Meteorology and Climatology Editing and Proofreading Services

The forecast said a seventy per cent chance of rain and it stayed dry. That is not a wrong forecast, and explaining why is one of the harder writing problems in the field — because the sentence that makes it clear to a reader has to convey what a probability means, over what area, in what period, and against what verification, without becoming a lecture.

We edit what meteorologists and climatologists write — forecast verification studies, model evaluation and intercomparison papers, climate projection reports, extreme event and attribution analyses, observational and reanalysis studies, warning and impact communication material, technical documentation for forecast products, and theses, articles and agency reports. Our editors work on probability statements and how they are verified.

A probabilistic forecast is a claim that can only be assessed over many forecasts, and the failure is a probability written as though a single outcome could confirm or refute it. Verification is what turns a forecast into a testable statement, and most writing in the field skips the step that makes that visible. We work through these so a probability is stated with its event definition, the area it applies to and the period it covers, since "70% chance of rain" means something different for a point and for a region and for six hours and for a day; so the verification is reported against a baseline, because skill is relative and a forecast that beats persistence or climatology is the only kind worth reporting; so the verification measure is named with what it rewards, given that reliability and sharpness pull in different directions and a forecast can be well calibrated and useless; so a single outcome is never used to assess a probabilistic statement, as the sample for verification is the set of occasions on which the forecast was issued; so ensemble spread is explained as a representation of uncertainty rather than as a range of outcomes; so climate projections are labelled with the scenario, the model set and the reference period; and so attribution statements name what they are conditioned on. Writing like this survives being quoted.

Everything you send is treated in confidence, including unpublished analyses, model output and reports before release. We are editors rather than meteorologists or climate scientists, and we offer no view on forecasts, models or projections. What we can do is make the probability statement precise and its verification visible.