Climate risk to crops is now measured district by district — and it varies by place

Rain in some countries, heat in others

A preprint on southern African maize during El Niño reports that rainfall deficits are the primary driver of El Niño-associated maize yield differences in Zimbabwe and Mozambique, while heat stress is the primary driver in South Africa, Namibia, and Zambia. The same event, the same crop, and the cause of the loss changes at the border — which means a forecast that treats the region as a uniform place will get the risk wrong in part of it. m:…9797

The preprint also reports that temperature and compound dry-hot forecasts retain predictive skill at longer leads better than rainfall forecasts in southern Africa. March rainfall forecast skill at one-month lead improves during El Niño in the northern part of the southern African domain. For planners, the practical point is that the forecasts worth trusting at longer range are the ones built on temperature, not rain. m:…9797

The split between rain-driven and heat-driven losses is the kind of detail that changes a response. A country facing a rainfall deficit needs water management and drought-tolerant seed; a country facing heat stress needs shade, timing, and varieties that tolerate high temperatures. A forecast cannot serve rain-driven and heat-driven losses alike. m:…9797

Crops maturing earlier, unevenly

A preprint on Southeast Asian commodity crops projects that crop maturity is advancing due to rising temperatures, and that phenological acceleration is projected to continue through the mid-century, covering 2026-2065. The magnitude of the shifts exhibits significant spatial heterogeneity and varies by species and seasonal timing, so the change is not a uniform shift of the calendar but a patchwork that depends on where the crop grows and when it is planted. m:…2517

An earlier harvest sounds like a small change until it meets a farming calendar built around a harvest date: planting, labor, storage, and sale all sit downstream of maturity. A shift that varies by species and season is harder to plan around than a shift that moves everything together, which is why the heterogeneity matters as much as the direction. m:…2517

A district-level record for India

A preprint introduces the Drought Intelligence Explorer, a dataset providing monthly, district-level drought indicators for 724 Indian districts from 1981 to 2025. The dataset includes four standardized z-score indices, SPI and SPEI at 1, 3, 6, and 12-month windows, and two percentage-based indicators, SMA and VCI. m:…7594

The dataset successfully reproduces documented drought episodes, including the 2002 Rajasthan drought and the 2015 Marathwada rainfall deficit. Reproducing known events is the test a dataset has to pass before anyone trusts it on events that are still unfolding, and passing it at district level means the record is detailed enough to say which districts were hit and which were spared. m:…7594

District-level detail is what turns a drought record into a usable tool. National averages hide the districts that suffered while the country as a whole looked fine, and relief, insurance, and credit all operate district by district. A record that reaches back to 1981 at that resolution gives those decisions a history to work from. m:…7594

Why it matters

The records add up to a change in how climate risk to food is handled. Forecasts are being judged by which variable holds skill at longer leads; crop calendars are being projected with their unevenness intact rather than averaged away; and drought is being recorded district by district over decades rather than summarized nationally. m:…9797 m:…2517 m:…7594

For anyone whose decisions depend on harvests — buyers, insurers, planners, lenders — the usable information is increasingly local: which district, which country, which variable, and at what lead time. The climate signal is no longer a uniform number for a region; it is a map, and the map is getting finer. m:…9797 m:…7594