Remote Sensing (Dec 2023)
Remote Sensing Analysis for Vegetation Assessment of a Large-Scale Constructed Wetland Treating Produced Water Polluted with Oil Hydrocarbons
Abstract
The identification and assessment of plant stress using wetland satellite images is a major task in remote sensing. In this study, one of the largest constructed wetlands (CWs) in the world, located in the Sultanate of Oman, was examined, assessed, and evaluated using remote sensor data from Sentinel-2. This CW system treats produced water generated during oil exploration activities in a desert environment; thus, CW vegetation is subjected to stress induced by oil hydrocarbons and water salinity. This study examined the plant stress and detected changes between the years of 2017 and 2019. Sentinel satellite images were evaluated for vegetation status extraction. The Normalized Difference Vegetation Index (NDVI), Modified Soil-Adjusted Vegetation Index (MSAVI), and Normalized Difference Salinity Index (NDSI) were used to evaluate the vegetation change. The results showed a comprehensive mapping identification of the plant stress and water flow parameter factors including oil in water contamination (OIW), dissolved oxygen (DO), water temperature (WT), and water conductivity (COND). Among the three indices, it was found that the NDVI showed a very good correlation with all parameters in both years with average R2 = 0.78, 0.67, 0.75, and 0.60 for OIW, DO, WT, and COND, respectively. The same trend was found for MSAVI but with R2 = 0.59, 0.48, 0.55, and 0.56 for OIW, DO, WT, and COND, respectively. This shows that the NDVI performed better than the MSAVI in evaluating the water flow parameters. On the other hand, the NDSI showed a strong correlation with one flow parameter, that is, water conductivity, especially at the outlet cells of the CW with R2 = 0.86 and 0.82 for winter time and summer time, respectively. The synchronization and correlation between the water flow parameters and remote sensing vegetation indices in this study lead to a new approach to large-scale landscape wetland monitoring that improves and helps predict any degradation or stress on vegetation growth. Furthermore, the results of this work can help decision makers potentially modify the wetland design and water flow path to improve future expansion phases. The mapping of such a critical and massive industrial CW should consider the use of high spatial resolution sensors where identifications and classifications are further improved. In summary, this research demonstrates that it is feasible to estimate vegetation stress within the constructed wetland using remote sensing techniques across extensive regions when an ample dataset comprising field data, satellite imagery, and supporting information is accessible.
Keywords
