Abstract

Climate change very likely has effects on vegetation so that trees grow faster due to carbon dioxide fertilization (a higher partial pressure increases the rate of reactions with Rubisco during photosynthesis) and that trees can be established in new territories in a warmer climate. This has far-reaching significance for the climate system mainly due to a number of feedback mechanisms still under debate. By simulating the vegetation using the Lund-Potsdam-Jena guess dynamic vegetation model, a territory in northern Russia is studied during three different climate protocols assuming a doubling of carbon dioxide levels compared to the year 1975. A back of the envelope calculation is made for the subsequent increased levels of emissions of monoterpenes from spruce and pine forests. The results show that the emissions of monoterpenes at the most northern latitudes were estimated to increase with over 500% for a four-degree centigrade increase protocol. The effect on aerosol and cloud formation is discussed and the cloud optical thickness is estimated to increase more than 2%.

1. Introduction

Climate change is a global problem facing humanity and at the same time it is of anthropogenic origin [1]. However, the mechanisms for global climate change are interlinked between all the components of planet Earth, including atmosphere, hydrosphere, and biosphere. Although climate change has begun to become more and more evident, there are still large uncertainties in future projections of the climate. In order to project the evolution of the climate change, we use general circulation models (GCMs), but the accuracy of the projection is dependent on both the input data and the compliance with real processes that affect the climate. One category of processes that is of particular importance concerns feedback mechanisms. A feedback mechanism is a chain of events that in the end has an impact on the initial step, either if it is an enhancing effect, so called positive feedback, or if it has an inhibiting effect, so called negative feedback.

Three examples of a positive feedback mechanism that result from an enhanced greenhouse effect with a subsequent increase in temperature are given by Mencuccini and Grace [2], Swann et al. [3], and Walter et al. [4] Mencuccini and Grace [2] discuss the leaf to sapwood area ratio in Scots pine as of a changed difference in water vapor pressure deficit in air. Swann et al. [3] propose one example of an enhanced positive feedback mechanism. It consists of two components: (1) broad-leaf deciduous trees assumed to invade the warming tundra and thereby alter the albedo to less reflective land by replacing snow covered land with darker forests and (2) increased release of water vapour through evapotranspiration compared to evergreen boreal needle forests contributing to the greenhouse effect. In a warming climate, the first component indicates that there is a change in type of vegetation with new establishment of broad-leaf deciduous trees in the tundra as well as a replacement of evergreen boreal needle forests. The latter has already been mentioned in component 2. The two components are interlinked through the change in vegetation type. Walter et al. [4] show the importance of emissions of methane from thaw lakes in Siberia through ebullition. The thawing permafrost along the lake margins accounted for the largest emissions, which was estimated to constitute an accelerating effect on climate warming.

One example of a negative feedback mechanism was proposed by Kulmala et al. [5], which links changes in growth of forests and emissions of volatile organic compounds (VOCs) with subsequent generation of aerosol particles and possible changes in the number of particles available for cloud droplet activation called cloud condensation nuclei (CCN) and cloud optical properties. Figure 1 shows a schematic representation.

Number 1 in Figure 1: global warming causally related to an increase of atmospheric carbon dioxide (CO2) and other so called greenhouse gases [1]. Number 2 in Figure 1: photosynthesis is driven by an increase of atmospheric carbon dioxide with a subsequent increase in the amount of chemical energy as biomass that primary producers create in a given length of time called the gross primary production (GPP) [6]. GPP increases when the boreal forest grows by both tree number and size. Number 3 in Figure 1: boreal forests contribute to the emissions of volatile organic compounds (VOCs). According to O’Dowd et al. [7] the oxidizing products of monoterpenes cis-pinonic and pinic acids are abundant VOCs in the organic fraction of newly formed ambient aerosol particles. Number 4 in Figure 1: an increased fraction of surface-active aerosol particles, which due to a decrease of surface tension [8] become efficient CCN, owed to an increase of VOC emissions. The surface-active properties of the VOCs oxidizing products have been discussed in several studies [912]. Even though Facchini et al. [9] may have oversimplified their estimates [1315], surface-active organic material remains to constitute a key component of atmospheric aerosols in cloud droplet activation [16]. A study by Dalirian et al. [17] indicates that the partitioning between water soluble and water insoluble matter in activating CCN is as important as the adsorption properties of the surface. Hings et al. [18] suggest that hydrophobic particles may be coated with slightly soluble material and thereby increase the activation process. The studies of both Dalirian et al. [17] and Hings et al. [18] therefore relate to the activation properties of the slightly water soluble oxidation products of monoterpenes. Number 6 in Figure 1: the increase in number of cloud droplets ultimately could alter cloud optical (microphysical) properties and form brighter clouds [19]. These brighter clouds are proposed to reflect the incoming solar radiation more effectively and thereby enhance surface cooling and by that provide a negative feedback mechanism.

