ARs occur most frequently in the extra-tropics (20°-60°), existing up to 15% of the time (Fig. 2a). ARs are more prevalent during winter months (not shown). GW19 flags 1500±100 ARgenesis events in the GISS-E2.1+MERRA2u/v annually using the 2018-2022 baseline 85^th% IVT. ARgenesis peaks in the subtropics/mid-latitudes at around 20°-40° (Fig. 2b) with around 150 ARgenesis events in the Atlantic basin, 200 in the Pacific basin, and over 500 events generated in these latitude bands globally.

We conditionally sort surface LHF based on GW19 flagged place and times of ARgenesis, AllARs (i.e. the entire AR lifetime, with various levels of maturity, after the genesis timestep), and nonARs. During nonAR, the distribution of surface CYGNSS LHF out of the ocean peaks at -135W/m² (Fig. 2c). This peak and the rest of the distribution of fluxes are larger (more negative) compared to the distributions for ARgenesis and AllARs (Fig. 2c), reflecting less evaporation and heat flux from the ocean into the atmosphere during ARs. GISS-E2.1+MERRA-2u/v has a consistent result, although the peaks in the distribution are less sharp, and the change in the mean of the distribution during AR times is ~-50W/m². For the model and CYGNSS LHF, the distributions during nonAR times differ enough that we can reject the null hypothesis of the Kolmogorov-Smirnov test at a 5% significance level: the differences could not be explained by random sampling from the same population. Reduced LHF during ARs suggests they are not fueled by local convection.

Next, we focus on spatial patterns of LHF differences. NonAR events have ~30-70 W/m² greater heat flux out of the ocean in the extra-tropics than AllARs (Supp. Fig 1). This pattern holds during ARgenesis, but the difference is greater for AllARs (Supp. Fig. 1a versus Supp. Fig. 1b). There is a clear transition from negative to positive differences at ~20°. However, the fraction of time of the year the tropics have as ARs is very small (Fig. 2a, b). The pattern here is highly consistent with what is known of ARs: they are a principal conduit for poleward heat transportation (Shinoda et al., 2019; Sodemann & Stohl, 2013).

3.2 Moisture fluxes from further upwind and equatorward, not directly below primarily fuel ARs

To facilitate our assessment of this change in LHF exhibited in GISS-E2.1, we apply VSD tracers to determine where ARs draw their moisture (and energy). VSDs allow us to determine both the average latitude of evaporation as well as the transport distance for water vapor during ARgenesis, AllAR, and nonARs.

An exemplar of water provenance during a single AR event that impacted Chile in January 2021 appears in Fig. 3. This event begins in a high frequency genesis region for ARs, in the subtropical Pacific northeast of New Zealand (Fig. 2a). During genesis, this AR is closer in shape (pink outline) to a typical ETC (Fig. 3a), and has elongated just enough to qualify as an AR using the GW19 algorithm. Moisture is drawn into the event from equatorward, with the highest density of moisture being drawn in from a bulls-eye centered on the upwind edge of the AR. Compare this ARgenesis moisture to climatological moisture from the footprint of this storm (Fig. 3c), which is centered on the downwind edge of the area and is more strongly overlapping with the footprint of the AR itself. The shift in latitude for this ARgenesis event to the climatological moisture for this AR is obvious, at about 15° equatorward.

This event became remarkable when it flooded Central Chile during the dry season (Valenzuela et al., 2022). As it comes onshore in Chile, some moisture continues to be sourced (Fig. 3b) as the same regions as during ARgenesis (Fig. 3a), with a long tail for moisture sources trailing along the AR’s path, 1000’s km away. In contrast, the climatological moisture source for an air parcel just off the coast of Chile with the same AR footprint remains largely local (Fig. 3d). This 2021 Chile AR event exemplifies how ARs source moisture from upstream and equatorward of where climatological or non-ARs moisture is sourced. Note that some of the moisture is sourced from the east of the AR during genesis (Fig 3a), which fits partly with the air-feeder mechanism of Dacre et al. (2019). By the mature phase, this example more resembles a river, with most of the moisture sourced from far upstream and equatorward (Fig. 3b).

In Fig. 4, we have repeated the calculation of moisture provenance and centroid of that provenance as shown in Fig. 3 for each of the ~7500 ARgenesis events over 2018-2022 and contrast with a calculation moisture provenance and centroid for each of the ARgenesis footprints during nonAR times and again during ARall times. This yields moisture provenance with attendant transport distance and average source latitude for each ARgenesis footprint at the time of genesis, climatological allARs, and climatological non-ARs.

Each location and event-type has its own transport pathway, yielding variability from event to event. We distill this variation using a zonal average and focus on the “average vapor source latitude equatorward” – defined as the number of degrees latitude between average source and ARgenesis footprint (Fig. 4a-c). AR events always have a vapor source latitude further equatorward, particularly in the mid- to high-latitudes. The distinction in moisture source latitude exists globally (includes land and ocean), but is also apparent when we restrict our view to only the Pacific and Atlantic basins with moisture sourced upwards of 10° further equatorward for allARs than non-ARs (Figs. 4b, c).

The moisture source transport distance for ARs is larger than it is for non-ARs, particularly poleward of 40°, with ARs transporting moisture nearly an additional 1000 km on average (Fig. 4d-f). This lengthening of transport distance is less obvious in the area overlapping with the CYGNSS retrieval area, due to the high amount of variability. Importantly, this increased moisture transport distance during AR events is consistent with CYGNSS and GISS-E2.1+MERRA-2u/v reduction in out of ocean LHF under AR events.

The VSD tracers here distinguish between cases where moisture is at equilibrium versus being transported - something satellite data cannot currently provide. A moist air mass may appear to have constant column water vapor in satellite observations because of multiple processes, such as moisture sinks like precipitation or entrainment of dry air at the lateral boundaries of the plume are balanced by a moisture source from LHF beneath the plume. In this case, the definition of the AR would not guarantee that the transport distance of moisture within the AR is large. Moreover, ocean evaporation and convection along the path of the AR might itself provide for excess IVT. Using the tracers, we have demonstrated that it is long-distance transport that creates the plurality, represented by the 10° latitude excess distance of transport for ARs over climatology.

We parse ARgenesis AllAR, and nonAR transport distances by season (Supp. Fig. 2) and find that the longer moisture transport distance of ARs (and ARgenesis) compared to nonARs in the Southern Hemisphere persists year-round; increased transport distance contrast for the Northern Hemisphere is greatest in winter, particularly at higher latitudes.