Simulation of organic aerosol, its precursors and related oxidants in the Landes pine forest in south-western France: Need to account for domain specific land-use and physical conditions
Abstract. Organic aerosol (OA) still remains one of the most difficult components of the atmospheric aerosols to simulate, given the multitude of its precursors, the uncertainty of its formation pathways and the lack of measurements of its detailed composition. The LANDEX project (The LANDes Experiment), during its intensive field campaign in summer 2017, gives us not only the opportunity to compare biogenic secondary OA (BSOA), but also its precursors and oxidants obtained within and above the Landes forest canopy, to simulations performed with CHIMERE, a state of the art regional Chemistry-Transport Model. The Landes forest is situated in the south-western part of France, and is one of the largest anthropized forests in Europe (1 million ha), composed by a majority of maritime pine trees, strong terpenoid emitters, providing a large potential for biogenic SOA formation. In order to simulate OA build-up in this area, a specific model configuration set-up, adapted to the local peculiarities was necessary. As the forest is inhomogeneous, with interstitial agricultural fields, high-resolution 1 km simulations over the forest area were performed. BVOC emissions were predicted by MEGAN, but specific land cover information needed to be used, chosen from the comparison of several high-resolution land cover databases. Moreover, the tree species distribution needed to be updated for the specific conditions of the Landes forest. In order to understand the canopy effect in the forest, canopy effects on vertical diffusivity, winds and radiation were implemented in the model in a simplified way. The refined simulations show a redistribution of BVOCs with a decrease in isoprene and an increase in terpenoid emissions with respect to the standard case, in line with observations. Corresponding changes on simulated BSOA sources are tracked. Very low night-time ozone, sometimes near zero, remains overestimated in all simulations. This has implications to the night-time oxidant budget, including NO3. Despite careful treatment of physical conditions, simulated BSOA is overestimated in the most refined simulation. Simulations are also compared to air quality sites surrounding the Landes forest, reporting a more realistic simulation in these stations in the most refined test case. Finally the importance of the see breeze system which also impacts species concentrations inside the forest is made evident.
Arineh Cholakian et al.
Arineh Cholakian et al.
Arineh Cholakian et al.
Viewed (geographical distribution)
Review of the manuscript “Simulation of organic aerosol, its precursors and related oxidants in the Landes pine forest in south-western France: Need to account for domain specific land-use and physical conditions”
This is overall a well-designed and comprehensive modelling study about the secondary organic aerosol formation over the Landes pine forest in south-western France. I think the manuscript can be well worth to be published in atmospheric chemistry and physics after some relatively minor but essential revision.
Since this is a model study about biogenic secondary organic aerosol (SOA) formation I think it is essential to describe the terpene gas-phase chemistry and SOA formation scheme used in the CHIMERE model in some detail. You state that you use the SAPRC-07A chemical scheme and provide several references to the used SOA scheme but I also think that it is appropriate to at least briefly describe the SOA formation scheme in the presented manuscript. Since the modelled BSOA formation is overestimated while the modelled terpene concentrations seem to be underestimated, at least in the surface layer at the measurement station in the Landes forest they most straight forward explanation that come to my mind would be too high BSOA yields in the model.
In figure 7 you present the modelled and observed monoterpene concentrations over the Landes forest and in Fig. 4 you show maps with total monoterpenes and isoprene emissions. However, I miss information about the actual “base case” and “final” BVOC species emission factors (e.g. the individual monoterpene species emission factors implemented into CHIMERE) for the maritime pine forest and how the BVOC emissions were handled over the non-forested patches (e.g. meadows an various fields with crops). Could you provide some statistics on this? On L323-326 you write:
“Therefore, modifications were made in the MEGANv3 model inputs in order to i) create a specific ecosystem for the Landes forest containing 95% of maritime pines and ii) modify the emission factors for the maritime pine to experimental values obtained during the Landex campaign (E. Ormeno, personal communication). The resulting emission factors were integrated into CHIMERE and a sensitivity case was run with this new configuration. The resulting emissions of isoprene and the sum of terpenoids (mono and sesqui terpenes) using these new emission factors are shown in Figure 4 for both sensitivity tests detailed in this section.”
Several of the figures have very small legends and units (e.g. Fig. 4, 13) or are smaller than they need to be (e.g. Fig. 5 and 11). Please consider if the font size could be increased. I also think that all subpanels should be named a, b, c, d … and all panel should be described briefly in the figure text. E.g. Panel a shows …
At several places in the manuscript you include links to datasets directly in the text but no references which describes the datasets. I think the webpage links should be complemented with references to publications where these datasets are described, if existing. E.g. I am sure that you can find some publication(report) that describes the EMEP anthropogenic emission inventories (L222) and Copernicus database on tree cover density (L276).
