Midlatitude Oceanic Fronts Strengthen the Hydrological Cycle Between Cyclones and Anticyclones

The Kuroshio‐Oyashio Extension and Gulf Stream oceanic frontal zones are characterized by enhanced activity of synoptic‐scale cyclones and anticyclones and vigorous air‐sea heat and moisture exchange in the cold season. However, the time‐mean air‐sea exchange attributed separately to cyclones and anticyclones has not been assessed. Here we quantify cyclonic and anticyclonic contributions around the frontal zones to surface turbulent heat fluxes, precipitation, and the associated hydrological cycle using atmospheric general circulation model experiments with observed and artificially smoothed sea‐surface temperature gradients. The evaluation reveals that precipitation exceeds evaporation climatologically within cyclonic domains while evaporation dominates within anticyclonic domains. These features as well as the net moisture transport from anticyclonic to cyclonic domains are all enhanced by the sharpness of the frontal zones. Oceanic frontal zones thus climatologically act to strengthen the hydrological cycle. These findings aid our understanding of the relationship between midlatitude air‐sea interactions on synoptic‐ and longer‐time scales.

The importance of synoptic-scale weather variability for driving surface turbulent heat flux (THF) fluctuations has been recognized (Ogawa & Spengler, 2019;Seo et al., 2023).Nevertheless, whether these synoptic-scale fluctuations are important for the time-mean midlatitude air-sea interactions, in particular the exchange of heat and moisture on the air-sea interface, has not been well understood.To assess the importance of synoptic-scale processes for the time-mean frontal air-sea interactions, Tsopouridis et al. (2021;hereafter TSS21) evaluated the contribution of extratropical cyclones to the climatological surface fluxes and assessed the impact of the sharpness of oceanic fronts over the North Pacific (NP) and North Atlantic (NA).They found that extratropical cyclones are mainly important for a response in precipitation and only play a secondary role in enhancing or suppressing THF over warm/cool SST anomalies, respectively.However, as the attribution of atmospheric fields to extratropical cyclones by TSS21 is based on a fixed-size circular mask centered on the position of a surface cyclone, their analysis does neither represent the actual size of individual cyclones nor capture their structure.Furthermore, TSS21 did not assess the potential contribution of anticyclones.Hence, we still lack insight into the relative contributions of cyclones and anticyclones, which limits our understanding of midlatitude air-sea interactions around oceanic frontal zones.
Recently, Okajima et al. (2021;hereafter ONK21) proposed a method to quantify contributions from cyclonic and anticyclonic domains to Eulerian statistics, demonstrating that instantaneous local curvature can be used to distinguish low-level migratory cyclones and anticyclones as well as upper-level pressure troughs and ridges.We apply the ONK21 methodology to a set of atmospheric general circulation model (AGCM) experiments with observed climatological-mean and artificially smoothed SST fields to quantify the cyclonic and anticyclonic contributions to THF and precipitation along the two major oceanic frontal zones over the NP and NA.Our results provide insights into the hydrological cycle along the SST front as well as into the moisture exchange between cyclones and anticyclones.

