Aerosol working group minutes

This page will hold minutes of meetings of the UKCA-GLOMAP-mode aerosol working group.

• Minutes of meeting held 22/11/2013 over teleconf

Summary of actions:

1) JM/AJ to try no-DMS run also to see how much the DMS is contributing to the AOD there. 2) GM to try sensitivity study where multiply sedimentation velocity by 2 (modal underpredict sedi?) 3) CJ to investigate precipitation rates in Southern Ocean region and whether wet dep may be inefficient there. 4) JM to report how bias looks at GA6 CheT+GLOMAP job when all bug-fixes have been applied. 5) CJ/BJ to try sensitivity run with convective scavenging coeffs reduced from 1.0 to 0.5 6) Next UKCA aerosol teleconf 2pm Thurs 12th Dec

Notes from UKCA aerosol teleconference held Friday 22nd November between Leeds & MO Present: Ken Carslaw, Graham Mann, Colin Johnson, Jane Mulcahy, Ben Johnson, Steve Rumbold, Alastair Sellar.

There were 2 main topics of discussion for the teleconference aimed as resolving the two remaining issues re: the current v8.4 GA4 CheT+GLOMAP job – which are: 1) AOD and sea-spray too high in Southern Ocean (and also Arctic) 2) AOD and BC, OC (also sulphate?) column-mass too low in biomass burning regions

1) S. Ocean AOD high bias At Sept 12th telecon, discussion was held about the very high AODs in the model in the S. Ocean. At that time job amtjn had large region with AOD>0.5 during SH summer (DJF) and max value of ~0.45 in JJA (winter). Had identified either sea-spray or DMS as potential causes. Had discovered S. Ocean DMS was very high-biased (b=0.29, 3.04, 3.31 at 3 S. Ocean sites). Had resolved to try changing Wanninkhof  Liss Merlivat sea-air transfer (known to generate much lower DMS emission). Equivalent job ajkzi (using LM sea-air) reduced DMS high bias substantially (b=-0.11,1.66,1.22 for same sites). AOD from ajkzi job (using LM sea-air) reduced AOD slightly in S. Ocean – peak now just below 0.5 in S. Ocean during DJF. Comment from marine biogeochemistry group was why use Liss Merlivat since it was so old. After a discussion MO folks resolved to try runs with Nightingale instead. But Leeds commented this falls within Wann & LM so this not expected to reduce DMS high-bias. Action: JM/AJ to try no-DMS run also to see how much the DMS is contributing to the AOD there. 2nd progress since last time was that SW had found factor-0.5 was being applied in fall velocity due to code-merge error. JM had circulated plots (5th Nov) showing impact of sedi bug-fix – DJF AOD in SO then peaks at 0.4 rather than 0.5. Bigger impact on AOD during SH winter when sea-salt is dominant contribution (SO peak at ~0.35 rather than ~0.45). But although AOD reduced, still much too high against collection-6 MODIS and AERONET which show AOD ~0.1 there). BJ explained he had checked AODs in each mode in the region and found coarse mode AOD contributed ~0.3 AOD on the annual mean so main source of high-AOD likely to be sea-salt not DMS. CJ explained budget analysis shows removal of sea-salt showed 45% by dry-dep-sedi and 55% by nucscav/imscav. This was surprising as coarse mode expected to show sedimentation as dominant removal process. Action: GM to try sensitivity study where multiply sedimentation velocity by 2 (modal underpredict sedi?). Action: CJ to investigate precipitation rates in Southern Ocean region and whether wet dep may be inefficient there. Note: CLASSIC sea-spray AOD was found to much higher at GA4 than GA3 or GA5 so expect reduction anyway GA4GA5/6. Action: JM to report how bias looks at GA6 CheT+GLOMAP job when all bug-fixes have been applied.

2) AOD too low in biomass burning regions GM had suggested at UKESM-UKCA meeting on Wed that BC/OC seemed reasonable in biomass regions so why AOD low? BJ had checked and although BC and OC mass burdens do show expected plumes in biomass regions, they are lower than at v7.3 which suggests AOD low-bias is consistent with BC/OC burdens so not an issue re: RADAER. BJ explained that spatial patterns in column-integrated SO4, BC and POM suggested too much scavenging in tropical regions --- likely that convective (plume) scavenging implementation was too aggressively scavenging. Omission of re-evaporation also likely to mean scavenging needs turning down. Action: CJ/BJ to try sensitivity run with convective scavenging coeffs reduced from 1.0 to 0.5.

