IGACtivities logo

Issue No. 17, May 1999

 

 

What have Lagrangian experiments accomplished?

  

Contributed by:
Barry J. Huebert, University of Hawaii, U.S.A and Donald H. Lenschow, National Center for Atmospheric Research, USA

Some problems involving reactive trace species in the atmosphere are simply so complex that there is no substitute for observing their rates in the actual atmosphere. A good example is the oxidation of sulfur dioxide (SO2) to non-sea salt sulfate (NSS). Four processes compete for SO2, and each could be significant in various circumstances: 1) dry deposition to the Earth’s surface in the marine boundary layer (MBL) may remove as much as half of the NSS produced; 2) homogeneous oxidation by OH is the only pathway that can produce the gas-phase H2SO4 needed to support the nucleation of new particles; 3) oxidation of SO2 in cloud droplets is now thought to be the major source of NSS; and 4) finally, under some circumstances seasalt aerosol water may also be an important oxidation medium. Its alkalinity may keep it from rapidly acidifying, thereby permitting extra oxidation by ozone.

Existing models cannot predict to within 10% how much SO2 is lost by each process. Some parameters are rather rough estimates; it is not hard to imagine that one of them could be off by a factor of 2, which would dramatically change the predicted relative importance of all four processes. Thus, if we want to know how much additional NSS would be formed by a 20% increase in SO2 emissions from the US or Asia, for instance, we need more certainty than the models can supply. To gain that certainty we must measure the rates of these processes in the actual atmosphere.

This can be done in two ways. One is to compare the output of chemical transport models (CTMs) with a time-series of concentrations in the area of interest. High-quality, comparable time-series data from many locations can constrain many factors in a model. The other approach is to measure the change of concentrations with time in a single airmass by following the airmass with a suite of measuring instruments as it moves. This Lagrangian approach is extremely powerful but difficult to implement. It requires multiple aircraft (with double crews to fly successive flights), ships, forecasting personnel, and a significant infrastructure. Is it worth it? We can address this question by looking at the experience gained from a series of Lagrangian experiments conducted over the past 12 years during major field campaigns: DYCOMS, FIRE Stratocumulus, ASTEX/MAGE, ACE-1, and ACE-2.

DYCOMS

One approach to Lagrangian observations started with an interest in the differences between free troposphere (FT) and MBL concentrations of nitric acid (HNO3) and SO2. To aid the analysis, a simple cloud-capped quasi-steady-state MBL regime was selected–the marine stratocumulus region at the eastern edge of the summertime subtropical high, where large-scale subsidence produces a well-defined lid to the MBL. This simplifies evaluation of trace species budgets in a Lagrangian framework.

The budget of any species, S, can be described by a continuity equation [Lenschow et al., 1981; Kawa and Pearson, 1989]:

Entrainment fluxes into the MBL, Jh(S), can be parameterized in terms of the entrainment velocity, VE (common to all species being mixed) and concentration differences across the capping inversion.

Repeated measurements in the volume can define d<S>/dt, the time-rate of change for each species. If the chemical formation and loss rates, F(S) and D(S), of this substance can be measured, the surface flux, J0(S), may be solved for as a residual. Alternatively, if the surface flux is measured, it may be possible to solve for the chemical formation and destruction rates.

One uncertainty in this analysis is determined by the magnitude of the concentration change during the observation period, d<S>/dt, relative to the uncertainty in species measurement systems. For many species the limited flight duration in DYCOMS made it impossible to remain in the study airmass long enough to detect a change [Lenschow et al., 1988].

FIRE Stratocumulus IFO

To overcome this measurement time limitation it is necessary to return to the target airmass for multiple flights. But how can you be certain that you sample the same air during successive flights? Neutrally-buoyant free-floating balloons can, in principle, mark the airmass. During the FIRE stratocumulus IFO experiment near San Diego [Albrecht et al., 1988], balloons deployed from a small boat near shore were tracked from a light aircraft, but were lost when they drifted into controlled military airspace. We learned that we needed a deep-water ship to launch the balloons and a transmitter on the balloons that the research planes themselves could detect from at least 150 km away. This experiment also provided examples of "decoupled" MBLs [Paluch and Lenschow, 1991], which we found in later experiments to be a common occurrence over the ocean.

ASTEX/MAGE

In ASTEX/MAGE, we were fortunate to have two ships–a balloon-launch ship at the start of the trajectories and a second ship to intercept them at the end. In addition to balloon launching, ships are great platforms for making continuous measurements of atmospheric and oceanic variables, and aerosol chemical size distributions. During the first Lagrangian, water accumulation due to drizzle from marine stratocumulus sank all the GPS-equipped balloons within 7 hours. Since observing the modification of aerosol by rain is an important objective, the marker balloons need to be able to survive modest rainfall.

