Tennessee Technological University

1
Tennessee Technological University
Sustainable Application of Water-Measuring
Satellite Missions for Water Resources Management
Past, Present and Future

• Faisal Hossain

Faisal Hossain Department of Civil and
Environmental Engineering Tennessee Technological
University
2
Tennessee Technological University
ACKNOWLEDGEMENTS

1. Former and Current Students- Amanda Harris,
   Preethi Raj, Nitin Katiyar, Jon Schwenk, Rahil
   Chowdhury, Ling Tang and Caitlin Balthrop.
2. Collaborators University of Connecticut,
   University of Mississippi, Ohio State University,
   NASA Goddard Space Center (Laboratory of
   Atmospheres and Hydrologic Sciences Branch),
   McNeese State University, University of Oklahoma,
   Oregon State University, University of
   California-Davis, University of Dhaka, Indian
   Institute of Technology-Kanpur.
3. Sponsors NASA Rapid Prototyping Capability
   Program NASA Precipitation Measurement Program
   NASA Earth System Science Fellowship TTU
   Research Initiation Grants, TTU Water Center,
   Ivanhoe Foundation, Mississippi Department of
   Environmental Quality.
4. International Partners Institute of Water
   Modeling (Bangladesh), International
   Precipitation Working Group (WMO), Bureau of
   Meteorology (Australia).

• Faisal Hossain

3
Tennessee Technological University
OUTLINE

1. Primary Research Area Scientific evolution of
   the concept of sustainability for
   water-measuring satellites for water resources
   management.
2. Overview of Complementary Research and Education
   Agendas.

• Faisal Hossain

4
Tennessee Technological University
WATER MEASURING SATELLITES - 101

• Hydrologic Remote Sensing- Microwave MW (1-20cm)
  and Infrared IR (lt 0.1cm) Wavelengths.
• Water Cycle Variables- Rainfall, Soil
  Moisture, Discharge.
• Water has a dipole and high dielectric constant.
• Orbiting and Geostationary platforms
  Passive/Active.
• MW sensors mostly orbiting higher accuracy,
  lower sampling frequency (space-time).
• IR sensors mostly geostationary platforms lower
  accuracy, higher sampling frequency (space-time).

• Faisal Hossain

5
Tennessee Technological University
WATER MEASURING SATELLITES - 101
TRMM
WaTER (SWOT)

• Faisal Hossain

Geostationary Orbit
HYDROS
6
Tennessee Technological University
WHAT IS SUSTAINABILITY FOR WATER MEASURING
SATELLITES?
Sustainability is a characteristic of a process
or state that can be maintained at a certain
level indefinitely. - Wikipedia
For water measuring satellites? To make
optimal use of satellite sensors capability to
measure water looking down over a large area
from the vantage of space.

• Faisal Hossain

Optimal Use? Identify, maintain and enhance the
realistic limits to which satellite hydrologic
data can be used for analysis, modeling and
monitoring of water resources.
7
Tennessee Technological University
The Conceptual Appeal of Water-Measuring
Satellites to the Hydrologist

• In-situ networks globally disappearing or
  absent expensive maintenance limited by
  point-scale.

• Faisal Hossain

Source Climate Prediction Center
Effective sampling area of the worlds rainfall
gages is the size of a few football fields!
Source USGS
8
Tennessee Technological University
The Conceptual Appeal of Water-Measuring
Satellites to the Hydrologist

• Global Hydrology (Earths Energy/Water Budget)
  Can only be supported by space-borne instruments
  (75 of surface is oceans).
• Flood prone Tropics sparse or non-existent
  network where floods are most catastrophic.

• Faisal Hossain

9
Transboundary Flood Forecasting The Story of
the Niger River
1. 4030 km long, 211,3200 km2
2. Flows through 5 countries
3. Drainage area comprised of 11 countries
4. Frequent river flooding induced by heavy
rainfall
Question How does one monitor early the
evolution of river flooding across political
boundaries of 5 nations, 11 administrations and a
diverse landscape?
5. Diverse climate, rainfall regime, soil
conditions, topography varying response of
landscape to rainfall
10
Tennessee Technological University
Transboundary Flood Forecasting The Global
Picture on International River Basins

• Hydro-political limitations worsen at the shorter
  time scales

Percentage Area Number of Countries
91-99 39
81-90 11
71-80 14
61-70 11
51-60 17
41-50 10
31-40 10
21-30 13
11-20 9
1-10 11
145 countries are associated in IRBs Accounts for
40 of total land surface. gt 50 of total
surface flow

• Faisal Hossain

214 International River Basins in 1979 UN
Register 261 in 2002 (Updated)
Source Dr. Aaron Wolf, Oregon State University
11
Tennessee Technological University
Appeal in terms of Future Scenario

