Aerosol Workshop — June 2-3, 1997

Session 5: Links Among Satellites, Models, In Situ and Surface Data

(Facilitator: John Ogren; Recorder: Ralph Kahn)

Aerosol Transport Models + Satellite Data

Steve Schwartz, Brookhaven National Laboratory

Published chemical transport models for aerosols or for specific aerosol species such as sulfate, especially those that are driven by monthly mean winds from climatology, exhibit smooth contours. The repeated publication of figures showing such smooth contours may have prejudiced thinking that this is in some way representative of the actual distribution of aerosol loadings. Seasonal composites of satellite-derived aerosol optical depth likewise exhibit rather smooth contours. Reality is very much otherwise. The short residence times of tropospheric aerosols are comparable to the time scale of variability in synoptic scale winds and precipitation that control the distribution of aerosols. This situation, together with the highly nonuniform distribution of sources of anthropogenic and dust aerosols, leads to a highly heterogeneous distribution of aerosol loadings (in great contrast to the rather smooth distribution of the long-lived greenhouse gases). The heterogeneity in aerosol loadings is readily manifest in time series of aerosol loadings at a given location. Likewise one should expect a similar variation in the spatial distribution, as is exhibited in models for which the controlling meteorology exhibits the short-time variability of actual synoptic scale variability, but the synoptic observational coverage required to develop this picture is lacking (one might imagine trying to get this from GOES, although the photometric resolution is marginal). Models can be highly valuable in trying to discern the aerosol loading in observations, provided they accurately represent the temporal heterogeneity that is responsible for the heterogeneity in loadings. For this it is necessary to drive the model by observation-derived meteorological data, not by climatological or GCM meteorology. This situation may be turned to advantage, because the inherent spatial and temporal variability of aerosol loading allows the possibility of experiments that can discern and quantify the aerosol influences, by comparisons between high and low loading situations.

Three-Dimensional Global Modeling with Cloud-Aerosol Physics

Steve Ghan, Pacific Northwest Laboratories

To facilitate comparison between simulated and observed aerosol, all aerosol models should be either driven by observed meteorology or by a GCM that is nudged toward observed meteorology.

Aerosol radiance (the difference between the true radiance and the radiance if aerosols were not present) should be used to evaluate aerosol models. Use of aerosol optical depth is not as closely related to aerosol radiative forcing and introduces uncertainties in the "observations".

There does not appear to be any way to determine cloud droplet number concentration from current or planned satellite measurements. The combination of a mm cloud radar and a microwave radiometer on a satellite would fulfill this critical need for estimating indirect radiative forcing.

Pinatubo Radiative Forcing and Climate Response

Alan Robock, University of Maryland

Scientists use a variety of definitions of "Radiative Forcing." What the climate system actually feels is in situ radiative heating of the atmosphere and changes of net shortwave and downward longwave flux at the surface. All radiative forcings should be evaluated in these terms to compare them. This can be illustrated by looking at the effects of Pinatubo aerosols on visible, near-IR, and thermal IR in both the troposphere and stratosphere.

Stenchikov et al. (Max Planck Inst. fur Meteorol., Rept. 231, Hamburg, 40 pp., 1997; submitted to J. Geophys. Res., March 1997) developed a detailed data set of stratospheric aerosol properties and distribution for the 2-year period following the 1991 Pinatubo eruption and produced heating rates and fluxes from the data, using the Max Planck Institute ECHAM-4 GCM. We suggest that other GCMs use these same data sets (aerosols, or heating rates and fluxes) as forcing, to compare the radiation schemes and climate response. Stratospheric aerosols should be an important part of the NRA, as they offer important episodic (event-based) forcings which are important for the climate system.

Factors Governing Aerosol Radiative Forcing

V. Ramaswamy, NOAA Geophysical Fluid Dynamics Laboratory

A principal factor affecting aerosol radiative forcing is the dependence on humidity. Thus far, most if not all estimates of anthropogenic aerosol forcing have been made using general circulation model (GCM) climatologies. Since sub-grid scale variation of humidity is a major uncertainty in GCMs, and as aerosol hygroscopicity and the accompanying growth are nonlinear functions of humidity (especially at relative humidities greater than 85%), this becomes an important element to consider in assessing the accuracy of the aerosol radiative forcing. Haywood et al. (GRL, 24, 143-146, 1997) demonstrate that the impact of considering a fine spatial resolution of the humidity field versus a coarse one can lead to significant biases in the forcing estimates. It is, hence, a factor to be kept in mind when attempting to verify model estimates with observations.

GCM simulations of anthropogenic aerosol concentrations, together with a model climatology of humidities and cloud distributions, have been used to investigate the sensitivity of the (negative) hygroscopic sulfate and (positive) hydrophobic soot aerosol forcings (Haywood and Ramaswamy, JGR, submitted, 1997). It is found that different model simulations of the anthropogenic sulfate aerosols differ considerably in the vertical distribution despite similar column burdens. This arises due to differences in the horizontal and vertical transport of species in the various models. These differences result in varying estimates of the radiative forcing. The sulfate aerosol forcing, being largely governed by the relative humidity profile, is greater when the major portion of the columnar mass is located lower in the troposphere. In contrast, black carbon yields a greater forcing when a substantial mass is present above low clouds; then, the multiple scattering process leads to an enhancement of the absorption.

