Skip to content
No. 047On holdInvasive water weeds and lake fisheriesUpdated Nov 19, 2026

Can a wind-and-current drift model, started from free radar satellite maps of where the water hyacinth mats lie, predict a few days ahead where the mats will move in Lake Victoria's Winam Gulf, so that fishers and beach managers in Kisumu know which landing beaches and fishing grounds will be blocked and when?

Good it would do
3 / 5
Can AI help
3 / 5
How neglected
4 / 5
Can we check it
4 / 5

Why it's on hold

Nobody forecasts hyacinth drift in the Winam Gulf, and a cheap public-data hindcast could test whether it can be done. But the blind second panel parked it, and its reason held up: the harm happens over hours, fishers already wait for the wind to turn, and no one has named a decision a 2-to-6-day forecast would change; KMFRI's 2018 'prediction map' is also unexplained.

Why it's stuck

The people who work the gulf say the problem is movement, not only how much weed there is: in March 2024 the chairman of Kisumu's Beach Management Unit told The Standard that 'Whenever the wind blows, it moves causing fishermen to be trapped in between the plants for hours', and a KMFRI scientist in the same article said 'the windy weather is pushing the weed to the shores'. The responses described there are manual clearing and dredging, not warnings.

What we'd try

Build a mat map time series for the Winam Gulf from free Sentinel-1 radar (which sees through cloud) and Sentinel-2 and Landsat optical images, then run ensembles of a drift model in which each mat moves with surface currents from a lake circulation model plus a fitted share of the wind, driven by ERA5 and forecast winds. Fit the wind share and mat behaviour on some seasons and test on held-out seasons: the check is whether the predicted mat outline 2 to 6 days ahead overlaps the next satellite-observed outline better than two simple baselines (the mats stay where they were; the mats move straight downwind).

We'd look again if

All three: (1) KMFRI, a beach management unit, Kisumu County or the port names a concrete decision a 1-to-6-day weed forecast would change (clearing crews or harvesters, landing site, ferry or port scheduling); (2) KMFRI confirms its 2018 'prediction map' does not already forecast drift, or shares it and it leaves the question open; (3) mats are present in the gulf again or expected. Then run the revised first test in the dossier (inventory, map check, frozen hindcast against strong baselines with issue-time forecast winds).

Source: https://www.standardmedia.co.ke/sports/amp/environment-climate/article/2001492064/water-hyacinth-destroying-livelihoods-on-the-lakeside

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.