Moisture control of tropical cyclones in high-resolution simulations of paleoclimate and future climate

doi:10.21203/rs.3.rs-2593770/v1

Download PDF

Article

Moisture control of tropical cyclones in high-resolution simulations of paleoclimate and future climate

https://doi.org/10.21203/rs.3.rs-2593770/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 13 Oct, 2023

Read the published version in Nature Communications →

Version 1

posted

You are reading this latest preprint version

The intensity of tropical cyclones (TCs) is expected to increase in response to greenhouse warming. However, how future climate change will affect TC frequencies and tracks is still under debate. To further elucidate the underlying sensitivities, we study TCs response to different past and future climate forcings. Using a high-resolution TC-resolving global Earth system model with 1/4^o atmosphere and 1/10^o ocean resolution we conducted a series of paleo-snapshot and future greenhouse warming simulations targeting the last interglacial (Marine Isotope Stage (MIS) 5e, 125 ka), glacial sub-stage MIS5d (115 ka), present-day (PD), and CO₂ doubling (2×CO₂) conditions. Our analysis reveals that precessional forcing creates an interhemispheric difference in simulated TC densities, whereas future CO₂ forcing impacts both hemispheres in the same direction. In both cases, we find that TC genesis frequency, density, and intensity are primarily controlled by changes in tropospheric thermal and moisture structure, with warmer hemispheres exhibiting a reduction in TC density.

Teaser

Precessional and CO₂ effects on tropical cyclone formation are primarily controlled by moisture-related processes.

Earth and environmental sciences/Climate sciences/Palaeoclimate

Earth and environmental sciences/Climate sciences/Climate change/Climate and Earth system modelling

Earth and environmental sciences/Climate sciences/Atmospheric science/Atmospheric dynamics

Landfalling tropical cyclones (TCs) can cause severe damage and impact human livelihoods and lives. Understanding what controls their frequency, tracks, and intensities is therefore of utmost importance. There is an emerging consensus between numerical models and theory suggesting that greenhouse warming will likely increase the intensity of TCs in the future ^1–6. However, there is still a large uncertainty in model-based projections of TC frequencies. Part of this uncertainty can be attributed to sensitivities with respect to external forcings, climate model physics, TC genesis classification (i.e., tropical cyclone seeds & it’s survival rate), and TC tracking algorithms^4,7−9.

To gain deeper understanding of how TCs respond to background climate state changes, it can be beneficial to study synoptic and mesoscale atmospheric variability in the context of paleo boundary conditions, representing in particular other warm climate episodes, such as the Last interglacial period (LIG, ~ 125 thousand years before present, ka )¹⁰, the Eocene (54,900 − 33,700 ka) and the Pliocene (5,300-2,600 ka)¹¹. Recent modeling studies ^11–13 document that past hothouse boundary conditions allowed TCs to form at higher latitudes. However, whether any of these paleo situations can be regarded as a “qualitative analogue” for the future is still an open question.

Apart from the question of “analogues” it is also important to elucidate how TCs are formed under different boundary conditions ^14,15 and identify the underlying physical mechanisms and controlling factors. The LIG, also referred to as Marine Isotope Stage (MIS) 5e, serves as an excellent test ground. Contrary to the present-day (PD) orbital configuration, the MIS5e was characterized by Northern Hemisphere (NH) summer perihelion conditions and warmer (colder) boreal (austral) summers. Understanding the TC response to an altered seasonal cycle may shed further light on the roles of interhemispheric forcings and seasonal changes in environmental factors such as sea surface temperature (SST), humidity, and vertical wind shear in tropical cyclogenesis. To obtain a detailed perspective on the relationship between large-scale climatic drivers and TC statistics, it is important to explicitly resolve TCs and their key mesoscale features in a climate model. This requires a relatively high atmospheric model resolution of at least 50 km ¹⁶ and it is also desirable to capture the coupling between atmosphere and ocean to account for cold wake effects, which may further influence the character and the statistics of TCs ¹⁷.

Here we set out to study the role of extreme orbital conditions (MIS5e and MIS5d) and greenhouse gas forcing (2xCO2) on TC frequencies and intensities using the high-resolution version ¹⁸ of the Community Earth System Model (CESM) 1.2.2 with 0.25° atmosphere and 0.1° ocean resolutions. To this end, we conducted two unprecedented, century-long simulations for MIS5e and MIS5d (~ 115ka) conditions (Fig. 1A) (Methods). These two periods represent extreme orbital conditions with low and high precession indexes respectively and large eccentricity. A low (high) precession index corresponds to NH summer perihelion (aphelion) condition and associated intensified (weakened) boreal summer solar radiation. This effect is visible only for high values of Earth’s orbit eccentricity and the intervals from 125 − 115 ka represent one of these periods. The orbital simulations are then compared with previously conducted experiments for PD climate and CO₂ doubling conditions (2×CO₂) ^6,19,20. We also employ two fundamentally different TC tracking algorithms (i.e., traditional vortex-based atmospheric conditions and phenomenon-based large-scale conditions; Methods text) to evaluate the sensitivity of the TC tracking schemes in the MIS5e, and MIS5d time-slice simulations, as well as the experiments with 2\(\times\)CO2 and PD conditions. To identify the linkage between large-scale and synoptic conditions, the results are then compared with the changes in the Genesis Potential Index (GPI) and each of its controlling factors, also referred to as GPI component analysis. Our study, therefore, focuses on understanding the dominant drivers of TC frequency and intensity changes under paleoclimate and future warmer conditions. This analysis can further enable us in determining whether there is a common link between the large-scale climate and changes in TC characteristics from the past to the future.

Climate response to orbital and greenhouse gas forcings

MIS5e and MIS5d correspond approximately to the minimum and maximum phase of the precession index (Fig. 1A) during the past 130 kyrs. Atmospheric CO₂ concentrations of ~ 275 ppm in both periods as well as the corresponding global mean tropical temperatures were similar to pre-industrial levels (Table 1, Fig. 1C), as shown by a reconstruction of tropical SST ²¹ and a transient climate model simulation ²² conducted with a lower-resolution model version of CESM1.2.2 (Fig. 1C). Therefore, the comparison between the corresponding CESM1.2.2 paleo time slice simulations reveal mostly the impact of precessional variability and the associated effect of seasonal and hemispheric insolation asymmetry on the climate system.

