Near Cloud Workshop



Download 16.04 Kb.
Date conversion08.07.2018
Size16.04 Kb.
Near Cloud Workshop
Marshak: Everything started from Koren’s 2007 Twilight Zone paper, which Sasha recommends reading. Remember that Charlson et al. (2007) came to the same conclusion from an entirely different direction, independently. Both papers should be read and referenced. As we start talking about twilight/continuum/transition zones, don’t forget co-varying with meteorology and satellite retrieval artifacts. People offer cloud contamination in the AOD as explanation, but cloud contamination would decrease spectral dependence of the retrieved AOD or correlate with the coarse mode, and often there is much stronger correlation with fine mode than with coarse mode, suggests contamination is NOT driving the correlation.
What about working in radiance space? This eliminates retrieval issues, but introduces an unconstrained geometry, which would dominate.
Quaas paper. Much less correlation between CF and aerosols found in models. Much of the correlation between AOD and cloud fraction is due to humidification, meaning that increases in AOD are dominated by aerosols swelling with humidity and moving towards more optically efficient sizes. Turn off humidification and correlation between CF and AOD plummets.
Can we use diagnostics of LES or other models to separate effects? You have to be careful using models. May not want to see insides of the sausage.
Varnai: What is relationship with CF and aerosol size? Can be positive or negative, while always positive with AOD. Dealing with growing fine mode. AE is a bad parameter. Can it be trusted especially at low AOD? Is FMW a better measure? Size is hardly addressed in the literature. Plot of CF-AE correlation seems to correspond to meteorological forcing of both. CALIPSO color ratio shows a shift to larger particles as function as distance to cloud.
What about downdrafts outside of clouds? How do particles accumulate when there is subsidence? Because subsidence is diffuse and small particles just don’e move down.
Ignatov 2005

Kaufman 2005

Zhang 2005

Charlson 2007

Eck 2014

Loeb and Manalo-Smith

Schuster/Loeb
Eck: Sudden increase in fine mode AOD at AERONET sites (0.3 to 0.6). No change in AE as function of time. Can’t be cloud contamination. If humidication there should also be decrease in AE. At the same time HRSL found real enhancement of backscatter near clouds. Did not show any change in particle size parameter as function of distance to cloud. In situ measurements found increase of aerosol concentration near clouds. They found new particle formation at one altitude and swelling at a different altitude. Two processes balance each to cancel changes in AE. PW changes slightly near cloud, which means humidity halos at cloud level must be huge. Non-preciptating clouds. Wouldn’t there be lower RH beyond the immediate halo as part of the subsidence. At what scales? Increase of AOD goes out to about 2.5 km from clouds. How do we isolate these two processes (swelling/new particles)? By altitude from lidar? HSRL didn’t do that analysis. MPL on the roof for statistics?
Eck: Cloud processing. Fog and stratus. Forms particles in the 0.4 um radius range. Mode has relatively short life time (3-7 hours). Shoulder doesn’t come all the way down. Processing sulfur dioxide in aqueous process. Low cloud fraction follows

AOD vs. fine radius (shift to more optically effective range). Stratus is acting for hours on the particles. All in the same layer.



