A Combination of Remote Sensing Data and Topographic Attributes for the Spatial and Temporal Control

1
A Combination of Remote Sensing Data and
Topographic Attributes for the Spatial and
Temporal Control of theSoil Wetness over the
Mackenzie River Basin
Presented by Dr Marouane Temimi
Co-authors Dr Robert Leconte
Dr Reza Khanbilvardi
NOAA-CREST Symposium 2008
2
Study Area The Peace Athabasca Delta (PAD)
Image of the PAD (http//imprint.uwaterloo.ca/)
Peace Athabasca Delta
(Leconte, 2004)
3
Soil moisture estimation
There are two ways to measure soil moisture
Direct Observations
Remote Sensing
Local measurements require more resources high
accuracy, but locally. Interpolation required
Extensive spatial and temporal coverage
The heterogeneity and the large extent of the
Mackenzie Basin strongly recommend the use of
remote sensing to estimate the spatial and
temporal distribution of soil moisture.
4
Use of remote sensing data
Approaches using passives microwaves
Two main categories
Approaches based on the use of a Radiative
Transfer Model
Simplistic approaches
Development of an Index as a surrogate for soil
moisture
Soil moisture estimates
5
Approaches for WSF Estimation

• Polarization Differences Ratio ?Tobs f.
  ?Twater (1-f) ?TSoil
• 
  (Sippel, 1994)

• Use of a Single Frequency WSF a Tb37H b
• 
  (Tanaka et al. 2003)

• Emissivity Differences Ratio WSF (e edry)
  /(ewet - edry)
• 
  (Fily et al. 2003)

• MultiFrequencies Differences
• 
  

(Basist et al. 2001)

• We propose to use the Polarization Ratio PR
  (Tbv Tbh) / (Tbv Tbh)

Brightness temperatures are provided by the
AMSR-E sensor at 37 GHZ frequency
6
The of use the Polarization Ratio
The fractional water extent estimated using
passive microwave is a combination of water body
area, flooded area and soil moisture.
PRobs ow. PRow f. PRf nf. PRnf sm PRsm
WSF(AMSR-E)ow f sm ow f sm nf 1
Where
nf non-flooded soil fraction f flooded soil
fraction ow fraction of the permanent open water
surface sm fraction of the soil moisture
contribution to the observed PR PRow, PRf, PRnf
and PRsm are respectively the PR of the open
water, flooded area, non-flooded area and
wetlands. PRow, PRf and PRnf were calibrated
using Brightness temperature measurements
7
Flooded Area Estimation
An example of flooded fraction map estimated
using AMSR-E 37 GHz Brightness temperature
measured over the PAD area on August 31, 2003
8
Definition of the Wetness Index
However, the water surface fraction derived from
visible images does not include the effect of
soil wetness, and exclusively provides an
estimate of the water body extent.
MODIS Images were used in this work in order to
estimate the water body extent over the PAD area
WSF(MODIS) ow f.
We propose to define a Basin Wetness Index (BWI)
based on the difference between the passive
microwave and visible responses.
BWI varies between 0 and 1. It presents the
fraction of the wetlands within the non-flooded
area.
WSF(MODIS) ??
9
Definition of the Wetness Index
An existing relationship between discharge and
WSF can be written as
WSF(MODIS) a Qb(t)
10
Definition of the Wetness Index
The availability of AMSR-E data and discharge
observations enables the estimation of a daily
value of the developed BWI, which can be written
as
11
Downscaling of the BWI
BWI provides an estimation of the spatial average
of the soil wetness over the entire PAD area
How can we predict the spatial distribution of
the estimated BWI ?
We suggest to use a Wetness Index (WI) based on
topographic attributes
The classic wetness index proposed by (Beven and
Kirkby, 1979) is WIln (a/tanß)
Where
a is the contributing area tanß is the terrain
slope
12
Modification of the classic WI
The vegetation cover affects significantly the
spatial distribution of the soil moisture
Therefore, we are suggesting modifying the static
index by weighting its members using the
vegetation canopy density to account for the
variable effect of the vegetation which largely
influence the spatial distribution of the soil
moisture.
WI V. ln (a) (1-V). ln(1/tanß)
V is the fractional vegetation cover which is
estimated using the following relationship
proposed by (Eaglson, 1982) V 1 exp(-µ.
LAI) Where LAI is the Leaf Area Index and µ is
the extinction coefficient.
13
Temporal variability of the vegetation density
over the PAD area
01/07
01/06
01/08
14
Classic WI vs. Dynamic WI
01/07
01/08
01/06
VS.
15
Downscaling of the BWI

• The WI provides information about the systematic
  spatial distribution of the soil wetness

• BWI will determine the temporal variability of
  soil wetness.

Hence, the basin is virtually saturated or dried
down depending on the BWI and WI values. Pixels
with the highest/lowest WI are wetted up/dried
down first.

• Two stations were selected to assess the
  reliability of downscaling approach

• the W values were averaged over 9 pixels (a
  window of 3x3 pixels) surrounding the stations.

16
Conclusions

• A satisfactory agreement was observed between the
  defined BWI, precipitation, and temperature
  values. Temperature has a dominant effect on the
  BWI variation

• The modified WI provided a better agreement
  between observed precipitation and estimated soil
  wetness

• It is expected that the correlation between the
  estimated wetness and soil moisture measurements
  will be better than that of the precipitation
  observations

• The results of this study are an improvement of
  the static indices reliability as the most of the
  proposed indices in the literature do not explain
  more than 50 of the spatial distribution of the
  soil moisture (Western et al., 1999).

17
Questions
?
Thank you