Use of multilinear regression for prediction of PH and aragonite saturation in the northwest Gulf of Mexico

• Main Authors: Jundt, EvaLynn, Hu, Xinping, Barbero, Leticia
• Format: Conference Object
• Language: English
• Published: 2022
• Subjects: Carbon cycle; mlr; data science; Ocean acidification
• Online Access: https://hdl.handle.net/1969.6/90592

The Gulf of Mexico (GOM) is home to large shellfish populations, and the northernmost tropical coral reefs in the contiguous US. Despite this, the progression of Ocean Acidification (OA) in the GOM is still poorly understood as historical carbonate chemistry measurements are scarce. Based on carbonate chemistry and hydrographic data collected from 2007, 2017, and 2021, we derived multilinear regression models built upon relationships between commonly measured hydrographic properties (salinity, temperature, pressure, depth, or oxygen) and aragonite saturation state (Ωaragonite) and pH. The resulting models robustly predict Ωaragonite with R2>0.92, RMSE<0.07 and pH with R2>0.66, RMSE<0.30 for four different scenarios including different subsets of the data based on area and depth.