For our second Lagrangian experiment, the balloons' ballast was put in a plastic bottle dangling from a line beneath the balloon. Using this approach in a heavily polluted European airmass, one balloon was tracked for 42 hours and sampled during seven research flights, four by NCAR's Electra (Figure 1). The MBL, which was about 2 km deep, contained a weak inversion at about 0.6 km which decoupled the upper part from the surface-based mixed layer (SBML), except for occasional intermittent transport. Unfortunately, our sampling strategy (the number and height of flight legs) was predicated on the assumption of a well-mixed MBL with a well-defined lid, as in DYCOMS. In retrospect, we should have made a point of chemically characterizing the separate layers much better than we did, but our understanding of decoupled layers was still in its infancy. The combined observations from two aircraft produced a picture of the evolution of a deep two-layered MBL as it moves over warmer water that has been extremely useful for understanding their dynamics and evolution [Bretherton et al., 1995].

 

Figure 1. Electra flights, ship locations, and balloon tracks (long, nearly north-south lines) during ASTEX/MAGE Lagrangian #2. SM is the airport at Santa Maria; OC is the location of the balloon-launch ship, R/V Oceanus; and MB is the location of the R/V Malcolm Baldrige.

Three significant chemical evolution papers resulted from this Lagrangian experiment. In one, based on essentially a single flight drifting with the airmass, Wingenter et al. [1996] found that the loss rate of some hydrocarbons was faster than OH alone could explain. The UC-Irvine group attributed the additional loss to atomic chlorine. Clarke et al. [1996] studied the evolution of aerosol populations in several layers, quantifying the impacts of mixing, coagulation and removal on aerosol properties. Zhuang and Huebert [1996] looked at the ammonia budget and quantified an ammonia source at the ocean's surface. Marine-fixed nitrogen is redistributed via the atmosphere, potentially moving hundreds or thousands of kilometers before being washed out in another region.

In each of these studies, our ability to quantify chemical and physical processes was limited by our ability to quantify the impact of vertical mixing. It is critical, therefore, that we improve our technologies for measuring the vertical transport and mixing of airmasses. One essential measurement is the rate of transport, or entrainment of mass into (and sometimes out of) the MBL across the capping inversion. A direct method for obtaining this measurement is to measure the flux profile of a tracer species through the MBL and extrapolate the profile to the top. The entrainment velocity, VE, is obtained from the ratio of the flux to the jump in tracer concentration across the top. Kawa and Pearson [1989] used a fast-response ozone instrument to measure the ozone flux across the inversion by eddy correlation in DYCOMS. Unfortunately, ozone is not the optimal tracer for this measurement.

 

Figure 2. Schematic illustration of the MBL concentration and flux of ozone (top panel), a species with a source aloft, a small sink at the surface and, in the remote MBL, a small (i.e., time scale of a week or more) photochemical sink within the MBL. The solid lines denote the more common case of greater ozone concentration aloft and a negative flux at the MBL top; the dotted lines denote a case of equal concentration within and above the MBL, where ozone is useless as a tracer of entrainment. The bottom panel illustrates concentration and flux of DMS, a species with a source at the surface and a photochemical lifetime in the atmosphere of about a day. This results in a minimal concentration above the MBL and thus a relatively large entrainment flux and well-defined and easily measured jump across the MBL top.

A schematic illustration of species concentration and flux profiles in the relatively simple DYCOMS regime is shown in Figure 2. Two different types of species are illustrated. The top panel shows the behavior of ozone, which has a small sink at the surface and a source above the MBL. As a result, the ozone concentration is higher above the MBL, where significant spatial variability exists, than within it. This leads to a variable (and sometimes very small) jump across the top which is difficult to measure accurately, and also a variable flux across the top. This can also lead to horizontal variability in the mean ozone concentration within the MBL. Thus, ozone can sometimes be useless as a tracer of entrainment.

Dimethylsulfide (DMS) is an ideal molecule for quantifying vertical exchange in the MBL. Concentration and flux profiles of DMS are shown in the bottom panel of Figure 2. DMS comes only from the surface (so there are virtually always quantifiable differences between layers), is negligibly soluble in clouds, and has a photochemical lifetime of only a few days so it decays in layers not in contact with the surface. If we could directly measure the DMS flux by eddy correlation, we could directly calculate entrainment velocities between layers. Other species fluxes could be accurately calculated by analogy, using the DMS-derived VE and easily-measured concentration differences across inversions. The challenge, then, is to develop a fast (10 Hz) DMS system for directly measuring the DMS flux from aircraft. Two fast DMS techniques offer promise: fluorine-induced chemiluminescence and atmospheric pressure ionization mass spectrometry; neither has been demonstrated in the field. Both should be given a high priority for development, since they would markedly increase the accuracy of Lagrangian-derived process rates.

