1-1 SWOT IGARSS July 27, 2011 INTERFEROMETRIC PROCESSING OF FRESH WATER BODIES FOR SWOT Ernesto Rodríguez, JPL/CalTech Delwyn Moller, Remote Sensing Solution Xiaoqing Wu, JPL/CalTech Kostas Andreadis, JPL/CalTech
Dec 13, 2014
1-1
SWOTSWOT
IGARSSJuly 27, 2011
INTERFEROMETRIC PROCESSING OF FRESH WATER BODIES FOR SWOT
INTERFEROMETRIC PROCESSING OF FRESH WATER BODIES FOR SWOT
Ernesto Rodríguez, JPL/CalTech
Delwyn Moller, Remote Sensing Solution
Xiaoqing Wu, JPL/CalTech
Kostas Andreadis, JPL/CalTech
1-2
SWOTSWOT
IGARSSJuly 27, 2011
Study AreaStudy Area
• ~1000 km reach of the Ohio River basin
• Drains an area of ~220,000 km2
• Topography from National Elevation Dataset (30 m)
• River vector maps from HydroSHEDS
• Channel width and depth from developed power-law relationships
• Explicitly modeled rivers with mean widths at least 50 m
1-3
SWOTSWOT
IGARSSJuly 27, 2011
Hydrodynamic ModelingHydrodynamic Modeling• LISFLOOD-FP raster-based model• 1-D solver for channel flow• 2-D flood spreading model for floodplain flow• Kinematic, Diffusive and Inertial formulations• Requires information on topography, channel
characteristics and boundary inflows• Needed to coarsen spatial resolution to 100 m
SWOT Hydrology Virtual Mission Meeting, Paris, 22 Sep 2010
• Simulation period of 1 month
• Boundary inflows from USGS gauge measurements
1-4
SWOTSWOT
IGARSSJuly 27, 2011
Data from the SWOT Land SimulatorData from the SWOT Land Simulator
Along-Track
Ran
ge
The SWOT simulator produces data with the correct signal to noise ratio, layover and geometric decorrelation scattering properties. Notice for SWOT the land SNR is low, while surface water stands out.
1-5
SWOTSWOT
IGARSSJuly 27, 2011
Challenges for near-nadir interferometry over land
Challenges for near-nadir interferometry over land
• Topographic layover and low land SNR makes conventional phase unwrapping approaches unfeasible• Notice that fringes are well defined over the water, since the water is flat and quite bright at nadir incidence.• The signal from topography may contaminate the signal over the water (see next slide)
1-6
SWOTSWOT
IGARSSJuly 27, 2011
Radar Layover and its Effect on Interferometry
Radar Layover and its Effect on Interferometry
€
δΦ=arg 1+PLand
PWater
γ Land
γWater
exp i ΦLand − ΦWater( )[ ] ⎡
⎣ ⎢
⎤
⎦ ⎥
Brightness Ratio (land darker than water)
Correlation Ratio (land less correlated than water)
Volumetric Layover (trees)
Surface Layover
Points on dashed line arrive at the same time
1-7
SWOTSWOT
IGARSSJuly 27, 2011
NASA SWOT Land Processing Approach
NASA SWOT Land Processing Approach
• Processing approach relies on having a fair estimate of topography and water body elevation– Estimate can be derived from a priori data or
previous SWOT passes (to account for dynamics)
• A priori information is used to generate reference interferograms for phase flattening and estimation of layover (to avoid averaging in land)
1-8
SWOTSWOT
IGARSSJuly 27, 2011
Interferogram after phase flattening with reference interferogram
Interferogram after phase flattening with reference interferogram
Noisy interferogram Noisy interferogram after flattening with reference
1-9
SWOTSWOT
IGARSSJuly 27, 2011
Layover region identificationLayover region identification
Noisy interferogram Noisy interferogram after flattening with reference
1-10
SWOTSWOT
IGARSSJuly 27, 2011
Land Processing Flow to GeolocationLand Processing Flow to Geolocation
1-11
SWOTSWOT
IGARSSJuly 27, 2011
Layover maskAll pixels with any layover are red
Layover maskAll pixels with any layover are red
Mid-Swath Near-Swath
1-12
SWOTSWOT
IGARSSJuly 27, 2011
What if we accept pixels whose expected error is < 5 cm?
What if we accept pixels whose expected error is < 5 cm?
1-13
SWOTSWOT
IGARSSJuly 27, 2011
From raw heights to hydrologic variables
From raw heights to hydrologic variables
Discharge
Width Flow Depth(height from bottom)
Slope
Manning’s equation
1-14
SWOTSWOT
IGARSSJuly 27, 2011
Water ClassificationWater Classification
1-15
SWOTSWOT
IGARSSJuly 27, 2011
River Channel MaskRiver Channel Mask
. Pavelsky and L. Smith, “Rivwidth: A software tool for the calculation of river widths from remotely sensed imagery,” IEEE Geoscience and Remote Sensing Letters, vol. 5, no. 1, p. 70, 2008.
1-16
SWOTSWOT
IGARSSJuly 27, 2011
Center Line MaskCenter Line Mask
1-17
SWOTSWOT
IGARSSJuly 27, 2011
Get Center LineGet Center Line
. Pavelsky and L. Smith, “Rivwidth: A software tool for the calculation of river widths from remotely sensed imagery,” IEEE Geoscience and Remote Sensing Letters, vol. 5, no. 1, p. 70, 2008.
1-18
SWOTSWOT
IGARSSJuly 27, 2011
Project to Local CoordinatesProject to Local Coordinates
s
c
• Spline interpolate center line to constant separation points downstream• Use spline to obtain local tangent plane coordinate system at each point
• For each point:-Use KDTree algorithm to find closest centerline point- Project point into local coordinate system to get along and across-track distance
1-19
SWOTSWOT
IGARSSJuly 27, 2011
Measured Noisy ElevationMeasured Noisy Elevation
1-20
SWOTSWOT
IGARSSJuly 27, 2011
Unsmoothed Elevation ErrorsUnsmoothed Elevation Errors
1-21
SWOTSWOT
IGARSSJuly 27, 2011
Height Error vs Downstream DistanceHeight Error vs Downstream Distance
1-22
SWOTSWOT
IGARSSJuly 27, 2011
Height Error vs Downstream Distancewith Downstream Averaging
Height Error vs Downstream Distancewith Downstream Averaging
1-23
SWOTSWOT
IGARSSJuly 27, 2011
Measurement ErrorsDownstream Averaging: 200 m
Measurement ErrorsDownstream Averaging: 200 m
1-24
SWOTSWOT
IGARSSJuly 27, 2011
Measurement ErrorsDownstream Averaging: 1 km
Measurement ErrorsDownstream Averaging: 1 km