loading page

Method Development for Remote Sensing of River Flow with Limited Ground-Based Measurements
  • Mackenzie Martin,
  • Kathleen May Glancey,
  • David M Kahler
Mackenzie Martin
Duquesne University

Corresponding Author:[email protected]

Author Profile
Kathleen May Glancey
Duquesne University
Author Profile
David M Kahler
Duquesne University
Author Profile

Abstract

Quantification of river flow is a significant component of water resource management needed to develop infrastructure, prepare for drought or flood, and ensure equitable resource use. Yet, these data have often been deficient in rural regions generally dependent on smaller river systems, especially in low- and middle- income countries, and in places where no remote gaging capabilities currently exist. A relationship between river flow and width has been previously demonstrated in attempts to record and monitor flow with only satellite-based measurements; however, those methods have focused on rivers with a width of > 30m and have shown extreme errors when compared to historical gaging records. A new technique for monitoring river flow remotely is presented here, which uses Manning’s equation for open channel flow coupled with width measurements from satellites. This technique allows for the creation of a remotely-sensed historical river flow record and uses the minimum number of ground-based measurements for the calibration of Manning’s equation. Cross-river depth profile, discharge, and slope are necessary for initial calibration. This method used four-band satellite images with 3m-resolution for width measurements, with images being adjusted for the Normalized Difference Water Index (NDWI) for better delineation of the water’s edge. Buffalo Creek, in Freeport, Pennsylvania and the Mutale River, in the Limpopo Province of South Africa, were used as study locations. An estimation of flow at each site was obtained using every available satellite image of those transects. Preliminary results indicated successful calibration of Manning’s equation for the individual river transects, and verification of this method yielded flow estimation with an average of 74% error at Buffalo Creek and 48% error at the Mutale River, when compared to historical data from gages present near the study sites. The primary challenge of using this method on smaller rivers is the determination of width from the satellite images. Automated techniques to delineate the water’s edge are presented as recommendations to increase accuracy. This method may be used as an additional tool in river flow monitoring for water resource management.