Analyses of averaged tropical climate variables relevant for TC dynamics over the oceans in these simulations reveal the following: atmospheric responses to incoming solar radiation in the orbital forcing simulations during the boreal summer and austral summer are out of phase (Fig. 2). Due to the meridionally asymmetric solar insolation, there is an increased solar radiation in the NH summer (July to September) and decreased radiation in the SH summer (January to March) during MIS5e compared to MIS5d. This anomalous distribution of solar insolation generates a warmer NH summer and autumn (as in the CO₂ forcing simulation at 200 hPa) and a colder austral summer and autumn (Fig. 2B). We observe increased temperatures with a lag of a few months both at lower levels and with a greater magnitude in the upper levels (Fig. 2B). The forced specific humidity variations are proportional to the temperature changes with higher values during the NH summer and lower values during the SH summer months in the MIS5e in comparison with MIS5d (Fig. 2C). As warm air can hold more moisture, the region with increased temperatures should correspond to reduced RH. Interestingly, the diagnosed changes in relative humidity (RH) do not show a decrease during the NH summer but indicate increases during the SH summer in the MIS5e than the MIS5d (Fig. 2D). There is no apparent trend in decreasing RH with an increase in temperatures in the NH summer due to regional variations in the moisture content associated with the regional-scale phenomenon (Fig S1).

In contrast to the orbital forcing simulations which have clear seasonal (i.e., hemispheric) differences in the temperature and moisture responses, the 2×CO₂ forcing simulation exhibits annual and hemispherically symmetric positive temperature anomalies relative to the PD climate conditions. The increased CO₂ concentration in the atmosphere increases tropospheric temperatures with significantly higher values in the upper levels compared to the lower levels, due to the moist adiabat response (red and blue dashed lines in Fig. 2B) ^12,23−25. The increased global air temperatures lead to year-round increased values of specific humidity and reduced RH in the lower troposphere across the tropical ocean basins. Although there are variations in each month, it is interesting to observe that the NH summer climatic response to orbital forcing (MIS5e - MIS5d) is qualitatively similar to that of the NH summer greenhouse warming response with higher temperatures. On the other hand, the SH summer orbital response is opposite to that of the CO₂ doubling forcing (Fig. 2B).

These differences in temperatures and moisture content of the atmosphere lead to changes in the stability and the moist-entropy deficit parameters (Fig. S2). It has previously been demonstrated that large-scale increases in atmospheric stability led to a reduction in TC frequency ^26–28. In addition, increased values of the moist-entropy deficit provide unfavorable conditions for TC formation which reduces the extent of the sustained atmospheric convection during the genesis stage of a storm ^29,30. The positive stability and moist entropy deficit anomalies are most pronounced for the 2×CO₂ simulation and they affect both hemispheres in their respective summer season. In contrast, in the MIS5e simulation, we observe higher stability and moist energy deficit during the NH summer as compared to the MIS5d simulation (Fig. S2A, C); but weaker values during the SH summer (Fig. S2B, D). Therefore, the environmental response to greenhouse gas forcing allows us to simply expect reduced TC frequencies in both hemispheres and reduced TC frequency in the NH summer and enhanced TC frequency in SH summer in response to orbital forcing (MIS5e). We will further test these hypotheses using the GPI analysis compared with two TC tracking schemes analysis and explore whether the MIS5e NH summer can be regarded qualitatively as an analogue situation to the greenhouse warming case.

Large-scale TC Genesis Environments

To understand the impact of orbital forcing and greenhouse gas forcing on TC genesis environments, we calculated the GPI (see Fig. S3 and Methods) for different climate conditions from paleoclimate to future climate^15,30−35. We also evaluated the relative importance of the GPI components in relation to the corresponding change in the large-scale environment for TCs genesis (i.e., GPI component analysis). As each basin has different seasons for TC activity, we calculated GPI changes in the peak TC seasons for the respective basins (JAS in the NH Pacific and Atlantic, OND in the North Indian (NI) Ocean, JFM in the SH basins) and combined them in one map (Fig. 3). In response to orbital forcing, the total GPI is reduced in key regions of TC genesis such as the Western North Pacific (WNP), the northwestern North Atlantic (NA), and equatorial side of the Eastern North Pacific (ENP), while is enhanced in the North and South Indian Ocean, subtropical ENP, Caribbean Sea, and some areas in the South Pacific (Fig. 3A). In the greenhouse gas forcing experiments, the total GPI is reduced over almost all the key TC genesis regions (Fig. 3F). In the paleoclimate simulations, the thermodynamical conditions (i.e., moist-entropy deficit and maximum potential intensity) can explain most of the TC frequency changes except for the Caribbean Sea where the vertical wind shear largely contributes to the changes in TC frequency. In future climate simulations, the moist-entropy deficit explains most of the changes in TC frequency. Therefore, both for the orbital and 2xCO2 simulations, the diagnosed changes in GPI are to first order consistent with the changes in the moist entropy deficit (Fig. 3D and 3I). Other GPI factors play only a secondary role in explaining the large-scale shifts in GPI. In contrast, a recent paleo-study focusing on the Paleocene–Eocene thermal maximum (PETM) epoch (characterized by increased CO₂ conditions and land-sea distribution) found that changes in vertical wind shear can explain the changes in the TC genesis under increased greenhouse gas conditions ¹². To further test the robustness of our results, we also calculated the conventional GPI Eq. (3⁶; Methods) which differs from the Korty ¹⁴ approach in two terms: (1) RH at 700 hPa instead of moist entropy deficit and (2) vertical velocity at 500 hPa. The conventional GPI component analysis also shows that moisture-related variables, i.e., 700 hPa RH, can capture most of the changes in the GPI (Fig. S4), thereby supporting our results from the other GPI component analysis.

