A New Perspective on the Atmospheric Rectifier Effect
PCTM team
In an important paper Denning et al., (1995) demonstrated that asymmetry in atmospheric transport of carbon dioxide exchanged by terrestrial ecosystem tends to result in higher CO2 concentrations near the ground than in the well mixed atmosphere. This occurs because the atmosphere tends to be stable and not well mixed when respiratory release of CO2 is the dominant activity of ecosystems (at night and in the winter), and the atmosphere tends to be well mixed when photosynthesis is active (mid day and summer). This collaboration between the atmosphere and the biosphere tends to withdraw CO2 from a large volume of air by photosynthesis. Later when these products of photosynthesis are respired, the CO2 is deposited in a smaller volume air that may stay near the surface. Denning et al., used a full atmospheric general circulation model with an interactive biosphere and a CO2 containing atmosphere to demonstrate that a seasonally balance biosphere coupled with this asymmetric mixing, termed the “rectifier” effect could result in an apparent accumulation of about 1.5 ppmv. CO2 (in the yearly mean) at high northern latitudes. They argued that this “extra” CO2 could be mistaken for a source of CO2 at these latitudes, and that this needed to be taken into account when analyzing CO2 concentration measurements to determine the geographic regions of the globe that are acting as net sinks for CO2 produced by human activity – the so called “missing sink”.
Following Denning et al., the rectifier effect has been assumed to be primarily associated with differences in vertical mixing of the atmosphere. This has the important consequence that CO2 that accumulates near the surface at high latitudes in winter due to rectification, can rapidly re-mix with the atmosphere when the sun returns. We report here on studies using a coupled global carbon cycle and transport model to analyze seasonal changes in the CO2 concentration of the global atmosphere. These simulations clearly show that there is a strong north-south movement of CO2 which results in the net accumulation of elevated CO2 levels in the polar atmosphere during winter and spring. This accumulation of CO2 occurs in a region of the globe that is remote from surface sources or sinks of CO2. While there is no net, long-term storage of CO2 in this region of the atmosphere, the slow remixing of this reservoir with the remainder of the atmosphere in summer and its refilling in fall and winter introduce a hitherto unappreciated lags in the response of the atmosphere to seasonal changes in biospheric activity. We estimate that this storage of CO2 associated with the seasonal rectifier can amount to 0.8 Pg (gigatons) of CO2. This is of comparable magnitude to the net CO2 storage activity of the biosphere or oceans of the northern hemisphere and needs to be considered as a dynamic component of the carbon cycle. Here we examine the mechanism storage and investigate its possible impact on studies of the seasonal cycle and efforts to measure regional scale CO2 fluxes.
Models and simulations ()
Polar projection plots of the CO2 concentration in the mid-troposphere on April 12 and August 10 of a simulated year are shown in Figure 1. These show the approximate peak and nadir of the seasonal oscillation in the far north. The region of highest seasonal difference in CO2 is typically over the Arctic Ocean, where there is no net terrestrial CO2 exchange. Clearly, the CO2 that accumulates in this region must be derived from terrestrial exchange processes occurring further to the south. In Figure 2 we plot the total CO2 exchange by the land surface (red) and the cumulative change in CO2 concentration of the overlying atmosphere (blue) for (0-15, 15-30, 30-45, 45-60, 60-75 and 75-90) bands of north latitude. The difference between these curves can be thought of as net import or export of CO2 from that zone over the course of the simulated year. For the entire globe, the surface exchanges and changes in atmospheric burden of CO2 are equal, as expected (data not shown). At the pole (north of 75N), there is essentially no surface CO2 exchange in this simulation, so all of the seasonal swing in the atmospheric burden of approximately 6 x1013 moles of CO2 must be imported. Interestingly the next lower latitude band (60-75) is approximately balanced but the atmosphere lags the surface exchanges by about 30 days. By itself net export from this cell is not sufficient to account for the swings in CO2 at the pole. However, the CO2 could be passing through this zone from the next (45-60N) zone which shows a substantial export of CO2 during winter and import of CO2 during the summer. The 30-45 zone is approximately balanced but there is a phase offset between surface exchanges and the atmosphere of about the same magnitude but opposite sign as the 60-75N zone. South of 30N the seasonal wave propagates to the equator – despite opposing surface fluxes. The substantial de-coupling between the surface flux of CO2 to the atmosphere and the consequent changes in atmospheric CO2 concentration in these zones indicate a substantial north-south movement of air masses containing different CO2 concentrations over the seasonal cycle. The magnitude of this processes is at least equal to the change in CO2 of the polar region. However, it seems likely that this is just the most obvious part of a larger oscillation. The Southern Hemisphere shows no evidence of this strong seasonal oscillation – presumably because of the absence of seasonal sources and sinks for CO2.
