P-ETSS Model - MDL

The Extra-Tropical Storm Surge (ETSS) model is developed and maintained by the National Weather Service’s (NWS) Meteorological Development Laboratory (MDL). It runs operationally four times a day to predict overland surge with tide guidance for extra-tropical storms along the U.S. East Coast, Gulf of Mexico, West Coast and all of the coasts of Alaska.

Like all forecasts, Storm surge guidance has various uncertainties associated with it such as (a) the atmospheric forcing (wind speed, wind direction and atmospheric pressure), (b) the initial water conditions, (c) the included physical processes, (d) the numerical scheme, etc.  While some of these can be reduced by enhancing the storm surge model, others, such as atmospheric forcing, rely on external inputs.  Uncertainty in atmospheric forcing is of particular importance as it is the main source of uncertainty in storm surge based inundation guidance.  Ensemble techniques combining atmospheric forcing and storm surge modeling are necessary to produce quantitative estimates of storm surge based inundation risk.

In 2017 (H. Liu and A. Taylor), MDL implemented one such ensemble technique in the form of the Probabilistic Extra-Tropical Storm Surge (P-ETSS) model.  P-ETSS uses the 21 ensemble members from the Global Ensemble Forecast System for atmospheric input to a storm surge and tide inundation model.  It then equally weights the resulting set of inundation guidance. More recently (H. Liu, A. Taylor and K. Kang, 2019), MDL enhanced P-ETSS by using the 42-member North American Ensemble Forecast System (NAEFS) instead of the 21‑member Global Ensemble Forecast System.

Since the inundation model does not currently account for components such as sea level rise, waves, river flooding and model error, a statistical post processing methodology similar to ETSS’ is used to enhance the overall guidance.  The results are then visualized via a website to facilitate user access. 

METHODOLOGY AND PRODUCTS

As the diagram shows, P-ETSS uses the 21 ensemble members from the GEFS as atmospheric input to the ETSS model.  In each grid cell, the surge + tide values from each ensemble run are sorted in ascending order.  The probability and exceedance products are created from this sorted set.

Specifically for the Probability of Surge + Tide greater than X products for X in {0, 1, 2, 3, 6, 7, 8, 9, 10, 13 and 16 feet above datum or ground level}, it counts in each grid cell the number (N) of ensemble runs greater than X.  The probability in that grid cell is then: N*100/21.  See Fig. 1 for an example where the probability of > 1 feet is 80%.

Fig. 1 is an example of using the number of values with surge + tide greater than a threshold to determine the probability

For the X% Exceedance Height products, for X in {10, 20, 30, 40, 50, 90}, it searches down the sorted list for the height (above datum or ground level) in the “correct” spot.  For example, the 10% exceedance with 21 members is the value that is matched or exceeded by 2.1 ensemble members, so we want the 2nd spot in the sorted list (Fig 2).  Thus the “correct” spots for {10, 20, 30, 40, 50, 90%} are {2nd, 4th, 6th, 8th, 10th, 19th} respectively.

The final P-ETSS product is the Ensemble Mean, Minimum and Maximum computed by taking the algebraic mean, minimum and maximum values in each grid cell respectively, in feet above datum or ground level.

Fig. 2 is an example of determining the 10% exceedance value from a specific element in the stored list of surge + tide values.