This supplementary material file describes (Section 1) the ocean model used to provide the

Transcription

1 P a g e The influence of ocean on Typhoon Nuri (2008) Supplementary Material (SM) J. Sun 1 and L.-Y. Oey *2,3 1: Center for Earth System Science, Tsinghua University; 2: IHOS and Atmospheric Science Department, National Central University; 3: Atmospheric & Oceanic Sciences Program, Princeton University *Corresponding Author: lyo@princeton.edu; Summary This supplementary material file describes (Section 1) the ocean model used to provide the SST for the WRF simulation (experiments pom* in main text table 1), including model validation using the Argo data; and (Section 2) changes in SST in South China Sea and western Pacific by vertical mixing by wind. More details are given in Sun et al [2015]. 1. The North Pacific Ocean model In order to account for the dynamical evolution of the upper-ocean state due to Typhoon Nuri, we ran the North Pacific Ocean model of Oey et al [2013, 2014] which then provided the SST for input into WRF. a. Ocean model descriptions: The model is based on the parallel version of the Princeton Ocean Model (POM). Air-sea coupling was not included, and we focused on separating the effects of SST from effects of internal variability. With coupling, winds produced by internal variability would modify SST s which in turn would change the internal variability of the storm, making it difficult to separate their effects. The domain included the entire North Pacific Ocean from 99 E-70 W and from

2 P a g e S-72 N (Fig.SM-1). The WRF domain is therefore wholly embedded within the large domain of the model ocean which therefore provided smooth and dynamically consistent ocean variables across the WRF s lateral boundaries during the integration. The ocean model has horizontal resolution and 41 vertical sigma levels, and the ETOPO2 topography was used ( The sigma cells were finer, logarithmically distributed near the free surface where the first grid cell was approximately 1~10 m below the surface depending on the water depth. A fourth-order scheme was used to minimize the sigma-level pressure-gradient error [Berntsen and Oey 2010]. The Mellor and Yamada s [1982] turbulence closure scheme was modified to include turbulence energy due to breaking waves near the surface [Craig and Banner 1994; Mellor and Blumberg 2004]. To account for mixing in stable stratification [e.g. internal waves; MacKinnon and Gregg, 2003], Mellor s [2001] modification of a Ridchardson-number-dependent dissipation was used. The Smagorinsky s [1963] shear and grid-dependent horizontal viscosity was used with a nondimensional coefficient = 0.1; the corresponding horizontal diffusivity is made 10 times smaller. The Oey et al s [2007] wind-drag formula with high wind-speed limit was used [Powell et al 2003] to specify the wind stress using six-hourly cross-calibrated multi-platform wind [CCMP; Atlas et al. 2011]. However, near the eye-wall, the CCMP wind was weaker than data from the IBTrACS, by as much as 50% (e.g. 20 vs 43 m s -1 ); a correction to the CCMP data was therefore applied during Nuri. The observed maximum wind speed, minimum sea level pressure (SLP), and location of the center of Nuri from the IBTrACS data were used in the vortex model of Holland [1980] to estimate the

3 P a g e typhoon s wind. The radius of maximum wind speed of 30 km, required in the Holland model, was also assumed; this value was close to the radius of the eye wall in the simulated Nuri (see below). The six-hourly synthetic wind was then merged with the CCMP wind using a Gaussian weight with e -1 -decay radius of 350 km determined by trials and errors, such that there was a smooth transition from Holland to CCMP wind at distances far from the typhoon s center. The merged wind was then linearly interpolated at each time step to force the model. Other details of the model can be found in Oey et al [2013, 2014]. b. Ocean initialization: The ocean model was run from August 6 th through 25 th 2008 to cover the period of Nuri (August 18 th - 22 nd 2008). Initial ocean fields on August 6 th were therefore required. Typhoon Nuri passed over a region known for deep Z 26 and active mesoscale variability [Oey et al 2013; Chang and Oey 2014]. These upper-layer ocean conditions were included by initializing the ocean model on August 6 th 2008 with the results of a POM run with data assimilation. Altimetry data from AVISO ( were assimilated at a 5-daily interval into the model using a simple optimum interpolation scheme described in details by Oey et al [2005], Lin et al [2007], Yin and Oey [2007] and Oey et al [2013]. In order to obtain SST approximating the observed conditions on August 6 th, we also incorporated SST observations into the model. After some experimentations, a simple and effective way was to nudge the model SST to observed SST from satellite using a time constant of 1/3 day -1. The satellite data used was AVHRR SST (Advanced Very High Resolution Radiometer, Multi-Channel Sea Surface