Reduced values of RH and increased stability of the atmosphere due to increased upper-level temperatures act as unfavorable conditions for TC formation. The RH is an indicator for water vapor given the maximum amount of water the air can hold at a temperature. Therefore, it is proportional to water vapor in the atmosphere and approximately inversely proportional to atmospheric temperature. In paleoclimate simulations, we observe a decrease in RH during the NH warm season and increases during its cold season indicating that temperature plays a larger role in determining the sign of RH (Fig. S5A-C). Similarly, in greenhouse warming experiments the sign of the monthly mean RH varies linearly with temperature. However, in contrast to the greenhouse warming effect (not shown), RH in the orbital forcing simulations shows regional differences (Fig. S1). The changes in the temperature and moisture across the NA, WNP, and all the SH ocean basins agree with the global changes, but the ENP and NI basins have higher RH during the NH summer, irrespective of the higher air temperatures. We notice stronger low-level monsoon westerlies in the MIS5e simulation (Fig. S6D) which can be partly related to the enhanced land-sea thermal contrast³⁷. This circulation is associated with increased moisture transport into the Arabian sea region as compared to the Bay of Bengal region (Fig. S6B). In addition, across the ENP and NI basins, we observe a northward shift of the ITCZ position in the MIS5e simulation leading to favorable conditions for TC formation (Fig. S7B & E).

Changes In Detected TC Genesis

The aforementioned changes in the large-scale conditions of different model simulations (i.e., orbital and CO₂ forcing) can lead to changes in explicitly detected TCs as the storm tracking schemes use thresholds dependent on background large-scale environmental conditions. One of the advantages of high-resolution fully coupled simulations presented here is that they explicitly resolve the structure of TCs and relevant air-sea interactions ⁶. As illustrated here for a category-4 storm (Saffir-Simpson) during the MIS5e simulation (Fig. S8), TCs have a pronounced eye in the center and an SST cold wake effect oftentimes larger than 3℃. It is also noted that the current version of the CESM under PD conditions can capture the observed TC genesis climatology with certain regional variations while underestimating TCs in the NA and WNP basins (Fig. S9 and S10). To study the effect of paleoclimate and future forcings on TC tracks, we detected TCs using the last 60 years of each simulation (Methods).

Changes in TC genesis density during MIS5e show an increased density in the SH summer across southwestern SI and SP ocean basins compared to the MIS5d (Fig. 4A) mostly due to favorable TC forming conditions. Although not represented by the GPI, the tropical eastern Indian Ocean near the Sumatra Island has a lower genesis density (Fig. 4A). In addition, during the MIS5e period, the genesis density and track density differences (Fig. 4A, C) show a reduced number of detections in NA and WNP regions whereas the NI (Arabian Sea) and ENP basins show higher frequencies. The simulated changes in global and hemisphere mean TC frequency (Fig. S9) agree with other orbital simulations conducted with coarser-resolution model simulations that employ GPIs to study the variations of TC formations during the mid-Holocene period ^14,15. However, these studies did not identify the inter-basin differences in genesis frequency and track density. Although there is a regional discrepancy between TC density and GPI in the SH summer during the MIS5e, it appears that the changes in TC genesis density during the NH summer are consistent with the changes in the GPI.

The future climate (i.e., greenhouse gas forcing) simulation shows reduced global annual TC genesis frequencies in both hemispheres compared to the PD (Fig. 4B, Fig. S9). The track densities indicate decreased densities in the tropics with a slightly increased track density in higher latitudes of the NH basins and in some locations of the South Pacific basin (Fig. 4D). In contrast to the MIS5e climate that exhibits a clear hemispheric asymmetry in the genesis frequency, the future climate simulation shows an overall decrease in the mean annual TC frequency across all the ocean basins in both hemispheres. Previous climate studies demonstrated that changes in the Hadley circulation (i.e., weakening of rising branches of Hadley cells in both the hemispheres) along with unfavorable environmental conditions can suppress TC formation in both hemispheres in response to CO₂ forcing ^4,6,11,38. In addition, the changes in TC genesis locations and tracks are insensitive to the use of traditional and phenomenon-based tracking schemes both in the paleoclimate and future climate thereby increasing the confidence of the simulated TC response (Fig. S11 and Methods).

It is interesting to note that the changes in GPI are qualitatively similar to the detected changes in TC genesis that are explicitly simulated by the model using different tracking schemes in most of the ocean basins. Previous studies on the GPI analysis showed that there is some disagreement between the model-simulated TCs and GPI-estimated TC genesis frequency in future warmer climates ³⁹. However, in the current study, the considerable agreement of these GPI results in most of the ocean basins using two GPI indices including the paleoclimate, present, and future climates with the explicitly simulated TCs by the high-resolution model is encouraging. Additionally, unlike the response to a future warmer climate, the TC genesis density is not clearly reduced in the NH summer during the paleo-warmer period (MIS5e). Furthermore, we compare the locations of TC genesis detected using TC tracking schemes to seasonal changes in the thermodynamical and dynamical conditions that favor TC formation.

The seasonal variations in the 700 hPa RH closely align with the TC genesis frequency changes in most of the ocean basins showing that moisture may be a key controlling factor for the orbitally induced changes between MIS5e and MIS5d (Fig. S6B). Similarly, another thermodynamical variable, maximum potential intensity (MPI), can capture the detected TC genesis frequency in most of the basins except ENP (Fig. S6E). Moving on to changes in the dynamical variables, the development of the localized cyclonic vortex depends on large-scale absolute vorticity in the lower atmosphere. The seasonal fluctuations in the vorticity between the MIS5e and MIS5d are consistent with variations in the TC frequency, especially in SH basins. On the other hand, some regions of the NH basins including the ENP, and NA basins show little linkage with simulated lower TC frequencies and higher low-level vorticity values (Fig. S6C). The mid-level vertical velocity (Fig. S6F) varies seasonally and correlates with changes in relative humidity; the sustained convection that results from the greater mid-level moisture increases the mid-level upward motion. The lower values of the vertical wind shear between the 850 and 200 hPa are important for the storms to maintain their strength, favoring the development of initial vortices to develop into storms and TCs. We further find that SH summer has higher shear in the MIS5e compared to the MIS5d in most of the ocean basins which would oppose the simulated higher TC activities there (Fig.S6G). Over the paleoclimate simulations, the mismatch in the anomaly patterns of shear and TC frequency in some NH basins suggests that vertical shear plays only a minor role in describing global ocean changes in TC frequency. This indicates that the variations in the thermodynamical variables such as RH are mainly responsible for the associated changes in TC formation across most of the global ocean basins in the paleoclimates, thereby outweighing the effects of other dynamical variables.

