Precip, and who is dry

Total precipitation from GFS, ECMWF, ECMWF AIFS, GEM — four models and three centres, named because a roster is a fact about a chart: two of these are ECMWF's, and where those two agree they agree partly for family reasons. Each model's window is the difference between two of its own accumulations from the same run, so this is what fell inside those 24 hours rather than since the run began. Raw model output: nothing here is bias-corrected, unlike the city forecasts elsewhere on this site.

Liquid-equivalent, which is what the models publish: frozen precipitation appears here as the water it would melt to, so a winter field is on this map without being named. A rain-versus-snow split is a later layer rather than a different number — this map's subject is whether anything falls, not what form it takes.

Unshaded ground is where all four keep it under 0.1 in. Most of what is coloured is an argument, which is the reason to draw it: the four models agree closely on how much falls over the map as a whole and disagree about where it lands, so counting who is wet shows a disagreement that mapping the amount would hide.

The threshold matters more than it looks. Against rain gauges these models produce measurable precipitation on 1.8× as many days as actually recorded any, falling to 0.6× at an inch — so a lower bar would fill the map in with the agreement least worth having, and a higher one empties it. At 0.1 in roughly a quarter to a half of the frame carries something, depending on the day.

Valid at 18Z, and the same hour every day so that successive maps compare. The bands are contoured between the counts rather than on them, and shaded ground is a forecast rather than a measurement — for who has actually been right about precipitation, the model leaderboards are the place.