This behavior is not entirely unexpected. Plumb and McConalogue (1988) showed using a 2-d tracer transport model that a tracer released in a zonal band centered at 45N tended to mix northward, filling to the pole. They noted that the tracer concentration at the pole, where there was no surface source, could be as high as it was in the source region at 45N. However, they did not specifically identify a mechanism associated with this transport. Recent work on the transport of pollutants to the Arctic troposphere (Klonecki et al., 2003; Hess et al., 2004) provide two plausible trajectories for movement of trace gases from the atmospheric boundary layer at temperate latitudes to the Arctic troposphere: a) Air masses conditioned in the boundary layer over Western Europe and Asia move east and northward over the continent, cooling by interaction with the landsurface which tends to trap this air in an isentropic well (Hess et al., 2004), and b) Air masses conditioned over Eastern North America becomes entrained in warm conveyor flow to the north and east associated with cyclonic storms in the North Atlantic, depositing this air in the upper troposphere at higher latitude. Presumably, the latter cools radiatively and subsides eventually mixing with the former. Air mass trajectory analysis (Stohl, 2001) and aircraft measurement campaigns (Atlas et al., 2003) demonstrate the accumulation of non-methane hydrocarbons, carbon monoxide, nitrogen oxides and others from industrial sources in North America and Europe in the Arctic troposphere. Presumably, these case studies are but examples of global scale transport associated with the movement of synoptic systems along the N. Hemisphere storm tracks. CO2 released by industrial activity or by ecosystem respiration at temperate latitudes around the globe would, like the pollutants, accumulate in the Arctic troposphere during winter.
The accumulation of depleted CO2 at the pole in summer is equally as strong as the winter accumulation of high CO2 air, but we are not aware of similar studies illustrating the transport processes. We speculate that summer storms associated with synoptic waves have some cyclonic character and transport boundary layer air northward and upward to the Arctic troposphere - just as do the larger and more intense winter storms. This could move CO2 depleted air to the polar region.
Transport of air masses to the Arctic is only half of the story. After lingering and mixing in the polar region, air masses from the Arctic must move to the south. A whiff of experimental support for a north-south return flow may be drawn from the spring maximum of ozone concentration typically observed at clean air sites around the N. Hemisphere. Photochemical transformation during southward transport of the polluted air masses accumulated and stored in the cold and dark of the Arctic troposphere during winter has been invoked to explain this vernal peak of ozone concentration (Monks, 2000). We would expect these air masses to also be high in CO2. Outbreaks of polar air in winter are a spectacular example of a trajectory that polar air masses may take to return to the south. Similar but less dramatic outflow could occur with synoptic high pressure systems and the cold conveyor flows also associated with cyclonic storms. The CO2 concentration of these air masses could be systematically different than those simultaneously transported to the north as a result of mixing in the Arctic troposphere. This could account for the seeming paradox of simultaneous south-north and north-south transport of air masses containing different CO2 concentrations.
Strong north-south differences in CO2 concentration and in the size and timing of the annual cycle of change in CO2 concentration are well known. But, it has generally been assumed that these differences were rooted in the seasonal pattern of activity of the biosphere; the large extent of land area at high latitude, and to some extent on vertical redistribution of CO2 within the atmospheric column as a function of latitude. The analysis presented here indicates that large-scale, meridional redistribution of CO2 is also a factor to be considered. As illustrated in Figure 2, this oscillation tends to de-couple changes in the CO2 concentration of the atmosphere from local exchanges of CO2 at the land surface. This not only affects the magnitude of the coupling but in some cases import or export of CO2 can change the sign of the relationship in a given latitude band. This is likely to be of significance to any studies which attempt to use changes in atmospheric CO2 concentration to infer the strength and location of CO2 exchange processes over the surface of the planet. It can be argued that this phenomenon is already included (unknowingly) in the models that are used to conduct global scale inversions. Therefore, the results of these studies are likely to be unaffected, but the interpretation of these studies may need to be adjusted as more is learned about the influence of this meridional transport on the coupling of atmospheric change to surface flux. For example, the prospect of being able to take advantage of zonal flow of the atmosphere to resolve the separate contributions of North America and Eurasia to the global CO2 balance seems less plausible – given the presence of this orthogonal redistribution of CO2.