In response to rising CO₂ levels, the reduced RH values (Fig. S12B) cause a decrease in TC frequency throughout the global ocean basins. We similarly observe reduced mid-level vertical velocity and large-scale vorticity in the SH summer in the 2xCO₂ experiment compared to the PD case (Fig. S12C, F). The fluctuations of these conditions, however, exhibit basin asymmetry in the NH basins. Therefore, the low-level vorticity/mid-level vertical velocity variables cannot explain the TC formation differences in both hemispheres. In the 2xCO₂ experiments, the vertical wind shear variations exhibit larger values in SH summer and asymmetric values in the NH ocean basins (Fig. S12G). Hence, the underlying environmental conditions for a future warmer climate also suggest that moisture (thermodynamical variable) is a key factor influencing changes in the projected TC frequencies. In the next section, we investigate the response of TC intensity across various simulations conducted under paleoclimate, current, and future climatic conditions.

Changes In TC Intensity

The diagnosed changes in the environmental conditions might have an impact on the TC intensity as well. Here we have further examined the changes in the lifetime maximum 10 m wind speeds and the mean-sea level pressure of the storms in different past, present, and future climate simulations (Fig. S13). Although the model generally underestimates TC intensity, the model can capture the TC wind-pressure relationship in all the simulations and up to category 4 storms (Saffir-Simpson’s scale) of the present climate and category 5 storms of future climate (Fig. 5, Fig. S13). From the MIS5d TC wind-pressure relationship, we can observe an increased frequency of intense storms compared to the MIS5e simulation. There is a clear intensification of storms across the globe in 2×CO₂ as compared to the PD climate.

The frequency of storms stratified by intensity category shows that there are less intense storms in the NH during the MIS5e, whereas more intense storms in the SH compared to the MIS5d. (Fig. 5). In contrast to the orbital forcing simulation which has hemispheric asymmetry in the frequency of the intense TCs, the greenhouse gas simulation shows a higher number of intense storms in both hemispheres. Furthermore, there is a reduced frequency (-20 to -470% across the category) of NH intense TC categories (Saffir-Simpson’s Category-2 to Category-4 storms) during the MIS5e and an increased frequency (+ 18 to + 90%) in the SH during the MIS5e (Category-1 to Category-4 storms) compared to the MIS5d. In the 2×CO₂ simulation, we observe a reduction in the weaker category storms (i.e., Tropical storms (TS) to Category-2 storms) and an increase in the frequency of the stronger storms (Category-3 to Category-5) in both the hemispheres as compared to PD condition. As there is an overall reduction in the global TC frequency for future warmer climate conditions compared to the past warmer climates (Fig. S10), we observe a more pronounced reduction in the weaker storms in future climates as compared to the paleoclimate time slices. Therefore, in addition to the changes in the TC frequency, we also observe a hemispherically asymmetric response of TC intensities in the paleoclimate and a symmetric response in a future warmer climate.

To further understand the TC intensity (maximum of 10m wind speeds), we examined one thermodynamical variable, the maximum potential intensity (MPI) which measures the convective instability of the atmosphere and gives the measure of maximum potential wind speeds a storm can attain under the background condition of the atmospheric temperature and moisture ¹. The seasonal differences in the MPI in general agree well with TC intensity change (Fig. S6E, Fig. S12E). In paleoclimate simulations, the seasonally varying radiation influences both atmospheric temperature and SSTs. The atmospheric temperatures respond faster than the underlying SSTs (Fig.S14) to the increases in radiation leading to reduced MPI (Fig. S15) in the NH summer of MIS5e as the upper atmosphere is warmer while the surface layer is still not that heated yet. The NH autumn (October and November), by which time surface waters have several months to adjust, develops a warmer surface layer than the atmosphere above with small differences leading to an increase in the MPI (Fig. S14 & Fig. S15). Therefore, MPI values in the NH summer are greater in the MIS5d simulation than those in the MIS5e simulation, whereas MPI in the SH summer is higher in the MIS5e simulation than the MIS5d. These variations in the MPI agree with earlier studies on paleoclimate that showed changes in TC intensity during Holocene or Mid-Holocene periods ^14,15. In contrast, the magnitude of changes in MPI for future climate simulation (2×CO₂) are generally small but are follows the sign of SSTs throughout the year (Fig. S14 & S15). Given these changes in the MPI caused by orbital and CO2 forcings, the NH summer in MIS5e cannot serve as an analogue to the 2xCO₂ simulation because the environmental response at lower and upper levels varies significantly to different forcings.

There is a clear consensus between the theory and the numerical modeling that TC intensity will increase in response to greenhouse warming ⁴. However, there is also no simple relationship between the changes in TC formation and large-scale climates ³⁹. Therefore, using high-resolution model experiments that explicitly resolve TCs up to at least Saffir-Simpson category 4 storms can improve our understanding of the linkage between background climate conditions and TC activity. Here we highlight the changes in TC genesis and intensity due to orbital and greenhouse gas forcings by explicitly simulating TCs under different background conditions using a mesoscale resolving coupled Earth system model. In response to orbital forcing, changes in TC frequency and intensity show a pronounced hemispheric asymmetry (Fig. 4, Fig. 5). Both frequency and intensity increase in the SH summer of MIS5e, whereas the NH shows decreases in both characteristics except that frequency increases in the NI and ENP. In contrast, greenhouse gas warming leads to suppression in frequency and an increase in intensity in both hemispheres (Fig. 4, Fig. 5).

An empirically derived GPI analysis captures most of the changes (except SI basin) in simulated TC frequency as obtained from two different tracking schemes qualitatively well under different background climate states (i.e., paleoclimate, present, and future) (Fig. 3, Fig. 4, and Fig. S11). This supports the notion that changes in the model’s large-scale climate variables can explain most of the associated changes in the TC activity. The seasonal composites of the large-scale climate variables show that in both the paleoclimate and future warmer climates the moisture-related variables (i.e., RH) are the most important variables for the associated changes in the TC formations. Similarly, the GPI component analysis also shows that moisture-related variables (RH & moist entropy deficit) explain the changes associated with TC frequency in paleoclimates and future climates (Fig. 3D and I, Fig. S4C and S4I).

