Data Assimilation in a Shallow, Island Coastal Environment

Ivica Janekovic and Brian Powell
University of Hawaii at Manoa

Data assimilation techniques are well established in global and regional modeling systems with spatial resolution of order 10-100km and temporal of days. However, applying the methods in the near-shore coastal regions, usually resolved with high spatial resolutions (~100m) grids simulating complex and highly dynamical environments remains a novel approach. We present our ongoing research related to the Western Oahu regional model based on ROMS by using variational data assimilation with limited observations (CTD and ADCP). The Modeling system is characterized by a variable mesh grid resolution of 600m in deeper regimes and 200m near the coast. Validation of TLM approximation used in variational methods for a such region is evaluated revealing that initial perturbations are quickly dissipated by the boundary conditions. With dominating tidal circulation, adjustment to the boundary conditions requires careful attention to avoid removal of the tidal signal. We present preliminary results from the assimilation of ADCPs recovered at the end of February, 2010 in this highly nonlinear regime.