Next UKCA aerosol teleconf meeting will be held at 2pm on Thursday 12th December. This is scheduled for 1 week ahead of UKESM-UKCA meeting on 19th Dec (target date for GA6 CheT+GLOMAP evaluation).

• Minutes of meeting held 12/09/2013 over teleconf

Notes from UKCA aerosol teleconference held Thursday 12th September between Leeds & MO Present: Ken Carslaw, Graham Mann, Colin Johnson, Jane Mulcahy, Mohit Dalvi.

From a general discussion over plots circulated in advance of the meeting the following points were discussed in detail

1) AOD in Southern Ocean seems too high in the v8.4 CheT+GLOMAP run (amtjn) -- in fact in all v8.4 runs (and v8.2 too?). 2 suggested causes here -- too-high sea-salt and/or too-high DMS. The AOD was higher in the SH summer (Dec) than in SH winter (Jun) which suggested the dominant cause could be DMS. 2) DMS: Comparing to the Dumont D’Urville (DD) DMS seasonal cycle showed the DMS was much too high during SH summer whereas we generally seem to get the DMS about right at DD in TOMCAT-GLOMAP (e.g. see Mann et al., 2010 and Spracklen et al. 2005).

Too high DMS could be caused by the use of the Wanninkhof sea-air transfer relation which tends to give a high DMS flux and may have been used for historical reasons with the CLASSIC scheme (which has lower chemical sink).

Could try switching from Wanninkhof to Liss & Merlivat which is also available in the UMUI as a different option. I recall us making this change in the v7.3 CheT+GLOMAP runs to address a Southern Ocean SO2/DMS/SO4 high-bias 3) sea-spray: The annual-mean comparison to the U. Miami surface NaCl observations suggests the S. Ocean sea-salt is a bit high – and for some reason there was very little drop-off of the sea-salt from the ocean to the Antarctic continen (particularly in SH winter, June). This was also the case in the Arctic winter. There seems in general to be very little removal of aerosol during the cold periods of the year. Could this be related to the way the scavenging shuts off if the cloud liquid fraction is zero? But then it was mentioned that actually this was a key control for transport to the Arctic for example. An alterative issue could be to do with Possible issue with dry deposition or sedimentation that is leading to too high sea-salt over the continents. Roughness issues? 4) Arctic AOD too high too? Could be similar issue to the high sea-salt over the – need to check use of roughness length and sub-optimal implementation of aerosol dry deposition (still using old approach inherited from TOMCAT-GLOMAP whereby roughness length is used to infer land-surface types for coefficients in aerosol-drydep routine). Or could there be an issue with the BL mixing taking too much aerosol out of the layer that is subject to dry-dep? We do the emissions simultaneously with the BL mixing but then the ddep is after that (could this lead to too inefficient dry-dep?) 5) Wildfire signal in AOD seems very low Don’t see any indications of biomass-burning /wildfire in AOD maps – for example SON AOD seems very low in biomass burning regions – could there be an issue with the emissions ancils being used? Suggest to look into this with biomass burning on/off experiments (could Ben Johnson look into this?). 6) General Arctic seasonal cycle against observations for v8.4 GA4 and GA5 CheST+GLOMAP runs. One thing seen in the GA5 CheST+GLOMAP run is that there seems to be more transport to the Arctic in GA5 than GA4. Jo Browse had done analysis of these 2 v8.4 runs against several sets of Arctic aerosol observations from her PhD. There was very encouraging comparisons of SO2 seasonal cycle against the observations at several sites and also the BC seasonal cycle looked good with both runs a little low against the measurements – but much improved compared to v7.3 runs. There was a general question-mark about the rather high AOD seen in the model during the winter and spring – this was in agreement with the seasonal cycle in the measurements but the AOD seemed too high at 0.3-ish? Perhaps could be related to an issue with the sea-salt transport being much too efficient over the continents in cold seasons? (as discussed in items 3 and 4).