4 P a g e Temperature; The assimilated and SST-nudging experiment was run for 6 months from February 6 th through August 6 th 2008 and was itself initialized from a 47-year non-assimilated run forced by the CCMP and NCEP wind and other surface fluxes, as detailed in Oey et al [2013] and Xu and Oey [2014]. c. Ocean model validation: As ocean dynamics play an important role in the evolution of Typhoon Nuri, it is necessary to validate the modeled upper-ocean thermal structure. To do that, we compared the POM temperature profiles against all available Argo data during Nuri period August 18 th 22 nd 2008 and in the Nuri study region east of the Philippines and in South China Sea (rectangle in main text Fig.3, reproduced here as Fig.SM-1). Figure SM-2a-f compare the profiles at 6 example locations ( + in Fig.SM-1): 3 in western Pacific east of the Philippines (d, e and f), 1 just east of the Luzon Strait (b) and 2 west of the Luzon Strait (a and c). Figure SM-2g compares all the data in a scatter plot. The Argo data show deep Z m at the southern-most location (f) and just east of the Luzon Island (e) where the SST reaches 29~30 o C, and shallower Z m at the northeast location (d) with cooler SST 29 o C. The agreements between model and Argo are good at these stations in the western Pacific. The Z 26 decreases below 100 m for the 2 profiles near the Luzon Strait, where SST 28 o C to the east (b) and 27 o C to the west (a) of the strait. For the profile in South China Sea (c), the SST is warmer ~29 o C and Z 26 thinner 70 m. The POM generally simulates these features well. However, the Z 26 is shallower, and there are larger discrepancies at 2 profiles near the Kuroshio (a and b) which are being compared

5 P a g e during times when Typhoon Nuri has just passed, and the thermoclines were undergoing strong inertial oscillations. In general, the scatter plot comparison shows that POM simulated temperatures are slightly cooler near the surface and slightly warmer at subsurface; i.e. the modeled profile is more diffusive. Otherwise the agreements are good: the slope of the regression line is 1.03, bias is -0.1 o C (POM is cooler), and the scatter is generally small with a high R 2 value. We also repeated the regression analysis using only the upper 500 m Argo and model data, and obtained very similar result; the regression formula becomes Argo SST = POM SST* We emphasize that the comparison of the simulated SST s with Argo s is for the period during the storm; the agreements are quite good considering that strong temperature variations existed as the storm passed. d. Ocean run summary: Satellite altimetry and SST data were assimilated into POM to initialize the ocean state on Aug/6 th, 12 days prior to Nuri. The model was then allowed to run without data assimilation from Aug/6th through Aug/25 th to simulate the upper-ocean response, forced by the merged IBTrACS-CCMP wind data. We therefore allowed 12 days (Aug/6-17) for the model to adjust from its initial data-assimilated state before applying the merged wind data on Aug/18. The resulting POM SST was then used in the WRF simulation (Exp#7 or pom_ks in main text Table 1). Additionally, by perturbing the various parameters in the ocean model simulation: IBTrACS track, minimum SLP, and radius of maximum wind, as well as the SST relaxation time constant and the interval of altimetry data assimilation, POM was repeated to produce a set of perturbed

6 P a g e SSTs. These SSTs were then used to produce five additional WRF experiments (Exp#8-12 in main text Table 1). 2. Change in SST due to vertical mixing To better understand why the SST cooling is stronger in South China Sea, we plot in Fig.SM-3 the along-track vertical section contours of potential temperature T and meridional velocity v. The T and v are shown at the initial time August 18 00:00, and represent the background fields before the passage of typhoon Nuri. Figure SM-3a shows that in the near-surface 100 m layer, the isotherms are approximately 2 times deeper east of Luzon in the western Pacific than to the west in South China Sea. The transition occurs at the Luzon Strait (121~122 o E) across the cyclonic, western side of the Kuroshio, as can be seen from the 116 v-contours in Fig.SM-3b. 1 Price [1981] has shown that the SST response tends to be largest where cold water is near the sea surface, i.e., where the initial mixed layer is thin and the upper thermocline temperature gradient is sharp. In other words, when the initial mixed layer is thin, it would take less energy to lower its temperature through mixing with the cooler subsurface water [see equation 2 in Oey et al 2006]. This is the case for South China Sea, and the mechanism may therefore explain part of the simulated strong cooling. To estimate the drop in temperature due to this mechanism, let the surface layer depth h 1 be much thinner than the subsurface layer depth h 2 so that the surface-layer temperature after mixing is approximately the temperature of the 1 Subsurface isotherms (down to 300~400m below the surface) slope upward from east to west in approximate geostrophic balance with the strong vertical shear of the Kuroshio at the entrance of the Luzon Strait (Fig.SM-3b).

