Trip Report: Solar Radiation and Climate (Richard Allan)

"The Gordon Research Conferences provides an international forum for the presentation and discussion of frontier research in the biological, chemical and physical sciences, and their related technologies" [GRC mission statement]

[Group photo]
I attended a GRC on Solar Radiation and Climate held at Connecticut College (equidistant between New York and Boston) lasting from 24-29 June. The meeting was joint hosted by Bruce Wielicki of NASA Langley and Ramaswamy of GFDL (see photo). The format of the conference was original with sessions running from Saturday to Thursday: Emphasis was placed on the most recent results with key speakers chosen by the chairs to deliver 40 minute talks followed by 20 minutes of discussion (23 talks in total). Lectures ran during the morning and (rather frighteningly) the evening with the afternoons free for poster sessions, informal discussion or recreation... or "soccer" in the case of this conference. A notable 7-6 Solar Radiation and Climate success against Mitochondria & Chloroplasts provided the clear highlight of the meeting! Informal discussion after the evening sessions included the tasting of probably some of the worst beers in the history of fermentation ("Mystic Beer" proved one of my personal least favourites and was indeed a complete mystery). [Rich pickings for the winners!]

The first Solar Radiation and Climate GRC conference was 2 years ago, fresh off the wave of debate created by the "Anomalous Solar Absorption" question which is still as yet unresolved (do clouds absorb more solar radiation than can be accounted for and does this limit our understanding of radiative processes and our ability to model climate adequately). This year the meeting was chaired by Bruce Wielicki of NASA Langley. Ramaswamy of the Geophysical Fluid Dynamics Laboratory (GFDL) was the vice chair and chairs the next meeting in 2003 (chosen among other reasons such that clashes with the European Championships or the World Cup are avoided!). The 2003 Solar GRC meeting will concentrate on Climate Modelling. Highlights of this years meeting are described below.

Dennis Hartmann - Climate and Cloud: The importance of low level stratus clouds to the net top of atmospheric radiation budget and the cancellation of shortwave and longwave cloud radiative forcing for deep convective clouds were illustrated using ERBE and ISCCP data. By compositing cloud amount by optical depth and pressure level it was shown that upper level cloud varies significantly in optical depth. One of the more interesting findings of the meeting concerned the results of a two box model whereby increased specific humidity supplied from the convective region to the dry subsiding region caused the boundary layer inversion to decrease in altitude due to the additional longwave cooling to the surface. Thus increased upward emission from the low level clouds was compensated by the decreased cooling to space and to the surface due to increased atmospheric humidity which thus had only a small effect on the SST. The concluding message was that the radiative properties of clouds are ill understood and cannot be validated globally precluding the accurate account of cloud feedback in climate simulations.

Jim Haywood - Climate and Aerosols: The only other Brit. Quality talk needless to say dealing with the latest IPCC estimate of radiative forcing most especially from aerosol. One source of difference between model aerosol radiative is the calculation of cloud based on relative humidity within the model. For example, high relative humidity with no cloud gives a large direct radiative forcing due to aerosols (e.g. an on-off cloud scheme rather than our RH crit parametrization). The recent measurements of the Saharan dust aerosol effect on the radiation balance off the West coast of Africa was discussed. Aerosol optical depths compared well with those measured by AVHRR.

Bill Rossow - Climate Feedbacks: By far the most depressing talk argued the futility of the linear feedback approach in Climate because: neither the forcing nor the sensitivities are constant in time, small feedbacks cannot be neglected and that there is no causal hierarchy of variables. Further a change in cloud amount is not required to produce a feedback: changing a clouds position (from land to ocean, or day to night) will impact the radiation balance without a change in properties/amount of cloud. A proper analysis should be multivariate, non-linear, non-stationary, non local in time and space and can be interpreted in terms of physical processes. By way of an example a Lorenz continuous model with 3 variables was used to illustrate these points. Ramaswamy argued that some non-linear feedbacks will cancel to a certain extent when averaging the system over space and time.

Steve Crueger talked on the GEWEX Cloud System Study (GCSS) while Tony Del Genio argued that concentrating on upper tropospheric humidity was inadequate and that cloud threshold humidity is not apparent from the observations. Dynamics plays a particularly large role, for example there is a large variation in Stratocumulus with vertical velocity. Nothing much new here...

Eugene Clothiaux - Cloud Frequency/Layering: Cloud profiling using radiometers, a micro-wave instrument and lidars were used from ARM sites to conduct atmospheric profiling. Thin cirrus at Nauru in the tropical West Pacific seemed ever present, detected by Lidar but missed by radar. Transmitter degradation is a likely source of error in trends in long data sets. Synergy with satellite instruments is encouraged. Interestingly small particle, thin cirrus cloud, it was argued, appears to be unrelated to tropospheric dynamics and may rather be determined by atmospheric waves propagating from the stratosphere.

