Amazon floodplain and Arctic.
Instead of cross-sectional methods, the ideal
solution is a spatial measurement of water
heights from a remote platform.
Non-Channelized Flow
100 Inundated!

• Many of the countries whose hydrological
  networks are in the worst condition are those
  with the most pressing water needs. A 1991 United
  Nations survey of hydrological monitoring
  networks showed "serious shortcomings" in
  sub-Saharan Africa, says Rodda. "Many stations
  are still there on paper," says Arthur Askew,
  director of hydrology and water resources at the
  World Meteorological Organization (WMO) in
  Geneva, "but in reality they don't exist." Even
  when they do, countries lack resources for
  maintenance. Zimbabwe has two vehicles for
  maintaining hydrological stations throughout the
  entire country, and Zambia just has one, says
  Rodda. Stokstad, E., Science, 285, 1199, 1999
• Operational river discharge monitoring is
  declining in both North America and Eurasia.
  This problem is especially severe in the Far East
  of Siberia and the province of Ontario, where 73
  and 67 of river gauges were closed between 1986
  and 1999, respectively. These reductions will
  greatly affect our ability to study variations in
  and alterations to the pan-Arctic hydrological
  cycle. Shiklomanov et al., EOS, 83, 13-16,
  2002

6
Resulting Science Societal Questions
How does this lack of measurements limit our
ability to predict the land surface branch of the
global hydrologic cycle? E.g., In locations
where gauge data is available, GCM precipitation
and subsequent runoff miss streamflow by 100
the question is unanswered for ungauged wetlands,
lakes, and reservoirs throughout the world.
What is the role of wetland, lake, and river
water storage as a regulator of biogeochemical
cycles, such as carbon and nutrients? E.g.,
Rivers outgas as well as transport C. Ignoring
water borne C fluxes, favoring land-atmosphere
only, yields overestimates of terrestrial C
accumulation
What are the implications for global water
management and assessment? The ability to
globally forecast freshwater availability is
critical for population sustainability. Water
use changes due to population are more
significant than climate change impacts.
Can we predict flooding hazards which could be
used to understand the consequences of land use,
land cover, and climatic changes for a number of
globally-significant, inhabited floodplains?
7
Why Use Satellite Based Observations Instead of
More Stream Gauges?

• Wetlands and floodplains have non-channelized
  flow, are geomorphically diverse at a point
  cross-sectional gauge methods will not provide
  necessary Q and ?S.
• Wetlands are globally distributed (cover 4
  Earths land 1gauge/1000 km2 X 40,000 230M)
• Declining gauge numbers makes the problem only
  worse. Political and Economic problems are real.
• Need a global dataset of Q and ?S concomitant
  with other NASA hydrologic missions (e.g., soil
  moisture, precipitation). Q ?S verify global
  hydrologic models.

8
Wetlands RequireSpatial View

• Only 1 of 8000 Amazon floodplain lakes has been
  measured for annual water balances!
• 7 Gauges on channels, how do they define flow
  across floodway?
• Annually inundated area in Amazon is 750,000
  km2!
• Gauge data is only sporadically available, if at
  all.
• Worlds largest river, yet Q and DS are poorly
  known.
• Situation is much worse for Congo and other
  remote basins.

9
Global Distribution of Wetlands and Lakes
Requires Satellite Perspective

• Wetlands are distributed globally, 4 of Earths
  land surface, but many locations known to be much
  larger than thought in this view (e.g., Amazon,
  Arctic).
• Mean interannual storage variability for 5 lakes
  in Africa is 200 mm averaged over all of Africa
  is 5 mm, about 1/10th the equivalent value for
  soil moisture. What is the summed effect of all
  smaller water bodies? ?S is not negligible and
  likely at least half that of soil moisture.