AGCM Experiments
We analyze the same 6-hourly outputs of the AGCM experiments as analyzed by TSS21.The AGCM is the version 3 of the AGCM for the Earth Simulator (AFES; Ohfuchi et al., 2004;Enomoto et al., 2008;Kuwano-Yoshida et al., 2010).The data period spans from 1 September 1981 to 31 August 2001 with a horizontal resolution of ∼0.5°(T239) and 48 vertical levels.In the control experiment (CNTL), the climatological-mean SST derived from the 0.25°daily OISST (Reynolds et al., 2007) was prescribed.In the SMTHK and SMTHG experiments, the prescribed SST fields have been horizontally smoothed over the western NP and NA, respectively.We regard the differences between the corresponding smoothed experiments and CNTL (i.e., CNTL-SMTHK and CNTL-SMTHG, respectively) as responses to the realistic KOE and GS fronts.Figures 1a and 1b show the differences in prescribed SST in CNTL relative to SMTHK and SMTHG, respectively.
We focus on wintertime (DJF) mean responses following TSS21.Statistical significance is assessed by a Student's t-test in setting the degrees of freedom to be the number of winter seasons available for our analysis.In Section 3.3, we calculate area-averaged THF, precipitation, and evaporation minus precipitation (E P) within the oceanic frontal zones over the NP (140°E-180°, 30°-50°N) and the NA (70°-30°W, 30°-55°N) to focus on regions of active synoptic-scale eddies, using the data only at grid points over the ocean.The key conclusions are not quite sensitive to the choice of the domains (Figures S1 and S2 in Supporting Information S1).Hereafter, a positive value of THF indicates an upward flux.

Separating Contributions From Cyclonic and Anticyclonic Domains
We determine cyclonic and anticyclonic domains by evaluating the two-dimensional curvature of the horizontal wind (see ONK21).We use the local instantaneous curvature of unfiltered 850-hPa wind to determine our cyclonic and anticyclonic domains, because near-surface wind is likely to be influenced by underlying SST directly through turbulent vertical mixing of momentum (Hayes et al., 1989;Wallace et al., 1989) or pressure adjustment mechanism (Lindzen & Nigam, 1987), which makes it rather difficult to extract the contributions from Geophysical Research Letters 10.1029/2023GL106187 synoptic-scale cyclones and anticyclones.We did not smooth the curvature horizontally, to prevent the influence of the topography from spreading.We set a curvature threshold of ±4.0 × 10 7 m 1 to determine cyclonic and anticyclonic domains, respectively, corresponding to a curvature radius of 2,500 km.Any grid point with a  (c, d) Climatological cyclonic contribution to E-P (shading in mm/day) in CNTL over the (c) North Pacific (NP) and (d) North Atlantic (NA) and climatological p c (contours every 5%, thick for 35%, only over the ocean).Panels (e-h) same as in panels (c, d), respectively, but for the (e, f) anticyclonic and (g, h) neutral domains' contributions and p a/n .Dashed boxes signify the domains in which the area-averaged contributions are calculated separately for the NP and NA.

Geophysical Research Letters
10.1029/2023GL106187 curvature radius larger than the threshold is thus named "neutral," because it is classified neither as "cyclonic" nor as "anticyclonic." We locally obtain additive climatological-mean contributions of cyclonic, anticyclonic, and neutral domains to a given variable X by accumulating its instantaneous values within those three types of domains separately and then dividing the accumulated values by the total number of times steps: where α, p, and N denote a typical amplitude of X, the probability and number of time steps, respectively, for the domains of a given type, including cyclonic, anticyclonic, and neutral domains as signified by subscripts c/a/n, respectively.Hereafter, the climatological contribution, as denoted by an overbar, and probability of a cyclonic, anticyclonic, or neutral domain are represented as X c/a/n and p c/a/n , respectively, unless otherwise denoted.
Evaporation (E) is estimated from surface latent heat flux by assuming the latent heat of vapourization as 2.5 MJ/kg.
Relative magnitude of the contribution from neutral domains to that from cyclonic or anticyclonic domains can depend on the threshold of curvature.Nevertheless, our key conclusions, especially about the contrasting E c P c and E a P a , are not very sensitive to setting the curvature threshold to zero (Figure S3 in Supporting Information S1) or ±1.0 × 10 6 m 1 (Figure S4 in Supporting Information S1).We also obtain similar results based on the curvature of wind at 925-hPa (Figure S5 in Supporting Information S1).
Furthermore, we have confirmed that the results based on CNTL are overall consistent with those based on the JRA-55 reanalysis (Text S1 and Figures S6-S8 in Supporting Information S1).

