Decreasing trend of tropical cyclone frequency in 228-year high-resolution AGCM simulations

Transcription

1 GEOPHYSICAL RESEARCH LETTERS, VOL. 39,, doi: /2012gl053360, 2012 Decreasing trend of tropical cyclone frequency in 228-year high-resolution AGCM simulations Masato Sugi 1,2 and Jun Yoshimura 2 Received 30 July 2012; revised 4 September 2012; accepted 6 September 2012; published 10 October [1] We conducted 228-year long, three-member ensemble simulations using a high resolution (60 km grid size) global atmosphere model, MRI-AGCM3.2, with prescribed sea surface temperature and greenhouse gases and aerosols from 1872 to We found a clear decreasing trend of global tropical cyclone (TC) frequency throughout the 228 years of the simulation. We also found a significant multidecadal variation (MDV) in the long term variation of Northern Hemisphere and Southern Hemisphere TC count in addition to the decreasing trend. The decreasing trend and MDV in the long term variation of TC count correspond well to a similar decreasing trend and MDV of upward mass flux averaged over the TC genesis region and active TC season. It has been shown that the upward mass flux decreases primarily because the rate of increase of dry static stability, which is close to that of surface specific humidity, is much larger than the rate of increase of precipitation, which is nearly the same as that of atmospheric radiative cooling. Thus, it is suggested that the decreasing trend of TC count is mainly caused by the decreasing trend of upward mass flux associated with the increasing dry static stability. Citation: Sugi, M., and J. Yoshimura (2012), Decreasing trend of tropical cyclone frequency in 228-year high-resolution AGCM simulations, Geophys. Res. Lett., 39,, doi: /2012gl Introduction [2] Recent models consistently project a reduction of global tropical cyclone (TC) frequency in the future due to global warming, although there is a large uncertainty in the projection of regional TC frequency changes [Knutson et al., 2010]. On the other hand, observational studies based on the past historical TC data indicate that it remains uncertain whether there is a statistically significant trend in the long term variation of global TC frequency in the past [Knutson et al., 2010]. There are two major reasons for the uncertainty in the past TC frequency trend. One is that even if any past trend in TC numbers in response to anthropogenic global warming exists, it is not as large as the trend projected in the 21st century and it is relatively small compared with the natural long-term variations of TC frequency. Another reason for the uncertainty is the availability and quality of 1 Japan Agency for Marine-Earth Science and Technology, Yokohama, Japan. 2 Meteorological Research Institute, Tsukuba, Japan. Corresponding author: M. Sugi, Japan Agency for Marine-Earth Science and Technology, Showa-machi, Kanazawa-ku, Yokohama, Kanagawa , Japan. (msugi@jamstec.go.jp) American Geophysical Union. All Rights Reserved /12/2012GL the historical observational records of TC frequency. For the North Atlantic region, where relatively reliable long historical TC records are available, some observational studies reported an increasing trend of TC frequency during the 20th century [Mann and Emanuel, 2006; Holland and Webster, 2007]. It was pointed out, however, that the increasing trend is significantly reduced if an adjustment is made for missing TCs in the record of pre-satellite years [Chang and Guo, 2007; Vecchi and Knutson, 2008]. Furthermore, Landsea et al. [2010] found that there is a slight decreasing trend for the Atlantic TC count, though it is not statistically significant, if the trend is considered only for TCs lasting more than two days along with an adjustment for the missing TCs. [3] Considering the projected reduction of TC frequency in the future due to global warming, and uncertainty in the observed TC frequency trend in the past, it is interesting to examine how the models, which project a reduction of TC frequency in the future, simulate the past long-term variation of TC frequency. In the present study, we explore this issue using the results of a 228-year long simulation of Meteorological Research Institute global atmosphere model, MRI- AGCM3.2, which projected a significant reduction of future TC frequency [Murakami et al., 2012a, 2012b]. We conducted 228-year long, three-member ensemble simulations using a high resolution (60 km grid size) MRI-AGCM3.2 with prescribed SST and GHG and aerosols from 1872 to We found a clear decreasing trend of global tropical cyclone (TC) frequency throughout the 228 years of the simulation. [4] The remainder of this paper is organized as follows. Section 