Introduction. Methane and carbon dioxide are the most important greenhouse gases, the increase in the concentration of which in the atmosphere is the main cause of climate change [Taylor and Penner, 1994; Drösler et al., 2014; Hoegh-Guldberg et al., 2019]. In addition to relatively constant sources of methane and carbon dioxide into the atmosphere (such as oligotrophic bogs of the boreal zone), there are sporadic sources (SS): intermittently flooded floodplains, boreal swamp forests, some intermittently swamp forests, etc. Despite the variability of SS as sources of methane, CH4 fluxes in floodplains and in swamp forests can reach 0.1–12.5 [Whalen et al., 1991; Van Huissteden et al., 2005; Terentieva et al., 2019] and 0.7 – 17.1 mgC m-2 h-1 [Moore and Knowles, 1990; Ambus and Christensen, 1995; Aronson et al., 2012; Koskinen et al., 2016; Glagolev et al., 2018], respectively. These values are comparable, and exceed those observed in bogs under certain conditions (a combination of soil moisture and temperature, and other factors) [Gulledge and Schimel, 2000; Vasconcelos et al., 2004; Ullah and Moore, 2011; Shoemaker et al., 2014; Christiansen et al., 2017; Torga et al., 2017; Glagolev et al., 2018; Mochenov et al., 2018]. Unfortunately, in Russia, studies of CH4 and CO2 fluxes from sporadic sources are extremely limited (one-time measurements were performed without reference to spatial, seasonal, and interannual variability of conditions) and were carried out mainly in Western Siberia [Sabrekov et al., 2013; Mochenov et al., 2018; Glagolev et al., 2018; Terentieva et al., 2019] and the European part of Russia [Kuznetsov and Bobkova, 2014; Ivanov et al., 2018; Glukhova et al., 2021; Glukhova et al., 2022]. In general, medium-scale (at the Federal subject level) studies of bogs and forests in Russia have not been carried out in all regions, although they are of particular interest due to the possibility of maintaining a balance between the detailing of estimates and the magnitude of spatiotemporal coverage [Zatsarinnaya and Volkova, 2011; Grishutkin et al., 2013; Baisheva et al., 2015; Ilyasov et al., 2019; Suslova, 2019]. Besides, estimates made throughout the country require clarification at the regional level [Vompersky et al., 2005]. The aim of our work was the simplest inventory of swamp forests of the Moscow region as sources of CH4 and CO2 using GIS mapping and field measurements.
Objects and methods. The basis for the map of swamp forests of the Moscow region (hereinafter, by this term we mean the total territory of Moscow and the Moscow region) was a mosaic of 6 Landsat-8 satellite images. The mapping was carried out using the Supervised Classification algorithm in the Multispec program (Purdue Research Foundation, USA). For each decryption class, at least 7 training polygons were set and the classification module was launched using the maximum likelihood estimation. After the classification, the decryption classes were combined into typological ones: “forest” (automorphic forests), “water surfaces” (rivers, lakes, other water bodies), “swamp forest” (excessively moist forests with a water table level (WTL), predominantly located on the soil surface or close to it) and “wet forest” (excessively moist forests with predominant WTL below the soil surface). We considered the classes of swamp forests and wet forests, regardless of the presence or absence of peat layer in them: the key criterion was WTL. To assess the accuracy of the classification, an error matrix was compiled. For that purpose, on the resulting map, the first operator identified 75 points evenly distributed in space within each typological class; the coordinates of these points without specifying the belonging to the class were randomly sorted and passed to the second operator. Further, the points were assigned to one of the mapped classes based on “blind” visual expert interpretation using ultra-high resolution satellite images. The overall classification accuracy was determined as the ratio of the sum of points, whose mapped and real classes coincide, to the total number of points (Table 1).
Measurements of carbon dioxide and methane fluxes were carried out from 2019 to 2022 in the Dorokhovo mixed black alder moist grass forest, located 66 km west of the border of Moscow, using the static chamber method [Hutchinson and Mosier, 1981; Terent'eva et al., 2017]. Opaque chambers were used in the measurements, so the term “CO2 flux” used in the paper implies the sum of the respiration of the soil-grass-moss cover. The calculation of the annual flux of methane and carbon dioxide from the swamp forests of the Moscow region was performed seasonally using the simplest inventory method [Glagolev, 2010]:
ФОРМУЛА НЕ РИСУНОК
where Aij – is the area (m2) occupied by the i-th source type in the j-th region; fi – is the surface flux density (mgC m-2 h-1), characteristic of the i-th source type; Tj – is the duration of the emission period (hour), characteristic of the j-th region. The duration of the methane emission period within individual seasons was taken on the basis of hydrothermal coefficients and the radiation index as follows: summer – 122 days (from June to September inclusive), autumn – 76 days (from October to mid-December), winter – 90 days (from mid-December to mid-March), spring – 77 days (from mid-March to the end of May). The surface flux density was calculated as the median (and also 1Q, 3Q) for the considered season based on all observations.
Results. The resulting map of swamp forests of the Moscow region is shown in Figure 1 and is characterized by the following areas of typological classes: “forest” - 2,157,716 ha, “water surfaces” 45,693 - ha, “swamp forest” - 58,384 ha, “wet forest” - 233,865 ha. Thus, the total share of forest ecosystems that are able to function as sources of methane - swamp forests and wet forests - is 1.2 and 5.0% of the region's area, respectively (in total 292,249 ha). According to the map, swamp forests are predominantly small ecosystems (from small ones with an area of 3-5 ha, which are extremely widespread, to larger ones, with an area of 30-50 ha, which are somewhat less common), which are exposed to excessive moisture as a result of their location on the outskirts of wetland massifs, near river floodplains, in small local relief depressions, as well as in elements of a ravine-gully planting (mainly in the southern part of the Moscow region). Wet forests are located in more drained areas, often associated with swamp forests in a single landscape structures, but they are much more widespread, and often occupy significantly larger areas: from 10–50 to 100–500 ha.
