Publications

Shanableh, A; Al-Ruzouq, R; Gibril, MBA; Jena, R; Hammouri, N (2025). Predicting Potential Vertical Mixing in a Semi-Enclosed Gulf: Insights from Satellite-Derived Data. JOURNAL OF THE INDIAN SOCIETY OF REMOTE SENSING.

Abstract
Given the semi-enclosed characteristics of the Gulf (specifically referring to the Arabian/Persian Gulf, hereinafter termed the Gulf), coupled with its contiguous desert landscapes and limited circulation dynamics, this region manifests as a significant sedimentary basin, characterized mainly by carbonaceous deposits. Numerous studies suggest that the cooling of the water's surface and wind-driven forces in the Gulf lead to the mixing of its water column and the resuspension of calcium carbonate sediments on its surface. Satellite-based remote sensing excels in capturing surface-level data from various aquatic bodies but intrinsically lacks the capacity for sub-surface observations. This study analyzed five years of daily remotely sensed data, and a mixing index (MI) was introduced to identify potential mixing zones within the Gulf. The multitemporal dataset included daily MODIS satellite images, hourly surface wind speed (WS), sea surface temperature (SST), and bathymetry. The occurrence of mixing within the Gulf was inferred from the presence of whiting-suspended calcium carbonate (CaCO3) particles-detected in the satellite data by estimating particulate inorganic carbon (PIC). Furthermore, an explainable artificial intelligence approach utilizing a hybrid deep learning model was employed to predict water mixing in the Gulf, and the influence of the examined factors on the predictive accuracy of vertical mixing was highlighted. Results indicated moderate correlations among PIC, bathymetry, SST, and WS, with a notably strong association between the MI and PIC concentrations. All evaluated factors were crucial in differentiating mixing from non-mixing instances from remotely sensed data. By integrating all the factors, the hybrid deep learning model achieved an accuracy of approximately 85% in predicting the Gulf's mixing potential. The introduced MI transpires as the leading determinant for mixing predictions, whereas SST predominantly influences non-mixing outcomes. This study underscored the efficacy of leveraging diverse remotely sensed data to predict potential vertical mixing in the Gulf, setting a foundation for further studies in the Gulf and contributing to enhanced marine conservation and resource management strategies.

DOI:
10.1007/s12524-025-02141-y

ISSN:
0974-3006