The study region includes the forest zone of ER (Fig. 1) between the forest–steppe transition zone in the south and forest–tundra transition zone in the north. The availability of the EEFCC dataset determines the eastern boundary of the study region that broadly coincides with the Ural Ridges.

We used the 250 m resolution map of the vegetation cover of Russia (Bartalev et al., 2016) to estimate forest-covered area and dominant forest species (Fig. 1). Forests cover 54.6 % of the study region. The most widespread dominant forest species are dark coniferous (Picea abies, Picea obovata, Abies sibirica), light coniferous (Pinus sylvestris), small-leaved (Betula pendula, Betula pubescens, Populus tremula), and some broadleaved species (Tilia cordata, Quercus robur and others) (Kalyakin et al., 2004). Secondary (regrown after logging or wildfires) small-leaved and mixed forests cover approximately 61 % of the total forested area. Old-growth dark coniferous forests are widespread on the western slope of the northern Urals and adjacent plain, and pine forests cover the largest area (> 100 000 km2) in the northwest of ER (Fig. 1).

We used multiple data sources to collect information on windthrow events for the 1986–2017 period. In particular, we utilized satellite data to delineate windthrow areas and determine a storm event type and used additional information to determine the dates of storm events.

We verified each windthrow area in the database using pre- and post-event Landsat/Sentinel-2 images, high-resolution images, and additional information. This verification was performed to ensure that the forest disturbance was caused by wind and to determine the type of storm that caused the windthrow. In total, we verified 54 % of windthrow areas with the HRIs mainly for the 2001–2015 period. Other events were verified using the Landsat images (22 % of windthrow), the Sentinel-2 images (9 %), and additional data sources like weather station and eyewitness reports (15 %). As a result, the probability that any forest disturbance was mistakenly referred to as a windthrow is minimal.

In addition, we used the last cloud-free Landsat or Sentinel-2 image obtained before a storm and the first image obtained after it to separate windthrow areas from other disturbances, mainly from logged areas. We removed forest disturbances that were not related to a storm event (Fig. 6). During the verification, we also found and delineated several storm-damaged areas that were missed in the GFC/EEFCC data. Such areas are located mainly in small-leaved or broadleaved forests. After the verification, we determined the type of windthrow depending on the weather phenomenon inducing this windthrow. We selected tornado-induced and non-tornado-induced windthrow areas; the latter were subdivided into those induced by convective and those by non-convective storms. In turn, non-convective storms include also snowstorms, which are indicated in the database but not analyzed separately further in the paper. By convective storms we mean squalls and downbursts; however, this more detailed division is lacking in the database.

To distinguish tornado-induced windthrow areas from other wind-related disturbances, we determined the direction of fallen trees using the HRIs. Indeed, the main signature of tornado-induced windthrow is the counterclockwise, or infrequently clockwise, rotation of the fallen trees (Beck and Dotzek, 2010; Shikhov and Chernokulsky, 2018). With the lack of HRIs, we considered three additional signatures of a tornado-induced windthrow, namely (1) a quasi-linear structure of a windthrow with a ratio of length and width ≥ 10 : 1, (2) a gradual turn of a storm track, and (3) a predominant total removal of forest stands (Shikhov and Chernokulsky, 2018; Shikhov et al., 2019b). Note that the ratio of length and width of a tornado track ≥ 10 : 1 is also typical for the United States (Schaefer and Edwards, 1999). Based on these three signatures and additional information from weather station reports, witness reports, photos, and videos, we assigned the high or medium degree of certainty of storm type determination for each windthrow (Table 4).

Windthrow areas, caused by non-convective windstorms or snowstorms, have specific geometrical features as well that are seen in satellite images. Specifically, windthrow areas related to non-convective windstorms typically have a damage track with an enormous length and width, up to 200 and 45 km, respectively, with, however, a slight or moderate degree of forest damage. Stand-replacing disturbances caused by non-convective windstorms usually occur in dark coniferous forests only (Dobbertin, 2002; Schmoeckel and Kottmeier, 2008). Since non-convective storms affect large areas and last for relatively long period, they are typically well-reported by weather stations, which simplify the attribution of related windthrow. In its turn, snowstorm-induced windthrow areas are distinguishable from other disturbances primarily based on the dates of occurrence: they happen usually in autumn; although – one severe snowstorm occurred in early summer. It is of note that we found none of the snowstorm-induced stand-replacing windthrows happening in winter.

After determining a storm event type, we excluded from the database tornado-induced windthrows with an area ≤ 0.05 km2 and other windthrows with an area ≤ 0.25 km2. We took into account the following reasons during the exclusion of such small-scale windthrow areas:

the difficulty to prove that these disturbances were actually caused by wind, especially with the lack of HRIs;

Only five squall-induced windthrows with an area < 0.25 km2 were stored in the database since they are associated with severe weather outbreaks with proven dates. It is of note that a typical tornado-induced windthrow consists of a relatively small number of EDAs with a total removal of forest stands that are well-detected by the Landsat images. In its turn, a typical, non-tornado-induced windthrow includes a larger number of small-scale (i.e., 2–4 Landsat pixels) areas of stand-replacing disturbances that are poorly detected by satellite images. This difference results in the necessity of using two distinct thresholds for tornado- and non-tornado-induced windthrow areas.

