Thank you for your feedback. We will ensure to revise the labelling of the Latitudes and Longitudes in the spatial maps to ensure that they are clear.

We thank the reviewer for their time and valuable suggestions to improve the manuscript. Here below, we respond to each individual comment and propose modifications to the manuscript (in italics) to accommodate the concerns raised.

Reviewer comment 1: The pair of concurrent extreme events defined in this paper represent two extreme events occurred within the same location in the same year, no matter if it occurred once or several times. The physics meaning of this definition is unclear. Usually the concurrent extreme events are defined based on daily data, i.e., a pair of extreme events occurring on the same day. The definition in this paper seems to be too crude to obtain physically-meaningful knowledge.

Response: We thank the reviewer for this comment. There are three  distinct motivations for our methodological choice to consider annual time steps in this study. The first concerns data availability: the available dataset from ISIMIP 2b only provides annual occurrences of the six categories of climate extreme events (as mentioned in one of the caveats in lines 79-80 of the manuscript). Unfortunately, no multi-hazard future data is available at daily resolution. Due to this limitation of the dataset, we define concurrent events as two events occurring in the same location, during the same time step (here being the same year). A second reason is that some of the climate extremes we consider play out over longer time scales: for example, droughts may last several months to even years, wildfires may rage an entire summer season, and crop failures may result from extreme conditions during the entire growing season. A third motivation is that the impacts of compound extremes may be larger than those for individual events even in the case where the concurrence is not on a daily timescale. These are sometimes termed temporally compounding extremes (e.g. Zscheischler et al., 2020), although for simplicity we opted not to make this terminological distinction in our study. For example, vegetation impacts of drought events can be aggravated by droughts in consecutive growing seasons (e.g. Bastos et al., 2021). Similarly, societal vulnerability to floods is modulated by the occurrence of successive flood episodes (Chacowry et al., 2018). We therefore argue that there is some relevance to considering concurrent extremes on yearly timescales. 

We nonetheless agree that being able to discriminate the timescale of concurrence of the extreme events we study would allow a more nuanced analysis, as illustrated in lines 23-30. Similarly, we agree with the values of knowing whether a given extreme has occurred once or several times in a given year. We propose to clarify these messages throughout the revised text in Sect. 2 lines 78-87, as follows:

• The dataset we use comes with a number of caveats. A minor caveat is that it does not contain crop failure projections under RCP8.5. More importantly, the data represents the occurrence of an extreme event category as a single event within a grid cell per year, no matter if it occurred once or several times within the same location in the same year. Finally, an extreme event such as a wildfire, river flood or tropical cyclone can only partly cover a given grid cell, whereas other extreme events (heatwaves, droughts and crop failures) are assigned by default to the entire grid cell. Thus, for the former three extremes, we consider that a grid cell is entirely affected when more than 0.5% of the 0.5°x0.5° grid cell area is simulated to be affected by the extreme event. Whilst these are limitations of the dataset, we have three distinct motivations to use it throughout our analysis: (i) the dataset is amongst the most detailed and complete of its kind, and provides information on the occurrence of extreme events within the study region over a very long time period (from 1861 until 2099); (ii) some of the climate extremes we consider play out over longer time scales, for example droughts may last several months to even years, wildfires may rage an entire summer season, and crop failures may result from extreme conditions during the entire growing season; (iii) the impacts of compound extremes may be larger than those for individual events even in the case where the concurrence is not on a daily timescale. These are sometimes termed temporally compounding extremes (e.g., Zscheischler et al. 2020b). For example, vegetation impacts of drought events can be aggravated by droughts in consecutive growing seasons (e.g., Bastos et al. 2021). Similarly, societal vulnerability to floods is modulated by the occurrence of successive flood episodes (Chacowry et al., 2018). We therefore use it as the backbone for this study.
• Bastos, A., Orth, R., Reichstein, M., Ciais, P., Viovy, N., Zaehle, S., Anthoni, P., Arneth, A., Gentine, P., Joetzjer, E., Lienert, S., Loughran, T., McGuire, P. C., O, S., Pongratz, J., and Sitch, S.: Vulnerability of European ecosystems to two compound dry and hot summers in 2018 and 2019, Earth Syst. Dynam., 12, 1015–1035, https://doi.org/10.5194/esd-12-1015-2021, 2021.
• Chacowry, A., McEwen, L. J., & Lynch, K. (2018). Recovery and resilience of communities in flood risk zones in a small island developing state: A case study from a suburban settlement of Port Louis, Mauritius. International Journal of Disaster Risk Reduction, 28, 826-838.

Reviewer comment 2: From fig.1,  among the 15 pairs of extremes, all the concurrent extremes including heatwaves show the strongest increase. It this related to the apparent global warming trend? And all the future years will become extreme years of heatwaves?