Matthews, E. and I. Fung, Global Biochemical
Cycles, 1, 61-86, 1987. Prigent, C., E.
Matthews, F. Aires, and W. Rossow, Geophysical
Research Letters, 28, 4631-4634, 2001. Sridhar,
V., J.Adam, D.P. Lettenmaier and C.M. Birkett,
American Meteo. Soc., Long Beach, CA, February,
2003
10
Typical Problems With Q From 2D Imagery
Iskut River, Alaska
Extreme Flood
Effective width determined from SAR imagery and
discharge for three braided rivers in the Arctic.
Discharge was determined from a gauge at a
downstream coalescing of channels. The three
curves represent possible rating curves to
predict discharge in the absence of gauge data.
Normal Flood
Critical Problems 1. Relies on in-situ
measurements to derive Q and DS, 2. Does not
provide h, dh/dt, dh/dx no hydraulics
Smith, L.C., Isacks, B.L., Bloom, A.L., and A.B.
Murray, Water Resources Research, 32(7),
2021-2034, 1996. Smith, L.C., Isacks, B.L.,
Forster, R.R., Bloom, A.L., and I. Preuss, Water
Resources Research, 31(5), 1325-1329, 1995.
11
Storage Change Discharge from Radar Altimetry
Presently, altimeters are configured for
oceanographic applications, thus lacking the
spatial resolution that may be possible for
rivers and wetlands.
Water Slope from Altimetry
Classified SAR Imagery

DS
Note loss of gauge data post 1997
Birkett, C.M., Water Resources Res.,1223-1239,
1998. Birkett, C.M., L.A.K. Mertes, T. Dunne,
M.H. Costa, and M.J. Jasinski,Journal of
Geophysical Research, 107, 2002.
12
Channel Slope and Amazon Q from SRTM
SRTM
Water Slope from SRTM
Channel Geometry from SAR
Observed at Manacapuru Gauge 96300
m3/s Estimated from SRTM and Mannings n 93500
m3/s
Hendricks, Alsdorf, Pavelsky, Sheng, AGU
Abstract, 2003
13
Predicted GRACE Detectability of Modeled Monthly
Changes in Terrestrial Water Storage
Orange bars are changes in total soil and snow
water storage modeled by the Global Soil Wetness
Project. Error bars represent the total
uncertainty in GRACE-derived estimates, including
uncertainty due to the atmosphere, post glacial
rebound, and the instrument itself. Modified
from Rodell and Famiglietti 1999.
14
?S and Floodplain Hydraulics from Interferometric
SAR
Interferogram showing water level increases of
12 cm over 44 days (dh/dt) in blue and land
surface in green
The ideal spaceborne technology would be capable
of measuring these hydraulics!
Perspective view of dh/dt
SRTM DEM
Amazon R.
Interferometric phase showing dh/dt from April 15
to July 12, 1996. Flow hydraulics vary across
this image. Arrows indicate that dh/dt changes
across floodplain channels.
Alsdorf et al., Nature, 404, 174-177, 2000
Alsdorf et al., Geophysical Research Ltrs., 28,
2671-2674, 2001 Alsdorf et al., IEEE TGRS, 39,
423-431, 2001.
15
ICESat Targeting of Lower Mississippi River
targeted path mode track 2.5 off-nadir
targeted path coincident w/ river reach
8-day reference track
22 km
16
Lower Mississippi River Extent, Stage Slope
2.5 Off- Nadir
17
Confluence of Rio Tapajos with Amazon

• Closer examination of flat area shows
  differences between GLA06 and GLA14 products
  (different analysis of digitized echo waveform)
• GLA06 based on max peak
• GLA14 based on entire pulse
• Indicative of vegetation
• Track crosses Amazon and upstream Rio Tapajos
• Region from 2.4? to -3.2? appears to be along
  Rio Tapajos

18
GLAS Precision Estimate

• Residuals to low degree polynomial fit to
  elevation on preceding chart represent GLAS
  precision
• Both GLAS data products give similar result (echo
  waveform is Gaussian)
• 40 Hz points shown (no averaging)
• Over this water surface, the precision is lt 3 cm
• May be remaining decimeter level altitude bias,
  but elevation slope is very accurate