Gerald "Jay" Mace told us that water content is dependent on temperature but size and concentration have the opposite sign dependence on temperature. Taneil Uttal, Polar Cloud Challenges: The predominance of cloud (about 90%) and Artic Haze dominates the Arctic radiation budget. Problems for cloud detection are caused by the lengthy polar night, sparse population, low contrast between snow and cloud, simultaneous ice and liquid clouds and temperature/humidity inversions. Longwave heating of the surface by clouds (35 Wm-2) dominates over the shortwave cooling (24 Wm-2) for all but summer day time. The Study of Environmental Arctic Change (SEARCH) is the latest project.

Herman Gerber spoke on new instruments while Jost Heintzenberg talked on airborne aerosol measurements. A big change in the vertical profile of aerosol had been noted after Pinatubo. Measuring aerosol from aircraft is limited because the flow field is altered and the aircraft pitches and rolls. Lagrangian measurements using balloons or airships are encouraged.

Oleg Dubovik - Surface Aerosol Measurements: The advantages of surface instruments of aerosol over satellites are that the direct solar path may be sampled thus removing the problems of angular dependence of radiation. However the limited spatial coverage is a problem and one which is being addressed by AERONET. Various examples of aerosol characteristics (single scattering albedo with wavelength of radiation) for particular sites were presented.

Yoram Kaufman - Aerosol Direct Radiative Effects: Some of the latest satellite instruments for measuring aerosol were described. For example the 36 channel Moderate Resolution Imaging Spectroradiometer (MODIS) and 9 view angle, 4 channel MISR are currently measuring aerosol (optical depth, particle size, refraction, etc), surface type and cloud/atmospheric structure. An important point was made with regards to assessing aerosol type. For example the CO tracer and its correlation in spatial extent with aerosol may be symptomatic of biomass burning aerosol.

Jim Coakley - Aerosol Indirect Radiative Effect: There is a huge uncertainty in measuring the indirect aerosol effect which is manifest as 3 separate effects of increased aerosol loading on cloud: more abundant but smaller cloud drops affecting cloud optical properties, modified size distribution thus affecting cloud lifetime and a possible change in precipitation efficiency. "Experiments" involving identifying ship tracks suggest that clouds lose water and thin slightly. While an indirect effect of aerosols on cloud appears to be active, the magnitude appears smaller than assuming constant liquid water path. Lengthy discussions followed concerning the limitations of ship track studies due to their questionable applicability to the global scale effect.

Norman Loeb talked about the errors inherent calculating average fluxes from satellite radiances. The different resolution of CERES compared to ERBE means differing response of the angular dependence - there is greater limb darkening in CERES radiances compared to ERBE. The synergy of GERB to LEO satellites was discussed (ie diurnal models from GERB, ADM's from CERES).

Tom Charlock - Surface and TOA Constraints on Atmospheric Solar Absorption The surface flux measurement error was discussed. Instantaneous shortwave uncertainty was thought to be 15 Wm-2 with longwave double this. Uncertainty in aerosol characteristics impedes the validation of the shortwave flux although a goal of 5 Wm-2 accuracy is achievable and there is agreement between separate instruments of 5-10 Wm-2. Problems with heating of the longwave pyranometer dome by shortwave radiation is now accounted for and accuracy of 10 Wm-2 is likely. A negative offset is often present in the shortwave measurements at night. It is thought that longwave heating by clouds affects the shortwave measurement thus introducing errors. There is still varying results with regard to the anomalous absorption problem. Recent results from Valero et all (JGR 105, p.4743) show no evidence of anomalous absorption beyond model/observation error. Using CERES and CAVE data (many surface based radiometers covering a CERES pixel) the atmospheric absorption of solar radiation is 227 Wm-2 for cloudy regions and 219 Wm-2 for clear-sky regions (using the Maximum Likelyhood Satellite derived cloud screening). This gives just 8 Wm-2 additional absorption in clouds which smaller than the model/observation error. A similar result is gained using Chuck Long's surface based cloud screening when scaled by Cos Zenith Angle.

Frederic Parol - New Multi-angle Polarization Cloud Measurement: POLDER-MISR are used to detect cloud properties. The region affected by "sunglint" is accounted for. Model results give reasonable agreement assuming 10 micron spheres for liquid cloud, but hexoids are more appropriate for ice clouds. Different effective radii and optical paths are used.

