Technical Comments

Response to Comment on "Oceanic Rossby Waves Acting As a `Hay Rake' for Ecosystem Floating By-Products"

+ See all authors and affiliations

Science  16 Apr 2004:
Vol. 304, Issue 5669, pp. 390
DOI: 10.1126/science.1095997

Anomalous chlorophyll concentrations detected from satellites propagate along with Rossby waves in all tropical oceans (1, 2). In the southeastern tropical Pacific, a biologically poor region, we found that chlorophyll maxima are located in convergence rather than divergence zones (3), and thus that these maxima can hardly be explained by upward supply of nutrients as previously hypothesized. We then proposed that, in absence of any obvious mechanism stimulating phytoplankton growth in that region, floating particulate organic carbon (fPOC), mostly composed of detritus, accumulates in the Rossby wave (RW) convergence zones and is falsely detected as chlorophyll by sea color algorithms. This “hay rake” effect was suggested to us by the accumulations of floating material that are commonly seen in oceanic fronts, coinciding with “lines” separating two water masses (4). This obvious coincidence made us neglect that RW-induced convergence zones do move at the same speed as the RWs, and that accumulated material might not necessarily stay concentrated as the waves propagate.

Killworth (5) pertinently points out that such material cannot be durably trapped in the convergences if the phase speed of the RWs is larger than the current anomalies generated by the waves. For a current varying as a sine function with amplitude a, and an RW phase speed of A, he introduces a ratio α = a/A, and demonstrates that small values of α, while allowing for particle concentration and dispersion, do not trap Lagrangian trajectories. Such trapping, and stronger concentration, occur only for α > 1, and they follow the maximum convergence zone, with a spatial lag that decreases when α increases. To first order, by contrast, we designed our conceptual model for fPOC dynamics without the westward-translation RW speed; hence, the model produced maximum concentrations in phase with maximum convergence.

To address the issues raised by Killworth (5), we use our two-dimensional hay rake model with the addition of a mean westward translation of the RW, thus allowing us to study the sensitivity to α. We made α vary from 0.5 to 1.75. In a first series of tests, fPOC was made conservative at the surface (decay rate = 0.001 d–1) to compare with the Lagrangian trajectories considered by Killworth. When α= 0.5, areas of fPOC concentration have a short lifetime and are weak. Larger values of α (0.75, 1, 1.25, 1.5, and 1.75) accumulate floating particles at the surface durably, and the fPOC maximum lags the convergence maximum by, respectively, 1.54, 1.32, 1.08, 0.88, and 0.78 radians. In a second series of tests, fPOC is biologically active and decays at a rate of 10% each day at the surface (3). Accounting for this decay slightly decreases the lag between maximum convergence and maximum fPOC concentration, giving 1.27, 1.12, 0.96, 0.84, and 0.75 radians for the abovementioned values of α (Fig. 1).

Fig. 1.

Distribution of fPOC at the sea surface across a propagating wave for various values of α. The scheme shows a complete RW, with (from west to east) the zone of maximum divergence, the sea level trough, the zone of maximum convergence, and the sea level crest, all separated by π/2.

Thus, with decaying fPOC, the lag decreases, but the decrease is noticeable only for α < 1.25. Higher decay rates would certainly decrease the lag even more, but would result in smaller amounts of accumulated fPOC, and hence would not explain the bias in chlorophyll retrieval by sea color algorithms. A decay rate near 10% per day (3) looks reasonable, and at this time we have no reason to change it. Fig. 1 also shows that the fPOC gradients generated by convergence increase with increasing α values. Overall, we state that whatever range we take for α, our hay rake model produces an fPOC maximum that lags the convergence maximum by less than π/2, a phase lag which is indeed observed in satellite data (6).

Obviously, a key point to be addressed here is the range of α, because the lags and the efficiency of concentration depend on such values. Because the signals one is seeking are small in these oligotrophic regions, there is a need for heavy data filtering in space and time (13, 6). Certainly, the analysis of filtered monthly satellite data does not permit an estimate of mean lag between physical processes and chlorophyll with much accuracy, especially because RW-induced current anomalies vary in space and time (3, 6). Thus, the ability of the concentration process by convergence to explain the variability of sea color depends on α being greater than about 1 (Fig. 1). So α must be estimated from observations.

Going back to the sea level anomaly data, we used the mean westward observed speed of the RWs at 20°S in our area, 5.6 cm s–1—a value compatible with observations (3, 6)— and we estimated the RW current anomaly extrema a in the 150°W to 80°W region for the period from 1993 to 2002. The resulting distribution of α is shown in Fig. 2. The majority of α estimates are below 1.25 and hence, lags between the maximum convergence zones and the maximum of apparent chlorophyll anomalies would still be generally lower than π/2, but fPOC concentration would be too flat. Such even profiles do not fit our hypothesis well (3).

Fig. 2.

Distribution of α values at 20°S, 80 to 150°W. Alpha is the ratio of amplitude of RW-induced variations of zonal component of surface current to RW phase speed. Surface currents have been estimated from filtered 7-day composites of sea level anomalies on the period 1993 to 2001. Amplitudes are half the peak-to-trough differences in current speed seen in the time series. Thin line gives estimates with observed RW phase speed of 5.6 cm s–1; thick line gives estimates with the mean current removed from RW phase speed.

However, in the south tropical Pacific, the mean background zonal speed is nonzero and is directed westward, so the image of a wave anomalous current acting on a mean zero background has to be corrected. The average zonal component of the mean current at 20°S and 150°W to 80°W is u = –3.05 cm s–1, as calculated from analysis of drifter trajectories (7), so that A should now be corrected to account for the relative movements of the waves to the water mass. Using this relative wave speed Au, we recomputed a distribution for corrected α; under these assumptions, α now often exceeds 1.5 (Fig. 2) and, thus, intense accumulation zones for fPOC can occur (Fig. 1), lagging the maximum convergence zone by π/4 or less.

The lag between the maxima of chlorophyll anomalies and crests of sea level has been recently estimated and mapped (6). West of South America at 20°S, the chlorophyll maxima are shown to lead the sea level crests by about π/4 on average, which is also what the discussion presented here with the hay rake mechanism indicates. Thus, there is agreement between our hay rake hypothesis and the lag necessarily induced by the translation of the waves: This lag amounts to π/4 both in observations (6) and in estimates based on RW phase speed and current anomalies.

Interestingly, another mechanism has been suggested to explain why maxima of sea color positive anomalies lead the sea level crests: In regions where a meridional gradient exists (chlorophyll-rich waters from the equatorial upwelling to the north, and oligotrophic waters to the south), poleward (and, hence, convergent) advection can bring in waters with higher chlorophyll concentration (6) at exactly the same place that, we argue, convergence concentrates fPOC (3). This would also explain the observed co-propagation of sea level and ocean color.

A final argument to support our hay rake mechanism is that the error in chlorophyll retrieval from sea color data depends on the sign of the convergence. In situ measured chlorophyll concentrations were lower than satellite-derived concentrations in convergence zones, and the opposite was true in divergence zones (3). Such behavior, in the area of the in situ observations, does not support the advection mechanism (6).

Beyond these specific questions, this debate underscores that one needs to go deeper into details to analyze the exact nature of the coupling between the biology and the dynamics. It also points out that a major process such as this one cannot be fully understood without the help of in situ observations. In the future, we plan to continue our work on the hay rake hypothesis, analyzing the spectral sea color signal, and modelling the potential distribution of fPOC. We also hope that the oceanograpic community will implement experiments on cruises to detect, identify, and quantify floating material, which has heretofore been undersampled.


Related Content

Navigate This Article