Comments: This study evaluated the precipitation forecast skill from dynamical forecast and ENSO teleconnections based on the set theory. The attribution of the forecast skill from the GCM and ENSO for different seasons is separated. Overall this study is well crafted with clear structures. Some comments need to be addressed before the potential publication of this study.

Lines 61-63: From this statement, it seems that the novelty of this study is the global scale investigation of the ENSO and GCM precipitation forecast. From the following statement (lines 64-65), the attribution of prediction skill seems to appear suddenly. Some improvement in the motivation of this study and associated novelty could enhance the argument.

Line 115: “associate P to forecast and ENSO”? This is not clear. Please clarify or revise.

Figure 2. In this figure, the R2(y~x1+x2) decrease with r(x1,x2). What is the meaning in the context of the GCM forecast and ENSO? (the coefficient of determination decreases with the correlation between GCM and ENSO?)

Lines 177-178: What is the lead time of this figure 4a (and the whole study)? In addition, please add the title or notation of the Figure 4(c-d) to make it clear.

Figure 8: The Venn diagrams for grid cells A and B seem to be close and are hard to distinguish.

Section 4.4. These patterns are interesting. Have you compared with other studies and see if these patterns are consistent with previous studies?

This work develops a statistical approach to attributing correlation skill of dynamical forecast to ENSO teleconnection. It can present regions where the forecast skill is attributed to its teleconnections with ENSO, can serve as an effective tool to investigate the source of predictability. The method and results sound reasonable. It is potentially publication if the following concerns are included, regarding to the lead-lag teleconnections and forecast.

1. This approach is just applied with concurrent correlation between Nino3.4 index and observations, to represent ENSO teleconnection. But ENSO also has some lead-lag impacts on the precipitation variations, which also involve in the forecast skill and sources of precipitation. How about of the attributions of these processes? More discussions on it are needed.
2. Lines 20-24, and 346-350, for the novelty of this work in Abstract and Conclusion, there is a repetition in writing. Please rewrite them.
3. Line 94, this work is just “paid to the latest forecasts”. It is also quite interesting to figure out that, are there any differences on the ratio of significantly positive anomaly correlation attributable to different types of ENSO teleconnections, with the increase of lead time?
4. For the extended analysis in Section 4.4, the percentage of significantly anomaly correlation in MAM, JJA and SON is largest in SON, but still lower than DJF. It suggests obvious seasonal differences. This may connect with the observed teleconnections with ENSO. It would be better giving more discussions on these differences and more comparisons with previous sections and studies.

None.