In table 2 you provide some statistics about how well the different model tests results agree with the observations in the Landes forest. I don’t think you specify what you mean with R. Is it the correlation coefficient?, What do you mean with “bias”, is it just the absolute difference between the arithmetic mean concentrations from the model and observations? I think you need to perform a bit more rigorous statistical analysis of how the different model versions performs. At least I think it is also important to report the Normalized Mean Bias (NMB) and the fraction of predictions within a factor of two of observations (FAC2). Then I also think you should refer back to table 2 in the text when you describe the results from the different model results in Sect. 4 and 5.
It is no problem for me to understand the general method or results of the present study, but I think the present work would benefit from a through English grammar check prior to the final publication. I have specified a few sentences which I think need to be revised below.
L29, “Henrys constant”. I think it should be Henry’s law constants.
L80 “improve (or not) agreement”. Consider to rephrase. E.g. influence the agreement
L103-105 “Regarding biogenic emissions, the site is quite homogeneous since the majority of the trees in the surrounding area are maritime pines. However, as mentioned above, since the forest is parcellated the density of the forest and the geographical distribution of these emissions have a certain degree of heterogeneity.”
Looking at satellite images over the area the landscape seem to be very patchy and not at all an homogeneous forest. Do you mean that almost all patches have planted maritime pines but of variable age? Please explain what you mean with that the site is “quite homogeneous” but at the same time “parcellated”.
L125-L126 “Since the purpose of the article is to focus on pollutant build-up from biogenic compounds” Consider if you should replace “pollutant build-up” with secondary organic aerosol formation. Too me it sounds a bit strange to call BSOA formation “pollutant build-up” even if I agree that it in potentially could be a health concern.
L160-163 “The parameters common to all sensitivity tests are explained in section 3.1. After
the description of the base simulation (section 3.2), sections 3.3 through 3.6 describe sensitivity tests changing inputs/variable calculations. Keep in mind that each modification is added on top of the previous ones. The implemented improvements are indicated in Table 1.”
I think the manuscript is well structured, but can you always state that the implemented changes are “improvements”? From the statistics presented in table 2, this is not always clear. I would replace “improvements” with changes or possibly “increasing complexity/resolution”
L279-281 “These changes are, on the average, about +30% forest growth in cells where the changes were positive and about -40% in cells where the changes were negative. Given these important changes, only the most recent (2015) data was considered.”
I don’t understand what you mean with these sentences. What do you mean with “cells”?
Figure 3. I miss a legend about the absolute tree density “shades of green” in the upper left figure.
L306 “only slightly lower densities”. Replace with some numbers about how much lower the densities actually are.
L317-318 “It was considered necessary to modify the tree type distribution in the MEGAN model since MEGAN 2.1 assumes 28% of maritime pine coverage for the Landes forest, while the BDForet database shows around 95% of the same species.”
Is it actually the MEGAN 2.1 model which assumes 28% of maritime pine coverage that is the problem in this case or just how the MEGAN 2.1 model has been setup? I guess the 28% of maritime pine coverage is not hard coded into the MEGAN 2.1 source code but provided as some input variables to the model or? In this case it should be enough to change how MEGAN v2.1 is setup and not the version of MEGAN, v2.1 or v3 or?
L326-328 “Figure 4 justifies the necessity of this modification, since as mentioned above the emissions of isoprene have dropped by about half, while the emissions of terpenoids have increased significantly, by about a factor of 2”
I don’t think you can state that this by just looking at the results in Figure 4. Possibly you can draw this conclusion when comparing the modelled monoterpene and isoprene concentrations with the observations from the station.
L360-362 “While the effect of the canopy on BVOC emissions is already taken into account in the MEGAN scheme, the canopy effects on wind speed and vertical diffusion inside the forest are not simulated in the CHIMERE model. This normally causes no issues, since most measurements are performed above canopy level”
Just because you normally do not compare the modelled concentrations inside the forest canopy with observations from inside the canopy this can still be a general model issue. E.g. when it comes to the modelled deposition of BVOC oxidation products in the forest canopy. Please consider reformulating this statement.
L399. In eq. 7 the altitude is denoted by λ but in eq. 1,2 and 6 you use z to denote altitude.
L401-402. “k is an extinction factor, which was measured to be 0.33 for the Landes forest (calculated by Ogee et al. (2003) using experimental data).”
Considering how patchy the Landes forest is, can you assume that the same empirical k parameter is valid over the whole Landes forest?
L437-440 “The sum of monoterpenes is the first group of species that we consider here from PTR-Tof-MS measurements (m/z 137 peak as their main fragment and m/z 81 values as their second most important one) and sum of simulated α-pinene, β-pinene, limonene, ocimene concentrations, see section 3.4) since these are main drivers of atmospheric chemistry for the given terpene emitting maritime pine forest.”
This paragraph needs to be clarified. I am not an expert on PTR-Tof-MS measurements so I cannot follow how the measured total monoterpene concentrations actually were derived from the mass spectrum. C10H16 should be closer to mass 136, but I guess you add a proton H+ for the detection (thus m/z 137) or? But then what about m/z 81? Why is this mass peak also counted as monoterpenes?
