| 01 Sep 2026
Suppression of local upwelling in the Gulf of Panama predicted a season in advance
Abstract. Wind-driven upwelling of subsurface ocean waters to the surface ensures nutrient-rich waters reach the epipelagic zone, supporting some of the most diverse ecosystems and largest fish stocks in the global ocean. In early 2025, the Gulf of Panama experienced a collapse of the local upwelling system not previously observed in historical records, with notable impacts on regional physical and biogeochemical conditions. Predictability of upwelling collapse in relatively local systems on seasonal timescales has yet to be assessed. Here, the predictability of upwelling in the Gulf of Panama is explored in an operational seasonal forecasting system. First, the extreme weakening of the northerly jets driving upwelling and the corresponding extreme sea surface temperatures of 2025 are shown to have been accurately predicted a season in advance. Crucially, hindcasts from the period 1993 to 2024 demonstrate high probabilistic and deterministic skill, and also capture the strong upwelling of 1998. By validating against a high-resolution global ocean reanalysis, an increase in temperature forecast skill with depth is uncovered, making the case for exploiting subsurface information for improved early-warning and description of upwelling events. Lastly, the occurrence of warm upwelling, specifically in 1998, raises a discussion of the most suitable upwelling indicators for the region. The seasonal forecast system sufficiently resolves wind stress-derived Ekman Pumping and the exceptionally weak vertical ocean velocity in 2025. This study demonstrates the capability of seasonal forecasts to provide early warning of upwelling collapse.
Summary
In this work, the authors are exploring the predictability of upwelling in the Gulf of Panama using the Euro-Mediterranean Centre on Climate Change Seasonal Prediction System version 4 (CMCC-SPS4). They find that useful prediction skill exists at lead times of up to 3 months in advance. Furthermore, the unprecedented weakening (strengthening) of the upwelling in 2025 (1998) had been accurately predicted, using the Ekman pumping has the main index to characterize the Gulf of Panama upwelling system.
In general, the manuscript fits the scope of Ocean Sciences and the research reported is of great interest for the readership.
Major comments
1) Analysis is needed regarding the characteristics of the Gulf of Panama ( GoP) Upwelling. What is the seasonality of SST and wind variability in the region? For this purpose, it would be helpful to see a figure with the seasonal cycle of SST and wind stress standard deviation in the target region. Furthermore, explain more to what extent the relatively high prediction skill in the region could be translate into useful information for fisheries.
2) Authors suggested that the exceptional SST events are attributed to a weak upwelling, but this relation seems not to hold in 1998 and 2016 (Fig.3). For that, It would be of interest to see a figure about the relative importance of thermodynamic, i.e the individual components of the net heat flux versus dynamic influences (Ekman pumping) on SST in the GoP. Also, Could coastally trapped waves be an additional factor complicating upwelling in the GoP? Could they be responsible for the lag between wind forcing and SST response mentioned in ll. 205-207?
3) The Ekman Coastal Upwelling cannot be ignored in the most of Eastern boundary current. I think it must be important to provide a figure of comparison between the transport based on Coastal Upwelling and Ekman pumping. For this purpose it would be a justification about the choice of Ekman pumping index only.
4) As noted, Ekman suction and coastal divergence are calculated separately (Eq.1 and Eq.2), but are not really independent in the calculation. Instead, one can unambiguously calculate the total divergence associated with Ekman divergence + Ekman suction by integrating the Ekman transport along all boundaries (north, south, east) of the region of interest. I don't think Ekman suction and coastal divergence should be treated separately anyway, as they overlap spatially. Thus, I recommend calculating the total divergence by integrating the Ekman transport along all boundaries.
5) Authors described briefly the role of ENSO on the GoP’s upwelling, but a figure of forecasting Ekman pumping and SST during some selected years of ENSO events would also be interesting. This helps readers better understand the role of ENSO on the GoP’s prediction.
6) Does the model allow to draw conclusion about a trend in upwelling intensity throughout the selected time scale (1993-2025) - does it increase/decrease or is there no trend? That would also be interesting to evaluate the predictability.
7) I gather that the authors use the wind stress provided by the atmospheric components for their calculations (Eq.1 and Eq.2). There is, however, also the wind stress actually seen by the ocean (i.e. interpolated to the ocean grid), which is available perhaps for the model and/or observations in your target region. I believe the authors should check for a few whether using ocean wind stress makes an appreciable difference.
8) There are many different observational data sets used for validation, and they have different spatial resolutions, cover different time periods, and use different methods (e.g., remote sensing vs.model reanalysis). Why not use a reanalysis that provides everything in a consistent way or interpolate all in the some grid.
Minorcomments
L29-30: In the EBUS, upwelling is predictable on seasonal timescales, as shown for the California (Siedlecki et al., 2016; Amaya et al., 2024) and Southern Canary Upwelling System (Sylla et al., 2026) systems. Not Benguela, please correct.
L43-44: Seasonal forecasting systems are typically reliable in the Eastern Tropical Pacific, where interannual variability is largely governed by the El Nino Southern Oscillation (ENSO, eg; Jacox et al.2017, 2026). Please add these references
L68: “the global ocean” rather than “the global oceans”
L84 :‘’Here, daily values of average of nighttime temperatures were calculated’’ instead of “was calculated”
L99: ocean and atmospheric analyses; for the operational period (2025 onwards). Please remove the semicolon (;) and start a new sentence from “For the operational’’…...
L107: local-scale upwelling indices (Capet et al., 2004), please add also Small et al. 2015 (
Small, J. R., Curchitser, E., Hedstrom, K., Kauffman, B. and Large, W. G. (2015) ‘The Benguela upwelling system: Quantifying the sensitivity to resolution and coastal wind representation in a global climate model’, Journal of Climate, 28(23), pp. 9409–9432. doi: 10.1175/JCLI-D-15-0192.1)
L120-126: Reference(s) needed here for Ekman Pumping and Ekman Coastal Upwelling calculation
L218-219: “ the wind anomaly for the upwelling season in 2016 was relatively weak (you mean intense?). That's not what I see in Figure 3, please clarify
L218: Enso influences SST in this region but not the wind whereas in the paper, this latter is described as the main drivers for SST variability. Please clarify
References
Jacox MG, Alexander MA, Stock CA et al (2017) On the skill of seasonal sea surface temperature forecasts in the california current system and its connection to enso variability.
Jacox, M. G., Edwards, C. A., Hazen, E. L.,& Bograd, S. J. (2018). Coastal upwelling revisited: Ekman, Bakun, and improved upwelling indices for the U.S. West Coast. Journal of Geophysical Research: Oceans, 123. https://doi.org/10.1029/
Jacox, Michael G., et al. "Seasonal-to-interannual prediction of North American coastal marine ecosystems: Forecast methods, mechanisms of predictability, and priority developments.Progress in Oceanography 183 (2020): 102307.
Small, J. R., Curchitser, E., Hedstrom, K., Kauffman, B. and Large, W. G. (2015) ‘The Benguela upwelling
system: Quantifying the sensitivity to resolution and coastal wind representation in a global climate model’, Journal of Climate, 28(23), pp. 9409–9432. doi: 10.1175/JCLI-D-15-0192.1