Minutes of the aerosol working group meeting 12/12/2013

From UKCA

Notes of UKCA aerosol teleconference held Friday 12th December 2013 between Leeds & MO

Present: Graham Mann, Jane Mulcahy, Ben Johnson, Colin Johnson, Mohit Dalvi, Steve Rumbold, Andy Jones.

Notes from teleconf 12th December 2013 (Graham Mann)

Summary of v8.4 results is that for aerosol microphysical properties, the model skill against observations is good (green light given). However, for aerosol optical depth, comparisons to AERONET and satellite observations suggest two important discrepancies:

  • In tropical South America and Central Africa, AOD too low against observations and there is a general AOD low-bias in the tropics.
  • Despite improved model skill with repair of doubled-sedimentation merge-glitch, AOD still too high remotely whereas v7.3 in much better agreement.

Tropical AOD low bias:
At previous teleconference, resolved to try several sensitivity tests to understand which process may be responsible for the tropical AOD low-bias:

  • weaken convective scavenging -- it had been argued the spatial pattern may be consistent with too much convective scavenging.
  • increase biomass burning emissions and also secondary organic aerosol production -- low-biased AOD regions strongly affected by both of these.
  • ageing rate -- Colin Johnson had discovered a possible bug with the ageing which may lead to too long ageing rate.

Sensitivity test for weakened convective scavenging
Jane Mulcahy had tried this sensitivity run (see results below) but found only slight increase in BC/OC burden and tropical AOD. Also she found that when convective scavening was weakened, this increased the sulphate and sea-salt AOD, worsening the remote AOD high-bias. It was agreed then at the meeting that the convective scavenging was unlikely to be the cause of the tropical AOD low-bias.

Sensitivity test for increased biomass burning (bb) emissions and increased SOA
Ben Johnson had tried sensitivity test where bb emissions were doubled and the molar yield to SOA (from monoterpene oxidation) was doubled. Results from these tests (see below) showed that when both these sources were doubled, the mass loading of BC and OC double and so did the AOD. The results indicate that biomass burning and SOA production influence tropical AOD much more than convective scavenging. However, despite doubling the AOD, there was still an AOD low bias against the observations. Nevertheless the tests indicated a further increase could achieve the goal of achieving good agreement with aerosol optical properties in tropical regions. It was also noted that there was good scientific justifications for increasing both these sources of aerosol mass. The biomass burning emissions being used in the GA4 job were for the year 2000 which was known to be a weak biomass burning year in tropical regions. For secondary organic aerosol it was known that the current 13% molar yield from monoterpenes was too low (e.g. see 2013 ACP paper by Cat Scott, Leeds). Model runs with GLOMAP-mode in the CTM had already tended to increase the 13% to 26% as this gave better agreements with surface aerosol measurements. Additionally, the model currently has no SOA coming from isoprene which would be a substantial source in these tropical regions. It was resolved to carry out a test where biomass burning and SOA were increased by another factor 2. This could be considered to be used as a temporary fix to ensure GLOMAP in GA6/7 agreed well with observations. This would allow the model to proceed while further tests identified the cause of the discrepancy. The issue of the size assumed for biomass burning emissions of BC/OC was noted as potentially being important for its scattering efficiency. This size was found to be important parameter for CCN in biomass regions in perturbed parameter ensemble simulations with TOMCAT-GLOMAPmode (e.g. Lee et al. 2013 ACP). Leeds were planning to carry out a comprehensive perturbed parameter ensemble with HadGEM-UKCA at v8.4 in Jan/Feb 2014 and this would help isolate the cause.

Sensitivity test for weakened ageing rate
Colin Johnson and Andy Jones had tried some sensitivity tests to decrease the ageing rate of BC/OC as CJ had found possible bug in ageing routine which would lead to lengthened ageing timescale. Results can be shown below, with the weaker ageing run leading to increased BC/OC in the Aitken-insoluble mode but only slightly increased BC/OC in the accumulation mode. The impact on AOD was thus small. This parameter was therefore unlikely to be the cause of the low tropical AOD.

