
RESEARCH LETTER Nonlinear Impacts of Surface Exchange Coefficient 10.1029/2019GL085783 Uncertainty on Tropical Cyclone Intensity and Air‐ Key Points: • Tropical cyclone ocean feedback Sea Interactions have a negative impact on intensity Robert G. Nystrom1, Xingchao Chen1, Fuqing Zhang1, and Christopher A. Davis2 but the extent depends on the air‐sea exchange coefficients 1Department of Meteorology and Atmospheric Science, and Center for Advanced Data Assimilation and Predictability • Both tropical cyclone intensity and 2 the model‐parameterized surface Techniques, Pennsylvania State University, University Park, PA, USA, National Center for Atmospheric Research, drag coefficient impact the degree of Boulder, CO, USA storm‐induced ocean cooling • Intensity forecast uncertainty is not necessarily reduced in coupled Abstract Tropical cyclone maximum intensity is believed to result from a balance between the surface simulations if uncertainty to the surface drag coefficient exists friction, which removes energy, and a temperature/moisture (enthalpy) difference between the sea surface and the air above it, which adds energy. The competing processes near the air‐sea interface are Supporting Information: controlled by both the near surface wind speed and the surface momentum (Cd) and enthalpy (Ck) exchange • Supporting Information S1 coefficients. Unfortunately, these coefficients are currently highly uncertain at high wind speeds. • Figure S1 • Figure S2 Tropical cyclone winds also apply a force on the ocean surface, which results in ocean surface cooling • Figure S3 through vertical mixing. Using coupled atmosphere‐ocean and uncoupled (atmosphere only) ensemble • Figure S4 simulations we explore the complex influence of uncertain surface exchange coefficients on storm‐induced • Figure S5 fi • Figure S6 ocean feedback and tropical cyclone intensity. We nd that the magnitude of ocean cooling increases • Figure S7 with storm intensity and Cd. Additionally, the simulated maximum wind speed uncertainty does not • Figure S8 necessarily decrease when ocean feedback are considered. Plain Language Summary The peak intensity of tropical cyclones is primarily the result of the Correspondence to: energy exchange between the ocean and atmosphere. Tropical cyclones obtain energy from the warm R. G. Nystrom, ocean they track over, while at the same time losing energy through surface friction. Additionally, strong [email protected] surface winds of tropical cyclones cause the ocean surface to cool—a negative impact on intensity. The complex tropical cyclone–ocean feedbacks are partially controlled by the surface wind speed and the Citation: efficiency of the energy exchange between the ocean and atmosphere. Unfortunately, air‐sea energy Nystrom, R. G., Chen, X., Zhang, F., & Davis, C. A. (2020). Nonlinear impacts exchange is highly uncertain, especially at high wind speeds, and ocean feedback have often been ignored in of surface exchange coefficient tropical cyclone modeling. Using coupled atmosphere‐ocean and uncoupled (atmosphere only) simulations, uncertainty on tropical cyclone we find that forecasted maximum wind speed uncertainty does not necessarily decrease when ocean ‐ intensity and air sea interactions. ‐ Geophysical Research Letters, 47, feedback are considered because of competing effects between ocean atmosphere energy transfer, ocean e2019GL085783. https://doi.org/ cooling, and storm intensity. 10.1029/2019GL085783 Received 10 OCT 2019 Accepted 10 JAN 2020 1. Introduction Accepted article online 13 JAN 2020 Tropical cyclones (TCs) present significant risk of strong winds, storm surge, and heavy precipitation to regions around the world (Knapp et al., 2010; Peduzzi et al., 2012). While the significant advancements in global numerical weather prediction over recent decades (Bauer et al., 2015) have helped to significantly improve TC track forecasts (Alley et al., 2019; Cangialosi, 2019), TC intensity forecasts have improved at a much slower rate (Cangialosi, 2019). The smaller realized improvements in TC intensity prediction are the result of a combination of factors including insufficient model resolution (e.g.,Davis et al., 2008; Jin et al., 2014; Nystrom & Zhang, 2019), errors in initial conditions (e.g.,Brown & Hakim, 2013; Emanuel & Zhang, 2016; Hakim, 2013; Nystrom et al., 2018; Torn, 2016; Van Sang et al., 2008; Zhang & Sippel, 2009), and errors in model physics (e.g.,Braun & Tao, 2000; Bu et al., 2014; Green & Zhang, 2013, 2014; Judt et al., 2015; Melhauser et al., 2017). While our understanding of TCs has evolved since originally proposed in the 1980s, the basic components of potential intensity (PI) theory (Emanuel, 1988; Emanuel, 1997; Emanuel & Rotunno, 2011) still provide a useful framework to understand the quasi steady‐state intensity of TCs (e.g., Bryan & Rotunno, 2009a, ©2020. American Geophysical Union. 