The earlier studies note that either increased saturation deficit/entropy deficit of the atmosphere ^29,30,40 or reduced upward mass flux ²⁶ influence the TC frequency statistics for future warmer climates. Our study agrees with earlier studies ^28,41, which note that both the mechanisms act simultaneously (i.e., increased moisture entropy deficit/reduced mid-level moisture reduces the upward mass flux) in explaining the response of TC frequency to climate change. In further examining NH summer of the MIS5e and 2×CO₂ simulations, we observe a striking difference between the two simulations in terms of the movement of the ITCZ. In response to CO₂ forcing, the ITCZ shifts equatorward ¹¹ whereas it shifts poleward in response to MIS5e orbital forcing (Fig. S7). This poleward shift of the ITCZ during the peak TC season across the ENP and NI basins may lead to increased TC frequencies in the MIS5e as compared to the 2\(\times\)CO₂ simulation (Fig. 4A). An earlier idealized study ⁴² noted an increase in the global TC frequency with warming due to poleward shifts of the ITCZ latitude which is larger than the reduction in TC frequency caused by the warming effect alone. The variations in the ITCZ position and the associated changes in large-scale thermodynamic variables during the NH summer of the MIS5e and 2\(\times\)CO₂ simulations explain the differences in the simulated TC genesis activity. Seasonal temperature fluctuations at the surface and in the troposphere are related to variations in the theoretical intensity (MPI) of paleoclimate TCs (Fig. S14 & Fig. S15). Whereas for future climate conditions, the changes in SST can explain the variations in TC intensity. These results imply that the controlling factors and mechanisms for the changes in TC frequency and intensity are fundamentally different from past to present climates.

Although the TC frequency response employing two fundamentally different tracking schemes is similar under various background climate conditions, there is an underdetection of the storms in the main development region of the NA basin. The prevalence of SST biases in the Atlantic basin could have an impact on large-scale disturbances like the African Easterly waves, and associated convective processes⁶. In summary, we infer that variations in tropospheric temperature and moisture are crucial attributes of the corresponding variations in TC activity in both paleoclimate and future climates.

Model Description

In the current study, we employ an high resolution CESM1.2.2 ⁶ to perform fully coupled high-resolution global Earth system model simulations for paleoclimate and future climates. The atmospheric component of the model is CAM5 with a spectral dynamical core with a horizontal resolution of 0.25\(^\circ\) and 30 vertical layers ⁴³. The ocean component is POP2 which has a horizontal resolution of 0.1\(^\circ\) (decreasing from the equator (11km) to poles (2.5km)) and 62 depth layers ⁴⁴. The land model is the Community Land Model version 4 ⁴⁵, and the sea-ice component is the Community Ice Code version 4 ⁴⁶. The time variations of the carbon-nitrogen cycle are switched off in the current model simulations. Chu et al.⁶ shows that the present-day simulation of TC climatology is quite realistic, with certain regional and seasonal differences in TC genesis frequencies (Fig. S5 of Chu et al., 2020) when compared with the observations (IBTrACS). They also show that the higher resolution reduces the mean SST bias (a common feature in some of the global climate models). The model can capture the localized air-interaction processes, mesoscale eddies, TC cold wakes, and convective features along the mountainous regions ^6,18. The details of the greenhouse gases, vegetation, and orbital years of the model simulations are shown in Table 1. The present-day simulation was initialized from a quasi-equilibrium state of the Small et al.¹⁸ and simulated for 140 years, the 2×CO₂ simulation was carried from the 71 years of the present-day simulation with a doubling of atmospheric CO₂ concentration and integrated for 100 years. Similarly, the two paleoclimate simulations (MIS5e, MIS5d) which were branched off from year 71 of the present-day simulation are integrated for 100 years. In the current study, we carry out the analysis for a period of the last 60 years to check the robustness of the results. The main difference between the two paleoclimate simulations is the precession configuration. In the MIS5e, the Earth receives more solar insolation during the Northern Hemisphere summer season and less during the Southern Hemisphere summer season, and vice versa during the MIS5d period.

TC Detection And Tracking Schemes

Here we employ two different tracking schemes. The first is a traditional tracking scheme that uses certain circulation-based thresholds to identify and track initial vortices resembling TCs. For the second scheme, rather than detecting TC-like vortices, a phenomenon-based tracking scheme uses the thresholds of the large-scale climate variables to identify the recirculating regions with stronger vorticity and weaker deformation that have the potential for TC formations. The description below gives the details of both tracking schemes.

A. Traditional TC Tracking Scheme

This TC tracking scheme ⁶ is based on the idea of the Bacmeister et al. ⁴⁷ except that they use the 3-hourly interval data with a 200 km search radius for TC tracking and a 50 km radius for 10 m wind speed threshold. The steps involved are as follows:

Firstly, we identify the low-pressure systems using a threshold of surface pressure anomalies lower than − 3hPa;
The 10m-wind speeds of the systems should exceed 10ms^-1 within a radius of 100 km from the low-pressure minima location otherwise those lows are discarded;
These detected lows at 6-hourly intervals are tracked forward in time based on the nearest location in a similar instance of time and the tracking is continued if there is a low within a 400 km radius.
A further check is involved to remove the duplicates of storm tracks and keep the tracks that have 17ms^-1 magnitude of 10m wind speeds during their lifetime and has a maximum duration of 2 days.
To obtain the global TC counts close to observations we impose a further threshold of 0.00145ms^-1 for the 850 hPa vorticity threshold.