Remote AOD high bias:
At previous teleconference, had suggested to try sensitivity test with sedimentation increased by a factor of 2. However, Steph Woodward and Alastair Sellar (AS) had discovered, with sedimentation being applied on the 1-hour UKCA timestep, a CFL limitation for the fall speed was being activated in the lowest few levels where gridboxes are quite thin. AS had tried sensitivity tests where the aerosol dry-deposition and sedimentation routine was called on a shorter timestep and found this had a big effect on simulated coarse mode components (modal dust and sea-salt). Although the impact of modal dust has been demonstrated (see below) the impact on AOD had not yet been assessed but was suggested to be substantial. This may therefore lead to reduced AOD in the Southern Ocean, one of the key regions where the AOD high-bias had been found. Colin Johnson had also carried out a test where he used a different sea-spray emissions parameterization which gave substantially less AOD.

Summary:

In summary the tests suggested increase biomass burning emissions and molar yield to SOA was an effective way to address the tropical AOD low-bias. And using a different sea-spray emissions parameterization (together with the shorter timestep for sedimentation) should fix the remote AOD high-bias.

The team wanted to stress however that these should be considered as temporary fixes and more work was needed to find the actual cause of both of these AOD biases, neither of which were apparent at v7.3.

Relevant Files for Discussion

Latest GA4 sensitivity results:
Scavenging coefficient sensitivity The scavenging coefficients for accumulation and coarse modes in convective regimes were reduced from 1.0 --> 0.5

  • Pdficon small.png Impact on AOD Info circle.png
    - does not have a large impact on tropical AOD, with detrimental increased to AOD in high latitudes
  • Pdficon small.png Impact on Burdens Info circle.png
    - small increases in BC/OC burdens but increases in accumulation and coarse mode sea salt and sulphate are most noticeable
  • Pdficon small.png Budgets Expt Info circle.png Pdficon small.png Budgets Control Info circle.png - reflects findings above, v. small impact on BC burdens/lifetimes while you see a large increase in the seasalt and sulphate burdens and lifetimes. A large increase in the nucleation mode Sulphate loss timescales??


Doubling BB emissions
In this experiment BB emissions and SOA yield from monoterpene were both doubled, as an attempt to increase AOD in the tropics.

  • Mass loading of BC and OC has increased by almost a factor of 2.
  • The aitken insoluble mass pretty much doubles
  • The accumulation soluble mode is pumped up a little, particularly over tropical continents.
  • AOD has moved in the right direction -now ~ 0.05 higher over tropical continents. I think we still need a further 0.05 - 0.1 increase to match observations.
  • The AOD also increases by a similar amount (~0.05) in some N. hemisphere mid-latitude regions where there are monoterpene and BB emissions.
  • This is a little suprising as the emissions of BB and monoterpene are much smaller in these mid-latitude regions
  • AOD increase in the tropics is quite limited to the continental source regions - indicating that the lifetime of the aerosol in the tropics is still quite low.

Pdficon small.png burden_ajkzi_anmad.pdf Info circle.png Pdficon small.png aod_maps_ajkzi_anmad.pdf Info circle.png

Decreasing ageing rates
Attached are the annual-mean results from the test to decrease the ageing rate of carbonaceous aerosols. The Control is amtj-u, the Experiment is amtj-w.

Bottom line: carbonaceous burdens are

  • decreased slightly in the nucleation mode (OC -11%)
  • increased in both Aitken modes (BC[sol] +28%; OC[sol] +23%; BC[ins] & OC[ins] both more than +200%)
  • increased in the coarse mode (BC +60%; OC +18%)

....but sadly, the accumulation mode burdens for both BC and OC are decreased (both c. -4%), so there's little impact on AOD, and such changes as there are could easily be due to interannual variability.
Pdficon small.png AODs_amtjw_cf_amtju.pdf Info circle.png Pdficon small.png burdens_etc_amtjw_cf_amtju.pdf Info circle.png

Initial results from Offline Chemistry: Pdficon small.png Offline_chem_GA4.pdf Info circle.png

GA6 Performance:

Pdficon small.png Latest GA6 AODs Info circle.png Pdficon small.png Earlier GA6 AODs with GA4 for comparison Info circle.png - for information, sorry uses different colorscale!


Sedimentation tests (Alastair Sellar)

Pdficon small.png Tests reveal issue with CFL limitation of fall speed: sub-step sedi Info circle.png



Summary of discussion:

In summary the tests suggested increase biomass burning emissions and molar yield to SOA was an effective way to address the tropical AOD low-bias. And using a different sea-spray emissions parameterization (together with the shorter timestep for sedimentation) should fix the remote AOD high-bias.

The team wanted to stress however that these should be considered as temporary fixes and more work was needed to find the actual cause of both of these AOD biases, neither of which were apparent at v7.3.