Evaluating Moisture Transport Between Cyclonic and Anticyclonic Domains
We evaluate the hydrological cycle between cyclonic and anticyclonic domains using two methods.One is an indirect evaluation by comparing E c P c and E a P a .The other is a more direct evaluation by calculating the moisture flux associated with high-pass-filtered wind fluctuations projected onto the upgradient direction of instantaneous flow curvature, which corresponds to the direction pointing from anticyclonic to cyclonic domains.We use this projected flux through cyclone-anticyclone transition zones (i.e., grid points where local curvature radius is greater than 2,500 km) as a measure of the net moisture transport from anticyclonic to cyclonic domains (see Text S2 in Supporting Information S1 for details confirming the equivalence to total moisture transport).

Cyclonic and Anticyclonic Contributions to the Climatological Hydrological Cycle for CNTL
Over the oceanic frontal zones in the NP and NA, E c P c is overall negative (Figures 1c and 1d), indicative of excessive precipitation compared to local evaporation within cyclonic domains in the storm-track core regions.In the storm-track entrance regions, however, a positive E c P c is evident along the warm currents, especially along the Kuroshio Current south of Japan and the Florida Current by Cape Hatteras.
In contrast, E a P a is overall positive (Figures 1e and 1f), especially equatorward of the storm-track cores and along the warm ocean currents.The large E a P a south of ∼30°-35°N is most likely related to the relatively high p a (contoured in Figures 1e and 1f; ONK21).The difference between the distributions of p c and p a is compatible with those of the densities of migratory cyclones and anticyclones based on Lagrangian tracking (Hoskins & Hodges, 2002;Okajima et al., 2023).In the storm-track entrance regions, the large positive E a P a is suggestive of the importance of cold-air outbreaks for air-sea heat exchange in these regions, acting as thermal damping for transient, subweekly fluctuations (Okajima et al., 2022).E n P n is overall positive but weaker than E a P a (Figures 1g and 1h), especially in the storm-track entrance regions and over the NA.

Local Response of the Climatological-Mean Hydrological Cycle to Oceanic Frontal Zones
In response to sharp SST gradients in the NP oceanic frontal zones, E c P c significantly decreases, especially over the cool SST anomalies along the main branch of the Oyashio Front and the front over the Japan Sea (Figures 1a and 2a).In the former region, this response acts to enhance the climatological-mean precipitation excess (P c E c ) by up to ∼30% (Figure 1c), with no apparent local change in p c (Figure 2a).In the latter region, the climatological-mean excess in evaporation (E c P c ) is reduced substantially together with a slight decrease in p c .In addition, the weaker negative E c P c response around the second branch of the Oyashio Front (around 40°N , 170°E) is likely associated with decreased p c , which is consistent with the cyclone density response in TSS21.Around the Kuroshio Extension, however, E c P c does not significantly change in response to the oceanic frontal zone, despite the underlying warm SST anomaly (Figure 1a).
Over the oceanic frontal zone in the NA, the negative E c P c response is even more evident over the pronounced cool SST anomaly on the poleward flank of the GS (Figure 2b).The positive E c P c response south of the GS front is overall modest despite the underlying warm SST anomaly of >3 K (Figure 1b), but it is marked along the Florida Current just off the U.S. east coast, where the enhancement of THF dominates over the precipitation increase (Figure S9 in Supporting Information S1).These E c P c responses are not associated with changes in p c .However, the positive E c P c response in the central NA (∼30°-40°W) is most likely associated with a decrease in p c , which is consistent with TSS21.
Conversely, the positive E a P a is significantly enhanced (by ∼20%-40%) in response to the NP and NA oceanic fronts, especially over the warm ocean currents and SST anomalies equatorward of them (Figures 1a, 1b, 2c,  and 2d).Meanwhile, the negative E a P a responses over the cool SST anomalies are substantially weaker than their cyclonic counterparts along the main branch of the Oyashio Front and the front over the Japan Sea (Figure 2c) and GS (Figure 2d).The enhanced E a P a is due partly to the increased p a , especially around the NP oceanic frontal zones (Figure 2c).The p a also increases in the climatological anticyclones southeast of the respective oceanic frontal zones.The neutral domains contribute moderately compared to the cyclonic and anticyclonic contributions (Figures 2e and 2f).The spatial patterns of the E c P c , E a P a , and E n P n responses in the NA bear greater similarity than those in the NP.
The differences between the responses of E c P c and E a P a are related to responses in both evaporation and precipitation (Figure S9 in Supporting Information S1).Within cyclonic domains, the suppression of evaporation over the cool SST anomalies as well as the pronounced precipitation increase over the warm SST anomalies yield the overall negative E c P c response, whereas the corresponding anticyclonic response in precipitation is much weaker.The E a response is somewhat greater than the E c counterpart over the warm SST anomalies around the NP frontal zones and along the Kuroshio, while the opposite is the case over the warm SST anomalies along the GS.This as well as the precipitation enhancement within anticyclonic domains leads to the closer resemblance between the patterns of the E c P c and E a P a responses over the NA than over the NP.