B. Phenomenon-based Tracking Scheme (OWZP)

The OWZP scheme ⁴⁸ uses the thresholds of large-scale climate variables such as relative humidity, specific humidity, vertical wind shear, and Okubo-Weiss Zeta parameter which is equivalent to vorticity but involves stretching and shearing deformations. This TC detection and tracking scheme shows better performance for different TC characteristics as assessed using the ERA-interim and ERA5 data ^48,49 and involves two steps:

In the first step, the “detection stage” involves identifying the grid points which satisfy the initial thresholds (see Table S1 of the supporting information) and merging the nearby grid points into clumps;
In the next step, the “tracking stage” the detected clumps are tracked forward in time until there exists no circulation by using the storm position determined by the 700 hPa steering wind. These clump positions along each of the detected tracks are further checked using the core thresholds of Table S1 and given True (thresholds are satisfied), and False (not satisfied). A particular track is considered a TC track if the core thresholds are satisfied for 48 hours consecutively. In 12-hourly data, there should be five consecutive True clumps along the track and the fifth True position is considered as the storm genesis location ⁴⁹.

TC Genesis Potential Index (GPI)

As TCs form under certain favorable environmental conditions, earlier studies developed certain empirical indices to identify the TC genesis locations. Emanuel and Nolan ⁵⁰, Emanuel et al. ²⁹, and Emanuel ³⁰ developed GPIs with a set of environmental variables. These variables include (i) elevated values of large-scale absolute vorticity which are required for the development of a strong localized TC-like vortex, (ii) higher mid-tropospheric moisture content which is required for persistent convection through inflow from the moist boundary layer favoring deep convection. The mid-level moisture availability is measured through a magnitude of moist entropy deficit estimated between the boundary layer and mid-troposphere and that changes with climate change even when the relative humidity doesn't change, (iii) lower vertical wind shear is necessary to maintain sustained vortex required for TC formation, (iv) potential intensity is the maximum attainable wind speed based on the available background thermodynamic environment of the atmospheric column (convective instability) involving the temperatures at the surface and outflow levels. Higher values of potential intensity indicate a more convectively unstable environment favoring the TC formation. Korty et al. ¹⁴ further developed this index with more sophisticated statistical analysis using similar environmental variables.

Earlier genesis potential indices are based on the sea surface temperatures (SSTs), absolute vorticity, relative humidity, and vertical wind shear variables. As the SSTs are not invariant with the background of changing climate conditions this index cannot be employed in studies that focus on other climates ⁵¹. The inclusion of 500 hPa vertical velocity can capture the future changes in the TC genesis activity ³⁶.

Here, \(\chi\) is the moist entropy deficit, s_b is the boundary layer entropy, sm is the mid-tropospheric (600 hPa) entropy, so* is the saturation entropy at the sea surface, \(\eta\) is the 850 hPa absolute vorticity, Vpot is the potential intensity, Vshear is the vertical wind shear between 850 and 200 hPa pressure levels, RH is the 700 hPa relative humidity, \(\omega\) is the 500 hPa vertical velocity.

The above-discussed GPIs are evaluated in paleoclimate, present, and future climate studies ^{14,15,30,31,33,35,52}. These studies note that the modified GPIs with the inclusion of moist entropy deficit/vertical velocity can capture the TC climatology and seasonal cycle in different climate conditions. A recent study by Camargo et al.³⁵ using a high-resolution 0.25\(^\circ\) model show that TC climatology estimated using GPI has not a strong relationship with the model’s explicitly simulated TC climatology. Most of the paleoclimate simulations are based on the changes in the GPI rather than TCs themselves, therefore in the current study, the comparison of the explicitly simulated TC climatology with the changes in the GPI can enhance our confidence in the paleoclimate TC variations and their associated atmospheric mechanisms.

GPI Component Analysis

To decipher the relative role of each GPI term, we adopt a variational GPI analysis technique ⁵³. The equation below represents the changes in the TC genesis potential estimated using a GPI is equal to the sum of changes in each term of the GPI index by assuming that the non-linear (correlation) terms are negligible. Here the first term is the change in the variable between two simulations multiplied by the mean values of other terms of the GPI index. The summation of each of the terms on the right-hand side equals to the overall changes in the GPI, that is positive values correspond to the increasing GPI and negative values reduce the GPI. In the current study, we conduct the GPI component analysis using both the GPIs which are described in the above section.

Acknowledgments:

Funding: This research was supported by the Institute for Basic Science, South Korea, under IBS-R028-D1. The model simulations are done using the CESM1.2.2 where the code is publicly available from the National Center for Atmospheric Research. Figures were generated by the NCAR Command Language (Version 6.4.0) [Software]. (2017). Boulder, Colorado: UCAR/NCAR/CISL/VETS. http://dx.doi.org/10.5065/D6WD3XH5.

The simulations were conducted on the IBS/ICCP supercomputer “Aleph”, 1.43 peta flops high-performance Cray XC50-LC Skylake computing system with 18,720 processor cores, 9.59 PB storage, and 43 PB tape archive space. We also acknowledge the support of KREONET.

Author contributions: R.P.H., J.E.C., and A.T. planned the study, performed data analysis, and prepared the figures. R.P.H., J.E.C., and A.T. drafted and wrote the main manuscript. S.S.L. conducted the CESM1.2.2 simulations with different forcings. K.J.E. provided editorial assistance in improving the manuscript. All authors reviewed and editted the manuscript.

Competing interests: The authors do not have any competing interests.

Data and Materials availability: The observed tropical cyclone data is from the IBTrACS archive available at the following link https://ncdc.noaa.gov/ibtracs/index.php?name=ib-v4-access) version 4. The HadlSST is obtained from the Hadley Centre, UK Meteorological Office, (http://badc.nerc.ac.uk/data/hadisst). The CESM code is publicly acessible from the National Center for Atmospheric Research. The model simulations using CESM 1.2.2 in the different climates are available upon request to the corresponding authors. All the data requried to produce the graphics in the main manuscript and the supplementary materials are available upon request. Figures are generated using the NCAR Command Language [version 6.4.0 (software), 2017, Boulder, Colorado: UCAR/NCAR/CISL/VETS; http://dx.doi.org/10.5065/D6WD3XH5] and MATLAB software.