• Faisal Hossain

WaTER (SWOT) Expected launch 2016 Q for major
rivers every 2-3 days
Expected launch 2013 3 hourly global rainfall
products at 10X10 km scale
12
Tennessee Technological University
Appeal in terms of Future Scenario

• Faisal Hossain

Hossain, F., N. Katiyar, A. Wolf, and Y. Hong.
(2007). The Emerging role of Satellite Rainfall
Data in Improving the Hydro-political Situation
of Flood Monitoring in the Under-developed
Regions of the World, Natural Hazards, Invited
Paper
13
Tennessee Technological University
Problems with Water Measuring Satellites

• OLD Issues (Relatively longer known and accepted)
• MW temporal sampling of rainfall was low until
  late 1990s. IR Rainfall data useful at gt
  degree-monthly scales.
• Scale Incongruity (satellite rainfall/moisture
  data too large for dynamic terrestrial
  hydrology).
• Soil moisture accuracy limited by the need for
  long MW wavelength (L-band).
• Passive MW (PMW) data for discharge estimation
  has been good only for large, steady (glaciers)
  rivers on monthly timescales.
• Historical solutions devised by hydrologists for
  handling coarse resolution data Spatial-Temporal
  downscaling.

Spatial Downscaling

• Faisal Hossain

14
Tennessee Technological University
Chronology of Scale and Accuracy of Satellite
Rainfall Data
1970 1980 1990 2000 -
2010

• Faisal Hossain

• IR Sensors on GEO platforms
• Good space-time sampling
• IR parameters weakly
• related to rainfall process

• PMW Sensors on LEO platforms
• Poor space-time sampling
• PMW parameters strongly related to rainfall
  process
• Merging or IR with PMW began

• More Merged Products
• Tropical Rainfall Measuring Mission- TRMM
• Anticipation of GPM
• 3 hourly and globally coherent rainfall data

1 Degree-Monthly
0.25 degree 3 hourly
15
Tennessee Technological University
Problems with Rainfall Measuring Satellites

• NEW Issues on Satellite Rainfall Data (Recent
  Insights post 2004 era)
• Existing frameworks and metrics (bias/rmse,
  correlation) inadequate for assessing hydrologic
  potential of satellite data.
• Satellite rainfall error more complex
  (multi-dimensional) than conventional network
  data.
• Complexity of error increases as scales
  (time/space) decrease non-negligible for
  dynamic hydrologic modeling.

• Faisal Hossain

16
All Overland Satellite Rainfall Algorithms are
Probabilistic at Hydrologically Relevant Scales
Four Possible Outcomes of a Rainfall Sensor at
any given time 1. Successful Rain
Detection/Delineation (HIT) 2. Unsuccessful Rain
Detection/Delineation (MISS) 3. Successful
No-Rain Detection/Delineation (HIT) 4.
Unsuccessful No-Rain Detection/Delineation (MISS)

• Rainy/Non-rainy area delineation has a distinct
  spatial structure
• Systematic error has a non-negligible
  spatio-temporal structure
• Random error has a spatial structure
• Regime Dependence of error structure on climate,
  location, season

17
Tennessee Technological University
Our Sustainable Solution and Framework for Old
and New Problems
HYPOTHESIS New approaches needed for
hydrologists that recognize scale incongruity.
As space and time scales become smaller, the
passive sensors precipitation measurement
characteristics become more complex and
random. Fine-scale hydrologic assessment of
satellite rainfall retrievals requires the
recognition of this increasing complexity of
satellite precipitation error structure.

• Faisal Hossain

Hossain and Lettenmaier, 2006, Water Resources
Research
18
We Need Hydrologic Process-based Understanding of
Scale Incongruity
Watershed Non-linear system yavg ?
f(xavg)
Time
Space
Thresholding
Non-linearity
An infiltration approach to surface runoff
modeling (physically-based) as follows
19
Our Generalized Framework for the Community
(IPWG- PEHRPP) For Rainfall
ONE Hydrologically Relevant Frameworks should
answer three key questions Q1. How does the
error vary in time? Q2. How does the error vary
in space? Q3. How off is the rainfall estimate
from the true value over rainy areas?
TWO Metrics should have Diagnostic and
Prognostic value Diagnostic Able to quantify
uncertainty on a specific feature/dimension of
precipitation. Prognostic Amenable for use in a
mathematical error model for synthetic generation
of high resolution satellite rainfall data.
Hossain, F. and G.J. Huffman. (2008).Investigating
Error Metrics for Satellite Rainfall at
Hydrologically Relevant Scales, Journal of
Hydrometeorology (In press)
20
Tennessee Technological University
Two-Dimensional (x-y) Satellite Rainfall Error
Model SREM2D

• Based on the concept of reference (ground
  validation) rainfall.
• Modular in design (collection of concepts) for
  any rainfall product.
• Total Error Metrics - 9
• Uses Error Metrics interpretable by both
  hydrologists and Data-producers.
• Currently used by other research groups (MSU
  UArizona OleMiss). Preferred by NASA Laboratory
  of Atmospheres.