The threshold values used in Sect. 4.1–4.2 have some subjectivity, and their modification may substantially change the number of allocated windthrow areas in the dataset. The optimization of the above-described threshold values can be evaluated in further studies that should involve ground-based data.

We used the Landsat data and the Landsat-based products GFC and EEFCC to estimate geometrical parameters of windthrow areas. We determined the path length (L), mean and maximum widths (Wmean and Wmax), and damaged area (A) for each windthrow using the technique that had been successfully implemented for tornado-induced windthrow areas (Shikhov and Chernokulsky, 2018). The calculation of these parameters was performed in the Lambert equal area and equidistant projection for North Asia to avoid possible projection-related distortions.

We calculated A in the ArcGIS 10.4 as the sum of the area of forest damaged plots which are attributed to one windthrow. We determined L as a length of the central line drawn through a damaged area, i.e., the distance between the two farthest points of a windthrow. The central line was created automatically (using a Python tool) as the distance between the two farthest points of a windthrow. It is insensitive to the allocation of patches to windthrow area.

We calculated Wmean as the mean length of several transects that are perpendicular to a storm track with a 200 m step; this step had been found optimal in terms of quality and counting efficiency (Shikhov and Chernokulsky, 2018). Only stand-replacing windthrow areas were taken into account in this calculation. In comparison to Shikhov and Chernokulsky (2018), in which Wmax was calculated manually using the HRI data, in this study, we assigned the length of the largest transect to Wmax because of the lack of HRIs for many windthrow areas.

In addition to windthrow characteristics, we estimated geometric characteristics of EDAs and those of storm tracks. In particular for EDAs, we calculated their area AEDA. For storm tracks, we estimated maximum and mean width (WTRmean and WTRmax), path length (LTR), and damaged area (ATR). We calculated WTRmean based on the same transects that were used to calculate Wmean but without excluding undamaged forests and treeless areas. Similarly, the length of the largest transect that includes undamaged forests and treeless areas was assigned to WTRmax (Fig. 7). If a track consists of two (or more) parallel windthrow areas, then its width was calculated within the outermost boundaries of these windthrow areas (Fig. 3). The same calculation was performed for LTR in the case of two (or more) subsequent windthrow areas (Fig. 3). Thus, the 10 km threshold used (see Sect. 4.1.4) may influence geometrical characteristics of a single windthrow area but do not affect those of a storm event.

We assessed the accuracy of GFC-based estimates of windthrow geometrical parameters by comparing them with the same parameters calculated manually with the use of HRIs. We performed just such a procedure for 10 windthrow areas caused by squalls, whose areas range from 0.26 to 6.09 km2 (Table 5). The distribution of their A is close to the one for the full dataset.

We delineated manually all EDAs within these 10 windthrow areas using the HRIs. In total, we found 837 and 947 EDAs according to the GFC and the HRI data, respectively. Owing to the relatively correct georeference of the Landsat data (USGS, 2019), we found no systematic spatial bias between contours of GFC-based and HRI-based windthrow areas. Despite their general matching, there is no complete overlap due to the different spatial resolutions of the GFS and HRIs (Fig. 8). For example, one GFC-based EDA may intersect with several HRI-based ones, and vice versa. We found that only 66.5 % of the total area is attributed to windthrow in both GFC and HRIs, while EDAs with a small area can be missed. In particular, 263 HRI-based EDAs with the total area of 0.97 km2 were completely missed in the GFC, while 146 GFC-based EDAs with the total area of 0.52 km2 were missed in the HRIs. For overlapped EDAs, we found that the mean absolute error and root mean square error of AEDA estimates amounted to 27.6 % and 13.1 %, respectively. We found that the relative error decreases for large EDAs and for those having a simple shape, i.e., quasi-circular. The user's and producer's accuracies increase from 20 %–25 % for EDAs with AEDA<0.01 km2 to 70 %–75 % for EDAs with AEDA>0.1 km2. In general, for the overlapped EDAs, the GFC overestimates their AEDA (by 4 % on average) primarily in coniferous forests. The mutual effect of more frequent omissions of small EDAs in the GFC compared to the HRIs and overestimation of overlapped EDAs results in the approximate equality in the total area of delineated windthrows, 17.11 and 17.13 km2, based on the GFC and HRIs, respectively.

For the entire windthrow area, we calculated the accuracy of their geometrical characteristic estimates as well. In particular, we calculated the user's and producer's accuracies of the GFC-based delineation for each of 10 selected windthrows. These accuracies are mainly determined by the complexity of windthrow shapes and composition. In particular, the accuracy is higher for a windthrow consisting of a relatively small number of simple-shape EDAs. Otherwise, the accuracy decreases down to 50 % for a windthrow having very amorphous spatial structure. In our sample, the GFC data tends to overestimate the area of a windthrow – 8 cases out of 10 were overestimated. The mean absolute percent error (MAPE) for A is 14.6 %. The major overestimation of A by the GFC data, as well as Wmean and Wmax, was revealed for relatively small windthrow areas. This is in line with the previous findings by Koroleva and Ershov (2012). They showed that the reliable estimate (with 15 % accuracy) of the damaged area using the Landsat images is possible only for windthrow areas exceeding 0.026 km2. It is of note that for tornado-induced windthrow areas, Shikhov and Chernokulsky (2018) found that the GFC data generally tend to underestimate A, with MAPE amounting for 17.9 %.

