sUAS Multispectral Remote Sensing to Evaluate Spatial and Temporal Distribution of Submerged Macrophyte Coverage for Silver Bow Creek in Southwest Montana
A thesis research paper submitted in partial fulfillment of the requirements for the degree of Masters of Science in Environmental Engineering.
Abstract
Macrophytes play a key role in aquatic systems as filters for detritus and nutrients and source of food and habitat for aquatic organisms; however, become overabundant due to excessive nutrient loading, as is the case in Silver Bow Creek, Butte, Montana. An overabundance of macrophytes can be detrimental, leading to large diurnal variation in dissolved oxygen (DO) and hindering recreational activities. Because of the importance of macrophytes in aquatic ecosystems, recent studies have focused on the efficacy of remote sensing techniques for mapping macrophytes in lakes, estuaries, and coastlines. Little research has been done on the application of remote sensing for mapping macrophytes in streams. The goal of this study was to collect the spatial and temporal data necessary to help determine the effect of macrophytes on DO concentrations using remote sensing techniques and determine if SBC is on a recovering trend in terms of DO concentrations. To determine diurnal and long-term trends in DO concentrations, multiparameter sondes were deployed at six locations along Silver Bow Creek throughout the 2021 macrophyte growing season (May-December). Along with this, water samples were collected for a one-day-period during baseflow conditions to capture the diel trend in nutrients, alkalinity, cBOD, and detritus. To collect spatial and temporal macrophyte coverage, nearly monthly (June-September) sUAS remote sensing campaigns were carried out upstream of DO monitoring locations during the peak of the growing season. High resolution multispectral images were collected using a small unmanned aerial system (sUAS) and classified using ArcGIS Pro Support Vector Machine (SVM) classifier. Correlation of the observed DO diurnal variations with macrophyte coverage upstream of the sonde location suggested the heavy influence of macrophytes on DO concentrations. In performing supervised classification, the spectral signature of macrophytes was evaluated. This study found that wavelengths approaching the near infrared, from 705 nm to 842 nm, showed consistent variation in signature profiles for submerged macrophytes and water. The best distinction could be made at the 717 nm wavelength. This determination assisted with the ArcGIS classifications. ArcGIS supervised classification of macrophytes produced overall accuracies ranging from 50% to 88% with an average of 76%. Kappa coefficients ranged from 0.20 to 0.80 with an average of 0.53. Only fair agreement was achieved for the classification of the orthomosaics for Whiskey Gulch, Beaver Dam, and Opportunity during June and July when macrophyte coverage was low. Agreement was best achieved for the August and September datasets when macrophyte coverage was higher. Monthly macrophyte coverage was compared with maximum DO diurnal variation during that month to demonstrate correlation. The percent macrophyte coverage of each classified orthomosaic is plotted against the maximum diurnal variation in DO which occurred during that month in the corresponding reach. This produced an R2 value of 0.61, showing strong correlation between macrophyte growth and maximum DO diurnal variation.