VERIFYING SUBMARINE GROUNDWATER FLOW DETECTION USING MATTSI (MODEL FOR ADVANCED THERMAL TIME-SERIES INVERSION)

Heat has been shown to be a reliable and relatively inexpensive tracer to determine groundwater flow velocity, but it can be complex and expensive in environments like the seafloor. Wilson et al. (2016) developed MATTSI, a thermal and inverse model that requires input of observed thermal profiles, to estimate the depth and extent of hydrodynamic exchange along with regional groundwater flow patterns. Sensitivity studies were limited because of the shallow depth of the thermal profiles. Now, we have tested the model on a far more extensive set of thermal observations based on a well field reaching depths of 1 - 3 m installed 5 - 15 km off the coast of Charleston, SC. Results from MATTSI were not particularly sensitive to errors in the depths of the thermal sensors, provided that the distances between the sensors were well known but were affected by some parameters. The tidal signal, for example, is much stronger in the nearshore (~5 km) thermal profiles and produces a noisy RMSE that is mitigated by smoothing the observations with a larger averaging window. Finally, new modifications to MATTSI provide initial guesses for groundwater velocities that vary with time.