Response: Yes, the increases shown in Fig.1 are due to the global warming trend. However, while our analysis shows that frequency of heatwaves will increase drastically, not all future years are projected as extreme heatwave years for every location (grid) in East Africa, even under the high warming scenarios (Illustration 1 in attached .zip). As reported by many studies, the frequency and magnitude of heatwaves is projected to increase in many regions of the world as a result of global warming (e.g., Russo et al. 2014, 2015, Thiery et al. 2021, IPCC 2021). This is generally confirmed in our analysis over East Africa (Table 2), whereby the mean percentage area affected by heatwaves is projected to increase by the end of the century under all the three future climate scenarios, with larger increases in the warmer scenarios. Furthermore, the increase in probability and spatial extent of concurrent extremes including heatwaves (in the pairs) is projected to increase, with increases in heatwaves identified as a main driver of these events. This is illustrated in lines 280 -287 of this manuscript. While we indeed find that the change in heatwave occurrence provides a strong contribution to the change in occurrence of concurrent extremes, this is far from our only result/conclusion. Indeed, we also highlight river floods and wildfires as a pair of extremes whose concurrence will increase sharply (Fig. 5), and we analyse other pairs of concurrent extremes not involving heatwaves (Appendix D). We propose to add the following text in the method (Sect. 3.3 Line 118) to clarify that the changes in the concurrent extreme occurrence are projected as a result of global warming.:  

• Considering that the processed impact model simulations account only for climate-induced changes in the extremes (as defined by Lange et al. (2020)), and not changes due to land-use change, we therefore only analyse the climate change-driven changes in concurrent extremes.

Reviewer comment 3: The method used in this paper to search drivers of concurrent extremes seems to be too shallow. So, the results look very obvious, i.e., the global warming is the most important driver. The paper lacks an analysis of the dynamic process that cause to concurrent extreme events 

Response:  Considering that the six extreme events considered each have different meteorological and physical drivers, and that we utilise extreme event data from processed impact model simulations, diving into the meteorological and physical drivers of the concurrent extreme events presents near-insurmountable challenges. We nonetheless believe that our analysis investigating the changes in the probability of occurrence of concurrent events under future climate scenarios in comparison to early-industrial conditions. This is because: (i) The models only account for climate-induced changes in the hazards and not changes due to land-use change (e.g., deforestation fires) (Lange et al., 2020). Therefore, we can only look at climate change/global warming as the driver of the changes in extreme event occurrence, and in turn the changes in concurrent extremes. (ii) We, furthermore, go deeper in our analysis by splitting the change in concurrent extreme event occurrence into: changes in only one variable per pair, and the changes in the (coupling) dependence between extreme events in a pair, as illustrated in Section 3.3. This provides a good basis to formulate some physical hypotheses on the drivers of the changes in concurrent extreme events as illustrated in Sect. 5.3. 

In the revised paper, we aim to improve the communication of the method that we use to identify the drivers of changes in the concurrent extreme occurrences, as well as clarifying in Sect. 5.3 the physical hypotheses to the changes in the concurrent extremes occurrence. -  

• We propose to add the following text in the methods section (Sect. 3.3, Line 118-122)  to clarify this point:   
  
      Considering that the processed impact model simulations account only for climate-induced changes in the extremes (as defined by Lange et al. (2020)), and not changes due to land-use change, we therefore only analyse the climate change-driven changes in concurrent extremes. At a given location, from a statistical perspective, the probability of concurrent extreme events can be affected by the effect of climate change on: (i) the probability of the individual extreme events and/or (ii) the dependence between the events (Bevacqua et al., 2020; Zscheischler et al., 2020b). To gain insights into the drivers of the changes, we compute the change in the probability of concurrent extreme events when assuming: (i) changes in the probability of extreme events in one variable only; and (ii) changes in the coupling between the variables only (Bevacqua et al., 2020).
• In the discussions (Sect 5.3), we also propose to add the following text to illustrate possible dynamical causes of change in river floods.
  
      According to Niang et al. (2014) & Seneviratne et al. (2021), the East African region is also projected to experience increased intense precipitation by the end of the century (with high confidence) under RCP8.5 scenario. This could be linked to the projected changes in events such as positive Indian Ocean Dipole and El Niño Southern Oscillation events, which play a role in increase in precipitation in parts of the East African region, and have already under present day conditions increased in frequency in relation to the pre-industrial period (medium confidence) (Seneviratne et al. 2021; Souverijns et al. 2021). However, in relation to river floods occurrence, IPCC reports that there is low confidence in the end-of-century projections of flood intensities and frequency due to inadequate data (Arias et al., 2021). Nonetheless, Alfieri et al. (2017) still projects that global warming will increase the frequency of river floods in the Nile and Congo basins, thereby greatly affecting DRC and Sudan (Alfieri et al., 2017; IPCC, 2022). The significant increase in the frequency of concurrent river floods & wildfires in the region by the end of the century can therefore be explained by the expected increase in frequency of river floods, which are the main driver of co-occurrence within this pair (Fig. 2a-c and 5a-c). While, as stated above, we find that heatwaves are a major driver for the increase in their joint occurrence with river floods by the end of the century under RCP8.5, increases in river floods themselves also shape these co-occurrences (Fig. 5d-e).
• Souverijns, N., Thiery, W., Demuzere M., and Van Lipzig N. P. M: Drivers of future changes in East African precipitation. Environ. Res. Lett. 11, 2016.