Global Impacts of Ho 2 Loss on Cloud and Aerosol Printer-friendly Version Interactive Discussion Atmospheric Chemistry and Physics Modeling Global Impacts of Heterogeneous Loss of Ho 2 on Cloud Droplets, Ice Particles and Aerosols Global Impacts of Ho 2 Loss on Cloud and Aerosol Printer-friendly Ver

Discussions This discussion paper is/has been under review for the journal Atmospheric Chemistry and Physics (ACP). Please refer to the corresponding final paper in ACP if available. Abstract The abundance and spatial variability of the hydroperoxyl radical (HO 2) in the tropo-sphere strongly affects atmospheric composition through tropospheric ozone production and associated HO x chemistry. One of the largest uncertainties in the chemical HO 2 budget is its heterogeneous loss on the surface of cloud droplets, ice particles 5 and aerosols. We quantify the importance of the heterogeneous HO 2 loss at global scale using the latest recommendations on the scavenging efficiency on various surfaces. For this we included the simultaneous loss on cloud droplets and ice particles as well as aerosol in the Composition-Integrated Forecast System (C-IFS). We show that cloud surface area density (SAD) is typically an order of magnitude larger than 10 aerosol SAD, using assimilated satellite retrievals to constrain both meteorology and global aerosol distributions. Depending on the assumed uptake coefficients, loss on liquid water droplets and ice particles accounts for ∼ 53–70 % of the total heterogeneous loss of HO 2 , due to the ubiquitous presence of cloud droplets. This indicates that HO 2 uptake on cloud should be included in chemistry transport models that already include 15 uptake on aerosol. Our simulations suggest that the zonal mean mixing ratios of HO 2 are reduced by ∼ 25 % in the tropics and up to ∼ 50 % elsewhere. The subsequent decrease in oxidative capacity leads to a global increase of the tropospheric carbon monoxide (CO) burden of up to 7 %, and an increase in the ozone tropospheric lifetime of ∼ 6 %. This increase results in an improvement in the global distribution when 20 compared against CO surface observations over the Northern Hemisphere, although it does not fully resolve the wintertime bias in the C-IFS. There is a simultaneous increase in the high bias in C-IFS for tropospheric CO over the Southern Hemisphere, which constrains on the assumptions regarding HO 2 uptake on a global scale. We show that enhanced HO 2 uptake on aerosol types associated with anthropogenic sources 25 could contribute to reductions in the low bias for CO simulated over the extra-tropical Northern Hemisphere.

We thank Dr. J.-F. Müller for his critical and insightful comments related to our manuscript. This critique has motivated us to examine our modeling assumptions once again. Upon inspection of the changes in the HO 2 mixing ratios due to heterogeneous loss on cloud droplets we feel that , indeed, our assumption of instantaneous mixing within a grid-cell, still leads to an over-estimation of the effect of cloud-uptake for grid-cell average values for instances when cloud cover (CC) is low. An improved assessment, as detailed below, suggests that in our current manuscript we over estimate the cloud effect on HO 2 loss on a global scale by approximately a factor 2-3. When accounting for this, a first estimate suggests that combined liquid and ice cloud uptake is no longer the dominating sink for HO 2 when compared to other sink terms on aerosol, but rather of equal magnitude. A scaling factor to the loss of HO 2 on cloud, which accounts for sub-grid scale (SGS) mixing, can subsequently be introduced to provide an updated evaluation of HO 2 uptake on all respective reactive surfaces.
