Impact of simulated SWOT Observations in a Multi-Scale Assimilation

Innocent Souopgui1, Joseph Matthew D'Addezio2, Clark David Rowley3, Scott R Smith4, Gregg A. Jacobs5, Robert William Helber2 and Max Yaremchuk6, (1)University of New Orleans, New Orleans, United States, (2)U.S. Naval Research Laboratory, Ocean Dynamics and Prediction, Stennis Space Center, United States, (3)Naval Research Laboratory, Oceanography, Stennis Space Center, MS, United States, (4)Naval Research Lab Stennis Space Center, Stennis Space Center, United States, (5)US Naval Research Laboratory, Ocean Sciences Division, Stennis Space Center, MS, United States, (6)Naval Research Lab, Stennis Space Center, MS, United States
Abstract:
This research builds on an existing a multi-scale, data assimilative, high-resolution ocean forecasting system that has demonstrated successful assimilation of simulated Surface Water Ocean Topography (SWOT) observations using a two-step 3DVAR analysis procedure. For this presentation, we examine the impact of the second step and the combination of observations going into that second step, in the multi-scale, data assimilation system. The first analysis step seeks to correct the large-scale, while the second step seeks to correct for the smaller-scale features present in the SWOT observations. An Observing System Simulation Experiment (OSSE) is used to explicitly calculate errors produced by competing single- and multi-scale analysis procedures. The first analysis step is the standard exploitation of observations with a 5-day data analysis period, horizontal correlation scales, vertical structure, and background errors consistent with mesoscale corrections. Experimentation with the observation window length of the second analysis step shows that a shorter time window produces lower analysis errors. This is consistent with the larger mesoscale structures having a longer time period and submesoscale structures having short Eulerian times being advected by the mesoscale. Therefore, a 24-hour observation window with a first guess at appropriate time (FGAT) for the second step was selected and sequential analysis/forecast cycles were performed for an entire year. The multi-scale analysis with SWOT observation in both steps produces less overall error than any other combination when analyzing both area-averaged errors and wavenumber spectral analysis. Therefore, the multi-scale assimilation is essential for most effectively utilizing the forthcoming SWOT observations, which resolve features across a much wider spectrum of horizontal scales than are observed by the current nadir altimeters.