Developing a high-resolution climatology for the Central California coastal region

Abstract

This work presents a procedure for developing a high-resolution, regional climatology estimate, named RClimo, off the coast of central California. This high-resolution climatology may provide an alternative way to initialize numerical nowcast/forecast exercises in coastal regions. The methodology includes two primary steps: (1) averaging available data on a high-resolution grid and (2) objective interpolating the resulting average profiles onto a regular grid. The first step involves the computation of averages over density layers in the vertical and allowing for data gaps in the horizontal if data are unavailable at a high resolution. The OA in the second step uses anisotropic correlation length scales derived from the data themselves and an averaging radius to preserve the scales and variability of the synoptic fields. The dataset used to compute this climatology includes the archived CalCOFI dataset, the Autonomous Ocean Sampling Network (AOSN) 2003 experiments near Monterey Bay, and many other previously undocumented profiles from various sources. As part of the climatology product, associated uncertainty is also generated through density averaging and employing the Gauss–Markov minimum error variance during Objective Analysis. The final climatology estimate is hence subject to greater error for larger mapping-grid size or lower data density, suggesting uncertainties that vary in space. The maximum value of the resulting error distribution for the RClimo estimate from the 50-km bins is less than 11% of the temperature estimate and 1% of the salinity estimate, whereas those for the 20-km-bin RClimo are 11% and 0.9%, respectively. We have conducted comparisons between the RClimo and 1/4° Levitus climatology fields via numerical simulations initialized with each field. Simulations were performed using Regional Ocean Modeling System (ROMS) and for the month of August, a peak period of upwelling-favorable wind, with and without the atmospheric forcing. The RClimo simulations produced reasonably realistic features like filaments, fronts, meanders, eddies, alongshore currents (∼50 cm s −1) and inshore undercurrents (∼15 cm s −1) under typical upwelling-favorable winds. In contrast, the Levitus simulations showed insignificant evolution over most of the domain, with or without winds. In the future, one could apply this methodology for other months and seasons in Monterey Bay, as well as develop similar high-resolution climatology for other coastal regions.