• Faisal Hossain

Hossain and Anagnostou (2006) IEEE Trans Geosci.
Remote Sensing, 44(4).
21
Tennessee Technological University
Transboundary Flood Monitoring New Questions
for Assessing Sustainability
General Science Question How realistic is the use
of satellite rainfall in overcoming the
transboundary limitations to flood monitoring?
Specific Questions What specific IRBs, and
downstream nations would benefit more than others
from GPM? Can we develop rules of thumb for
application of satellite rainfall data in
ungauged IRBs?

• Faisal Hossain

22
Ball Park Assessment for NASA product 3B41RT
Improvement
Major
Minor
Negative
Fully Distributed Open-Book Hydrologic Model
KANPUR 1.0 by Katiyar, N. and Hossain, F. 2007
Environ. Mod. Software, vol. 22(12).
23
Speculations on IRBs where Satellite Rainfall
Data will be Sustainable for Flood Monitoring
Preliminary Speculation - Setting aside ALL
assumptions
Name of down stream country International River Basin of Total Basin Area
Cameroon Akpa/Benito/Ntem 41.8
Senegal Senegal 8.08
Ivory Coast Cavally 54.1
Benin Oueme 82.9
Botswana Okovango 50.6
Nigeria Niger 26.6
Bangladesh Ganges-Brahmaputra-Meghna 7.0
Brunei Bangau 46.0
Laos Ca/Song Koi 35.1
Cambodia Mekong 20.1
Improvement
Negligible Improvement
24
More Intelligent Speculation
Based on Koppen Climate Classification
Source Encyclopedia Britannica
25
Speculation on IRBs (Contd.)
Cfa Cwa Humid Subtropical Bsh-
Semi-arid Ganges River Bangladesh (45) ? Yalu
and Tomen Rivers North Korea (20)?Limpopo
River Mozambique (35)? Senegal River
Senegal (42)?La Plata River Uruguay (45)?
26
Tennessee Technological University
Spatial Downscaling of Satellite Rainfall Data
New Questions for Assessing Sustainability

• Spatial downscaling based on scale invariance.
• Preserves the mean of rainfall.
• Stochastic in nature yields equi-probable
  realizations.
• Mimics the expected variance of rainfall at
  downscaled resolution.

• Downscaling schemes preserve the mean and mimic
  the expected variance. Is that good enough for
  flood prediction needs for GPM?
• Satellite rainfall data has scale-dependent
  complex error
• i) Does the downscaling scheme add artifacts to
  downscaled satellite rainfall data?
• ii) What are the hydrologic implications of using
  a spatial downscaling scheme for satellite
  rainfall on flood prediction uncertainty?

• Faisal Hossain

27
An end-to-end system for NASA real-time satellite
rainfall data analysis
End-to-End system conceptualized, developed and
tested over Upper Cumberland River basin in
Kentucky.
Upper Yazoo Basin
28
Downscaling of 3B41RTIncreases streamflow
simulation uncertainty(?!)
0.0625 degree
0.125 degree
0.03125 degree

Stream flow simulation uncertainty using
downscaled 3B41RT data
29
Tennessee Technological University
Physically-based Investigation of Spatial
Downscaling on Overland Runoff Generation

• Faisal Hossain

Downscaling

• Rainy grid boxes can be non-rainy
• Non-rainy grid boxes can be rainy
• Redistribution and bias of downscaled rainfall
  can be significant

Upscaling
30
Tennessee Technological University
Physically-based Investigation of Spatial
Downscaling on Overland Runoff Generation

1 i1
1 i1
f (avg. rainfall)
i
yavg
2 i2
2 i2

• Faisal Hossain

f (avg. rainfall)
yavg
Watershed Non-linear system yavg ? f(xavg) What
role does C subgrid rainfall variability play
in runoff simulation?
31
Tennessee Technological University
Physically-based Investigation of Spatial
Downscaling on Overland Runoff Generation
Ksat - field
Rainfall- field

High spatial Correlation-200km
Clayey Loam
Medium spatial Correlation-100km
Silty Loam

