Meetings: Documents

Impact of Satellite Sea Surface Salinity Observations on ENSO Predictions from the GMAO S2S Forecast System
[14-Dec-2018] Hackert, E.C., Kovach, R.M., Marshak, J., Borovikov, A., Molod, A., and Vernieres, G.
Presented at the 2018 AGU Fall Meeting
El Niño/Southern Oscillation (ENSO) has far reaching global climatic impacts and so extending useful ENSO forecasts would be of great benefit for society. However, one key variable that has yet to be fully exploited within coupled forecast systems is accurate estimation of near-surface ocean density. Satellite Sea surface salinity (SSS), combined with temperature, help to identify ocean density changes and associated mixing near the ocean surface. We assess the impact of satellite SSS observations for improving near-surface dynamics within ocean analyses and how these impact dynamical ENSO forecasts using the NASA GMAO Sub-seasonal to Seasonal (S2S Version 2) coupled forecast system (Molod et al. 2018 - i.e. NASA's contribution to the NMME project). For all initialization experiments, all available along-track absolute dynamic topography and in situ observations are assimilated using the LETKF scheme similar to Penny et al., 2013. A separate reanalysis additionally assimilates Aquarius V5 (September 2011 to June 2015) and SMAP V4 (March 2015 to present) along-track SSS data.
We highlight the impact of satellite SSS on ocean reanalyses by comparing validation statistics of experiments that assimilate SSS versus our current prediction system that withholds SSS. We find that near-surface validation versus observed statistics for salinity are slightly degraded when assimilating SSS. This is an expected result due to known biases between SSS (measured by the satellite at ~1 cm) and in situ measurements (typically measured by Argo floats at 3 m). On the other hand, a very encouraging result is that temperature, absolute dynamic topography, and mixed layer statistics are improved with SSS assimilation.
Previous work has shown that correcting near-surface density structure via gridded SSS assimilation can improve coupled forecasts. Here we present results of coupled forecasts that are initialized from the GMAO S2S reanalyses that assimilates/withholds along-track (L2) SSS. In particular, we contrast forecasts of the overestimated 2014 El Niño, the very strong 2015 El Niño, and the weak 2016 La Niña. For each of these ENSO scenarios, assimilation of satellite SSS improves the forecast validation. Improved SSS and density improves the mixed layer depth leading to more accurate coupled air/sea interaction.