For this purpose, we have reconstructed a HO 2 concentration field that is a best estimate of the assumption of no-mixing within a grid-cell. This assumption is likely more realistic, considering the relatively large mixing time scale between cloudy and non-cloudy air. The reconstruction is done by combining two independent 1-day simulations with the C-IFS, where HO 2 concentrations are modeled to be representative either for within the cloud ( , the reaction rate representative for in-cloud HO 2 loss is applied without any scaling to the whole grid-cell. Next, the two resulting instantaneous HO 2 concentration fields are scaled with CC to obtain a grid-cell average concentration, according to Eq. 1, which could serve as a best-estimate of grid cell average HO 2 in the situation of no-mixing. This reconstructed HO 2 field contrasts to the 'instantaneous mixing' (IM) approach as followed in our present manuscript, where reaction rates are scaled with cloud fraction, rather than the resulting HO 2 concentrations. Resulting mean HO 2 mixing ratios of the various approaches for a single day (1 April 2008), given as a function of CC, are presented in Fig 1. For practical reasons a no-mixing approach (NM) i.e. where calculations are performed both for in-cloud and cloud-free chemistry separately within each grid cell, is difficult to achieve in a CTM.
With the availability of the reconstructed HO 2 field we can now quantify the estimated error of the IM approach with respect to NM. The net loss of HO 2 at 800hPa for IM is over-estimated by, on average, ~35% for grid cells with CC>0.01. This lower limit for CC is also applied in the C-IFS for activating cloud chemistry and scattering in the photolysis routine. When averaging over all grid cells containing the full range of CC values, the difference between the two methods is ~17%. Fig. 1 illustrates that the mean HO 2 mixing ratios in IM still follows the same general shape as the reconstructed HO 2 , representative for NM. For high CC (> 0.8) the resident HO 2 mixing ratios become fairly similar between approaches, where both are more reasonable than when assuming no HO 2 uptake on cloud at all. However, the frequency distribution of CC at global scale (not shown) reveals a large contribution with CC < 0.1. Specifically, over ~50% of all grid cells with CC>0.01 have a value below 0.1. In summary, the change in HO 2 at low CC contributes most to the global average, which is where the IM approach makes the largest error.
Also when comparing the average decrease in daily mean HO 2 mixing ratios with respect to the case of no HO 2 cloud uptake at all, this is ~21% at 800hPa for IM (as applied in run AER-CLD in our manuscript) and ~7% for NM. This again is largely defined by the contributions of grid-cells with low CC, which occur mostly in the tropics. Note also that here photochemical activity, and thus HO 2 mixing ratios, are higher than in the extra-tropics. Figure 2 shows a zonal mean in the HO 2 loss between approaches, where one can see that the reconstructed HO 2 field shows a relatively small percentual decrease versus the situation without HO 2 uptake on cloud.
In the final version of our manuscript we propose to improve our methodology to account for SGS effects in the model by introducing an empirical SGS scaling factor to the HO 2 loss rate on cloud, as function of CC. This scaling factor can easily be tuned to match the simulated HO 2 concentrations towards the reconstructed field, representative for NM, hence providing an empirical parameterization to account for SGS effects.
An additional question posed by J.-F. Muller concerned our effective liquid and ice cloud radii in C-IFS, which were computed as function of the LWP and IWP, respectively. Additional to the description in the paper we should have mentioned that we limit , to between 4 and 10 to avoid erroneous values. For ice radii, we do not limit the particle radius. Examining the distribution in C-IFS reveals we obtain values up to 45 maximum. Figure 3 of this response shows a zonal mean for both cloud and ice to allay fears of very large values in our study. Figure 1. The average HO 2 volume mixing ratios at 800hPa for 1 April 2008 in C-IFS as sampled from instantaneous 6-hourly fields, and binned for cloud fraction ranges of 0.1. Colour key: Black: No HO 2 uptake in cloud is assumed ( 2, ), orange: HO 2 uptake reaction scales with cloud fraction ('instantaneous mixing'), blue: HO 2 uptake reaction is not scaled, i.e. modeled HO 2 is representative for the value within the cloudy fraction ( 2, ) and red: Reconstructed HO 2 according to eq. 1, as best estimate for a 'no-mixing' assumption. Figure 2. The zonal mean percentual decrease of HO 2 at 800hPa for 1 April 2008, on the assumption of instantaneous mixing (black) and no mixing (blue), as compared to the reference situation without any cloud HO 2 uptake.