Monitoring Multi-Depth Suspended Sediment Loads in Lake Erie's Maumee River Using Landsat 8 and Unmanned Aerial Vehicle (UAV) Imagery

Monitoring Multi-Depth Suspended Sediment Loads in Lake Erie's Maumee River Using Landsat 8 and Unmanned Aerial Vehicle (UAV) Imagery
Author: Matthew David Larson
Publisher:
Total Pages: 292
Release: 2017
Genre: Remote sensing
ISBN:

Download Monitoring Multi-Depth Suspended Sediment Loads in Lake Erie's Maumee River Using Landsat 8 and Unmanned Aerial Vehicle (UAV) Imagery Book in PDF, Epub and Kindle

Suspended sediment in water bodies is a considerable environmental concern. Traditional sampling methods for suspended sediment are time-consuming as they involve vertical and spatial point-sampling. Remote sensing (RS) is an alternative to in-situ measurements and it is capable of monitoring suspended sediments in shallow waters spatially at large scales. Use of RS technology to map suspended sediment concentrations (SSC) depends on sensor type and its capability `to see through' the water column at given surface and water column conditions. This study examined the capabilities of RS technology to spatially quantify SSC at multi-depth intervals within the Maumee River, Ohio. Water samples were collected and analyzed for SSC in May, June, and October at depths of 0.5 ft., 2 ft., 3 ft., and 6 ft. Landsat 8, surface hyperspectral measurements (aggregated to simulate sensors), and MicaSense Sequoia camera onboard an unmanned aerial vehicle (UAV) were used. Single spectral bands, ratios, and multiple bands/ratios were examined in developing algorithms relating RS and field measurements. Linear regression models provided the best relationship for surface, Landsat 8, and UAV data throughout all depths. A 6 ft. depth had the highest correlation for surface (R2adj=0.93) and Landsat 8 (R2adj=0.79) data. For UAV a 3 ft. depth provided the best relationship (R2adj=0.52). Band ratios using nonlinear fitting provided good relationships (surface R2adj=0.72 and Landsat 8 R2adj=0.54) at 6 ft. as well. Results showed Landsat 8 more accurately measured suspended solids at 6 ft. than shallower depths. Regression equations and band ratios showed increasing relationships with SSC with increasing depth for Landsat 8 with an exception for 3 ft., which can occur due to stratification. UAV measurements produced best results for 3 ft. Algorithms with best results included ultra blue, blue, and green bands which are not typically used for quantifying SSC. Shorter wavelength bands (400 nm-550 nm) should be considered in waters with small suspended sediments as those found in the Maumee River. Equations were not transferable from one day to another. It is surmised that concentration thresholds of 40-60 mg/L play a role in equation derivation, as well as meteorological factors.