In this study we extend the Kulmala et al. [5] study to also discuss possible implications for the albedo of low-level clouds in the high Arctic as of an increase of VOCs and forest growth in Siberia causally related to global warming due to increase of levels of atmospheric carbon dioxide. Figure 2 gives a schematic representation of the proposed negative feedback mechanisms under study. Main differences from the feedback mechanism proposed by Kulmala et al. [5] are marked with orange.

To estimate changes of GPP upon a doubling of CO2 levels relative to the year 1975, we used the Lund-Potsdam-Jena (LPJ-GUESS) dynamic vegetation model [21, 22] The simulation was started in 1975 and carried out for a period of 200 years over a terrestrial territory in the Siberian inland of northern Russia. During the time of simulation a linear increase in temperature of 2, 3, and 4 degrees centigrade, respectively, resulted after doubling of the atmospheric carbon dioxide level. From the estimates of GPP subsequent emissions of monoterpenes were calculated. At last, the simulations and levels of emissions were discussed in terms of cloud formation and possible climate effects.

2. Simulation Method to Estimate GPP

To simulate GPP the dynamic vegetation model LPJ-GUESS was used [20, 2229]. LPJ-GUESS can be viewed as a framework for ecosystem-modelling which has used some elements from the equilibrium terrestrial biosphere model (BIOME) family of models [29]. The LPJ-GUESS model shows an intermediate response and sensitivity to atmospheric CO2 concentrations and temperatures [27, 28]. Large-scale dynamics of terrestrial vegetation and interactions between the biosphere and atmosphere in terms of exchange of carbon dioxide and water vapour are represented in the model. Processes like photosynthesis and evapotranspiration are simulated daily while slower processes, like biomass growth, are implemented annually. The input data to the model consist of climate parameters, atmospheric CO2 concentrations, and a soil code. The climate parameters are monthly or daily values of mean diurnal air temperature, precipitation, and incoming shortwave radiation (sunshine). The soil code is used for parameters such as hydrology and thermal diffusivity of the soil. Climate forcing, that is mean monthly temperature, precipitation, and cloud fraction, was using the Climate Research Unit (CRU) TS 3.0 data set [26]. Atmospheric CO2 concentrations were taken from observations [20]. Each simulation was started with a 1000-year initialisation to establish vegetation and soil budgets for the equilibrated climate of year 1975. CRU climate data cycled repeatedly to force the model during the initialisation using CO2 concentrations of the year 1975 (330 ppm(v)). After the initialisation, a production simulation was applied. During the production simulation, the temperature increased monotonically with adding an offset of 2–4 degrees centigrade to the baseline climate dataset. The simulations were performed across a 0.5° × 0.5° (latitude–longitude) grid [25]. Vegetation in a grid cell is described in terms of the fractional coverage of populations of different plant functional types (PFTs). Each PFT is assigned bioclimatic limits, which determine whether it can grow and spread under the current condition during the simulation. We used LPJ-GUESS version 2.1, which includes the PFT set and modifications described in [24].

2.1. Simulated Area and Global Warming Protocols (1 in Figure 2)

The simulated area is located in Siberia in northern Russia. We simulate GPP for a specific geographical coordinate. This is done in a grid-box representing the selected area. Because of the spherical shape of Earth, the grid-box in the model is not rectangular, but rather a trapetzoid as a first simplification.

This study’s model domain is marked in Figure 3(b). This is a part of the permafrost region, wasteland with permanently frozen soil; see Figure 3(a). The total area simulated is 8 424 510 km2. This can be compared to the total area of Siberia, which is about 13.1 × 106 km2. The simulated area is thus about 64% of the total aerial extent of Siberia. The total simulated area that was forested after the initial equilibration of 1000 years was 6 115 464 km2. This value corresponds to the vegetation in the year 1975. Results by Esser et al. [23] show that the historical carbon storage in the boreal forests is located in the western part of Siberia from the years 1860 to 2002.

