In 2025, we observed a failure of seasonal upwelling in the Gulf of panama.

Typically (left), wind jets form across the Isthmus of Panama between Jan-March as the ITCZ migrates south.

These wind jets push surface waters in the Gulf of Panama south which are then replaced by deep, cold, nutrient-rich waters, supporting biodiversity and fisheries.

In 2025, the southerly winds were neither strong enough nor did they blow for sufficiently long enough for the upwelling to kick-start.

We published our observations in this paper: https://www.pnas.org/doi/10.1073/pnas.2512056122

the 2026 upwelling season

We are closely tracking the upwelling in collaboration with STRI’s Marine Monitoring Program. Will upwelling fail again in 2026 or will it return to more normal conditions?

We will update in the graph below how the thermocline evolves through the year as directly-measured data comes in:

About this graph

The thermocline marks the boundary between warm surface waters and the cold, nutrient-rich waters below. We define it here as the shallowest depth where temperature drops below 25°C. During upwelling, strong winds push surface waters offshore, allowing deep waters to rise and bringing the thermocline closer to the surface.

The blue line shows the typical seasonal pattern based on weekly measurements from 2017-2024, with the shaded area representing 95% confidence interval. The coloured lines track how 2025 (green) and 2026 (pink) compare to this baseline. When the thermocline moves upwards towards the sea surface (0m) it indicates upwelling. When it impinges on the surface this show upwelling is strong.

Temperature profiles are directly measured by weekly CTD at Station 4 of STRI’s Bay Panama Monitoring stations (located: 8.735626, -79.265925). The baseline uses a generalised additive model (GAM) to smooth the seasonal pattern, whilst 2025 uses locally weighted regression (loess) to capture the year’s trajectory. For 2026, we connect the points directly until sufficient data accumulates for smoothing. Code available on request.

Data and more information on sampling available here: https://doi.org/10.60635/C35S59

Thanks to Andrew Sellers, Anabell Cornejo and all those who support the monitoring program.