1. Effects of Fluvial Morphology on Remote Sensing Measurements of Discharge American Geophysical Union, Fall Meeting, 2010 Session on “H66, The Remote Sensing of Rivers” G. R. Brakenridge 1, A.J. Kettner 1, I. Overeem 1, S.V. Nghiem 2, T. De Groeve 3, J.P. Syvitski 1 1 Dartmouth Flood Observatory and the CSDMS Facility, University of Colorado 2 Jet Propulsion Laboratory, California Institute of Technology 3 Joint Research Centre of the European Commission

2. Can Available Orbital Remote Sensing provide useful measurements of global river discharge variability? The Advanced Microwave Scanning Radiometer - Earth Observing System (AMSR-E) is a twelve- channel, six-frequency, passive-microwave radiometer. It measures horizontally and vertically polarized brightness temperatures, including at 36.5 GHz. AMSR-E was developed by the Japan Aerospace Exploration Agency (JAXA) and launched by the U.S. aboard Aqua in mid-2002. Monitoring of River Discharge Changes: A Passive Microwave Approach

3. From USGS field data For this gauging station From Smith et al, 1996, WRR. Effective width = flow area/river length

5. From Gage (4.5 ft rise) To Discharge (via a rating curve) There is a nearly 4x increase in discharge, over a period Of 9 days.

6. One day of AMSR-E data collection (high latitudes revisited most frequently)

7. To measure surface water change: Traditional Method : Polarization Ratio PR=(T bV  T bH )/(T bV +T bH ) C M M is the Measurement cell containing the river C is the Calibration cell with no river New Method : Pair Ratio HR=T bH (C)/T bH (M) VR=T bV (C)/T bV (M)

8. Emission Model Analysis Pair Ratio H-pol: HR=T bH (C)/T bH (M) Silt-loam soil: 27% clay, 62% silt Water fraction: f w =10-90% Soil moisture %: Rain in M, not C m v =10 M, 5 C (thin) m v =20 M, 5 C ( bold ) Rain in M and C m v =10-35 ( medium ) in both M and C 36.5 GHz

9. Use H polarization at 36.5 GHz (6.92/10.65 RFI, 18.7/28.3 GHz water line, 89 GHz oxygen line) C is near M to be within correlation length scale of physical temperature T C and M are located to minimize differences in data acquisition time, and over similar ground cover types. For calibration to be unaffected by river, C is chosen at location outside of river reach M M is selected to conform with river reach to maximize surface water change Use rating curve to calibrate HR (the paired ratio estimator) to stream flow Protocol for River Flow Measurement

10. T b dry wet T b dry wet Dry pixel Wet pixel Influence of other factors (clouds, ground temperature, etc) is much reduced by comparing dry and wet signal Water has a lower brightness temperature than land 1 2 3 123123 flood signal

11. Example: Wabash River near Mount Carmel, Illinois, USA Black square shows Measurement pixel (blue line in next plot) White square is calibration pixel (green line in next plot) Dark blue colors: Flooding mapped by MODIS

12. Scatter Plot of AMSR-E HR ratio (calibration site/measurement site) versus Wabash River discharge. Plot is uncorrected for any lag times. As discharge rises above bankful state, HR increased from 1.05 to 1.1.

13. Wabash River in U.S. Near New Harmony, Indiana Brown = estimated discharge from AMSR-E data Black= USGS in-situ stream gauge data Green = brightness temperature of calibration site Blue= brightness temperature of river measurement site.

14. Portion of the estimated discharge time series (black) compared to local gauging station discharge (blue line). Satellite-estimated discharge lags gauging station discharge..

15. Estimated discharge time series (black) compared to local gauging station discharge (blue line), with a lag applied.

16. No time lag. With time lag.

17. Flood in March 2006 along the Wabash River recorded by the gauging station (lower plot, bold black line) and by HR from AMSR-E data (lower plot, thin black line). Due to inundation dynamics, peak discharge was reached at the gauging station 2–3 days before peak discharge as inferred from the AMSRE. From Brakenridge et al (2007) WRR.

18. River Watch Site 496, White River near Augusta. Channel is 170 m in width, with sinuous meanders. Circle is area of river measurement reach.

19. Green line is upwelling microwave radiance within 5 km radius of a calibration land target Blue line is radiance within 5 km radius of a river measurement site target Brown line is their ratio and is used to estimate discharge via a rating equation.

20. Preliminary rating curve for this site. This equation used to estimate discharge, using remote sensing data (in black, above)

21. Monthly Runoff, calculated from daily mean discharge, ground station Monthly Runoff, calculated from AMSR-E data

22. Annual Runoff (mm) 60 274 298 351 169 367 549 608 482 Annual Runoff (mm) 118 307 299 365 151 384 702 633 440 Yearly Runoff, calculated from daily mean gauging station discharge Yearly Runoff, calculated from AMSR-E data 2002 2003 2004 2005 2006 2007 2008 2009 2010 2002 2003 2004 2005 2006 2007 2008 2009 2010

23. Measured rainfall (blue), modeled river Q (green), gauged Q (red) and AMSR-E Q (orange) for the Okavango R, Botswana. AMSR-E calibrated to discharge by a simple linear equation. An improved rating equation would correct the over- estimation of low flow

24. Can Available Orbital Remote Sensing provide useful measurements of global river discharge variability? Constraints 1.Data are less precise than in-situ gauging. 2.Conversion of the estimator to actual discharge requires ground-based data. Benefits 1.The signal alone, without any calibration to discharge, still allows prompt characterization of anomalies (floods and droughts). 2.Adequate calibration to actual discharge has been obtained at many sites via comparison of monthly mean, maximum and minimum values: so, existing global runoff data can be used, now, to calibrate a global array of sites. 3.There is clear potential for an accurate global river measurement system that can be updated in real time, and, via SSMI microwave data, extended back into the 1980s.

25. Location of river gauging stations on the grounds, with various data archived at the Global Runoff Data Centre. Periods of record, type of information varies widely. Many usage restrictions apply. Very little real time data are available on international basis.

26. Location of presently measured sites using AMSR-E data. Data can be updated daily; are consistent through time, extend back to mid-2002, could be extended to mid 1980s using DMSP/SSMI 36 GHz data. NASA’s GPM will provide abundant new data forward in time.

27. Why not use what is available, from ground based data, in order to calibrate, produce rating curves, for satellite-based river measurement sites? This is optimum use of available data.

28. October 12, 2010 River Watch Measurement Sites Red: Purple: Blue: Yellow: Major flooding, > 5yr recurrence Moderate flooding, >1.3 yr recurrence Normal flow Low flow or ice-covered

29. November 7, 2010

30. November 17, 2010

31. December 8, 2010