Area-Averaged, Net Contributions to the Hydrological Cycle
The area-averaged net (viz.sensible plus latent) THF c is substantially (by ∼30%-50%) larger than THF a over both the NP and NA (Figures 3a and 3b).With the curvature threshold of 2,500 km, THF n is roughly comparable with THF c and THF a , with p c , p a , and p n being comparable (Figures 1c-1h).In contrast, P is associated predominantly with cyclonic domains (Figures 3a and 3b).The additional contribution from neutral domains may be associated with atmospheric fronts, cold-air outbreaks, or planetary waves.The positive E a P a and E n P n as well as negative E c P c are indicative of their distinct roles in the climatological hydrological cycle (Section 3.1).The result is consistent with that based on the JRA-55 (Figure S7 in Supporting Information S1).
The oceanic frontal zones significantly increase area-averaged THF a over the NP and NA (Figures 3c and 3d).The corresponding response of THF c is weaker than THF a , especially over the NP (Figure 3c).This suggests the significance of anticyclones in the restoration of near-surface baroclinicity, which is essential for storm-track maintenance (e.g., Hotta & Nakamura, 2011;Nakamura et al., 2004Nakamura et al., , 2008;;Nonaka et al., 2009;Papritz & Spengler, 2015;Taguchi et al., 2009).Meanwhile, the oceanic frontal zones significantly amplify area-averaged P c over the NP and NA (Figures 3c and 3d).Over the NA, precipitation is enhanced slightly also within anticyclonic and neutral domains (Figure 3d), which may be associated with atmospheric fronts and cold-air outbreaks.Those THF c and P c responses are compatible with TSS21.
As a net response of the hydrological cycle to the realistic oceanic frontal zones, E a P a and E c P c increases and decreases, respectively (Figures 3c and 3d), while the decrease in E n P n is insignificant.The contrasting features of the cyclonic and anticyclonic contributions to the E P response are more pronounced in the NP than in the NA, which is compatible with the greater similarity in the distributions over the NA between E c P c and E a P a (Figures 2b and 2d).Note that the horizontal averaging over a relatively large area including both positive and negative SST anomalies associated with the smoothing is likely to yield a weaker response due to a partial offset between the opposing responses.Still, the significant area-averaged responses are indicative of the rectified E P response to the oceanic frontal zones due presumably to the nonlinearity of the Clausius-Clapeyron equation.Note that the relative roles of E c P c and E a P a to the response to the sharpness of the oceanic fronts are not affected qualitatively by the corresponding response of p c and p a (Figure S10 in Supporting Information S1).