Michael King - New Multi-spectral Cloud Measurements (e.g. MODIS): Products are Cloud Mask, Cloud radiative and microphysical properties (cloud top pressure and temperature, optical depth, emissivity, phase, effective radii), aerosol optical depth and size distribution, atmospheric temperature and humidity gradients, column water vapour available on a 1x1o grid as daily, 8 day and monthly means, standard deviation and PDFs.

Pat Minnis - Validation of Cloud Properties Using ARM Time Series: Validation of CERES cloud properties using surface measurements at the Southern Great Plains and Tropical West Pacific. Optical depth is generally 18% less for satellite retrievals than the surface. Radar data picks up thin cirrus cloud not detectable by radiometric means. Geostationary data (GOES) gives effective radius larger than surface values.

Frank Evans talked on Far-IR and Sub-mm Measurements of Ice Water Path/Partical Size. David Winker talked on the impressive satellite "train" consisting of AQUA (was EOS PM), CloudSat and PICASSO and their potential for observing direct aerosol forcing, the indirect aerosol effect, cloud feedback and longwave heating rates. Cloud overlap strongly affects the surface longwave direct aerosol forcing.


[Group bonding session!] The final session in fact began with a group bonding session with all speakers and the discussion leader (Ramaswamy) plus an additional aerosol expert (Jim Haywood) undertaking a scientific trip to the Casino in a convertable. Inevitably, the first talk dealt with statistics...
[Winnings=donuts!]
Betsy Wetherhead - Climate Trend Detection Betsy argued that there must be robust physical understanding attached to climate trends to avoid statistical artifacts or instrument errors. For example a repeating set of 10 identical annual cycles gives a statistically significant negative correlation depending on the initial and final points. Standard statistics is limited because the real world provides non-normal distributions and autocorrelation of the variables. To detect a trend a representative size, stable instrument with full documentation are required and the trend must be larger than the background noise. A low noise, low autocorrelation dataset is best for trend detection. MSU mid-tropospheric temperature is highly autocorrelated in the tropics but isn't as noisy as higher latitudes. It was pointed out in the discussion that noise is in fact a valid climate parameter as well (e.g increased hurricane frequency, etc).

Brian Soden - Interannual Variations of radiation and climate parameters: GCM's generally all predict an increase in column water vapour with temperature of 7% per Kelvin (in line with the Clausius Clapeyron relationship). However there is disagreement on the change in precipitation with temperature but with all models producing a smaller increase per Kelvin than column water vapour. This suggests a longer residence time of water which contradicts the proposal (e.g. Lindzen) that a warmer world will produce greater convective overturning. The model net surface radiation is strongly correlated with precipitation suggesting that radiative cooling to the surface is balanced by latent heating. Models generally agree on the variation of atmospheric temperature, clear-sky OLR and column water vapour over the ENSO cycle and this agrees with SMMR, SMM/I and ERBE observations. Decreases in column water vapour using TOVS statistical retrievals of column moisture (e.g. Schroeder and McGuirck) are thought to be artifacts of the satellite calibration with radiosonde data. Model precipitation and surface net longwave fluxes do not agree well with observations although it is not certain that the observations are of sufficient quality for use in interannual variability studies. A question is, how can models accurately simulate atmospheric temperature while the precipitation, and hence latent heating, is erroneous? Brian argued that the latent heating of the atmosphere cools radiatively to the surface thus explaining the much larger observed surface longwave flux variability than is simulated. I argued that the surface longwave flux "observations", which are derived from satellite, are not of sufficient quality for use in such an argument. Further, because temperature and column water vapour are well simulated by models, discrepancy in the surface longwave flux must be caused by a much larger cloud variability than models. This would then alter the solar fluxes, further complicating the issue. ERBE reflected shortwave fluxes are not dissimilar to model simulations over the 1985-1990 period analysed leaving the only explanation as a variability in cloud base altitude.

Richard Allan: Didn't take any notes on this one - sorry! Talked about cloud and water vapour feedback, their dependence on the atmospheric distribution of relative humidity and the changes in the wide field of view/ scanner observed fluxes since 1985. Models appear to conserve relative humidity much more than the ERA (although ERA column moisture variability is likely erroneous) and have a smaller interannual variability in cloud amount as diagnosed from the top of atmosphere tropical longwave and shortwave radiation budget.

Overall this was a good meeting marred by the clash with the final week of Euro 2000 football. Unfortunately some talks were without solid conclusions perhaps due to the emphasis on most recent results being taken too literaly in some cases (i.e. plots snatched off their students desk the previous morning). Also some talks described useful cloud and aerosol products without actually completing a useful analysis. The format of the meeting was excellent: Almost all participents stayed the entire duration (even until my talk!) and discussion was encouraged not only by the length of time allocated after each talk, but also by the informal nature of the conference compounded by the provision of plenty of recreation time. [Banquet]