Strategy for Generating Rating Curves:

A very limited amount of international river discharge measurements are available online at:

http://www.grdc.sr.unh.edu/index.html (UNH/GRDC Composite Runoff Fields, B.M. Fekete, C.J. Vorosmarty, W. Grabs; see “Station Data Explorer” on that site). These data were obtained from the Global Runoff Data Center, Federal Institute of Hydrology, Koblenz, Germany. They are monthly means for the period of record, and maximum and minimum values, only.

Within the U.S., Canada, Australia, and several other nations, daily mean discharge data are also available.

We use daily discharge and AMSR-E “C/M” data at co-located gaging reaches and gaging stations to test the streamflow and runoff estimation methodology. These sites allow accurate calibration of the calibration pixel/measurement pixel ratio to discharge. This information will also be used to produce confidence intervals for the discharge estimates.

Where in situ daily data are not available, there are two possible approaches to calibration of C/M values to discharge:

1) Attempt to model the flow from remote sensing of changing river widths, assumptions about channel bathymetry, and topographic data. Such an approach may produce discharge estimates accurate to about +/- 20% (Bjerklie D.M., Dingman S.L., Vorosmarty C.J., Bolster C.H., Congalton R.G.: Evaluating the potential for measuring river discharge from space, Journal of Hydrology, 278, 17-38, 2003).

2) Use existing streamflow data even when only monthly means are available and the periods of record are variable. For many of the gaging reaches, this is the best approach. Where flow is strongly seasonal, C/M ratios track the seasonal high and low flow states. Several years of C/M values (monthly mean values) are compared to the monthly mean at the station to produce a rating curve. The instrumental maximum and minimum discharges also provide useful additional calibration and are matched to the extremes so far measured by AMSR-E.

These are less precise calibrations compared to that possible with daily data, but still provide better information than that obtainable by modeling. Note that regression coefficient (r squared) values given in the rating curve plots range between .3 and .9. These do not, however, provide any information directly on the accuracy of the resulting rating equations. Low r-squared values are expected in some regions because of the mismatch of the periods of record and known interannual variability of streamflow.

Upcoming Corrections:

Where the calibrating discharge stations are long distances from the gaging reach, an adjustment to the monthly mean discharges based on the different contributing areas must be made. In early 2006, we have not yet accomplished such adjustments, and such will result in retroactive adjustments to our discharge and runoff measurements. We also hope to obtain additional ground-based data to further improve many of the rating equations.

Runoff Values:

Finally, monthly and annual runoff values are also now being computed. Because of the beneficial effects of averaging, and the high temporal density of the remote sensing data, these values are tracking very closely those measured by discharge stations. We hope soon to provide appropriate error estimates and confidence intervals for both AMSR-E estimates of daily discharge and for the monthly and annual runoff results.

Dartmouth Flood Observatory Homepage