1 Storm Surge Modeling in Puerto Rico in Support of Emergency Response, Risk Assessment, Coastal Planning and Climate Change Analysis Report prepared for the Caribbean Coastal Ocean Observing System (CariCOOS)/NOAA University of Puerto Rico/Mayagüez, P.R. and Puerto Rico Coastal Zone Management Program Department of Natural and environmental Resources by Jose Benítez (Ph.D. candidate) and Aurelio Mercado Irizarry (Professor) Department of Marine Sciences/University of Puerto Rico/Mayaguez July 2015 2 TABLE OF CONTENTS Content Page Sponsors ……………………………………………………………………………………………………………… 3 Methodology ………………………………………………………………………………………………………. 4 Computer Models Used …………………………………………………………………………… 4 Hurricane Wind Model Used ……………………………………………………………………. 4 Hurricane Headings Used …………………………………………………………………………. 6 Computational Unstructured Mesh …………………………………………………………. 13 Domain Decomposition …………………………………………………………………………… 18 Frictional Dissipation ……………………………………………………………………………….. 19 Steric Effects on Sea Level in Puerto Rico …………………………………………………. 25 Astronomical Tide Validation ………………………………………………………………………………. 26 Validation Against Storm Surges ………………………………………………………………………….. 26 Results (Maps of Synthetic Hurricanes) ……………………………………………………………….. 41 Caveats ………………………………………………………………………………………………………………... 57 Acknowledgments ……………………………………………………………………………………………….. 58 References …………………………………………………………………………………………………………… 58 Appendix 1 …………………………………………………………………………………………………………… 59 Appendix 2 ……………………………………………………………………………………………………………. 60 TABLES Table Page 1 Hurricane statistics used for the five Saffir-Simpson categories 9 2 Manning coefficient classification of the National Land Cover Class shown In Figure 21 20 3 Manning coefficient classification of the benthic habitats shown in Figure 22 21 3 SPONSORS: • The Caribbean Coastal Ocean Observing System (CariCOOS), a NOAA sponsored program established at the University of Puerto Rico at Mayaguez, P.R. • The Puerto Rico Coastal Zone Management Program (PRCZMP), a NOAA sponsored program established at the Department of Natural and Environmental Resources of Puerto Rico. SCOPE OF WORK Figure 1 – Map of the archipelago of Puerto Rico, including its two island municipalities of Vieques and Culebra (to the east). The uninhabited Isla de Mona to the west was not included in the flood mapping. • First a note on the nomenclature to be used. Each hurricane simulation outputs spatial and temporal information about the ocean’s and wind waves’ response. One of these is the maximum sea surface elevations at every computational node irrespective of the moment in time when it is produced. This is called the Maximum Envelope Of Water, or MEOW. There is something similar for Hs, the significant wave height. In addition, when parallel hurricane tracks are run with a given constant heading, we will then have a collection of MEOWs, and based on them we compute the Maximum of the MEOWs, or MOM. This MOM then gives, for the sea surface heights, the maximum storm surge elevation at each computational node irrespective of the hurricane track(s) which generated it (see Figures 8 to 11). Henceforth, each set of (constant) hurricane headings will have a set of MEOWs for each track in the set, and one associated MOM showing the maximum storm surge elevation irrespective of the individual hurricane track(s) that produced it. In this study, as will be discussed below, four different headings were chosen, so for each hurricane category there will be four MOMs. Next we combine all of the MOMs into a SuperMOM, which will show the maximum storm surge 4 elevations at each computational node for that given category irrespective of the track(s) and hurricane heading(s) that produced it. So the principal output will be MEOWs, MOMs, and SuperMOMs. • The SuperMOM for each hurricane category will include results of all chosen tracks and storm headings (to be mentioned below), with the exception of a heading from the southwest to the northeast, which has been included due to Hurricanes Lenny and Omar. That heading has been chosen since it will be a worst case scenario for the west coast of the island, but since no hurricane has ever made landfall on the west coast, it is not included in the SuperMOM, but the results are there just in case they are ever needed. And, besides, they could be useful for coastal planning along the west coast of Puerto Rico if one wonders what could be expected from a hurricane making landfall along the west coast. • The mandate on the part of the CariCOOS program was to develop a storm surge atlas for Puerto Rico (Figure 1) based on present sea level conditions. The atlas is prepared by running multiple hurricanes along different tracks, with different headings, this being repeated for each of the five Saffir-Simpson hurricane categories. It is very important to highlight that by storm surge it is implied the stillwater elevation produced by the passing hurricane due to pressure setup (inverted barometer effect), wind setup (piling up of water