Compile atmospheric data for research

I have the atmospheric sensor logs and the draft data table for the research paper. Send the compiled dataset to Dr. Alvarez and to the lab manager for validation, signers in that order, with a Thursday deadline so the analysis can start — but first confirm sensor timestamps are in UTC and that any flagged gaps are documented.

Before I hand this to Dr. Alvarez and the lab manager, make the dataset easy to validate — put UTC timestamps in the first column, summarize missing-data windows up top, and flag any sensors whose readings drift by more than 5 percent compared with neighboring stations during the collection period.

Station B12 shows a steady overnight pressure rise of 8 percent compared with all other stations flat for the same hours. Dr. Alvarez is out in the field and wants results fast. I fear it is sensor drift, but if I discard it we lose a possible real event. What is the most likely diagnosis and the best immediate step that preserves the integrity of the paper and keeps the Thursday analysis start date?

Across several datasets we repeatedly lose days because timestamps are mixed time zones or sensors drift without a documented calibration. I curate these datasets and want one habit change and one procedural tweak that would halve the time spent chasing metadata before analysis begins.