L441 “BVOCs has”, BVOCs have …
L469. Consider replacing “heavily” with substantially.
L473-476 “It is also important to keep in mind that in the canopy test case, the changes seen in the concentrations arise almost entirely from the Kz modifications and not from the swrd ones (as they do not affect the emission of BVOCs since their effect is already taken into account in the emission factor calculations).“
Consider to reformulate this sentence. What do you mean with “swrd ones” ?
L486. What do you mean with “mean bias” and “0.43 for correlation”. Is it the normalized mean bias (NMB) and correlation coefficient?
L500-502 “While simulations with the standard model version correctly reproduce the day-time O3 maxima, the significantly overestimate the O3 minima, which never decrease below to 20 ppb.” Correct this sentence
L503-505 “Daily O3 maxima are only slightly changed in sensitivity tests, the largest impact being noted for the test with refined emissions, leading to enhanced NOx emissions and an increase in the O3 peak on July 7 from about 60 to about 70 ppb (Figure 6).”
Too me it looks like the major change in the maximum O3 concentrations occur already between simulation 1 (base case) and simulation 2 (changed meteorology).
L521 “seem well-simulated” replace with some statistical measure of the model performance.
L524 “dry deposition speed”. Maybe this is OK but I think you normally call it dry deposition velocity and not dry deposition speed.
L524 “Another plausible candidate for it might be an underestimation of deposition of O3 over forested areas.” Maybe reformulate this sentence.
L540 “maxima (up to 0.1 ppt) in contrast due to our simulated night-time maxima.” I don’t understand this sentence.
L548 “diminution of OH” Do you mean decreasing OH concentration?
L553 and figure 8. I suggest that you also report the modelled OH concentration in unit molecules/cm^3 and not in ppb.
L568-571 “It is concluded from the model results and known OA sources, that most of it is biogenic SOA (84% at Salles-Bilos over the campaign period). For instance, the concentration of anthropogenic OA (primary and secondary is quite low at this site (average of 0.11 μg.m−3 overall test cases).”
How can you know the OA sources and that 84 % is biogenic SOA? Are these numbers coming from model results or observations? Please explain in more detail and provide some reference. I would suggest that you actually perform a model sensitivity run without BVOC emissions over the Landes forest in order to quantify the modelled BSOA originating from the local BVOC emissions.
L576-580 “Changing the Kz parameterization and in canopy radiation decrease both OA and BSOA, as the concentrations of the three oxidants (OH, NO3 and O3) participate in the biogenic OA formation process, while increasing the concentrations of terpenes (its major precursors). This test reveals an interesting feature: for BSOA formation around Salles-Bilos, changes in oxidation rates have a larger effect than changes in precursors (i.e. changes in terpenoids and isoprene).”
Do you mean changes in precursor emissions or changes in precursor concentrations in the canopy? It is obviously the case that decreasing mixing and decreasing in canopy radiation will result in higher surface (in canopy) terpene concentrations and decreasing biogenic OA formation since the integrated total boundary layer oxidation rates of the terpenes will decrease and thus also the biogenic SOA formation in the whole boundary layer. If less terpenes are oxidized in the boundary layer you get less biogenic SOA.
Figure 10. Can you give some explanation to the higher O3 concentrations over the ocean compared to the ozone concentrations over the Landes forest?
L596 “outstanding in this group” Consider reformulating this sentence.
L608 “Differences with respect to our study can be explained by many factors which we do not attempt to quantify here:”
What other study do you compare with? What is the difference between?
Section 5.2. I think the case studies focusing on the see breeze fronts are very interesting but the onset of the OA peak and the rapid drop in OA seem to be a bit shifted between the model and observations, at least with a few hours. In the observations the OA mass seem to drop before noon while in the model it occurs later in the afternoon. Can you provide some potential explanation to this? Could it also have to do with the planetary boundary layer height (PBLH)? Maybe the model underestimates the PBLH?
L721-723 “The scenario representing physical changes to the simulations of vertical diffusivity, wind speed and radiation penetration inside the canopy seems to have a more realistic view of what the measurements show about the atmospheric chemistry inside the forest, especially for terpenes, radicals and BSOA formation.” I don’t understand this sentence completely. Please consider to reformulate it. What do you mean with “realistic view”
L724-727 “Using the final test case, we also simulated the formation of BSOA from different precursors and oxidants. This showed us that the chemical pathways behind the BSOA formation for the base case scenario is inaccurate in the base case since the majority of BSOA is formed through the oxidation of isoprene, while this is changed to terpenes (specifically sesquiterpenes and β-pinene) in the final simulations.”
Is it the chemical pathways behind the BSOA or the BVOC emissions which are inaccurate in the base case simulations? Can you make any definite statement about the accuracy of the chemical pathways of the BVOCs and the SOA formation in this work?