• Faisal Hossain

Low spatial Correlation-50km
Sandy Loam
32
Tennessee Technological University
Physically-based Investigation of Spatial
Downscaling on Overland Runoff Generation
Estimation Bias () Estimation Bias () Estimation Bias () Estimation Bias () Estimation Bias () Estimation Bias () Estimation Bias () Estimation Bias () Estimation Bias () Estimation Bias ()
Rainfall 50 KM (LOW) 50 KM (LOW) 50 KM (LOW) 100 KM (MEDIUM) 100 KM (MEDIUM) 100 KM (MEDIUM) 200 KM (HIGH) 200 KM (HIGH) 200 KM (HIGH)
Soil Clay Silt Sand Clay Silt Sand Clay Silt Sand
Ponding Time Scale Effect -98.2 -97.8 -79.5 -89.6 -94.8 -98.1 -97.9 -88.8 -89.9
Ponding Time Downscaling Effect -90.1 -90.0 10.0 -91.1 -69.9 -77.8 -98.0 -51.2 -17.4
Runoff Volume Scale Effect -75.3 -75.5 -80.5 -75.0 -75.1 -75.8 -75.1 -75.3 -77.0
Runoff Volume Downscaling Effect 0.1 1.13 -3.46 -7.6 -7.6 -8.8 -0.1 -1.1 4.8

• Faisal Hossain

• Spatial downscaling technique improves the
  estimation of accumulated runoff parameters when
  compared to estimates derived from lower
  resolution rainfall data.
• Not suitable for improving the estimation of time
  sensitive runoff parameters such as the time to a
  flood peak.

33
Tennessee Technological University
Discharge Estimation of Braided
RiversSustainability of the SWOT Mission
What is the uncertainty of satellite
interferometry (SRTM) -based discharge estimation
of large braided rivers?

• Faisal Hossain

SRTM Overpass Feb 20, 2000
34
Tennessee Technological University
Discharge Estimation of Braided RiversValue of
SWOT Mission
Estimated dry season discharge comparable to the
natural low-flow variability.

• Faisal Hossain

Hamski et al (2008) ASLO Conference March 2-7,
Orlando, Florida.
35
Tennessee Technological University
The Future on Sustainability of Application of
Water Measuring Satellites
NASAs vision for the post-GPM era (2013) - To
produce routine high-level uncertainty
information of their global and real-time
rainfall products for users to identify
sustainable application on their own (George
Huffman of Laboratory of Atmospheres-NASA).

• Faisal Hossain

The Unresolved Paradox Satellite rainfall will
be most useful over ungauged (non-GV) regions
so how can we generate routine uncertainty
estimates for satellite data over those regions ?
36
Tennessee Technological University
The Future on Sustainability of Application of
Water Measuring Satellites
Our Strategy for Solving the Paradox

• Faisal Hossain

Study Regions over US 8 years of data Radar as
ground validation
Rainfall Climatology over US
Global similarity of US climate zones
37
Tennessee Technological University
Overview of Other Research AgendaNew Paradigms
for Improving Spatial Mapping

• Development of NLDMAP 1.0
• (Non-linear Dynamic Mapping) for rural settings.
• Test cases 1)Arsenic contamination of
  groundwater in Bangladesh 2) USGS monitored
  aquifers in Connecticut.
• Improving geostatistical (kriging) methods using
  Chaos Theory and Neural Networks.
• Chaos and ANN analysis are complete (merging of
  schemes on-going collaboration with Dept. of
  ECE/TTU).

• Faisal Hossain

Hossain,, F., and B. Sivakumar. (2006). Spatial
Pattern of Arsenic Contamination in Shallow
Tubewells of Bangladesh Regional Geology and
Non-linear Dynamics Stochastic Environmental
Research and Risk Assessment, vol 20(1-2), pp.
66-76
38
Overview of Education Agenda on Water Resources
Engineering Education

1. Uncertainty is omni-present in natural or
   man-made water resources systems.
2. Need good understanding of Stochastic Theory,
   e.g. Random Functions, Geostatistics, Time
   series, to model/predict the variability.

• More and more research conducted at graduate
  level involving stochastic theory applications.
• Blooms learning level of entering graduate
  students should be Analysis or Application.
• Are we doing a good job with instruction of
  stochastic theory in CE/Water resources?

39
Tennessee Technological University
Overview of Education AgendaStochastic Theory
Education through Visualization Environment
Total Number of Universities Surveyed 67
Number of Universities with www listing of relevant courses 57
Total number of courses identified (having the generic terms stochastic, statistics, numerical etc in CE curricula) 241
Graduate(Dual listed) and Undergraduate 84(4.5)11.5
Number of schools with integrated courses on Stochastic Theory 40
Number of courses on Stochastic Theory 84 (35)
Number of courses on Stochastic Theory in Water Resources and Environmental Engineering 27 (11.2)
Number of courses on Stochastic Theory in Water Resources only 23 (9.5)

• Faisal Hossain

Schwenk, Hossain and Huddleston (2008) Computer
Applications in Engineering Education (In press)
40
A Long-term Vision
http//iweb.tntech.edu/saswe
41
THANK YOU!
QUESTIONS?