On model validation for meso/submesoscale currents: Metrics and application to ROMS off Central California

ROMS with horizontal grid spacing of 3.5 km for the region off Central California was compared to RAFOS float observations and satellite altimetry on meso/submesoscales. The approach introduced and used two new metrics for model-data comparison, as well as suggested how to calculate these metrics for different spatio-temporal scales. The first metric consisted of the first two moments of exit time and was used to compare ROMS against RAFOS float observations at mid-depths (between 300 m and 350 m). Exit time is the time a float launched at a point takes to leave a domain for the first time. The second metric was spectral entropy and was used to estimate how well ROMS reproduced variability of the sea surface height (SSH) anomaly field extracted from an AVISO data set (1992-2007) for specified temporal and spatial scales. Calculations showed that ROMS reproduced the mid-depth mesoscale/submesoscale currents next to the coast in a very accurate manner (low-order exit time statistics of floats were reproduced by ROMS with an accuracy better than 95%); but ROMS overestimated the speed of westward drift of floats by as much as 20-30% at distances greater than 350 km from the coastline. ROMS predicted the variability of the mesoscale (100-400 km) SSH anomaly field for temporal scales of 1-12 months with a reasonable accuracy. A wavelet transform modulus maxima technique applied to the spectral entropy of SSH anomaly also demonstrated good agreement between ROMS and satellite altimetry for mesoscales characterized by singular exponents and multi-fractal spectra for 1-12 month time scales.