19
Problems with Currently Operating Technologies

• Low Spatial Resolution
• The spatial resolution of currently operating
  radar altimeters is low and not capable of
  accurately measuring water surface elevations
  across water bodies smaller than 1 km.
• GRACE spatial resolution is 200,000 km2
• Between track spacing of radar and lidar
  altimeters is much greater than 100 km, thus
  easily missing many important lakes and
  reservoirs.
• Low Temporal Resolution
• Interferometric SAR requires two data-takes, thus
  typical ?t is one month or much greater.
• SRTM operated for just 11 days in February of
  2000.
• Special Requirements
• Interferometric SAR measurements of dh/dt only
  work with double-bounce travel path which
  results from inundated vegetation.
  Interferometric SAR does not work over open water
  (i.e., dh/dt measurements are not possible).

20
Why Altimetry?

• Only method capable of high resolution water
  surface elevation measurements
• can provide h, dh/dx, and dh/dt
• Is technology evolution, not revolution
• Both radar and lidar altimetry have already been
  used in space
• Does not require double-bounce like
  interferometric SAR
• The water surface is highly reflective, thus
  should be easily measured at nadir

21
Summary of Future Technologies

• Lidar Altimetry
• Along-track spatial resolution is already 70 m
  and should be capable of 20 m samplings (GSFC).
• Height resolutions are already 3 cm and better
  (U.Texas CSR).
• Images of heights might be achieved through
  multiple lidar beams because off-nadir returns of
  2.5º are demonstrated from space (GSFC).
• Cloud and vegetation penetration are possible,
  but density limits are not known (GSFC).
• Radar Altimetry
• Through delay-doppler and SAR-like processing,
  radar altimetry is capable of along-track 100 m
  samplings (Johns Hopkins APL).
• Waveform histories and tomography should allow
  height resolutions approaching 1 cm (JPL).
• An interferometric altimeter should be capable of
  providing an image of elevation values, if
  off-nadir returns at K-band frequencies are
  possible (JPL with CNES partnering).

22
Future Work with ICESat Surface Water Data

• Quantify signal return amplitude from surface
  water as a function of off-nadir angle,
  atmospheric cloud/aerosol optical depth, and
  surface water roughness
• Examine frequency-of-surface-return global
  climatology
• spatial and temporal variability of observing
  through clouds/aerosols
• Evaluate surface water retrieval capability for
  inundated forests
• Quantify accuracy of slope measurements for along
  channel profiles, multiple crossings of river
  meanders, and areas of inundation
• Establish absolute accuracy of stage retrievals
  by comparison to in-situ gauges
• Demonstrate stage change retrieval from repeat
  profiles

23
(No Transcript)
24
Delay-Doppler Advantages

• Better open-ocean performance
• SSH, SWH, WS precisions 2-times better
• More averaging (multi-looking)
• Full spatial control over Doppler cells being
  averaged
• Small along-track footprint
• Typically 250 meters at 30 Hz rate,
  independent of SWH
• Smaller spacecraft altimeter
• Less transmitter power required
• Specular scatterers identified in Doppler
  spectrum
• Enhanced response to small inland (calm) water
  surfaces
• Robust measurements in water/terrain areas
• Small footprint, Doppler-smart tracker
• Flight-proven
• NASA Incubator airborne D2P

25
Tomographic Height Estimation

• Use entire range/time power history instead of
  single waveforms
• Use imager to obtain water mask and geolocation
• Generate simulated waveform templates and
  optimize fit with data by varying river height
  and reflectivity
• The problem can be recast as a Maximum Likelihood
  or MAP estimation problem for a limited set of
  model parameters
• Formal estimates of measurement errors can be
  obtained by error propagation

