Turbidity in Apalachicola Bay, Florida from Landsat 5 TM and Field Data: Seasonal Patterns and Response to Extreme Events

• Published in: Remote sensing (Basel, Switzerland) Vol. 9; no. 4; p. 367
• Main Authors: Joshi, Ishan D., D’Sa, Eurico J., Osburn, Christopher L., Bianchi, Thomas S.
• Format: Journal Article
• Language: English
• Published: Basel MDPI AG 01.04.2017
• Subjects: Archives & records; Atmospheric models; Classification; Data acquisition; Data mining; Diurnal; Estuaries; Human settlements; Landsat; Landsat satellites; Monitoring; Principal components analysis; Rainfall; Reflectance; Remote sensing; River discharge; River flow; River mouth; Satellite imagery; Seasonal variations; Seasons; Shallow water; Statistical analysis; Statistics; Turbidity; Water quality; Wind speed; Florida; United States > US; Apalachicola Bay
• ISSN: 2072-4292; 2072-4292
• DOI: 10.3390/rs9040367

Synoptic monitoring of estuaries, some of the most bio-diverse and productive environments on Earth, is essential to study small-scale water dynamics and its role on spatiotemporal variation in water quality important to indigenous marine species and surrounding human settlements. We present a detailed study of turbidity, an optical index of water quality, in Apalachicola Bay, Florida (USA) using historical in situ measurements and Landsat 5 TM data archive acquired from 2004 to 2011. Data mining techniques such as time-series decomposition, principal component analysis, and classification tree-based models were utilized to decipher time-series for examining variations in physical forcings, and their effects on diurnal and seasonal variability in turbidity in Apalachicola Bay. Statistical analysis showed that the bay is highly dynamic in nature, both diurnally and seasonally, and its water quality (e.g., turbidity) is largely driven by interactions of different physical forcings such as river discharge, wind speed, tides, and precipitation. River discharge and wind speed are the most influential forcings on the eastern side of river mouth, whereas all physical forcings were relatively important to the western side close to the major inlet, the West Pass. A bootstrap-optimized and atmospheric-corrected single-band empirical relationship (Turbidity (NTU) = 6568.23 × (Reflectance (Band 3))1.95; R2 = 0.77 ± 0.06, range = 0.50-0.91, N = 50) is proposed with seasonal thresholds for its application in various seasons. The validation of this relationship yielded R2 = 0.70 ± 0.15 (range = −0.96-0.97; N = 38; RMSE = 7.78 ± 2.59 NTU; Bias (%) = −8.70 ± 11.48). Complex interactions of physical forcings and their effects on water dynamics have been discussed in detail using Landsat 5 TM-based turbidity maps during major events between 2004 and 2011. Promising results of the single-band turbidity algorithm with Landsat 8 OLI imagery suggest its potential for long-term monitoring of water turbidity in a shallow water estuary such as Apalachicola Bay.