Thermistor Chain Data Assimilation to Improve Hydrodynamic Modeling Skill in Stratified Lakes and Reservoirs

Abstract

Results from a three-dimensional hydrodynamic model of a stratified lake show that the computed structure of the pycnocline changed rapidly due to numerical diffusion, thus altering the vertical mixing dynamics and introducing a positive feedback that quickly drives model predictions off course. To negate the numerical diffusion a pycnocline filtering method is proposed that assimilates high-resolution thermistor chain data and adaptively adjusts to minimize the discrepancy between observed and computed temperatures. The adaptive pycnocline filter ensures that the computed temperature gradients in the metalimnion at the position of the thermistor chain remain within the bounds of the measured values so the computation preserves the spectrum of internal wave motions that trigger diapycnal mixing events in the deeper reaches of the lake.