The error matrix of the resulting map is presented in Table. 1. The overall classification accuracy (the ratio of the sum of the elements of the main diagonal of the error matrix to the sum of checkpoints by class) is 76%. Water surfaces with the highest possible producer’s accuracy (100%) are most accurately identified. The “other” class has the same user’s accuracy as water surfaces (93%), but poorly less producer’s accuracy (74%). In general, the classes of swamp and wet forests are the least accurately defined (36–46%): they have significant intersections with all classes except that for the open water surface, and, most importantly, with each other. In order to achieve a reasonable classification accuracy and to make further calculations of the regional flow, we combined the “swamp forest” and “wet forest” classes into one: in this case, the user’s accuracy of the combined class was 65%, and the producer’s accuracy was 74%, which allows us to fairly accurately predict the location of forests of varying degrees of waterlogging when they are considered together.
Generalized results of measurements of methane and carbon dioxide fluxes by seasons and their brief statistical characteristics are presented in Table. 2. The simplest inventory based on the proposed approach makes it possible to estimate the methane flux from the soils of swamp forests with different degrees of waterlogging at 6666 tC yr-1 (1Q – 407; 3Q – 38790); carbon dioxide at 1.5 MtC yr-1 (1Q – 0.6; 3Q – 2.7). Taking into account the 100-year global warming potential for methane equal to 28 [Drösler et al., 2014], the total emission of methane and carbon dioxide from the soils of swamp forests with different degrees of waterlogging was 5.7 MtCO2-eq yr-1 (1Q – 2.2; 3Q – 11.4)[1]. More detailed information obtained on the basis of the simplest inventory presents in table 3.
Discussion. According to the data of the Great Russian Encyclopedia [Osipov et al., 2004], the area of automorphic forests in the Moscow region in 2015 amounted to 1,896,000 ha, which is in good agreement with the data obtained based on the current classification (the area of the “forest” class amounted to 2,157,716 ha). The distribution of swamp forests in the north of the Moscow region, observed on the resulting map, corresponds to swamp black alder, downy birch forests, as well as forests with gray alder on the map of G.N. Ogureeva et al. [1996]. In the southeastern part of the Moscow region, the areas occupied by swamp forests, according to the results of satellite data classification, are identical to the distribution of downy birch and pine-spruce-long-moss-sphagnum forests along the edges of wetlands. Wet forests are located to the south of the Ruza Reservoir correspond to spruce forests with gray alder, whereas those located to the northwest of the town of Klin are associated with black alder forests and pine-spruce forests with black alder (Ogureeva et al., 1996). The area occupied by swamp and wet forests identified in the current work is comparable to that of distribution of forests with black and gray alder (5.01 and 1.44% of the area of the region) provided in (Kotlov and Chernenkova, 2020), which indirectly confirms the assessment adequacy of the share of the territory occupied by wetland forest ecosystems identified in our work.
One of the main problems of GIS cartography based on remote sensing data is the poor availability of ground-based data or the inability to check map errors by field methods due to the wide coverage of the study area. However, the classification accuracy of 60-70% is the rule rather than the exception [Kotlov and Chernenkova, 2020] and is considered satisfactory. We anticipate that GIS mapping that combines multiple cartographic sources at its core (for example, by calculating a median estimate based on multiple maps) will improve the final result in the future.
Conclusion. The total area of swamp forests and wet forests in the Moscow Region is 292,249 ha. The emission of methane from these ecosystems is 0.25 (1Q – 0.02; 3Q – 1.45) MtCO2-eq per year, whereas that of carbon dioxide is 5.40 (1Q – 2.16; 3Q – 9.92) MtCO2 per year. The highest total emission of methane and carbon dioxide from wetlands is observed in the summer-autumn period, gradually decreasing by the beginning of winter and increasing again (to the level of autumn values) in spring. The value of the total emission of the main carbon-containing gases from the soils of swamp forests of the European part of the Russian Federation should be taken into account when quantifying all significant sources and sinks.
[1] The annual total methane flux was calculated as follows: the median of measurements for each of the season (0.14, 0.74, 0.02 and 0.25 mgC m-2 h-1, for summer, autumn, winter and spring, respectively) was multiplied by the number of hours in days, by the corresponding length of the season (122, 76, 90 and 77 days), then by the wetland forest area (2.922×109 m2), and finally by a correction factor (10-9) to convert mgC to tC. The annual total carbon dioxide flux was calculated in a similar way (the difference was in the value of the correction factor, which was 10–15 for converting mgC to MtC). When converting the CH4 flux (expressed in tC yr-1) to MtCO2-eq yr-1, the original value was multiplied by 16/12 (the ratio of the molar mass of CH4 to the molar mass of C), then by 28 (100-year global warming potential) and, finally, by a correction factor (10-6) to convert tons to megatons. To calculate the total flux consisting of emissions of CH4 (MtCO2-eq year-1) and CO2 (MtC year-1), the latter was multiplied by 44/12 (the ratio of the molar mass of CO2 to the molar mass of C) and added.