2 briefly describes the model and experiment. Section 3 and 4 present the simulated long-term variation of TC frequency and upward mass flux. In Section 5 we discuss a relation between the long-term variations of TC frequency and upward mass flux. Finally, a brief summary is given in Section Model and Experiment [5] The model we used for the 228-year simulation is a high resolution (horizontal grid size of 60 km) version of the new Meteorological Research Institute (MRI) Atmospheric General Circulation Model, MRI-AGCM3.2 [Mizuta et al., 2011]. The model has 64 levels in the vertical with top at 0.01 hpa. The model was developed from the Japan Meteorological Agency (JMA) Global Numerical Weather Prediction (NWP) Model ( jma-center/nwp/outline-nwp/index.htm) and shares many physical process subroutines, such as radiation scheme, PBL scheme, land surface scheme, and so on, with the NWP model, but some new physics schemes are incorporated. For MRI-AGCM3.2, we employed a new cumulus convection scheme developed by H. Yoshimura based on Tiedtke 1of5

2 Figure 1. Long-term variation of (a) global, (b) Northern Hemisphere and (c) Southern Hemisphere TC frequency simulated in the ensemble simulation. Thin curves indicate inter-annual variation of one of the three members of the ensemble. Thick smoothed curve is the 11-year running average of the TC counts of each member. [1989]. This convection scheme considerably improved the climatology of tropical convection [Mizuta et al., 2011]. For stratiform cloud scheme, the Tiedtke [1993] cloud scheme is used. In this cloud scheme, cloud water and cloud amount are the prognostic variables. We do not use the subtropical marine stratocumulus scheme [Kawai and Inoue, 2006] used in the JMA NWP model. The direct effect of aerosols is calculated in the MRI-AGCM3.2, but indirect effect of aerosols is not considered in the present simulation. [6] The 228-year simulation from 1872 to 2099 was conducted using the 60 km resolution version of the MRI-AGCM3.2 with prescribed sea surface temperature (SST) and atmospheric concentration of greenhouse gases (GHG) including CO 2 and aerosols. For the years from 1872 to 2003, the observed SST and sea-ice (HadISST1) [Rayner et al., 2003] are prescribed. For the years from 2004 to 2099, the ensemble average SST increase projected by Coupled Model Inter Comparison Project Phase 3 (CMIP3) [Meehl et al., 2007] models based on the IPCC Special Report on Emission Scenario (SRES) A1B scenario [Nakicenovic et al., 2000] are added to the de-trended observed SST from 1979 to In this procedure, the 25-year interannual variation of SST anomalies during is repeatedly added to the CMIP3 SST projection for consecutive future 25-year periods. Similarly, the reduction of sea-ice is included based on the ensemble average sea-ice change projected by CMIP3 models. Historical and A1B scenario GHG data ( gov/modelforce/ghgases/gcm_2004.html) are prescribed in the 228-year simulation. We used 3-dimensional natural and anthropogenic aerosol data calculated by MRI Earth System Model [Yukimoto et al., 2011] based on historical and A1B scenario aerosol emission data. Aerosols from volcanic eruptions are not included except for the Mt. Pinatubo eruption in The three-member ensemble simulations were conducted using the same model, the same SST and GHGs and aerosols, but slightly different atmospheric initial conditions. [7] The algorithm we used in the present study for detecting TCs in the model simulations is the same as the one used in Oouchi et al. [2006], in which a TC center is defined as a gird point with local minimum of surface pressure, with relative vorticity at 850hPa above sec 1 and with some other additional criteria. The algorithm is similar to the one used in Murakami et al. [2012a, 2012b], but there are some differences in the threshold values used in selection criteria. As a result, the simulated TC counts are a little different. For example, the annual average global TC count for the present climate simulation for the period is 71.9 in the present study, while it is 83.5 in Murakami et al. [2012b] in which the selection criteria is further adjusted so that the simulated TC count agrees better with the observed TC count (84.8). 3. Long-Term Variation of TC Count [8] Figure 1 shows the long-term variation of TC frequency during the 228-year simulation from January 1872 to December The global TC count shows a clear decreasing trend (Figure 1a). The average global TC count is about 83.0 during the last thirty years of 19th century, while it decreases to about 59.6 during the last thirty years of 21st century. The decreasing trend of the global TC count is 12.4/century. We can also see the decreasing trends in the long term variation of TC counts both in the Northern Hemisphere (NH) and the Southern Hemisphere (SH) (Figures 1b and 1c). The decreasing trends of TC count in NH and SH are 8.7/century and 3.7/century, respectively. [9] In addition to the decreasing trend, we can see a significant multidecadal variation (MDV) in both hemispheres. In the NH there are maxima of MDV at around 1990 and 1950 and minimum at around , while in the SH there are a maximum of MDV at around 1910 and a minimum at around The period of MDV is about 60 years 2of5