Three different emission protocols were used to model the response in forest growth. We assumed a doubling of the CO2 level compared to the year 1975 (330 ppm(v)) for all of the protocols after the simulated period of 200 years. The temperature increase corresponds to different levels of climate sensitivity, namely, 2, 3, or 4 degrees centigrade increase. The emission protocols are presented in Table 1. The level of CO2 increase could also be varied, so that higher levels than a doubling can be obtained. It is also possible to keep the CO2 at a constant level during temperature increase and thereby determine how much of the effect is attributed to the CO2 doubling. These types of assumptions will not be considered in this study but could be recommended for future studies.

3. Tree Growth Established in New Areas

In the LPJ-GUESS model larch forest is referred to as boreal needle summergreen (BNS) vegetation. In Siberia 46% of the forested area is made up by larch forest [30]. Moreover, spruce and pine trees are named boreal needle evergreen (BNE) vegetation in the model. The grass is not contributing significantly to the emission of monoterpenes and is therefore not considered in this study. In this study also broadleaved summer green trees are not considered as a source of monoterpene emissions, even though there are exceptions such as oak trees [31].

3.1. Simulations Indicating Melted Permafrost (2 in Figure 2)

To see whether any new vegetation would be established, as a first survey we carried out LPJ-GUESS simulations over the permafrost region using the three emission protocols listed in Table 1. Typical results are shown in Figure 4. The simulations resulted in the fact that new vegetation was established in the wasteland regions as well as in changes of the composition of the forest. We were encouraged by the results from this first survey that a more detailed study could be considered, and that the primary region of interest should be Siberia, located in northern Russia. This was motivated by that the landscape and vegetation were both highly affected by climate change in this region.

3.2. Simulations of Forest Growth (3 in Figure 2)

Based on the three emission protocols in Table 1 we calculated the relative growth of two types of forests, BNE and BNS, defined as the difference in biomass (kgC/m2) between the simulated year 1975 and after 200 years.

In Figures 611, the simulated BNE relative growth in kgC/m2 for northern Russia is listed for each of the protocols given in Table 1. Light shading represents the relative growth as defined; no colour indicates a moderate relative growth (0–3 kgC/m2); yellow colour indicates a substantial relative growth (>3 kgC/m2); green colour indicates a very high relative growth (>5 kgC/m2); and red colour indicates a relative decrease in growth (<0 kgC/m2). Dark shade marks biomass change in percentage; no colour indicates a moderate change in percent (0–200%); orange colour indicates a substantial change in percent (>200%); green colour indicates a very high change in percent (>500%); and red colour indicates a negative change in percent (<100%). The symbol “#” marks locations where there was a growth of forest on land with previously not established forestation. The relative growth in percent is given as the percentage of the biomass in the year 1975; that is, 100% after 200 years equals the biomass level in the year 1975. The background image is the box showed in Figure 3(a).

As can be seen from Figures 6, 8, and 10, the relative growth (%) of BNE forest is located in the northwest region of the investigated area, which is in northern Russia. Further, there is a belt of increased growth of BNE forest stretching from the northwest to southeast of the region, and this is at the same time, at least in the southeast part, and at an expense of BNS forest, which is decreasing in biomass. The BNE forest is decreasing in biomass in the southwest corner of the region. This is according to the model due to a change of vegetation type where conifer forest is growing at an expense of BNE forest. Similar trends are shown in all the three climate protocols: A, B, and C. The trend is however most visible for the emission protocols with the highest temperature increase (C in Table 1).

From Figures 7, 9, and 11 we can conclude that BNS forest is growing in the northeast region and decreasing in biomass in the southwest region as well as in the belt described above. All in all, the general trend is that the boreal forest of northern Russia is growing and moving northward as a result of an increase in temperature and atmospheric CO2. This result is supported by the findings of Xu et al. [32].

4. Emissions of Monoterpenes and Cloud Droplet Activation

4.1. Estimated Emissions of VOC and CCN (4 in Figure 2)

In order to perform a back of the envelope calculation of the increase in VOC emissions in the sub-Arctic region that would follow forestation of the tundra, VOC emission rates of spruce, pine, and larch were used. Petersson [33] estimated the fraction of needles to total tree weight to be 9.2% for spruce and 4.2% for pine. An average of these number fractions was used in the simulated BNS and BNE forests.

