In physical oceanography it is believed that wind stress drives the ocean, while in dynamical

meteorology the ocean surface is treated as a bottom boundary that influences the

atmosphere. The interaction between the sea surface temperature (SST) and wind stress,

respectively characterizing the sea and the atmosphere at the interface, has become of

enormous interest. In this paper, the authors applied a causality analysis which is built on a

firm physical ground, in contrast to other statistical formalisms, to the study of this problem,

and obtained intriguing new results. Specifically, they found that that the ocean surface (SST

and SST tendency) strongly drives changes in the lower atmosphere (THF) and that the lower

atmosphere also has an important influence on the ocean surface in many regions of the

world, different from the traditional view that ocean-driven regimes largely exist in western

boundary currents and atmospheric-led regimes dominate in the open ocean. In recognition of

the importance of the finding, I hence recommend publication of this manuscript. The

following are just some points that the authors may pay some attention.

l.1， l.13, True. But in this paper, the usage of information transfer/information flow (IF) in

studying the interaction is actually more fundamental. It is the exchange of

entropy/information rather than energy. In statistical physics, entropy plays a role in distributing energy.

l.97, the last term may also represent the effect from unobserved processes.

l.100, While mathematically this is correct in terms of Shannon entropy, you may want to be more cautious in interpreting the sign, as it actually may not be explained using the well-known physics.

ll. 130-135. ). “This suggests that SST variability generally increases THF variability, while

THF variability mainly constrains SST variability.” This is good. But be cautious.

ll.189-190. To include more additional variables, make sure they are not nearly parallel;

otherwise the singularity of the covariance matrix could numerically deteriorate the result.

Section 3.3. In studying lagged transfer of information, be careful that only the IF in one way

makes sense—Causality cannot be from the future to the past.

ll.226-229. Indeed ocean-atmosphere interactions become more pronounced at larger time

scale. So these results do make sense.

See the attached pdf file.