7 P a g e subsurface layer. We assume that the temperature in the surface layer linearly decreases from T s at the surface z = 0 to T b at the base of the surface layer z = -h 1 ; we simplify further and set the subsurface-layer temperature equal to T b (Fig.SM-4a). Now let h 1SCS be the thickness of the surface layer in South China Sea that is mixed by the Nuri wind. Then the drop in temperature in South China Sea is: 129 T SCS (T b T s ). (SM2.1a) Now, typhoon Nuri reached a maximum intensity in Luzon Strait, and its intensity and duration before and after entering South China Sea are approximately the same, as seen in main text Fig.1b. The wind power is therefore also approximately equal in both basins, and it mixes approximately to the same depth h 1SCS in western Pacific. However, in the western Pacific, the T b is at a deeper level z = -h 1WP (< -h 1SCS ) below the surface (Fig.SM-4b). Therefore, the corresponding drop in temperature after mixing in western Pacific is: 136 T WP (T b T s )(h 1SCS /h 1WP ) (SM2.1b) where we have used the fact that the SST (i.e. T s ) in the two basins is initially approximately the same (Fig.SM-1). Then, 139 T SCS / T WP h 1WP /h 1SCS 2. (SM2.2) An estimate of T WP may be obtained from the red contours of Fig.7c of the main text, which yield T WP -1 to -2 o C. However, that estimate includes cooling processes by horizontal processes such as upwelling and near-inertial internal waves [Oey et al 2008; Chiang et al 2011]. A more accurate estimate is obtained by re-running the POM experiment with the same initial

8 P a g e conditions but in the vertical z-direction only at each grid point of the same model domain. This is called the vertical mixing experiment or POMZMix; it most closely corresponds to the simple situation described above. The result is shown in Fig.SM-5b which is also compared with the result of full POM experiment (Fig.SM-5a); the T WP is approximately -1 o C (red contour in Fig.SM-5b). From (SM2.2), we estimate therefore that the cooling in South China Sea as a result of its thinner surface layer is T SCS -2 o C, as stated in the main text. 150

9 P a g e References Atlas, R., et al., 2011: A cross-calibrated, multiplatform ocean surface wind velocity product for meteorological and oceanographic applications. Bull. Amer. Meteor. Soc., 92, Berntsen, J., & Oey, L. Y. (2010). Estimation of the internal pressure gradient in σ-coordinate ocean models: comparison of second-, fourth-, and sixth-order schemes. Ocean dynamics, 60(2), Chang, Y.-L. & L.-Y. Oey, 2014: Instability of the North Pacific subtropical countercurrent. J. Phys. Oceanogr. 44, Chiang, Wu & Oey, 2011: Typhoon Kai-Tak: a perfect ocean s storm J. Phys. Oceanogr. 41, Craig, P. D., & Banner, M. L. (1994). Modeling wave-enhanced turbulence in the ocean surface layer. Journal of Physical Oceanography, 24(12), Holland, G. J. (1980) An analytic model of the wind and pressure profiles in hurricanes. Monthly weather review, 108(8), Lin, X.-H., L.-Y. Oey & D.-P. Wang, 2007: Altimetry and drifter assimilations of Loop Current and eddies. JGR, 112, C05046, doi: /2006jc003779, MacKinnon, J. A., & Gregg, M. C. (2003). Mixing on the late-summer New England Shelf-solibores, shear, and stratification. Journal of Physical Oceanography, 33(7),

10 P a g e Mellor, G. L., & Yamada, T. (1982). Development of a turbulence closure model for geophysical fluid problems. Reviews of Geophysics, 20(4), Mellor, G. L. (2001). One-dimensional, ocean surface layer modeling: a problem and a solution. Journal of Physical Oceanography, 31(3), Mellor, G., & Blumberg, A. (2004). Wave breaking and ocean surface layer thermal response. Journal of physical oceanography, 34(3), Oey, L. Y., Ezer, T., Forristall, G., Cooper, C., DiMarco, S., & Fan, S. (2005). An exercise in forecasting loop current and eddy frontal positions in the Gulf of Mexico. Geophysical Research Letters, 32(12). Oey, L.-Y., T. Ezer, D.-P. Wang, S.-J. Fan and X.-Q. Yin, 2006: Loop Current warming by Hurricane Wilma, Geophys. Res. Lett., 33, L08613, doi: /2006gl Oey, L. Y., Inoue, M., Lai, R., Lin, X. H., Welsh, S. E., & Rouse, L. J. (2008). Stalling of nearinertial waves in a cyclone. Geophysical Research Letters,35(12). Oey, L.-Y., T. Ezer, D.-P. Wang, X.-Q. Yin and S.-J. Fan, 2007: Hurricane-induced motions and interaction with ocean currents. Cont. Shelf Res. 27, Oey, L.-Y., Chang Y.-L., Lin Y.-C., M.-C. Chang, F.-H. Xu, and H.-F. Lu 2013: ATOP-the Advanced Taiwan Ocean Prediction System based on the mpipom Part 1: model descriptions, analyses and results, Terr Atmos Ocean Sci, 24, Oey, L.-Y. Y.-L. Chang, Y.-C. Lin, M.-C. Chang, S. Varlamov, and Y. Miyazawa, 2014: Cross flows in the Taiwan Strait in winter. J. Phys. Oceanogr., 44,