2009b). Within the basic PI framework, the maximum potential intensity (in terms of maximum wind speed) All Rights Reserved. a TC can achieve is determined by NYSTROM ET AL. 1of10 Geophysical Research Letters 10.1029/2019GL085783 − ÀÁ 2 Ck Ts To * V max ¼ k −k ; (1) Cd Ts * where Ts is the sea surface temperature (SST), To is the outflow temperature, (k − k) is the air‐sea enthalpy disequilibrium, and Ck/Cd is the ratio of the surface enthalpy exchange coefficient (Ck) to the surface drag coefficient (Cd) (Emanuel, 1997). We would therefore expect the maximum wind speed to increase with the ratio of Ck/Cd. Arguably, the biggest unknown when applying the PI theory to real cases, however, is how to estimate the surface exchange coefficients (Ck/Cd). Accurate estimates of the surface exchange coef- ficients at high wind speeds have been challenging to obtain in both the laboratory and nature, because of the dangers of collecting in situ observations within the turbulent TC boundary layer. As a result, the surface exchange coefficients are highly uncertain and recent studies often disagree on the sign of the relationship between wind speed and the exchange coefficients (Black et al., 2007; Bell et al., 2012; Chen et al., 2018; Donelan, 2018; Donelan et al., 2004; Holthuijsen et al., 2012; Hsu et al., 2017; Komori et al., 2018; Powell et al., 2003; Troitskaya et al., 2012, 2016; Takagaki et al., 2016). Bell et al. (2012) attempted to estimate the surface exchange coefficients and suggested that the current uncertainty may be larger than 40%. Additionally, Nystrom and Zhang (2019) suggested that the impacts from uncertainty in the surface drag coefficient can be as influential as initial condition uncertainty in limiting the intensity predictability of a strong TC. Further complicating the problem, TC‐induced ocean feedback, specifically the ocean surface cooling induced by vertical mixing, can have significant negative consequences on TC intensity (Bender & Ginis, 2000; Chen et al., 2007; Cione & Uhlhorn, 2003; Mogensen et al., 2017; Price, 1981; Schade & Emanuel, 1999; Zarzycki, 2016), but the implications for the PI have been largely ignored. A few recent studies (e.g., Balaguru et al., 2015; Lin et al., 2013; Miyamoto et al., 2017) have attempted to include the negative ocean feedback into PI theory but have neglected the uncertainty in the surface exchange coefficients, which will likely impact TC intensity and ocean feedback in currently unknown ways (e.g., Fan et al., 2009). In this study, we will focus on the influence of uncertain surface exchange coefficients on the intensity of TCs and ocean feedback using both atmosphere‐ocean coupled and uncoupled (atmosphere only) simulations. Our primary goal is to determine the influence of uncertainties within the modeled surface exchange coeffi- cients on TC intensity prediction in coupled and uncoupled simulations. 2. Modeling Methodology To examine the impacts of uncertainty in the surface exchange coefficients of momentum and enthalpy fluxes, we investigate both coupled atmosphere‐ocean and uncoupled (atmosphere only; fixed SST) simula- tions of Hurricane Patricia (2015). The ocean model used for the coupled simulations is the Regional Oceanic Modeling System (Shchepetkin & McWilliams, 2005), the atmospheric model is the Weather Research and Forecasting version 3.9.1 model (Skamarock et al., 2008), and the two models are coupled through the Coupled Ocean‐Atmosphere‐Wave‐Sediment Transport Modeling System (Warner et al., 2010). A figure depicting the domain configuration is shown in Figure S1 and further details of the model setup are also included in the supporting information. All simulations in this study have identical initial and boundary atmospheric and oceanic conditions, the details of which are included in the supporting information. The perturbations to Cd and Ck are generated by introducing three parameters (α,VC, and β) which modify the model representation of the surface exchange coefficients presented in Chen and Yu (2016, 2017) for Cd and Weather Research and Forecasting isftcflx option 2 for Ck. The model Cd representation from Chen and Yu (2016, 2017) is chosen due to its consistency with recent laboratory experiments (e.g., Donelan et al., 2004; Donelan, 2018; Takagaki et al., 2012, 2016; Troitskaya et al., 2012, 2016). We have intentionally chosen not to also couple with a wave model, which may be more physically realistic, because it would not allow us to control Cd explicitly as a function of the wind speed. In this study, the surface drag coefficient is a function of wind speed and is described as ÀÁ 2 –4 Cd ¼ α × max 5−0 :014 minðÞ V C; U10 þ 1 :033 minðV C; U10 þ 4 :895Þ ×10 ; (2)
Details
-
File Typepdf
-
Upload Time-
-
Content LanguagesEnglish
-
Upload UserAnonymous/Not logged-in
-
File Pages10 Page
-
File Size-