Moisture Transport Between Cyclonic to Anticyclonic Domains
The preceding section suggests that the oceanic frontal zones enhance the moisture supply from the ocean mainly within anticyclonic domains with an enhanced transport into cyclonic domains.To verify this, we evaluate moisture transport between these domains as described in Section 2.3.
The climatological net moisture transport through cyclone-anticyclone transition zones in CNTL is overall positive for both the NP and NA (Figures 4a and 4b), indicative of the net moisture transport from anticyclonic to cyclonic domains.This is consistent with the results obtained in Section 3.3 and the results based on the JRA-55 (Figure S8 in Supporting Information S1).This result is also compatible with Bui and Spengler (2021), who suggested the importance of feeding airstreams taking up moisture ahead of a cyclone.The moisture transport is particularly strong equatorward of the precipitation maxima (Figures 4a and 4b), where E P is positive in anticyclonic domains (Figures 1e and 1f).
The oceanic frontal zones strengthen the climatological moisture transport from anticyclonic to cyclonic domains (Figures 4c and 4d), which is compatible with their distinct contributions to the E-P response (Figures 3c and 3d).The more significant enhancement in the moisture transport over the NA is most likely related to the stronger positive SST anomaly over the NA (Figure 1b).Increased specific humidity around the warm ocean currents and positive SST anomalies equatorward of the oceanic frontal zones as well as intensified low-level storm-track activity are the most likely causes for the enhanced moisture transport (Figure S11 in Supporting Information S1).

Conclusions
We assessed the role of cyclones and anticyclones in air-sea interactions over midlatitude oceanic frontal zones in the wintertime Northern Hemisphere by quantifying cyclonic and anticyclonic contributions to the climatological THF, precipitation, and E P as well as their responses to the oceanic frontal zones based on AGCM experiments.In addition, we delineated the climatological moisture transport between cyclonic and anticyclonic domains and their corresponding response to the influence of the SST fronts.
We have demonstrated that synoptic-scale, sub-weekly disturbances play an important role in midlatitude air-sea interactions on a climatological time scale, bridging our understanding of midlatitude air-sea interactions from Geophysical Research Letters 10.1029/2023GL106187 synoptic to longer time scales.When smoothing the SST gradients, THF is climatologically reduced when compared to realistic oceanic frontal zones.This reduction mainly occurs within anticyclonic domains, while precipitation is climatologically enhanced predominantly within cyclonic domains.Consistently, the net moisture transport from anticyclonic to cyclonic domains is strengthened when realistic oceanic frontal zones are present.These changes are mainly attributable to a moisture increase around the anomalously warmer waters as well as enhanced storm-track activity, yielding an overall strengthened climatological hydrological cycle around midlatitude oceanic frontal zones.
Our results thus emphasize that variations in synoptic-scale THF and precipitation are modulated by midlatitude frontal zones and SSTs around them.The modulations in heat and moisture release along oceanic frontal zones can affect storm-track activity and a westerly jet response (e.g., Kuwano-Yoshida & Minobe, 2017;Nakamura et al., 2008), though this requires further studies to pinpoint the mechanisms including the enhancement of moisture transport from anticyclones to cyclones.The difference in the distinctiveness of the E c P c and E a P a responses between the NP and NA also requires further investigation, especially with respect to the distance between oceanic fronts and upstream continents as well as the amplitude of the SST anomalies.
The contributions of cyclonic, anticyclonic, and neutral domains to THF can be compared with the results of previous studies that showed the importance of cold air outbreaks for vigorous THF behind cyclones where cold, dry, continental air spreads over oceanic frontal zones (e.g., Shaman et al., 2010).The substantial THF contribution of neutral domains (Figure 3) is compatible with the importance of cyclone-anticyclone transition zones for large THF (Rudeva & Gulev, 2011;Tilinina et al., 2018), while cyclones dominate the precipitation response (Figure 3).While the majority of the precipitation in cyclones is most likely associated with atmospheric fronts (Small et al., 2023), oceanic fronts influence these atmospheric fronts both directly and indirectly (Bui & Spengler, 2021).The direct effect of sensible heat flux along the flanks of oceanic fronts, however, is small and mostly detrimental in airstreams with atmospheric fronts (Reeder et al., 2021).The indirect effect is related to the moisture transport from to cyclones (Figure 4), fueling the hydrological cycle.Stationary atmospheric fronts, or fronts with local genesis along oceanic fronts, are not generally attributed to cyclonic domains.