ACE-1

Several improvements were made to the technology in ACE-1 [Bates et al., 1998], although we had only one aircraft and were thus unable to observe the airmass continuously. The balloons were made "smart," so that they could adjust their own buoyancy in response to drizzle. We were careful to map out layers in the target airmasses and to get useful samples in each. Also, we developed a strategy for measuring VE by three independent techniques, each of which can be implemented using the same flight track [Lenschow et al., 1999]: First is the inversion flux/concentration-jump technique described above. Second is evaluating budgets of trace species for which the chemical source/sinks and surface fluxes are known, and solving for the entrainment flux. Third is the divergence technique. By flying constant-level sampling legs in roughly 60 km diameter circles, the divergence of the air mass can be computed directly as the closed integral of the cross-track winds around the circles. Mean vertical velocity can then be calculated by integrating the divergence with height above the surface. A key development that made this approach possible is the advent of GPS for improved accuracy measuring horizontal airplane velocity. Lidar observations of the inversion height permit a calculation of VE.

The 1998 NCAR Publication Prize was awarded to several ACE-1 PIs for implementing these techniques to determine VE. Russell et al. [1998] measured the submicron aerosol size distribution with a radial DMA and found different shapes of this distribution in adjoining layers of the atmosphere. They then used the changes in the aerosol spectrum in the SBML to estimate entrainment of air from the overlying air (which they called the buffer layer, BuL) into the SBML. Similarly, they used changes in the DMS concentration in the BuL to estimate entrainment of air into that layer from the SBML. Combining this with calculations based on the divergence and flux techniques gave them estimates of three entrainment rates: entrainment from the SBML to the BuL, from BuL to the SBML, and from the FT to the BuL.

During Lagrangian A the northeast side of our circular legs began to develop cloudiness and was clearly different from their southwest side. This heterogeneity provided evidence about the evolution of aerosol in a cloudy region [Varner, 1997]. This confirmed the importance of cloud processing over the Southern Ocean. In Lagrangian B we were able to conduct three flights while all three balloons remained aloft. Although both the wind direction and speed changed dramatically, the balloons maintained their relative positions (Figure 3). This argues that the target airmass remained coherent for the 30 hours of our observations, although there remained some differential advection due to changes of wind with height.

 

Figure 3. Circular C-130 flight tracks and balloon trajectories during ACE-1 Lagrangian B. X marks the location of the R/V Discoverer, the balloon-launch ship. H marks Hobart, Tasmania, where the aircraft was based.

The evolution of sulfur species concentrations confirmed our picture of remote sulfur chemistry: DMS increased in the nighttime and dropped during the brightest sunlight, while SO2 increased in the day and dropped at night [Suhre et al., 1998]. Photochemically-derived sulfate and MSA aerosol both increased in the daytime and held relatively constant at night [Huebert et al., 1998]. This is an excellent set of data against which to test marine sulfur models, but for one major problem: while it is likely that much of the NSS was being formed on seasalt [Sievering et al., 1999], our inability to quantitatively collect particles larger than 1 µm from an aircraft made it impossible to close the budget and assess the importance of this reaction pathway. In view of the potential importance of large particle fluxes, it is critical that we be able to sample total aerosols without artifact to close budgets during Lagrangian experiments.

Among the most important conclusions from the ACE-1 Lagrangians is that the evolution of the DMS concentration was consistent with a DMS surface flux as calculated by the Liss and Merlivat (L&M) parameterization [Mari et al., 1998; Russell et al., 1998]. In view of the very different flux predicted by the Wanninkhof relationship, it is significant that under this set of conditions L&M was more effective at predicting the evolution of DMS concentrations.

ACE-2

The strategy was further refined during ACE-2, in which balloons were supplemented with inert chemical tracers during three cloudy Lagrangian experiments of three UK C-130 flights each. Although ACE-2 was too recent to have comparisons with models in print, there is some very interesting behavior to challenge models. The first Lagrangian was in clean air, in which the accumulation mode concentration increased by fourfold as the Aitken mode concentration decreased. Seasalt is being considered as one of the possible sources of the increased accumulation mode number. The other two cloudy Lagrangian experiments were in polluted air. In one case the SO2 concentration dropped from 800 to 40 pptv, while sulfate fraction increased and cloud processing decreased the Aitken mode concentrations. The other is apparently a case in which the precursor gases were consumed prior to a series of cloud encounters, such that no change occurred over the course of one day. A series of cloud-free pseudo-Lagrangian experiments between surface measurements at Sagres, Portugal and a ship offshore present cases in which dry continental air evolves as it moves over the ocean. The ACE-2 Lagrangians will no doubt help modelers refine several process rates in their regional aerosol models.