3 in NH and 80 years in SH. The MDV in the global TC count is not clear, because the MDVs in NH and SH are out of phase at around The MDVs in both the hemisphere are likely to be caused by the MDV of the SST prescribed for the 228-year simulation. It should be noted that we do not have MDV of TC counts in the 21st century, because the prescribed SST for the 21st century does not have MDV as observed in the 20th century. [10] In Figure 1, the inter-annual variation of TC count is plotted for one member of the three-member ensemble simulation by a thin curve, and the decadal variation is plotted for each member by thick smoothed curves. These short time scale variations are considered to be mainly associated with the SST variations on the same time scale. However, the three curves for the three ensemble members show a considerable difference in the decadal scale variations, even though the three members are forced by exactly the same SST. This indicates that the TC counts are not fully determined by the SST even on the decadal time scale. Therefore, it is appropriate to use the ensemble mean to examine the long-term TC count variation forced by the SST and the radiative forcing associated with the changes in GHG and aerosols. In the following section, we used ensemble mean data to explore the cause of the decreasing trend of TC counts. 4. Long-Term Variation of Upward Mass Flux [11] In this section we examine the possible cause of the decreasing trend of TC count as shown in Figure 1. Previous studies have indicated that the reduction of global TC frequency in the future due to global warming is closely related to the reduction of upward mass flux in the tropics [Sugi et al., 2002; Yoshimura and Sugi, 2005; Held and Zhao, 2011; Sugi et al., 2012]. One may argue that the reduction of TC count is the cause of the reduction of upward mass flux and not the result of it, because a substantial amount (3 11%) of tropical precipitation (and upward mass flux) is associated with TCs [Jiang and Zipser, 2010]. However, Zhao et al. [2012] estimated that the contribution from TCs to the changes in mass flux is not so large (0.15% mass flux change for 1% TC count change). Furthermore, Sugi et al. [2012] showed that primary reason for the reduction of upward mass flux is the increase in dry static stability due to global warming. They argue that the upward mass flux decreases because the rate of increase in dry static stability is larger than the rate of increase in precipitation. [12] From an approximate thermodynamic equation in the tropics, we can derive the following relation between fractional changes in precipitation, stability and upward mass flux, Dw w DP DS P S where w is upward mass flux (magnitude of upward p-velocity at 500 hpa), P is precipitation and S is dry static stability (potential temperature difference between 200h Pa and 850 hpa). Equation (1) indicates that upward mass flux may decrease even when precipitation increase, if the fractional increase in stability is larger than the fractional increase in precipitation. We examined the changes in precipitation, dry static stability and upward mass flux in the 228-year simulation, and found that the fractional increase in stability is indeed larger than the fractional increase in precipitation and upward ð1þ mass flux is decreasing (see auxiliary material for further detail) Relation Between TC Count and Upward Mass Flux [13] We noted that the simulated upward mass flux starts to decrease at the beginning of 21st century (see auxiliary material), suggesting that the change in upward mass flux cannot explain the decreasing trend of TC counts during the 20th century as shown in Figure 1. In order to further explore the possible relationship between the TC count and upward mass flux, we examine the relationship in the TC region and TC season. Figures 2a 2c shows the NH TC count in the NH active TC season (July October) and the changes of upward mass flux averaged over the NH TC genesis region (ocean region between 5 N and 30 N) and NH active TC season. A similar plot for SH TC genesis region (ocean region between 5 S and 30 S) and SH TC season (January