Because this oscillation tends to de-couple changes in the CO2 concentration of the atmosphere from local exchanges of CO2 at the land surface, it may have particular significance to studies which attempt to measure surface CO2 flux at regional scales. Studies of the carbon cycle are beginning to emerge at these scales in an effort to address the large gap in scale that separates the two major nodes of carbon cycle studies: a) those based on global inversions which can resolve areas on the order of 1 x 107 km2 , and b) those using eddy covariance and other methods to study ecosystem exchange of CO2. The average footprint, or area of surface flux integration of the latter, typically does not exceed 1 km2. It is obvious that an intermediate scale of measurement would be useful, and considerable interest has developed in the possibility of developing methodologies that could be applied the regional scale (broadly defined as 104 to 106km2). These approaches must rely a quantitative relationship between changes in atmospheric composition and surface flux.
Several papers have attempted to address this possibility by making use of CO2 concentration measurements in the atmospheric boundary layer (ABL) and adjacent free troposphere (FT). Betts et al., (2004) provide a simplified framework to understand the approach. They posit that over time sequences of several days the properties of the ABL can be viewed as changing toward a state of balance where fluxes of trace gases between the ABL and the adjacent FT are equal and opposite to the fluxes between the surface of the planet and the ABL – at which point the ABL could be said to be in equilibrium with the atmosphere and surface. Betts et al., used a simple 1-dimensional model to show how the concentrations of CO2, water vapor and radon could be related to the corresponding fluxes. Bakwin et al., (2004) and Helliker et al., (2004) used measurements of CO2 concentrations in the ABL and FT together with estimates vertical exchange between the ABL and FT to calculate the apparent surface CO2 flux, and they showed that these estimates were in general agreement with extrapolations based on eddy correlation studies in the areas of the measurements. Lin et al., (2003) and Gerbig et al., (2003a, b) propose an approach for estimating surface fluxes in the midst of distributed sources and sinks over land based on use of a modeling tool, STILT (Stochastic, Time-Inverted, Lagrangian Transport) and data on atmospheric transport from a mesoscale weather forecasting model. This is designed to take into account horizontal and vertical advection and sub-grid scale dispersion to obtain much higher spatial resolution than would be possible using a 1-dimensional approach.
All of these approaches are conceptually similar in that they attempt to separate the variation in CO2 concentration observed at a receptor into a near-field influence (due to local fluxes) and a far-field influence of background atmosphere that is also responding to the global carbon cycle. One needs to know the far field influence before one can quantify the near field influence and estimate the flux. One approach has been to associate the far field effect with the FT. It is argued that the free troposphere is well mixed and tends to be dominated by zonal flow. Therefore, measurements of CO2 concentration at mountain stations or in the remote, marine boundary layer can be used to specify the properties of the FT as a function of latitude over the continents, thus providing boundary conditions for exchange with the continental ABL and to specify the initial properties of the ABL at the western continental margins. Given this assumption, the surface CO2 flux over the path that an air mass takes to a measuring site in the continental boundary can be obtained by contrasting the concentration of CO2 in the sampled air with that of its presumed source in the marine boundary layer to the west. On the surface this seems quite plausible. However, it is of interest to note that this approach carries an implicit assumption of conservation of CO2 in a given latitude band. Meridional transport of air masses of contrasting CO2 concentration and mixing of these is likely of complicate the problem of specifying the far-field influence, and therefore the ability to resolve the regional-scale CO2 flux, at receptor locations in the continental ABL.
(Perhaps a paragraph on the failure of the 1-d inversion at WLEF as an example, if we can understand it)
There is no complete solution to this problem. If the regional flux is unknown, the far-field influence of the background atmosphere must be in error. Therefore the near-field effect and the local flux can’t be resolved. However, a way forward might be to take an approach which strives to simultaneously reduce the errors in specification in the far-field influence while using estimates of the near-field influences to improve estimates of the global CO2 fluxes that specify the far-field effect. This could be accomplished by integrating both influences in the same modeling environment. For example, a forward model (such as those used in this study) could be used to approximate the concentration of CO2 of the global atmosphere as a function of assimilated weather and the best estimates of surface CO2 fluxes. Time-reversed trajectory analysis or data assimilation approaches could then be used to improve modeled CO2 flux influencing the CO2 concentration at measurement sites distributed around the globe. Several groups around the world are beginning to use this approach. This study is of significance to these efforts in that it shows that the coupling of atmospheric CO2 to surface flux is even more complex than previously appreciated, and that local or regional scale studies which do not simultaneously take a global view of the carbon cycle and the atmosphere are likely to fail.
References:
Figure 1. Polar Plots of mid troposphere CO2 concentration.
Figure 2.