In this manuscript, the authors demonstrate the importance of extratropical disturbances in the development and persistence of South Pacific Convergence Zone events. The innovative methodology combines the objective identification of cloud band events and PV structures, providing a robust link between the extratropical forcing and the organized convective activity in the subtropics. The results also advance the understanding of the dynamic environment necessary to form and sustain persistent cloud bands and could be extended to other regions. However, the manuscript could benefit from a more in-depth discussion of the main findings in light of previous research. I also have some minor questions related to details in the methodology, with clarification of these points facilitating the future reproducibility of the results. Thus, I would recommend the manuscript undergo minor revisions before acceptance and publications.  

Major Comments 

1. Section 3.3. In this section, the authors propose a mechanism for the formation of the cloud bands based on the presence of PV structures. On the poleward side of the CB centre, cyclonic PV anomalies destabilise the lower troposphere and induces convection and ascending motion. On the other hand, the diabatic heat released by the enhanced convection contributes to the anticyclonic PV anomalies ahead of the cloud band axis. The mechanism is similar to that proposed by Van der Wiel et al (2015). According to those authors, the propagation of the Rossby wave trains forming the SPCZ is subjected by two main forcings: horizontal advection of cyclonic vorticity anomalies, dictating the propagation of the RW into the tropics, counteracted by anticyclonic vorticity tendencies from vortex stretching due to the enhanced convection and ascent at the SPCZ axis. Similar mechanisms were also observed in the SACZ region. Thus, it would be informative to discuss the results obtained by the authors with those from previous literature. This discussion could also be included in the conclusions. 
2. Zilli and Hart (2021) demonstrated that transient cloud bands over South America occur poleward of the persistent ones. They also showed that the relationship between the vorticity tendencies and the local environment are different for these events when compared to the more tropical persistent cloud bands. In the manuscript, the authors also mentioned this in section 3.6.2. However, the main analysis in sections 3.3 and 3.4 does not take this into account. Results in Fig 5 also suggest a clear seasonality in the transient cloud bands, with austral spring and summer events occurring at similar latitude as persistent events, but austral fall and winter transient events located primarily south of 20^oS. To what extent does the local environment of the cloud bands could increase the variability and affect the composite results in Figs 3 and 4?
3. One of the findings of the manuscript is that “... long-lived cloud bands are distinguished [from transient events] by a higher frequency of upstream PV structures persisting throughout the cloud band lifecycle.” (lines 323-333). The frequency of upstream PV structures depends on the upstream area of the cloud band, as illustrated in Figure A1. However, this area can vary massively among cloud band days, with the events in Figure 1 serving as a great example to that. Thus, what is the relationship between the persistence of the cloud band and the upstream area, and to which extend this can affect the frequency of PV structures associated with the events.   

Minor Comments 

1. The analysis considers ERA5 data since 1959. How reliable are ERA5 data, and in particular OLR, for the years before 1979? Would the same results be obtained if considering only data after 1979?
2. The caption suggests that the contours in Fig 2a represent the number of days with cloud bands. However, a maximum of ~50 days per year is below what would be expected for the region.  Previous papers (Haffke & Magnusdottir, 2013) estimated that the SPCZ is present in 30-40% of the NDJFMA months, with would be equivalent to ~70 days per year. From Figure 6a, the detection algorithm identified ~3000 cloud band events which, considering 63 years (1959-2021), results in 48 events per year, on average. This number is close to the numbers represented by the contours in Fig 2a. Thus, I would like to double-check if the information in Figs 2a and 2b are cloud band days or cloud band events? Reference: Haffke, C., & Magnusdottir, G. (2013). The South Pacific Convergence Zone in three decades of satellite images. Journal of Geophysical Research: Atmospheres, 118(19), 10,839-10,849. https://doi.org/10.1002/jgrd.50838 
3. A suggestion for figure 2c: repeat the contours from Fig 2a here, for reference. 
4. The caption in Fig 3 refers to more panels than included. Also, the sentence “The white contour represents the 210 Wm−2 cloud band detection threshold.” refers only to panels a-c, so it would be better to place it with the description of these panels. 
5. In Figures 3 and 4, the composites for day +2 are centred in the centroid of the cloud bands objects detected in day +2? If so, I would expect to have some OLR values below 210 W.m-2 in Fig 4f. Furthermore, what is the centroid considered for those events persisting only one day? If not, what is the reference considered? Please, clarify this in Session 2.3.2.
6. In section 2.1.3, cloud band events lasting less than 3 days are called “transient” while those persisting 4 or more days are called “persistent”. However, in some parts of the manuscript, “transient”, “short” or “short-lived” events are used interchangeably. See, for example, the caption of figs 3 and 4, as well as the label of the first row in Fig 4. Similarly, “persistent” events are also referred to as “long” or “long-lived” (as in line 332). Consider using only one of the terms. 
7. Figure 6: please, include a description of the significance tests used here (Wilson score and the bootstrap confidence interval) to the methodology section.