Africa's rain-gauge network has shrunk sharply since the 1970s and 1980s; which closed gauge sites in sub-Saharan Africa, if reopened or replaced, would most reduce the error of the satellite-plus-gauge rainfall estimates that drought early warning and index insurance rely on?
- Good it would do
- 2 / 5
- Can AI help
- 3 / 5
- How neglected
- 4 / 5
- Can we check it
- 4 / 5
Why we stopped
Reject. For CHIRPS, the main gauge-blended product, the value of a gauge at any site is already given by a published distance-based formula and monthly variance-explained maps (CHIRPS v3, 2026), and public archives cannot tell a closed station from a running but unshared one, so a 'reopen' list would be wrong in ways compute cannot fix; gauges exist or vanish because of upkeep, data sharing and siting rules, and no user chooses sites by error value (Good done 2). The outside skeptic argued for Park; a narrower question, testing CHIRPS v3's station-influence model against withheld gauges, is recorded for the scout.
Why it's stuck
The stations were not shut down by any decision anyone studied; they faded out through budget cuts and staff shortages, so nobody owns the question of which ones were worth most. The decline is large: Ghana's rainfall station network fell from 518 stations in 1976 to 87 in 2021 (Asare et al., Frontiers in Climate 2025), an Eos feature (October 2025) reports that Africa now has less than one third of the weather stations it once had and that the density of stations meeting WMO standards is 8 times lower than recommended, and Dinku (EMS 2019) notes that African stations are sparse, declining and mostly located along major roads.
What we'd try
Run large data-denial and data-addition experiments: rebuild a gauge-blended rainfall field (for example a CHIRPS-like or regression-kriging blend of satellite rainfall with gauges) many thousands of times with different sets of gauges withheld or added, using public station records (GHCN-Daily, which holds precipitation from more than 100,000 stations worldwide, plus any open national or TAHMO data), and measure how much each historical site's return would cut error in rainfall and in derived drought indices over farming and pastoral areas. Check: hold out gauges that are still reporting, predict how much the blend degrades without them, and compare with the actual degradation; also test whether past station closures line up with measurable drops in the skill of satellite products.
Source: https://www.frontiersin.org/articles/10.3389/fclim.2025.1524029/full
Think we got it wrong?
Ruled-out problems can come back. Tell us what we missed.
Members can read the full record for this problem, check its sources and comment. Open the member view → Not a member? Here is how to join.