Real-Time Ocean Forecast System Adjoint Model

Comparing Data Observing Systems' Impacts on the RTOFS Model during Hurricane Franklin

 hurricane-franklin-err-sens-temp-140m.png

Figure 1: Plot illustrates the error change in RTOFS after temperature data assimilation at 140m. Negative values indicate a reduction in model error due to temperature data assimilation while positive values indicate an increase in error.

       The Real-Time Ocean Forecast System (RTOFS) adjoint model quantifies how the assimilation of each data observing system’s observations impacts the error in RTOFS model. The data observing systems in the analysis include, but are not limited to, satellite altimetry, gliders, Argo floats and expendable bathythermographs (XBTs). This project allows NOAA to evaluate the performance of each data observing system (DOS) in specific spatial distributions and timeframes. Adjoint model applications allow the Environmental Modeling Center (EMC) to predict the optimal spatial distributions of data observing systems, and empower EMC to tailor these distributions to individual incoming tropical cyclones.

 

hurricane-franklin-all-impacts-by-DOS

Figure 2: The left and right plot showcase the change in error by DOS when the temperature and salinity data are assimilated, respectively. Negatives values indicate that data assimilation reduced error in RTOFS.

 

The adjoint model is able to isolate regions in a lat/long box, enabling analysis in specific regions and at several spatial scales. The following focus will be on the Argo float analysis to demonstrate the adjoint model's basic capabilities.

hurricane-franklin-argo-spatial-temp-and-salt.png

Figure 3: The left and right plots are for temperature and salinity, respectively. Each plot showcases the Argo floats' spatial distribution in observations and each observation's beneficial impact on the RTOFS model. The term beneficial impact is used to illustrate that an observation reduced RTOFS error after assimilation. It is designated by cooler colors and negative values in this analysis.

hurricane-franklin-argo-all-depths.png

Figure 4: The left and right plot are for temperature and salinity, respectively. Each plot shows how RTOFS error reduction varies as a function of depth after Argo data assimilation.

 

For further information about the RTOFS adjoint model and its applications, contact:

Corbin Brooks

(301) 873-7882

corbin.brooks@noaa.gov

Top