11 P a g e Powell, M.D., Vickery, P.J., Reinhold, T., Reduced drag coefficient for high wind speeds in tropical cyclones. Nature 422, Price, J. F. (1981). Upper ocean response to a hurricane. Journal of Physical Oceanography, 11(2), Smagorinsky, J. (1963). General circulation experiments with the primitive equations: I. The basic experiment*. Monthly weather review, 91(3), Sun, J., Oey, L. Y., Chang, R., Xu, F., & Huang, S. M. (2015). Ocean response to typhoon Nuri (2008) in western Pacific and South China Sea. Ocean Dynamics, 65, DOI /s Xu, F.-H., L.-Y. Oey, 2014: State analysis using the Local ensemble transform Kalman Filter (LETKF) and the three-layer circulation structure of the Luzon Strait and the South China Sea, Ocean Dyn., 64, Yin, X. Q., & Oey, L. Y. (2007). Bred-ensemble ocean forecast of Loop Current and rings. Ocean Modelling, 17(4),

12 P a g e Fig.SM-1 The North Pacific Ocean, POM model domain showing the simulated SST (shading in o C) on Aug 16 th The blue rectangle is WRF fine-grid model domain used for Nuri simulation. The pluses + inside the WRF domain show the positions of example Argo [free-drifting profiling floats that measure the temperature and salinity of the upper 2000 m of the ocean; stations where detailed profiles are compared against the model in Fig.SM-2a-f. They and all Argo data during Nuri in the WRF domain are used in the regression plot in Fig.SM-2g. This figure was reproduced from Fig.3 of the main text.

13 P a g e (a) (b) 216 (c) (d) 217 (e) (f)

14 P a g e (g) Fig.SM-2 (a-f): comparison of POM temperature profiles with Argo at the six locations shown as + in Fig.SM-1 inside the Nuri study domain: (a), (b) & (c) in South China Sea and Luzon Strait; (d), (e) & (f) in western North Pacific. (g) scatter plot comparison of all POM and Argo profiles from m deep within the study domain (rectangle in Fig.SM-1) for the Nuri period from Aug 18 th 22 nd The red line is the regression line and black line is the reference perfect match line. Additional profile comparisons are given in Fig.SM-2 (i-n).

15 P a g e 15 (i) (j) 231 (k) (l) 232 (m) (n) Fig.SM-2 (i-n). As in (a-f) at six other indicated locations.

16 P a g e (a) (b) Fig.SM-3 Along-track (Nuri track; see main text Fig.1) and vertical section contours of (a) the potential temperature T ( o C; contour interval is 1 o C above 26 o C (thick contour) and is 3 o C below), and (b) the meridional velocity v (m s -1 ; contours are -0.1, +0.1, 0.3,.., white is negative and zero is omitted). Both for Aug/18, 2008, the initial time of typhoon Nuri. Abscissa is longitude ( o E) and ordinate is z (m). Nuri track is approximated by a straight line passing through its positions on Aug/18/09:00 and Aug/21/21:00.

17 P a g e Fig.SM-4 A schematic of temperature profiles with T = T s at surface and T = T b at the base of a thin surface layer of thickness h 1 overlying a much thicker h 2 (>> h 1 ) deep layer with uniform temperature T b. (a) is for South China sea with a thin h 1 and (b) is for the western Pacific with a thicker h 1 (but still << h 2 ).

18 P a g e Fig.SM-5 SST change (shading) defined as the SST on Aug/22/12:00 minus SST on Aug/18/00:00 for (a) POM and (b) POMZMix The observed typhoon Nuri track from IBTrACS is plotted with daily positions shown as dots. Red contours (-1 and -2 o C) are SST changes before Nuri entered South China Sea: i.e. SST on Aug/20/12:00 minus SST on Aug/18/00:00.

19 P a g e 19 (A) 258 (B) Fig.SM-6 (a) Typhoon Nuri minimum SLP from observation (IBTrACS data ; black line) and from experiments using the same NCEP SST but different microphysics schemes indicated (various colored lines) (see Table 1), from Aug/18-23, Vertical bars show standard errors of the minimum center pressure from the IBTrACS data. (b) RMS errors in minimum SLP plotted as a function of microphysics schemes for NCEP & POM SST experiments. While the observation showed a monotonic weakening of the storm after Aug/20, the simulated typhoons with different microphysics all incorrectly re-intensified and achieved lower SLP s in South China Sea, suggesting that other factors controlled the intensity change.