Measuring the 2022 Po drought against Italy's own rice loss survey

The basin authority ordered a 20 per cent cut in irrigation withdrawals on 20 June and suspended every environmental flow derogation for irrigation from 22 July. In the rice district those cuts landed unevenly. The Est Sesia consortium, which serves about 300,000 hectares across 254 municipalities, was down to a third of its normal flow by June and, in its president's words, could not rotate at all because of the length of its canals. Wetting turns that should come every few days stretched to fifteen, twenty, in the worst cases thirty five.

Rice is grown in standing water. National yield fell from 6.59 tonnes a hectare in 2021 to 5.81 in 2022 and production dropped 15.2 per cent, on the smallest sown area since 2000. The regional split is stark: Lombardy's rice production fell 28.2 per cent against Piedmont's 5.4.

Every decision that summer ran against a deadline. A farmer with water for part of the farm had to choose which parcels to keep before the next turn came round. The consortia rationing the water had no view of where their cuts were actually landing. Compensation turned on a per parcel test, damage above 30 per cent of ordinary gross saleable production, assessed by regional technical bodies.

Only 9.4 per cent of Italy's sown area carried crop insurance, and the national catastrophe fund did not become operational until January 2023, so it paid nothing for 2022. Lombardy alone counted €417.6 million of claims across 9,109 farms. Pavia province accounted for €172.4 million of that across 2,315 farms, the largest provincial figure in the region.

The rice authority then ran a satellite classifier calibrated against ground checks and the farms' own aid declarations, and published hectares lost by district and by commercial class. An official, field level loss figure from the body that regulates the crop makes this one of the few droughts anywhere with a real target to measure against.

Areas are the sum of connected patches of half a hectare and larger. The 0.35 column is the one that corresponds to the rice authority's own definition of an unproductive field.

At the authority's threshold the drought returns 18,353 hectares. The same test between two ordinary seasons returns 1,618. Their published figure for the same two districts is 15,186 to 16,608 hectares, so ours sits about 12 per cent above the top of that range, in the direction the difference in coverage predicts: our box is a superset of their declared rice parcels and includes maize, which fell 24.7 per cent in Lombardy the same year.

The separation strengthens as the test gets stricter. At a 0.15 fall the drought is 9.5 times the control, at 0.25 it is 10.9 times, at 0.35 it is 11.3. A confound that scaled with the threshold would not behave that way.

The two districts are published at 22 to 24 per cent loss and 10 per cent loss, and they are contiguous fields in the same image on the same day. Any error in atmospheric correction, phenology or viewing geometry would apply to both, so the contrast between them cannot come from those. The national statistics reach the same geography by a completely different route: Lombardy's rice fell five times harder than Piedmont's.