The Future

Whether the aerosol optical depth is controlled by chemical growth or the mixing of airmasses with different relative humidities, we need to know what is controlling it in various regions so it can be modeled realistically. If indeed much of the aerosol evolution takes place near source regions, Lagrangian experiments should be conducted as close as practical to the sources, so we can observe the maximum evolution of species concentrations. One essential refinement for Lagrangians in airmasses moving off Asia is to use regional aerosol CTMs to predict where the various species will undergo the most measurable changes under a variety of scenarios. Models can suggest distances from sources and times where the changes are likely to be the most measurable. In that way we can gather data sets that will constrain the models most tightly. We also need to develop fast DMS instruments to improve our knowledge of entrainment rates, and thus to significantly reduce the uncertainty in derived process rates.

There are some needed insights that can only be gained by observing the changes in airmasses over time. By steadily and incrementally improving the technology for doing Lagrangian studies, we have reached a point where unique and valuable insights are being derived from this observational strategy. As we implement further improvements to this strategy, the error bars will continue to shrink and more process rates will be amenable to measurement.

References

  1. Albrecht, B.A., et al., Observations of Marine Stratocumulus Clouds During FIRE, Bull. Amer. Meteorol. Soc., 69, 618-626, 1988.
  2. Bates, T.S., et al., The International Global Atmospheric Chemistry (IGAC) Project’s first aerosol characterization experiment (ACE-1) - Overview, J. Geophys. Res., 103, 16,297-16,318, 1998.
  3. Bretherton, C.S., et al., Cloudiness and marine boundary layer dynamics in the ASTEX Lagrangian experiments. Part II. Cloudiness, Drizzle, Surface Fluxes and Entrainment, J. of Atmos. Sci., 52, 2724-2725, 1995.
  4. Clarke, A.D., et al., Lagrangian evolution of an aerosol column during the Atlantic Stratocumulus Transition Experiment, J. Geophys. Res., 101, 4351-4362, 1996.
  5. Huebert, B.J., et al., Filter and impactor measurements of anions and cations during ACE-1, J. Geophys. Res., 103, 16493-16510, 1998.
  6. Kawa, S.R., and R. Pearson, Jr., Ozone budgets from the Dynamics and Chemistry of Marine Stratocumulus Experiment, J. Geophys. Res., 94, 9809-9817, 1989.
  7. Lenschow, D.H.,et. al, III, Dynamics and chemistry of marine stratocumulus (DYCOMS) experiment, Bull. Amer. Meteorol. Soc., 69, 1058-1066, 1988.
  8. Lenschow, D.H., P.B. Krummel, and S.T. Siems, Measuring entrainment, divergence and vorticity on the mesoscale from aircraft. J. Atmos. and Oceanic Tech., 1999, in press.
  9. Lenschow, D.H., R.J. Pearson, and B.B. Stankov, Estimating the ozone budget in the boundary layer by use of aircraft measurements of ozone eddy flux and mean concentration, J. Geophys. Res., 86, 7291-7, 1981.
  10. Mari, C., K. et al., Physico-chemimcal modeling of the First Aerosol Characterization Experiment (ACE 1) Lagrangian B, 2, DMS emission, transport, and oxidation at the mesoscale, J. Geophys. Res., 103, 16,457-16,474, 1998.
  11. Paluch, I.R. and D.H. Lenschow, Stratiform cloud formation in the marine boundary layer. J. Atmos. Sci., 48, 2141--2158, 1991.
  12. Russell, L.M., D.H. Lenschow, et al., Bidirectional mixing in an ACE-1 marine boundary layer overlain by a second turbulent layer, J. Geophys. Res., 103, 16,411-16,432, 1998.
  13. Sievering, H., et al., O3 oxidation of SO2 in sea salt aerosol water and the size distribution of non-sea-salt sulfate during ACE-1, J. Geophys. Res., in press, 1999.
  14. Suhre, K., et al., Physico-chemical modeling of the first Aerosol Characterization Experiment (ACE 1) Lagrangian B, 1, A moving column approach, J. Geophys. Res., 103, 16,433-16,456, 1998.
  15. Varner, Z.L., Subsidence and cloud processing of condensation nuclei in the first Lagrangian experiment during ACE-1, M.S. thesis, University of Hawaii, Honolulu, 1997.
  16. Wingenter, O.W., et al., Hydrocarbon and halocarbon measurements as photochemical and dynamical indicators of atmospheric hydroxyl, atomic chlorine, and vertical mixing obtained during Lagrangian flights, J. Geophys. Res., 101, 4331-4340, 1996.
  17. Zhuang, L., and B.J. Huebert, A Langrangian Analysis of the Total Ammonia Budget during ASTEX/MAGE, J. Geophys. Res., 101, 4341-4350, 1996.
 

IGAC home
IGACtivities
IGACtivities #17: "Aerosols and Climate"