10.1029/2023GL106187
These fronts can also play a role in shaping time-mean atmospheric structures (Masunaga et al., 2020a(Masunaga et al., , 2020b;;Reeder et al., 2021).An analysis combining a frontal analysis with our cyclonic and anticyclone domains could shed further light on the intricate roles of these synoptic systems in shaping the time-mean precipitation distribution.
contributions to air-sea heat and moisture exchange are quantified around midlatitude oceanic frontal zones • Oceanic frontal zones enhance surface turbulent heat fluxes within anticyclones and precipitation within cyclones, respectively • Sharpness of midlatitude oceanic frontal zones strengthens the net moisture transport from anticyclones to cyclones Supporting Information: Supporting Information may be found in the online version of this article.

Figure 1 .
Figure 1.(a, b) Climatological winter-mean differences in prescribed sea surface temperature (SST) (shading in K) in CNTL relative to (a) SMTHK and (b) SMTHG.Contours denote climatological SST prescribed in CNTL (contour every 2 K, thick for every 10 K).(c, d) Climatological cyclonic contribution to E-P (shading in mm/day) in CNTL over the (c) North Pacific (NP) and (d) North Atlantic (NA) and climatological p c (contours every 5%, thick for 35%, only over the ocean).Panels (e-h) same as in panels (c, d), respectively, but for the (e, f) anticyclonic and (g, h) neutral domains' contributions and p a/n .Dashed boxes signify the domains in which the area-averaged contributions are calculated separately for the NP and NA.

Figure 2 .
Figure 2. (a) Cyclonic contribution to the response (CNTL-SMTHK) of the climatological E-P (shading in mm/day).Stipples signify statistically significant signals at the 90% confidence level by a Student's t-test.Contours denote p c (every 1.5%, zero contours omitted; dashed for negative values).Panel (b) same as in panel (a), but for CNTL-SMTHG.Panels (c-f) same as in panels (a, b), respectively, but for the (c, d) anticyclonic and (e, f) neutral domains' contributions and p a/n .Dashed boxes signify the domains within which area-averaged contributions are calculated separately for the North Pacific and North Atlantic.

Figure 3 .
Figure 3. (a) Area-averaged climatological-mean net turbulent heat flux (THF; W/m 2 ), total precipitation (TPRAT(P); mm/ day), and E-P (mm/day) over the North Pacific frontal region (rectangular domain marked in Figure 1) based on CNTL.Red, blue, and green bars represent the contributions from cyclonic, anticyclonic, and neutral (neither cyclonic nor anticyclonic) domains, respectively.Whiskers signify standard errors of the means.Panel (b) same as in panel (a), but for the North Atlantic frontal region.Panels (c, d) same as in panels (a, b), respectively, but for the response to oceanic frontal zones as extracted in (c) CNTL-SMTHK and (d) CNTL-SMTHG.

Figure 4 .
Figure 4. (a) Climatological-mean net moisture transport (from anticyclonic to cyclonic domains) integrated from the surface to 100 hPa (shading in mm m/s) over the North Pacific for CNTL.Brown and green contours denote E and P (mm/day), respectively.Panel (b) same as in panel (a), but for the North Atlantic.Panel (c, d) same as in panels (a, b), respectively, but for responses as (c) CNTL-SMTHK and (d) CNTL-SMTHG.Stipples signify statistically significant signals at the 90% confidence level by a Student's t-test.Black contours indicate the climatological-mean moisture exchange (mm m/s) in CNTL (spatially smoothed).