Integrating Aircraft Measurements, Satellites and Models

Antony Clarke, University of Hawaii

Assessing global radiative forcing and related climate issues depends upon analysis of validated satellite data. One approach to validation is through comparison to calibrated surface measurements (e.g., AERONET), but by itself such a comparison has many ambiguities. Effective linking and interpretation of satellite observations and model simulations require knowledge of the vertical characteristics of the aerosols, implying the need for aircraft and/or lidar measurements. A strategy for integrating limited resources to best support a satellite/model validation and interpretation effort should consider the use of light 2-4 seat aircraft or remotely piloted vehicles (RPVs), which offer cheap, flexible and frequent sampling opportunities, as well as the use of traditional major aircraft campaigns.

Aircraft data need to be linked appropriately to both satellite and modeling needs. Given aerosol variability, appropriate spatial and temporal scales must be established better, as well as the most uncertain or sensitive parameters needed by the modeling community. Lidar measurements, which in the past decade have revealed complex layering of aerosols, can be especially useful for placing in situ measurements in context, thus helping to link in situ data to models and satellite retrievals. These considerations lead to a suggested strategy for enhancing the integration of models, satellite data, and other measurements [see viewgraph of Clarke in Panel C (Strategy)].

Aircraft data must play a role, in combination with models and satellite data, in analysis of the challenging indirect aerosol forcing issue. A large number of factors influence cloud albedo and lifetime, in addition to possible effects of aerosol perturbations, which thus complicates interpretation of even well-conceived field experiments. Anomalous aerosol conditions, as exist in large well-defined plumes (such as the volcanic plume from Kilauea, Hawaii or possibly an isolated city plume over pristine ocean) can provide exaggerated conditions to study indirect effects on clouds. It would be useful to find situations in which there is an extended persistent gradient in aerosol properties in a region with reasonably uniform cloud fields, dynamics and surface temperature. Interpretation will probably depend upon having satellite data over many years to verify the existence of statistically robust cloud effects, consistent with the notion of "events in global decadal context".

Regional Modeling with Data Assimilation

Doug Westphal, Naval Research Laboratory

I suggest that we evaluate our ability to model aerosols (sources, transport, sinks) by applying models (global or regional) to observed events. Questions to be answered using this approach include:

  1. What resolution (spatial or temporal) is required to accurately model sources, the near-source nonlinear microphysics and chemistry, and sinks (precipitation).
  2. Does the dynamical model capture the characteristics of the meteorological forcing, including RH, PBL depth, precipitation, easterly waves, etc.?
  3. How can we combine data from disparate satellite retrievals into a single product? What is the accuracy?

Session 5 Summary by Recorder
(Ralph Kahn)

John Ogren (as Facilitator). Regarding the link between models and observations: (1) Observations provide estimates of aerosol properties, their variability and uncertainties. (2) Models provide sensitivity of climate to aerosol properties. So the two communities must interact.

Steve Schwartz. Aerosol properties and amounts vary on the space and time scales of synoptic meteorology, which governs aerosol transports. "Averaged" values, (e.g., smoothed contours) eliminate the peaks and valleys, and are inadequate to represent aerosol non-linear effects, such as those involved in aerosol interactions with clouds. Both models and observations must capture aerosol variability on synoptic scales.

Steve Ghan. Satellite observations of radiance are usually converted to aerosol properties, which are used in a model to generate radiative forcing. This procedure involves a set of assumptions that can be avoided if model-produced radiances are compared directly to satellite-derived radiances.

Ghan also emphasized the variability of aerosols on synoptic scales.

Alan Robock. Robock recommends using the same forcing in multiple models to explore model behavior. Stenchikov and Robock developed a detailed data set of stratospheric aerosol properties and distribution for the 2-year period following the 1991 Pinatubo eruption. They produced heating rates and fluxes from the data, using the Max Planck Institute ECHAM-4 GCM. They suggest that other GCMs use these same data sets (aerosols, or heating rates and fluxes) as forcing, to allow comparison of the radiation schemes and climate response.

V. Ramaswamy. The vertical distribution of aerosols matters, e.g., for hygroscopic aerosols in a model where relative humidity varies with height, or for absorbing aerosols when there is cloud in the atmospheric column. The horizontal distribution of aerosols matters, e.g., for hygroscopic aerosols in a situation where relative humidity varies horizontally.

Ramaswamy also pointed out the danger of representing sub-grid-scale variability with "averaged" values, because of the non-linear dependence of aerosol properties on relative humidity when RH >~ 80%.

Tony Clarke. In some situations, vertically integrated aerosol properties, e.g., those measured by satellite, are related to vertically resolved properties obtained by in situ measurements - but not always. LIDAR profiles and associated relative humidity offer promising ways to establish and interpret column aerosol properties and structure.

Clarke also pointed out variability of aerosol properties on synoptic spatial and temporal scales. There are some relatively inexpensive aircraft packages to measure aerosol properties in situ. These can be used to identify the cases when surface aerosol properties are or are not representative of those in the entire column.

Doug Westphal. It is important to resolve aerosol sources and sinks at the correct scales. For example, you need the detailed, local wind pattern to understand dust sources in the Gulf of Oman region. Weekly "averaged" AVHRR aerosol retrievals show a dust "plume" off West Africa whereas the dust is actually transported in eddies mixed with clouds.

Westphal recommends an "events-based strategy" in which attention is paid to the scales necessary to resolve key attributes of the event, such as looking at "bright spot" dust sources over deserts in the TOMS daily data.

Bill Rossow. Pointed out that event-specific data handling operations may be helpful in characterizing phenomena of interest in the observations, such as studying all the easterly-phase waves off West Africa for the Sahara dust transport over the Atlantic.

Workshop Homepage * Agenda * Summary
Sessions: 1, 2, 3, 4, 5, 6
Panels: A, B, C, D * Participants