April) is shown in Figures 2d 2f. The mean mass flux intensity plotted in Figures 2b and 2e indicates the mean intensity of the upward mass flux, while the total mass flux plotted in Figures 2c and 2f indicates the mean intensity of mass flux multiplied by the area of the upward mass flux. Thus, the changes in total mass flux include the changes in mean intensity of the mass flux and the changes in the area of the upward mass flux. [14] In Figures 2b and 2c, we can see that both the mean mass flux and total mass flux show a clear decreasing trend corresponding to the decreasing trend of TC count shown in Figure 2a. In Figures 2b and 2c, the mean intensity and total upward mass fluxes calculated by equation (1) are also plotted. These mass fluxes agree well with the simulated mass fluxes. In Figures 2e and 2f, we also find that both the mean and total mass flux show a decreasing trend corresponding to the decreasing trend in TC count as shown in Figure 2d. In addition, we note a large amplitude MDV in the total mass flux in Figure 2f which corresponds to the MDV of TC count as shown in Figure 2d. As the amplitude of MDV is not large in the mean mass flux intensity, a variation in the upward mass flux area is mainly responsible for MDV, while the mean intensity change is mainly responsible for the decreasing trend. We also note that in Figures 2d 2f the mean intensity and total upward mass fluxes calculated by equation (1) agree well with the simulated mass fluxes as in Figures 2a 2c, indicating that the changes in mass fluxes are explained by the changes in precipitation and stability. Therefore, we can conclude that there is an excellent agreement between the long-term variation of TC count and upward mass flux not only in the decreasing trend but also in the MDV. This suggests that the decreasing trend of TC count is caused by the decreasing trend of upward mass flux associated with the increasing stability. 6. Summary [15] We conducted 228-year long, three-member ensemble simulations using a high resolution (60 km grid size) MRI- AGCM3.2 with prescribed SST and GHG and aerosols from 1 Auxiliary materials are available in the HTML. doi: / 2012GL of5

4 Figure year running average of ensemble mean of (a) TC count for NH TC genesis region (ocean between 5 N and 30 N) and for the active NH TC season (July October), (b) mean upward mass flux intensity and (c) total upward mass flux. Orange curve in Figure 2b and green curve in Figure 2c indicates upward mass flux calculated by equation (1). (d f) Same as in Figures 2a 2c but for SH TC genesis region (ocean between 5 S and 30 S) and active SH TC season (January April) to We found a clear decreasing trend of global tropical cyclone (TC) count throughout 228 years of the simulation. We also found a significant MDV in the long term variation of TC count in NH and SH in addition to the decreasing trend. In order to explore the cause of the decreasing trend of TC count, we examined the long-term variation of upward mass flux. We can see a decreasing trend of tropical mean annual average upward mass flux during the 21st century, but the decreasing trend is not so clear in the 20th century. However, the decreasing trend and MDV is clearly seen if the mass flux is averaged over the TC genesis region and active TC season. It has been shown that the decreasing trend and MDV in the long term variation of TC count well correspond to a similar decreasing trend and MDV of upward mass flux averaged over the TC genesis region and active TC season. [16] We examined a possible reason for the decreasing trend of upward mass flux. It has been shown that the upward mass flux decreases primarily because the rate of increase of dry static stability, which is close to that of surface specific humidity, is much larger than the rate of increase of precipitation, which is the same as that of atmospheric radiative cooling. Thus, it is suggested that the decreasing trend of TC count is mainly caused by the decreasing trend of upward mass flux associated with the increasing stability. [17] So far, there is no observational evidence indicating a clear decreasing trend of global or hemispheric TC frequency as simulated by the model. We have noted that in our simulation the decreasing trend of western Northern Pacific TC count is the main contributor to the NH TC count decreasing trend, while the MDV in the North Atlantic TC count is mainly contributing to the MDV of NH TC count (figure not shown). Considering the large uncertainties in the regional TC frequency trend in the models as well as the observation, we should first compare the model simulation and observation in western North Pacific and North Atlantic regions, where relatively reliable long historical observation data are available. 4of5