Lind [34] reported the carbon content of dry mass in spruce and pine to be close to 50%. These numbers are used in this study to convert the biomass carbon into weight of needles per km2. A typical value of 15 kgC/m2 would correspond to 30 kg dry mass per m2, which is  kg/km2. This can be compared with the numbers given by Albrektson [35]; for 194 450 trees per km2 with an average dry mass of 172 kg makes  kg/km2 dry mass.

Ruuskanen et al. [30] report emissions of monoterpenes at 30°C of 5.2–21  for Siberian Larch (Larix sibirica). Emissions from Scots pine (Pinus sylvestris) and Norwegian spruce (Picea abies) of monoterpenes at 20°C were reported by Janson [36] to be and , respectively. Similar levels of emission are reported for other members of the Pinaceae family; see Table 2 [31].

However, emissions of monoterpenes are temperature dependent. To account for this we used the following equation given by Kesselmeier and Staudt [31]:Here, is the temperature difference between ambient temperature and 20°C. The factor , which is multiplied with the emission rate at 20°C, varies as a function of temperature.

The factor for calculating the temperature dependency of monoterpene emissions was estimated from the average monthly temperature at longitudes E47.25, E92.25, and E152.25, respectively. The temperature increase according to the protocols A, B, and C (the increase in temperature due to climate change and thereby the deviation from 20°C) is compensated by an increase of the factor proportionally by 1.20, 1.32, and 1.45, respectively. A typical temperature profile is shown in Figure 5.

The number of days is divided into bins of temperature. Each bin corresponds to a factor () for which the emission rates are corrected, shown in Table 3.

The number of days accounted for in each temperature bin is based on one month (30 days) for which the temperature is within the interval. On average, the model and the temperature profile are in agreement.

All days when the temperature exceeds 0°C have been assumed to emit monoterpenes. The emission rates by Ruuskanen et al. [30] and Janson [36] given above were used in the calculations. To compensate for lower temperatures in the area the numbers of days were additionally corrected with a factor and done in the same way as for the increased growth listed in Figures 611. The resulting estimates of the annual increase of monoterpenes for the three emission protocols are as follows:for A: 14.5 Tg y−1,for B: 16.4 Tg y−1,for C: 17.0 Tg y−1.

As a comparison, Laothawornkitkul et al. [37] estimated the annual global emission of monoterpenes to range from 33 to 480 TgC y−1. Corresponding to the year 1975 (330 ppm(v) CO2) the annual total emission of monoterpenes, in the area under study, is calculated to be 26.7 Tg. The relative increase is 55%, 62%, and 64% for the protocols A, B, and C, respectively. These results were compared well with the estimates by Xu et al. [32] that report a relative greening of the boreal region from 34 to 41%. Most striking is the estimated emission of monoterpenes at the most northern latitude investigated (N69.25), which increased from 0.54 Tg y−1 with 258% for A protocol, 355% for B protocol, and 496% for C protocol, respectively.

4.2. CCN Production (5) and Cloud Droplet Activation (6) in Figure 2

The number of CCN available for cloud droplet activation is hard to estimate because the process of condensational growth of the newly formed particles involving monoterpenes is not fully understood [38]. Given the complexity, Spracklen et al. [39] estimated that a factor of 5 increase in monoterpenes yields up to a factor of 3 in the enhancement of CCN from new particle formation. Again, at present the production of CCN from monoterpenes can only roughly be estimated as the processes involved are not fully understood.

In the boreal forest Kerminen et al. [40] observed nucleating particles, on average <150 cm−3, followed by subsequent rapid growth. This produced a large number of particles in sizes larger than 50 nm in diameter. 12 hours later, a cloud entered the region and >60% of the particles in the size range 50–100 nm diameter had been activated into cloud droplets. 67% of the cloud droplets were estimated to originate from particles below ca 15 nm in diameter.

5. Discussion of the Negative Feedback Mechanisms Postulated in Figure 2

5.1. The Feedback Mechanism

Kulmala et al. [5] proposed that a global fertilization of atmospheric CO2 with subsequent temperature increase would favour photosynthesis and the growth of Scots pine trees. The increased biomass would emit larger amounts of VOCs adding to particle formation and growth into CCN-sizes. A higher aerosol load would contribute to net cooling of the land surface both due to reflection and scattering and to the indirect effect of aerosol particles [19]. There are however four issues regarding the feedback mechanism proposed that we believe they are worth discussing in view of the results obtained in this study.