on the coast due to onshore winds), and wave setup (due to the accumulation of water when wind waves break in the near shore). The high frequency wind-forced waves riding on top of the stillwater and capable of producing wave runup are not included and would require another type of model. Coastal flooding due to wave runup is a very important component in the overall coastal flooding problem. But its inclusion is beyond the scope of this work. • The mandate on the part of the PRCZMP was to repeat all of the above but now for the scenarios of 0.5 and 1.0 meters of sea level rise, which is simply given to the model as an input parameter • Images and figures of the inland flooding are to be presented in three formats: kmz, shapefiles, and png. • Finally, as mentioned in the section titled Caveats at the end, flooding maps near nearshore inland water bodies, and rivers, are not reliable due to the fact that for the lack of information about them, the National Geophysical Data Center (NGDC – know known as the National Centers for Environmental Information - NCEI) decided to put a “lid” over those water bodies in the Digital Elevation Model (DEM) for Puerto Rico. This “lid” lies at an elevation up to 0.5 m above Mean High Water (MHW), which is the reference vertical datum. METHODOLOGY Computer Models Used: • We used version 50.99 of the tightly-coupled hydrodynamic and wind wave models ADCIRC+SWAN (called by some PADSWAN, the P standing for the parallel version), both running in the same unstructured mesh. This allows the computation of the three storm surge components: pressure, wind, and wave setups. Originally the model was run on a 48 nodes workstation. Later on we used a 448 nodes Linux cluster. 5 Hurricane Wind Model Used • For the production runs used in the preparation of the maps use was made of the internal hurricane winds model available in ADCIRC, version 50.99 Stable. In this version the Holland B parameter is an output of the model. The model is the so-called asymmetric model in which, every six hours, the model reads 1-minute averaged winds at 10 m height at the four quadrants of the storm, given at a certain specified distance from the storm center. With this information the radius of maximum winds (RMW) is computed. In our case, we specified maximum winds corresponding to the chosen hurricane category and at a distance equal to a chosen radius of maximum winds. Based on the given input of central pressure, maximum winds, and radii, the hurricane wind model outputs Holland’s B value. This was made via the use of the fort.22 formatted input files, which are shown in Appendix 2, together with other fort.** files needed by ADCIRC/SWAN. • For the validation of the models, re-analyzed winds (Automated Tropical Cyclone Forecasting – ATCF - system's best track) were used for hurricanes Georges (1998), Omar (2008), Earl (2010), and Irene (2011). The information was obtained from the following links: • GEORGES 1998 aal071998.dat http://ftp.nhc.noaa.gov/atcf/archive/1998/ • OMAR 2008 aal152008.dat http://ftp.nhc.noaa.gov/atcf/archive/2008/ • EARL 2010 aal072010.dat http://ftp.nhc.noaa.gov/atcf/archive/2010/ • IRENE 2011 aal092011.dat http://ftp.nhc.noaa.gov/atcf/archive/2011/ • The information in the links was also supplied in the fort.22 format. • Table 1 shows the actual hurricane parameters used for each category. The central pressures are dictated by the range assigned to each Saffir-Simpson category. There is some latitude in choosing the central pressure for each category. In this study what was done was to take approximately the value falling at position 3/4 starting from the highest pressure defining the hurricane category. For example, category 2 is defined by the limits 979 to 965 mb, encompassing 15 mb. 75% of 15 is 11.25, and counting down to the eleventh position from 979 gives 969 mb, and this is the value used. This implies that for category 2 hurricanes 75% of the pressures defining its range fall above 969 mb (less intense), with approximately 25% within the category 2 pressure interval being more intense than the value of 969. The same for the maximum wind speeds (Vmax). Category 5 is open-ended. Knowing that in 1928 we were struck by a category 5 hurricane (called San Felipe in Puerto Rico and the 1928 Okeechobee hurricane elsewhere), with 1-minute sustained winds topping 160 mph (the San Juan cup anemometer reported sustained winds of 160 mph before failing; since the anemometer was 30 miles north of the storm center, winds near the landfall point were unofficially estimated as high as 200 mph - https://en.wikipedia.org/wiki/1928_Okeechobee_hurricane), and lowest central pressure 6 estimated as 929 mb, it was decided to assign a central pressure of 900 mb, and maximum winds of 150 kn (173 mph) for the simulated category 5 hurricanes. ≤ • The hurricane forward speed, Vf, is a free variable that is assigned based on typical values for the region. The value of 10 kn is such a value. Later on tests can be made with different Vf values.