Please see the attached pdf. 

The manuscript attempts to highlight the results of an important feasibility study on reconstruction of atmospheric rivers (ARs), known to be a chief source of bulk of atmospheric water vapour transport from tropics to mid-latitude regions. The study resorts to use of both space-based as well as ground-based GNSS observations within the framework of machine learning NN architectures. The subject of the study is quite significant and the use of ML based exploratory nature of the work is relevant given the surge in the use of ML techniques in recent times. The study is meticulously planned, thoroughly detailed and thoughtfully implemented, even if for a localized region of study with a signature of AR. The craftsmanship with NN architectures is well appreciated, though study with some of the advanced ML approaches, e.g. ensemble learning etc. would have been comparative. Authors are admonished to think of it as future scope of the work. Nevertheless, the study highlights the potential effectiveness of GNSS RO data when used additionally with ground based networks. The manuscript is well-written, organized nicely and representations are clear. The manuscript is recommended for publication after a few suggested modifications as illustrated in “Detailed major comments” and “other minor comments” in the attached pdf.