Eck et al. 2012
Eck: Paper in prep. Aerosol-cloud situation in east Asia. Around Beijing and South Korea. Likely meteorological co-variation. Big influxes of pollution associated with week frontal passage, comes with cloud system. Hazy in clear slots. AOD spectra in the one obs in that day shows extreme curving for fine mode domination. AOD = 2.4. AE =0.8. But AE is not meaningful for this much curvature. Mixed typical modes (Fresno + humidification) and was able to match observed AOD spectra. A significant number of AOD events are associated with cloud cover. Very hard for AERONET and satellite to monitor these events. Version 2 might knock this ob out because of triplet in the UV. Version 3 will weight triplet variability towards coarse mode, and will keep more data when it is fine mode dominated.
General discussion: What do we want to get out of today and for the future? Started from thinking of a interdisciplinary proposal. Should we get together again in 3 months? Be prepared for next call for Center proposals. STG. Maybe we can send recommendation for specific observation and modeling needs. Maybe we can use some modeling results to show that this is important on a global basis.
Levy: eMAS gives fine resolution AOD retrievals during SEAC4RS. eMAS vs. MODIS. When collocated, eMAS is a little bit low biased. But eMAS retrieves a lot more than MODIS, and those histograms are very different. Now comparing to AERONET. eMAS is biased very high as compared with AERONET. If you include more background, then high bias goes down. Plotting as distance to cloud, see major enhancement starting about 5 km from cloud. There is some difference with solar geometry. Now using CPL, plotting eMAS and CPL-AOD vs. cloud density, eMAS is dependent, but not CPL. Done with constant lidar ratio. Suggests 3D effect dominates. How does this translate to aerosol radiative effects? What kind of modeling or measurements that plot the RH as function of cloud?
General discussion: These 3D effects are real. There are photons escaping to space that side scatter from clouds, into adjacent areas and then to space. They are part of radiative effects. Rotem’s paper. This looks at aerosols-clouds as a single field, without worrying about which is which. The goal is the overall radiative effects of the field.
Patadia: Pixel level uncertainties in DT retrieval. For enhanced AOD retrievals, go back into 500 m data, and removed all pixels at various distances from cloud. Removing closer to cloud pixels gave both + and – difference from original retrieval. Did not eliminate high bias. If in a very clear region, then Gausian distribution of reflectances, but if near clouds, the reflectances skew both high and low because not homogeneous. Not every near cloud case does this, but broken cloud fields do.
Sayer: We should hire a real statistician. Currently asking “what is the linearity between the two quantities?” when we want to ask, “what is the relationship between the two quanities?” Should not use Pearson correlation. Suggests using rank correlation. Also be careful about the need for normal distributions. Be aware of uncertainties as part of the analysis.
Zamora: MISR group over view. Current capabilities: Aerosol plume and cloud height. aerosol light absorption. Infer hygroscopic behavior. Potential future directions: reduce uncertainties, upwind/downwind plume properties, HSRL-2 to constrain change in aerosol properties near cloud and MISR further from clouds, stratifying satellite obs and model simulations by relevant parameters, field campaigns, lab experiments, daytime/night time cases.
Limbacher: Use HSRL and other lidar to look at aerosols near clouds, but you don’t have extinction with other lidar, and with HSRL you have extinction but not spectral extinction. With HSRL-2 gives spectral extinction. Lets look at changes in aerosol size parameter as you approach cloud and as you leave cloud. No 3D or straylight effects. With MISR characterize background aerosol. Modify MISR retrievals from background as we approach cloud. Take 3D effects out of the equation. HSRL-2 is necessary because of spectral extinction. Need more coincidence between HSRL-2 and Terra. Work fast before Terra dies. Have you considered trying Guoyong’s technique to remove 3D effect from MODIS/MISR? The proposed would offer a more quantified attempt at the same thing lots of other people have been doing . Polarization will also help separate out the 3D effect.
Jethva: Aerosol above clouds. Validation of technique applied to EPIC, validated with HSRL-2. Several times per day. Don’t know of diurnal variation of aerosol from that area. ORACLES region. Important data set is available. EPIC product will be release in 2 weeks. This would be UVAI. Before AGU will release cloud product.
Zhao: 3D cloud structures. History comes out of the Glory team. Capture polarized 3D effects and how do these influence cloud retrievals and retrieving aerosol properties in the vicinity of clouds. POLDER is hard because of limited angular resolution. New polarimeters may give better angular resolution. Increasing complexity of cloud representation in models, climaxing in LES simulations. Simplest demonstration of 3D effects with step model. Can see the enhancement and depression (leaking effect) of reflectance because of illumination geometry. But looking at the polarimetric field has very narrow/sharp illuminating effect but no depression (leaking effect). Polarization has a 3D effect, but different in nature. Will affect microphysics retrievals. One step up in complexity is fractal cloud. Q field is only sensitive to microphysics. 3D effects can shift magnitude up and down, but does not shift the polarization effects (bows). In future will have polarimetric simulator. Down the road we can put in aerosol between cloud. Maybe at some angle we can retrieve aerosol and at another angle we can retrieve cloud. But what about ice? Ice has no polarization. Maybe we can retrieve aerosol below ice cloud.
Da Silva: ACEPOL. 4 polarimeters, HSRL-2, AirMSPI on ER-2. No in situ measurements.

Sub-grid variability from modeler perspective. models are now down to 3.5 km. Still not at a cloud-resolving model at global scale. Clouds are sub-grid phenomena. Need parameterization of sub-grid variability (water vapor, liquid water relationship to total water). Assimilating clear-sky aerosol data. Right now aerosol model is dry and must be humidified to get to AOD. Must relate measured AOD to dry mass. Want to stay away from cloud. If you go to the MODIS 10x10 approach, we throw out all data with 70% or more cloudiness. Must remove twilight zone. Interested in Falguni’s results. Want estimate of AOD that is closest to dry mass. Calls for need to separate the different processes: 3D, humidity, etc.


General discussion. What about MAIAC? Have they done anything? Account for clouds/shadows differently than we do. Report distance to cloud, but not yet studied.
Summary: 1) List literature. 2) We represent different approaches. Should add MAIAC. Compare how address this problem in retrievals, different instruments, algorithms, including ground-based. If successful, talk to Maring/Bontempi to put a call out for aerosol-cloud interaction or retrieval. 3) Next year for STG call. 4) Come back in 3 months. 4) Sasha will make review of literature.
Why are we interested in this? DRE? Aerosol processes? Retrievals? Assimilation models? Do we include or not include this enhancement near clouds in estimating DRE/F? Do we need a new vocabulary to address cloud-aerosol fields? Have not talked about microphysical stuff like CCN. Is it fair to use aerosol near clouds to investigate cloud-aerosol action? Yes. If you can take out adjacency effects. How much are these aerosols growing in the presence of clouds? Want TOA radiance because its real, but it isn’t dry mass. Maybe an AGU session. Goddard is uniquely positioned to lead in this area. AGU is where you share results, but not where you solve problems. Should invite HSRL or GISS folks.


The database is protected by copyright ©dentisty.org 2016
send message

    Main page