Firstly, the increase in average temperature and the increase of CO2 in the atmosphere as a result of climate change not only contribute to an increase of numbers of trees in the forests, as stated by Kulmala et al. [5], but it would also be possible for new forests to establish in former wastelands such as the subpolar permafrost region in the north of Russia, Scandinavia, Greenland, Canada, and Alaska. Further, if the permafrost will melt in the region, vegetation can start to grow. This effect would also be amplifying the effect described by Kulmala et al. [5], since more forests would emit more VOCs contributing to the population of aerosol particles.

Secondly, the effect on surface tension and thereby the lowering of supersaturation of water vapour needed for condensational growth by water vapour of droplets according to Köhler theory [8] is questioned by Sorjamaa et al. [13] and Kokkola et al. [14]. They state that the partitioning between surface and bulk reduces the effect of surface tension reduction. At a first assumption Li et al. [41] show that the natural surfactant cis-pinonic acid (an oxidation product of monoterpenes) greatly can reduce the surface tension of nanosized aerosol particles by accumulating at the surface; however, Hede et al. [11, 12] showed that cis-pinonic acid can form micelle-like aggregates inside the nanoaerosol, but the surface tension reduction is not limited by this behaviour in accordance with experimental results by Schwier et al. [42]. Therefore, the effect of surface tension depression is not irrelevant for cloud formation processes, implying that a greater number of cloud droplets can form as a result of an increase in the number of available particles consisting of surface tension reducing surfactants. The clouds that are formed are thus “whiter” (as of changed microphysical properties) with a higher albedo and reflecting more of the incoming solar radiation. This effect is amplifying the negative feedback process proposed by Kulmala et al. [5].

Thirdly, monoterpenes evaporated from boreal trees react quickly with oxidizing reactants like ozone and hydroxyl radicals producing low-volatile substances such as cis-pinonic acid that take place in gas-to-particle conversion processes [43, 44]. According to Boy et al. [45] the growth rate of nucleation mode particles is from 30 to 50% explained by hydroxyl radical oxidation. According to Riipinen et al. [46] the growth of nanoparticles is governed by the presence organics. Given time in the atmosphere newly formed particles may grow to form CCN. Depending on the synoptical scale meteorology, CCN generated and grown from VOCs originating over Siberia could potentially reach the high Arctic pack ice region.

In the high Arctic summer there are few of available particles for cloud droplet formation, since the air is clean with limited influences from manmade activities [4749]. An increase in particles due to regional transport would have an impact on the microphysical properties of the clouds being formed. Generally, clouds over the pack ice constitute a warming factor [50, 51], but adding particles when  cm−3 will have a net cooling effect [52]. This is again an amplification of the negative feedback mechanism.

The fourth issue is the most important. Kulmala et al. [5] use an assumed moderate estimate of the increase of VOC emissions of 10%. This value of VOCs emissions increase is corresponding to an increase of optical thickness of the clouds by 1 to 2%. Kulmala et al. [5] also state that a doubling (100%) of increase of VOCs emissions corresponds to 20% increase in optical thickness of the clouds. In the present study we have found that the increase of emissions of VOCs is higher than 10%. This result is not based on a moderate estimate but on simulations using the LPJ-GUESS vegetation model and calculations of emission rates reported in literature. Even though there are approximations included in both simulations as well as in the calculations, we hope that the results derived from this study are more accurate than a moderate estimation. That our estimates of the relative increase of VOCs are higher than the estimates by Kulmala et al. [5] would work in the direction towards an amplification of the feedback mechanism under study.