5 [18] Acknowledgments. This work was conducted under the framework of the Projection of the Change in future Weather Extremes using Super-high-resolution Atmospheric Models supported by the KAKUSHIN Program of the Ministry of Education, Culture, Sports, Science, and Technology (MEXT). The calculations were performed on the Earth Simulator. We would like to express our thanks to the KAKUSHIN program Extreme Events Projection Team members who developed the model and conducted the 228-year long ensemble simulation experiments. References Chang, E. K. M., and Y. Guo (2007), Is the number of North Atlantic tropical cyclones significantly underestimated prior to the availability of satellite observations?, Geophys. Res. Lett., 34, L14801, doi: / 2007GL Held, I. M., and M. Zhao (2011), The response of tropical cyclone statistics to an increase in CO 2 with fixed sea surface temperature, J. Clim., 24, , doi: /jcli-d Holland, G. J., and P. J. Webster (2007), Heightened tropical cyclone activity in the North Atlantic: Natural variability or climate trend?, Philos. Trans. R. Soc. A, 365, , doi: /rsta Jiang, H., and E. J. Zipser (2010), Contribution of tropical cyclones to the global precipitation from 22 eight seasons of TRMM data: Regional, seasonal, and interannual variations, J. Clim., 23, , doi: / 2009JCLI Kawai, H., and T. Inoue (2006), A simple parameterization scheme for subtropical marine stratocumulus, Sci. Online Lett. Atmos., 2, 17 20, doi: /sola Knutson, T., J. McBride, J. Chan, K. Emanuel, G. Holland, C. Landsea, I. Held, J. P. Kossin, A. Srivastava, and M. Sugi (2010), Tropical cyclones and climate change, Nat. Geosci., 3, , doi: / ngeo779. Landsea, C. W., G. A. Vecchi, L. Bengtsson, and T. R. Knutson (2010), Impact of duration thresholds on Atlantic tropical cyclone counts, J. Clim., 23, , doi: /2009jcli Mann, M., and K. Emanuel (2006), Atlantic hurricane trends linked to climate change, Eos Trans. AGU, 87, 233, doi: /2006eo Meehl, G., C. Covey, T. Delworth, M. Latif, B. McAvaney, J. Mitchell, R. Stouffer, and K. Taylor (2007), The WCRP CMIP3 multi-model dataset: A new era in climate change research, Bull. Am. Meteorol. Soc., 88, , doi: /bams Mizuta, R., et al. (2011), Climate simulations using the improved MRI- AGCM with 20-km grid, J. Meteorol. Soc. Jpn., 90A, Murakami, H., R. Mizuta, and E. Shindo (2012a), Future changes in tropical cyclone activity projected by multi-physics and multi-sst ensemble experiments using the 60-km-mesh MRI-AGCM, Clim. Dyn., doi: / s x, in press. Murakami, H., et al. (2012b), Future changes in tropical cyclone activity projected by the new high-resolution MRI-AGCM, J. Clim., 25, , doi: /jcli-d Nakicenovic, N., et al. (2000), Emissions Scenarios, A Special Report of Working Group III of the Intergovernmental Panel on Climate Change, 599 pp., Cambridge Univ. Press, New York. Oouchi, K., J. Yoshimura, H. Yoshimura, R. Mizuta, S. Kusunoki, and A. Noda (2006), Tropical cyclone climatology in a global-warming climate as simulated in a 20 km-mesh global atmospheric model: Frequency and wind intensity analysis, J. Meteorol. Soc. Jpn., 84, , doi: /jmsj Rayner, N. A., D. E. Parker, E. B. Horton, C. K. Folland, L. V. Alexander, and D. P. Rowell (2003), Global analysis of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century, J. Geophys. Res., 108(D14), 4407, doi: /2002jd Sugi, M., A. Noda, and N. Sato (2002), Influence of the global warming on tropical cyclone climatology: An experiment with the JMA global model, J. Meteorol. Soc. Jpn., 80, , doi: /jmsj Sugi, M., H. Murakami, and J. Yoshimura (2012), On the mechanism of tropical cyclone frequency changes due to global warming, J. Meteorol. Soc. Jpn., 90A, Tiedtke, M. (1989), A comprehensive mass flux scheme for cumulus parameterization in large-scale models, Mon. Weather Rev., 117, , doi: / (1989)117<1779:acmfsf>2.0.co;2. Tiedtke, M. (1993), Representation of clouds in large-scale models, Mon. Weather Rev., 121, , doi: / (1993)121<3040: ROCILS>2.0.CO;2. Vecchi, G. A., and T. R. Knutson (2008), On estimates of historical North Atlantic tropical cyclone activity, J. Clim., 21, , doi: / 2008JCLI Yoshimura, J., and M. Sugi (2005), Tropical cyclone climatology in a highresolution AGCM Impacts of SST warming and CO 2 increase, Sci. Online Lett. Atmos., 1, Yukimoto, S., et al. (2011), Meteorological Research Institute Earth System Model version 1(MRI-ESM1) Model description, Tech. Rep. 64, 88 pp., Meteorol. Res. Inst., Tsukuba, Japan. Zhao, M., I. M. Held, and S.-J. Lin (2012), Some counterintuitive dependencies of tropical cyclone frequency on parameters in a GCM, J. Atmos. Sci., 69, , doi: /jas-d of5