POSTER Session 2

BUILDING A NEW REGIONAL ALGORITHM TO ESTIMATE TOTAL SUSPENDED SOLIDS IN THE SOUTH ATLANTIC BIGHT

Total Suspended Solids (TSS) play a vital role in water quality and turbidity of inland and coastal waters. Existing TSS algorithms are often only regional, or if they are global, do not always perform well in all regions. As a preliminary study, four widely-used algorithms were tested and applied to the South Atlantic Bight (SAB) showing equally poor results in this region. This study aims at building a new SAB regional algorithm and responding to two research questions: (a) Does our SAB TSS algorithm improve by using hyperspectral remote sensing reflectance (Rrs) data?, (b) Does algorithm performance improve by incorporating geolocation and temporal information? Two in situ hyperspectral remote sensing reflectance (Rrs) SAB datasets were used for algorithm development and validation: 1. The “GLObal Reflectance community dataset for Imaging and optical sensing of Aquatic environments” (GLORIA, n = 465), and 2. Dr. Rivero-Calle’s Bio-Optical and Satellite Oceanography Lab (n = 70) dataset. To build the algorithm, Least Absolute Shrinkage and Selection Operator (LASSO) and Random Forest Regression (RF), Generalized Linear Models (GLM), and Generalized Additive Models (GAM) were tested. With respect to Question 1, incorporating the full hyperspectral Rrs data showed no significant improvement in LASSO or RF approaches due to the high correlation (r > 0.99) between the Rrs bands, GLM and GAM cannot handle 402 bands. In terms of the second question, incorporating geolocation and temporal information did not significantly improve algorithm performance. Future work includes increasing the number of in situ observations and applying Machine Learning methods.

*Md Masud-Ul-Alam, University of Georgia, USA, https://orcid.org/0000-0003-3036-8167

Sara Rivero-Calle, University of Georgia, USA

Code of Conduct
Privacy Policy