Emanuel, K. A. The Maximum Intensity of Hurricanes. Journal of the Atmospheric Sciences 45, 1143-1155 (1988).
Korty, R. L., Emanuel, K. A., Huber, M. & Zamora, R. A. Tropical Cyclones Downscaled from Simulations with Very High Carbon Dioxide Levels. Journal of Climate 30, 649-667 (2017).
Bhatia, K. T. et al. Recent increases in tropical cyclone intensification rates (vol 10, 635, 2019). Nature Communications 10 (2019).
Knutson, T. et al. Tropical Cyclones and Climate Change Assessment: Part II: Projected Response to Anthropogenic Warming. Bulletin of the American Meteorological Society 101, E303-E322 (2020).
Gori, A., Lin, N., Xi, D. Z. & Emanuel, K. Tropical cyclone climatology change greatly exacerbates US extreme rainfall-surge hazard (vol 12, pg 171, 2022). Nature Climate Change 12, 491-491 (2022).
Chu, J. E. et al. Reduced tropical cyclone densities and ocean effects due to anthropogenic greenhouse warming. Science Advances 6 (2020).
Roberts, M. J. et al. Projected future changes in tropical cyclones using the CMIP6 HighResMIP multimodel ensemble. Geophysical research letters 47, (2020).
Yamada, Y. et al. Evaluation of the contribution of tropical cyclone seeds to changes in tropical cyclone frequency due to global warming in high-resolution multi-model ensemble simulations. Progress in Earth and Planetary Science 8, 1-17 (2021).
Raavi, P. H. & Walsh, K. J. E. Sensitivity of Tropical Cyclone Formation to Resolution-Dependent and Independent Tracking Schemes in High-Resolution Climate Model Simulations. Earth and Space Science 7 (2020).
Otto-Bliesner, B. L. et al. Large-scale features of Last Interglacial climate: results from evaluating the lig127k simulations for the Coupled Model Intercomparison Project (CMIP6)-Paleoclimate Modeling Intercomparison Project (PMIP4). Climate of the Past 17, 63-94 (2021).
Studholme, J., Fedorov, A. V., Gulev, S. K., Emanuel, K. & Hodges, K. Poleward expansion of tropical cyclone latitudes in warming climates. Nature Geoscience 15, 14-+ (2022).
Kiehl, J. T., Zarzycki, C. M., Shields, C. A. & Rothstein, M. V. Simulated changes to tropical cyclones across the Paleocene-Eocene Thermal Maximum (PETM) boundary. Palaeogeography Palaeoclimatology Palaeoecology 572 (2021).
Chikira, M., Yamada, Y., Abe-Ouchi, A. & Satoh, M. Response of convective systems to the orbital forcing of the last interglacial in a global nonhydrostatic atmospheric model with and without a convective parameterization. Climate Dynamics 59, 1617-1648 (2022).
Korty, R. L., Camargo, S. J. & Galewsky, J. Variations in Tropical Cyclone Genesis Factors in Simulations of the Holocene Epoch. Journal of Climate 25, 8196-8211 (2012).
Koh, J. H. & Brierley, C. M. Tropical cyclone genesis potential across palaeoclimates. Climate of the Past 11, 1433-1451 (2015).
Walsh, K. J. et al. Hurricanes and climate: The US CLIVAR working group on hurricanes. Bulletin of the American Meteorological Society 96, 997-1017 (2015).
Li, H. & Sriver, R. L. Tropical cyclone activity in the high‐resolution community earth system model and the impact of ocean coupling. Journal of Advances in Modeling Earth Systems 10, 165-186 (2018).
Small, R. J. et al. A new synoptic scale resolving global climate simulation using the Community Earth System Model. Journal of Advances in Modeling Earth Systems 6, 1065-1094 (2014).
Wengel, C. et al. Future high-resolution El Nino/Southern Oscillation dynamics. Nature Climate Change 11, 758-+ (2021).
Huang, L., Lee, S. S. & Timmermann, A. Caspian Sea and Black Sea Response to Greenhouse Warming in a High‐Resolution Global Climate Model. Geophysical Research Letters 48, (2021).
Herbert, T. D., Peterson, L. C., Lawrence, K. T. & Liu, Z. H. Tropical Ocean Temperatures Over the Past 3.5 Million Years. Science 328, 1530-1534 (2010).
Timmermann, A. et al. Climate effects on archaic human habitats and species successions. Nature 604, 495-+ (2022).
Steiner, A. K. et al. Observed Temperature Changes in the Troposphere and Stratosphere from 1979 to 2018. Journal of Climate 33, 8165-8194 (2020).
Thorne, P. W. et al. Probable causes of late twentieth century tropospheric temperature trends. Climate Dynamics 21, 573-591 (2003).
Santer, B. D. et al. Human and natural influences on the changing thermal structure of the atmosphere. Proceedings of the National Academy of Sciences of the United States of America 110, 17235-17240 (2013).
Sugi, M., Murakami, H. & Yoshimura, J. On the Mechanism of Tropical Cyclone Frequency Changes Due to Global Warming. Journal of the Meteorological Society of Japan 90a, 397-408 (2012).
Walsh, K. J. E. et al. Real World and Tropical Cyclone World. Part II: Sensitivity of Tropical Cyclone Formation to Uniform and Meridionally Varying Sea Surface Temperatures under Aquaplanet Conditions. Journal of Climate 33, 1473-1486 (2020).
Raavi, P. & Walsh, K. Response of Tropical Cyclone Frequency to Sea Surface Temperatures Using Aqua-Planet Simulations. Oceans 2021, 2, 785–810. Tropical Cyclone Future Projections, 91 (2021).
Emanuel, K., Sundararajan, R. & Williams, J. Hurricanes and global warming - Results from downscaling IPCC AR4 simulations. Bulletin of the American Meteorological Society 89, 347-+ (2008).
Emanuel, K. Tropical Cyclone Activity Downscaled from NOAA-CIRES Reanalysis, 1908-1958. Journal of Advances in Modeling Earth Systems 2 (2010).
Camargo, S. J., Emanuel, K. A. & Sobel, A. H. Use of a genesis potential index to diagnose ENSO effects on tropical cyclone genesis. Journal of Climate 20, 4819-4834 (2007).
Korty, R. L., Camargo, S. J. & Galewsky, J. Tropical Cyclone Genesis Factors in Simulations of the Last Glacial Maximum. Journal of Climate 25, 4348-4365 (2012).
Pausata, F. S. R. et al. Tropical cyclone activity enhanced by Sahara greening and reduced dust emissions during the African Humid Period. Proceedings of the National Academy of Sciences of the United States of America 114, 6221-6226 (2017).
Yan, Q., Korty, R., Zhang, Z. S. & Wang, H. J. Evolution of tropical cyclone genesis regions during the Cenozoic era. Nature Communications 10 (2019).
Camargo, S. J. et al. Characteristics of Model Tropical Cyclone Climatology and the Large-Scale Environment. Journal of Climate 33, 4463-4487 (2020).
Murakami, H. & Wang, B. Future Change of North Atlantic Tropical Cyclone Tracks: Projection by a 20-km-Mesh Global Atmospheric Model. Journal of Climate 23, 2699-2721 (2010).
Kumar, K. K., Rajagopalan, B. & Cane, M. A. On the weakening relationship between the Indian monsoon and ENSO. Science 284, 2156-2159 (1999).
Sharmila, S. & Walsh, K. Recent poleward shift of tropical cyclone formation linked to Hadley cell expansion. Nature Climate Change 8, 730-736 (2018).
Walsh, K. J. et al. Tropical cyclones and climate change. Wiley Interdisciplinary Reviews: Climate Change 7, 65-89 (2016).
Rappin, E. D., Nolan, D. S. & Emanuel, K. A. Thermodynamic control of tropical cyclogenesis in environments of radiative‐convective equilibrium with shear. Quarterly Journal of the Royal Meteorological Society 136, 1954-1971 (2010).
Sugi, M., Yoshida, K. & Murakami, H. More tropical cyclones in a cooler climate? Geophysical Research Letters 42, 6780-6784 (2015).
Merlis, T. M., Zhao, M. & Held, I. M. The sensitivity of hurricane frequency to ITCZ changes and radiatively forced warming in aquaplanet simulations. Geophysical Research Letters 40, 4109-4114 (2013).
Neale, R. B. et al. Description of the NCAR community atmosphere model (CAM 5.0). NCAR Tech. Note NCAR/TN-486+ STR 1, 1-12 (2010).
Smith, R. & Gent, P. Reference manual for the parallel ocean program (POP). Los Alamos unclassified report LA-UR-02-2484 (2002).
Lawrence, D. M. et al. Parameterization improvements and functional and structural advances in version 4 of the Community Land Model. Journal of Advances in Modeling Earth Systems 3 (2011).
Hunke, E., Lipscomb, W., Turner, A., Jeffery, N. & Elliott, S. CICE: The Los Alamos Sea ice Model Documentation and Software User’s Manual Version 5 (Tech. Rep. LA-CC-06–012). Los Alamos, NM: Los Alamos National Laboratory (2015).
Bacmeister, J. T. et al. Exploratory High-Resolution Climate Simulations using the Community Atmosphere Model (CAM). Journal of Climate 27, 3073-3099 (2014).
Tory, K. J., Chand, S. S., McBride, J. L., Ye, H. & Dare, R. A. Projected Changes in Late-Twenty-First-Century Tropical Cyclone Frequency in 13 Coupled Climate Models from Phase 5 of the Coupled Model Intercomparison Project. Journal of Climate 26, 9946-9959 (2013).
Bell, S. S., Chand, S. S., Tory, K. J. & Turville, C. Statistical Assessment of the OWZ Tropical Cyclone Tracking Scheme in ERA-Interim. Journal of Climate 31, 2217-2232 (2018).
Emanuel, K. & Nolan, D. S. in 26th conference on hurricanes and tropical meteorolgy.
Rotunno, R. & Emanuel, K. A. An air–sea interaction theory for tropical cyclones. Part II: Evolutionary study using a nonhydrostatic axisymmetric numerical model. Journal of Atmospheric Sciences 44, 542-561 (1987).
Yan, Q., Korty, R., Zhang, Z. & Wang, H. Evolution of tropical cyclone genesis regions during the Cenozoic era. Nat Commun 10, 3076 (2019).
Murakami, H. Substantial global influence of anthropogenic aerosols on tropical cyclones over the past 40 years. Science advances 8, eabn9493 (2022).