5.2. Simplifications and Uncertainties

In any kind of attempt to simulate feedback mechanisms there are both simplifications made and uncertainties to consider. For Step  1 in Figure 2, we do not know the magnitude of the global warming. The Intergovernmental Panel on Climate Change [1] concludes that there likely is an increase in average global temperature, and that the increase is somewhere between 0.3°C and 0.7°C for the period of 2016–2035 compared to the period 1986–2005. For the period 2081–2100, different climate model scenarios (Representative Concentration Pathways, RCPs) project the average global temperature increase as 0.3°C–1.7°C (RCP2.6), 1.1°C–2.6°C (RCP4.5), 1.4°C–3.1°C (RCP6.0), and 2.6°C–4.8°C (RCP8.5) [1]. In future studies, the RCPs can be used to model the temperature increase during the simulations. For Step  2, the present knowledge is insufficient to know how much warmer the tundra region must be for the permafrost to melt permanently. For Step  3, we rely on the model simulations to be correct, and as mentioned the model is simplified and the result comes accompanied with uncertainties. For Step  4, the rate of emissions of VOCs is temperature dependent and also different for different types of vegetation. For Step  5, even if we knew the number of newly formed organic particles from emissions of VOCs, the number of particles available as CCN is hard to predict because the process of condensational growth of the newly formed particles is not fully understood [38]. We have seen that the conventional theory used, Köhler theory [8], lacks in the description of organic compounds for sizes below 5 nm and complex morphology as the CCN. However, we have previously shown that there are methods for better describing such systems [11, 53]. For Step  6, the aerosol-cloud-radiation-albedo relationship over the Arctic pack ice is discussed by Leck and Bigg [49] and was concluded to be more complex than elsewhere.

For the calculations of increase in emissions of monoterpenes from forest growth, the resolution of the simulated area is of vital importance. The more accurate the resolution of the calculated locations is, the higher the degree of information would be. To meet the need for highly resolved simulations we use the 50 km × 50 km horizontal grid resolution provided by the LPJ-GUESS model. By this we increase the resolution by 5 to 10 times relative to using atmospheric GCMs. However, the uncertainties in the assumptions used in the simulations described above limit the level of accuracy in the calculations. An even higher resolution would therefore not favor the final results.

The temperature increase that we set to 2, 3, and 4 degrees centigrade is corresponding to an increase globally in the LPJ-GUESS model evenly distributed. This is, however, not the case since we can see an amplification of the warming in the Arctic region and a global mean of 2-degree centigrade increase could actually correspond to a higher temperature in the Arctic region. This effect is not considered in our simulations. This suggests that the enhancement of the forests-aerosol-climate negative feedback mechanism is most likely even more pronounced at a lower global temperature increase.

6. Summary and Conclusions

In this study model simulations of the vegetation in northern Russia were carried out for the climate of the year 1975 as a reference and for three possible future projections: two-, three-, and four-degree centigrade increase of mean global temperature and with a doubling of the atmospheric carbon dioxide concentration over 200 years. The simulations were carried out using the LPJ-GUESS model. The growth of the forest was converted to emissions of monoterpenes based on literature values and compensating for the number of days of sunlight and to a temperature factor.

The results show the following.(i)There was a growth of the boreal forest both for spruce and pine (in the central and western parts of the forest area) and larch (in the north-eastern part of the forest), as well as new forestation of previously wastelands of the tundra. Based on the simulations the boreal forest will move northward as a result of global warming.(ii)The emissions of monoterpenes at the most northern latitudes were estimated to increase with close to 500% for a four-degree centigrade increase protocol.(iii)The likelihood for organic vapours and particles to be transported northward over the central Arctic Ocean is highest for this northern latitude at the same time as the projected increase in precursors for particle formation is huge. Kulmala et al. [5] suggested that the cloud optical thickness would increase 1 to 2% for an increase in VOC emissions by 10% and the cloud optical thickness would increase 20% for an increase in VOC emissions by 100%. As we report an overall increase in VOC emissions by 64%, the cloud optical thickness would then increase more than 2%. However, a back of the envelope calculation indicates that the optical thickness may be up to 9%. For more accurate calculations of cloud optical thickness, a future study could incorporate a box model for radiation and convection. The preliminary results however indicate that the cloud deck to a greater extent would reflect solar radiation back to space and thereby cool the surface of the Arctic region more than what was previously expected. Kurtén et al. [54] estimate the global contribution from boreal aerosol formation to radiative forcing to be from −0.03 to −1.1 Wm−2 and therefore an increase of VOC emissions would be expected to increase the radiative forcing of the same magnitude. This can be compared to the radiative forcing from surface albedo changes which has a global mean value of −0.20 Wm−2 or 0.35 Wm−2 radiative forcing caused by an atmospheric CO2 increase of 19 ppm(v) since 800 AD [55].

Conflict of Interests

The authors have no conflict of interests to declare.