When coastal researchers, supported by the NOAA Climate Program Office Climate Observations and Monitoring (COM) Program, developed extreme sea level indicators for the contiguous U.S. coast, they knew their work was only the first step in improving coastal flood risk assessment and management. These indicators, presented in Nature Scientific Data, describe variations in observed extreme sea levels. University of Central Florida researchers Mamunur Rashid and Thomas Wahl are now working to take the next step—improving predictive skill of future sea level. In their new COM-funded study, they developed and evaluated models built to predict extreme changes in sea level by linking storm surges to large-scale climate variability. The study was recently published in the Journal of Geophysical Research: Oceans.
Specifically, Rashid and Wahl ask how well can we predict the storm surge climatology (SSC) indicators developed in their previous study. SSC indicators are measures that quantitatively reflect the role different major weather and ocean forces have played in affecting extreme sea levels in coastal areas around the country—independent of mean sea level rise. While change in mean sea level rise is most often the phenomenon considered in coastal flood risk, long-term changes and variations in the storm surge activity can also have significant implications for coastal adaptation efforts. Therefore, developing models to predict SSC variability is an important factor for coastal management plans.
In their study, Rashid and Wahl set up three modeling experiments, each using a different set of climate indices—known as “predictor data”— as the basis of their model predictions to assess how well each model can replicate past storm surge variability along the contiguous U.S. coastline from the first half of the 20th century until now.
Why use three different predictor data sets? Each unique data set tells Rashid and Wahl something new about their prediction models. First, the use of traditional climate indices, which can be thought of as “off-the-shelf” data related to variables like precipitation and global temperature, allows them to see if their basic method for predicting SSC variability works. Second, using tailored climate indices derived from NOAA sea level pressure (SLP) and sea surface temperature (SST) reanalysis data enables researchers to see what improvements, if any, tailored indices bring to predictability. Finally, tailored indices, derived from decadal model simulations of SLP and SST (NOAA’s GFDL-CM21), tell us if output from these simulations can be used to predict observed SSC variability. If the data from decadal climate model simulations can successfully predict the behavior of observed SSC indicators, then they could also be used to forecast future storm surge behavior.
Investigating changes over time, however, comes with a major complication; the models’ climate data represent phenomena occurring at very different time scales. To account for this, Rashid and Wahl leveraged a powerful mathematical tool known as discrete wavelet transform (DWT) to split their predictor data sets into low frequency (longer timescales) and high frequency (shorter timescales) sub-series. Previously, the relationship between SSC indicators and predictor variables has often appeared weak and insignificant because the relevant phenomena occur at such different time scales. By using DWT, Rashid and Wahl were able to reveal much stronger relationships and pinpoint significant predictor variables to use in their models.
The results of the study show that all three model types can reproduce the overall extreme sea level variations observed in certain coastal regions. The models using tailored climate indices do show improved prediction skill. Rashid and Wahl, however, caution that their work is far from complete. They emphasize that further study of the physical mechanisms connecting SSC variability and large-scale processes is necessary. In addition, they indicate that modeling frameworks must be refined before they can be used for the operational forecasting of storm surge variability.