Table 1

Model experimental details of atmospheric gases, orbital year, and surface data.
Model Parameters	Orbital forcing experiments		Greenhouse gases forcing experiments
Model Parameters	MIS5e (Eemian)	MIS5d	Present-day	2×CO2
Orbital Year	125ka	115ka	1990	1990
CO₂ (ppm)	276.016	275.623	367.0	734.0
Other greenhouse gases (ppm)	CH₄ = 640.417e-9 CFC-11 = 0.0 CFC-12 = 0.0 N₂O = 263.084e-9	CH₄ = 472.015e-9 CFC-11 = 0.0 CFC-12 = 0.0 N₂O = 251.334e-9	CH₄ = 1760.0e-9 CFC-11 = 653.45e-12 CFC-12 = 535.0e-12 N₂O = 316.0e-9	CH₄ = 1760.0e-9 CFC-11 = 653.45e-12 CFC-12 = 535.0e-12 N₂O = 316.0e-9
Aerosol forcing	Pre-Industrial (PI) condition		Present Day (PD) condition
Surface data	Pre-Industrial (PI) condition		Present Day (PD) condition
Urban effect	No		Yes
Crop	No		Yes

There is NO Competing Interest.

NCTCsupplement16022023.docx

Download PDF

Journal Publication

published 13 Oct, 2023

Read the published version in Nature Communications →

Version 1

posted

You are reading this latest preprint version

Moisture control of tropical cyclones in high-resolution simulations of paleoclimate and future climate

Status:

Journal Publication

Version 1

Abstract

Figures

Introduction

Results

Climate response to orbital and greenhouse gas forcings

Large-scale TC Genesis Environments

Changes In Detected TC Genesis

Changes In TC Intensity

Summary And Discussion

Materials And Methods

Model Description

TC Detection And Tracking Schemes

A. Traditional TC Tracking Scheme

B. Phenomenon-based Tracking Scheme (OWZP)

TC Genesis Potential Index (GPI)

GPI Component Analysis

Declarations

References

table

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1