Abstract's details

Ocean reanalysis skill during the S-MODE pilot campaign

Joseph D'Addezio (Naval Research Laboratory, United States)

CoAuthors

Gregg Jacobs (Naval Research Laboratory, United States); Brent Bartels (Peraton, United States); Christopher DeHaan (Peraton, United States); Bruce McKenzie (NAVOCEANO, United States); Matthew Kuhn (NAVOCEANO, United States); Chad Kramer (NAVOCEANO, United States)

Event: 2022 Ocean Surface Topography Science Team Meeting

Session: Science III: Mesoscale and sub-mesoscale oceanography

Presentation type: Type Poster

An altimeter and glider assimilating ocean model was recently used to support NASA’s S-MODE pilot campaign: a science project designed to quantify mesoscale to submesoscale processes off the San Francisco Bay. There are several goals we intend to reach. The first is to demonstrate ocean prediction is limited by present observations. We have shown in recent publications that the variability in high-resolution ocean forecasts is not constrained in the assimilation of presently available observations, which leads to large errors. The S-MODE campaign deployed 10 NAVOCEANO ocean gliders on Sep 1, 2021 and recovered them on Nov 9 2021. The gliders in the area were controlled by an automated system called the Global Heterogeneous Observations STrategies (GHOST). The system used the ocean forecast system assimilation analysis error variance to guide the gliders to locations of high expected model errors. These were typically where few satellite observations had been taken in recent days. With these gliders, we have a high-resolution in situ dataset that can resolve ocean features much smaller than resolved by the regular in situ observations and the satellite sea surface temperature (SST) and nadir altimeters.
A set of ocean forecast experiments is conducted in a hindcast. A 1 km horizontal resolution 100 vertical layer ocean model is nested successively inside the Global Ocean Forecast System (GOFS). A daily assimilation cycle uses observations to correct the ocean model state prior to making a subsequent forecast. In these experiments, observations beyond an analysis time do not influence the analysis, which simulates performance that would occur in a real forecast system. The first experiment utilizes all regular observations and withholds the data from S-MODE. This sets the baseline performance. The optimal settings for the system are based on prior sensitivity tests relative to extensive observing systems deployed in the Gulf of Mexico. The most influential components found through these studies were the estimation of background error variance and the horizontal length scale that prescribes the distance an observation will influence the correction to the model.
The second experiment includes the glider temperature and salinity profile data using the same parameter settings as the first experiment. We evaluate results in terms the scales at which the experiments have skill. We define the scale based on the point where the errors of the model relative to the glider data are less than the variance at the scale. This is found to be about 275 km wavelength. The use of the glider data in the experiment leads to some improved prediction, though the advancement is small. The reason is the scales in the assimilation process did not adapt to the higher density in situ data.
The third experiment adapts the horizontal scales to correct features resolved by the in situ observations. In the vicinity of the dense glider deployment, the separation geometry of the gliders defines the resolved scales, which is about 40 km. The decorrelation scale in the assimilation process is specified as a spatially varying field. The values used in the first two experiments are used throughout the domain with the values in the vicinity of the dense glider data reduced to those resolved. The results indicate a greater skill in the scale of features constrained in the ocean forecast.
A second goal is to demonstrate the accuracy with which we can presently predict the ocean in the presence of a high density in situ observing system that will be important during the SWOT cal/val period. The concurrent model runs and observation gathering has proven instrumental in preparing for future altimeter missions, such as SWOT.
 

Poster show times:

Room Start Date End Date
Mezzanine Tue, Nov 01 2022,17:15 Tue, Nov 01 2022,18:15
Mezzanine Thu, Nov 03 2022,14:00 Thu, Nov 03 2022,15:45
Joseph D'Addezio
Naval Research Laboratory
United States
joseph.daddezio@nrlssc.navy.mil