26
Altimeter Instrument Concept

• Use Ka-band frequency (8 mm wavelength)
• 1.5 m reflector antenna gt 4.3 km beam limited
  footprint
• 500 MHz bandwidth (30 cm range resolution) gt 650
  m pulse limited footprint
• Use preset tracker based on known topography
• Reduce number of onboard averaging to minimize
  distortion
• Use full-deramp processing to reduce data rate
• For SAR mode, use bursts to reduce PRF, and
  onboard SAR compression (e.g., K. Raneys
  delay-doppler)
• Required transmit power (10W) available from
  solid-state technology
• SAR mode requires onboard processor, higher
  complexity and digital subsystem power

27
Surface Water Interferometer Concept

• Ka-band SAR interferometric system with 2 swaths,
  50 km each
• Produces heights and co-registered all-weather
  imagery
• 200 MHz bandwidth (0.75 cm range resolution)
• Use near-nadir returns for SAR altimeter/angle of
  arrival mode (e.g. Cryosat SIRAL mode) to fill
  swath
• No data compression onboard data downlinked to
  NOAA Ka-band ground stations

28
What Needs to be Done

• Determine spatial and temporal sampling
  resolutions required to answer hydrologic
  questions.
• For example, the regular Amazon floodwave may
  need a ?t of just a few weeks, but the sudden
  Arctic Spring melt requires a much more frequent
  observation.
• Are profiles of h from an altimeter sufficient to
  measure hydraulics, or do we need an image of h
  values? How wide of image is required?
• What are the cost vs. science trade-offs
  represented by varying spatial and temporal
  resolutions? Is there a cut-off below which no
  valuable science can be gained?
• Are both discharge and storage change required?
• Surface water velocities measured from space will
  be flawed by wind-induced waves, instead use
  water slope and Mannings equation. But, still
  requires some knowledge of water depths (i.e.,
  channel cross sectional area).
• ?S is a simple spaceborne measurement, but is ?S
  sufficient to constrain water and energy cycle
  models?
• Technology Demonstrations
• What is the capability to penetrate clouds and
  vegetation?
• Does the instrument provide reliable off-nadir
  measurements of h?
• Need funding opportunities for such
  demonstrations.
• Is surface water science sufficient to support an
  entire satellite mission?
• What is the cost of a surface water mission?
• Which space agencies and related groups would
  participate, with funds?
• If other science is joined with a surface water
  mission, what technology and orbital compromises
  are required to ensure a healthy mission for all
  participating science groups?

29
The Virtual Mission

• The VM is a synthetic hydrologic model of a
  continental-scale basin with an embedded
  floodplain and channel hydraulics model. By
  controlling the various hydrologic parameters
  (precipitation, evaporation, infiltration, energy
  balances, etc.), the runoff related boundary
  conditions of the channel and wetlands hydraulics
  models are known which thus allows a known
  relationship between samplings of various channel
  and wetland morphologies to water cycle science.
• Science, technology, and cost trade-offs will be
  determined by sampling the modeled water surface
  at various resolutions related to alternate
  configurations of existing and space-ready
  technologies.
• The VM will identify exact water cycle, carbon
  cycle, and natural hazards questions that can be
  answered from hydraulic measurements collected by
  a spaceborne platform.
• The VM will establish trade-offs between
  measuring storage changes versus measuring
  discharge.

30
Conclusions

• Lack of Q and ?S measurements cannot be
  alleviated with more gauges (e.g., wetlands
  diffusive flow political economic problems are
  real).
• This lack leads to poorly constrained global
  hydrologic models and unknown carbon fluxes from
  wetland water surfaces.
• Conceptually, the ideal solution is a satellite
  mission with temporal and spatial resolutions
  compatible with planned hydrologic missions and
  modeling efforts.
• Radar and lidar altimeter instruments are capable
  of measuring hydraulics, but spatial and temporal
  resolutions required to answer science questions
  are not known.
• The Virtual Mission will identify the resolutions
  required to answer important hydrologic science
  questions.

www.swa.com/hydrawg