The 3rd Northern Forum FWG Meeting:
Assessment of the accuracy of river forecasts and decision management in the Northwest region of the United States

By Robin Radlein
Director, Alaska-Pacific Hydrology Center

A Funded Project of NASA's Decisions 2005 CAN (Integrated Systems Solution, Disaster and Water Management Applications)

G. Robert Brakenridge, Elaine Anderson, Dartmouth Flood Observatory
Son V. Nghiem, Jet Propulsion Laboratory
Pedro Restrepo, NOAA Office of Hydrologic Development

Three Enhancements:

1. Rapid Response Flood Inundation Mapping
2. Gauge-Centered Reach Maps for Inundation Prediction
3. Satellite Ice-out and Flood Detection in Alaska and Yakutsk

First, overview of activity at Dartmouth …..on which the enhancements are based

The Observatory detects, maps, measures, and analyzes major flood events world-wide using satellite remote sensing. The record of such events is being preserved as a World Atlas of Flood Hazard.

DFO also provides yearly catalogs, large-scale maps, and images of river floods, from 1985 to the present.

The Atlas is an accumulating observational record of flood inundation in GIS format

Same information, at larger scale. All such records enter the Atlas

AMSR-E operates in polar sun-synchronous orbits at 98.2 deg inclination. The swath width is 1445 km providing full global coverage in 3 days. The AMSR-E radiometer has six frequencies measuring brightness temperatures at an Earth-incidence angle of 54.8 deg.

The low-frequency channels at 6.92 and 10.65 GHz have very coarse resolutions and suffer significantly from radio frequency interference [ Njoku et al ., 2005]. The resolutions of the 18.7 and 23.8 GHz channels are still coarse and these channels are close to the water vapor spectral line [ Skou , 1989] and susceptible to atmospheric conditions. The 89.0 GHz channel has the best resolution of 6 x 4 km ; however, it is close to the oxygen spectral line [ Skou , 1989] and large atmospheric contaminations. For our purpose, the 36.5 GHz channel is optimal.

U.S. examples offer the opportunity to compare remote sensing to actual discharge obtained at a gauging station. We ratio a measurement pixel (“m”) to a calibration pixel (“c”) to isolate the signal of changing in-pixel surface water area.

When Red River experiences high discharge, measurement pixel (blue line) microwave emission values decline, but calibration pixel values do not. AMSR-E microwave detects such water area changes with high sensitivity.

Ratio of the two pixel values produces the brown line curve (here converted via a rating equation to discharge). Black line is the gauging station discharge.

Along many rivers, AMSR-E can detect floods and also tracks periods of normal flow; on a near-daily basis, and without interference by cloud cover.

Why does this work?

1. Rivers and floodplains form landscapes with complex low-relief topography.

2. When rivers rise, flow width, and water surface area, increases. Thus: more water, less land in mixed water/land pixels.

3. The spatial imaging resolution obtained by the sensor is less important than scene-to-scene calibration and strong contrast between water and land (choose best spectral band)

Third Enhancement: Satellite Ice-out and Flood Detection

(AMSR-E sees ice-out along pre-selected reaches; yellow squares indicate ice cover)

White squares: to be implemented.

Lena River , Dartmouth 's Global Hydrographic Array Reach 001.

M and C pixel dimensions shown in local UTM projection.

MODIS-based inundation area changes are shown for several spring floods

Rating curve using monthly mean discharge data from Stolb

This station is far downstream but exhibits similar seasonal hydrograph to that at Zhigansk.

Actual discharge values to be corrected by catchment area method).

The brown line is the C/M ratio (AMSR-E) values transformed into discharge by use of the rating equation. On the day of ice-break up, the measurement pixel suddenly drops in value.

Note the reversal of values during the winter months when the river is ice-covered. Same technique is being applied to Alaskan rivers.

Discharge values are from AMSR-E data.

AMSR detects ice-out and also estimates inundation extent (here mapped by MODIS).

Preliminary results, Susitna River , Reach 365

Questions:

Similar ice-out/flood detection and mapping capability is available for Alaska .

No detailed information on ice-dams, etc.

Daily AMSR-E information from a series of reaches, along rivers of interest, can provide ice-out warning and flood detection.

1. Which rivers are of most interest? How closely-spaced should measurement sites be?

2. Trial tests of utility of this approach in May of 2006 and 2007?

3. How will results be disseminated?

4. Which locations in Siberia are of most interest to NOAA colleagues in Yakutsk ?

. Is there interest in Yakutsk in getting this information?

back