The assessment of geometrical parameters of windthrow areas appearing before 2000 and being found by the EEFCC is challenging due to the low availability of the HRIs or other independent data sources, e.g., the data of forestry services. Windthrow areas induced by storm events that occurred > 20 years ago can be delineated by the HRIs only if a storm passed through old-growth forests that have not been affected by other disturbances, i.e., timber harvesting or wildfires, in subsequent years. Such forests are widespread only in the northeastern part of ER (Pakhuchiy, 1997). We found five EEFCC-based windthrows occurring between 1998 and 2000 that were most well-detected by the HRIs: four tornado-induced and one storm–induced windthrow. We delineated them with the EEFCC and HRI and compared their characteristics (Table 6). We found a general overestimation of A, Wmean, and Wmax in the EEFCC that was larger than in the GFC. It may be related to the inclusion into a windthrow of not only the real wind-damaged area but also the surrounding pixels where trees died after a storm event mostly because of bark beetles (Köster et al., 2009). The intensity of this mortality is highest during the second year after a storm event (Köster et al., 2009).

We aimed to establish the exact date or even the exact time for each windthrow appearance. However, due to data constraints, dates of some windthrow events were determined with 6 months accuracy. We iteratively refined a date, or a date range, by using different data. The process, related to the determination of the date of a tornado-induced windthrow only, was described previously in Shikhov and Chernokulsky (2018).

First, the year of a windthrow can be obtained directly from the Landsat products but with some limitations. In the GFC, forest disturbances are accompanied by information on the year of the event's occurrence. However, the exact year is determined correctly only for 75.2 % of forest loss pixels; for 24.8 % of them, the date can be either 1–2 years earlier or later (Hansen et al., 2013). In the EEFCC, a year of windthrow occurrence is not explicitly determined and came within the ranges of 1986–1988 and 1989–2000.

Next, we refined a range of dates based on all available images from the Landsat and Sentinel-2 satellites. The accuracy of such refinements depends on the frequency of observations and cloudiness. The availability of cloudless Landsat images varied from year to year. The lowest number of cloud-free images (2–4 images a year on average) is available for 2003–2006 and 2012, when only Landsat-7 data after scan line corrector failure are available (Potapov et al., 2015). Hence, the worse accuracy of windthrow date determination is typical for these years. On average, 8–10 images per year can be used for windthrow date determination. Due to the launch of the Sentinel-2A satellite, the number of images per year had an abrupt increase after the summer of 2016. We used images taken throughout the year. Despite the frequency of cloudless images in autumn and winter being lower than in the summer season, it was sufficient for analysis. Thus, wintertime images (of land covered with snow) were successfully used for windthrow identification, especially if a storm occurred at the end of the summer season, and the autumn season lacked cloud-free images.

Further, given the satellite-derived range of possible event dates, we made the subsequent analysis using additional data, such as weather station observations, various databases and reviews on hazardous weather events, damage reports, photos and videos in the media and social networks, and reanalysis data (see Shikhov and Chernokulsky, 2018, for details). This analysis allowed us to establish the exact dates for 48.4 % of all windthrow events, including 39.2 % and 59.7 % of tornado- and non-tornado-induced windthrow events, respectively.

The dates of storm-induced windthrow events were defined more successfully than those for tornado-induced ones due to the local nature of convective storms, especially of tornadoes, and the relatively large distance between Russian weather stations. Specifically, the average and median distance between the nearest weather stations within the study area amounted to 53.7 and 49.9 km, respectively. Consequently, many storm events were reported by weather stations located on a storm path at a distance of 50–100 km from a windthrow, while the closest stations did not report strong wind gusts since they were away from the storm path. In total, we matched storm reports of weather stations, namely reports with wind gusts ranging from 15 to 34 m s-1, with only 34.5 % of windthrow events with known dates.

Another reason for the more successful determination of dates of appearance for large-scale windthrow areas than for small-scale ones, e.g., tornado-induced, is an increase in the probability that a corresponding storm passes through a settlement(s) and is covered in the media. In total, we used media reports, information from regional weather services, witness photos and videos, and existing scientific literature (e.g., Dmitrieva and Peskov, 2013; Petukhov and Nemchinova, 2014; Shikhov and Chernokulsky, 2018; Shikhov et al., 2019a) to specify the date and time of 29.7 % of windthrow events.

Dates and times of some cases (7.8 % of all cases) were established using images from the meteorological satellites Terra/Aqua MODIS and METEOSAT-8 and Russian weather radar data (Dyaduchenko et al., 2014). However, the routine usage of these data is time-consuming and limited due to some access restrictions. The subsequent clarification of the exact time of windthrows can be carried out in further studies.

The compiled database includes three shapefiles (.shp) corresponding to three hierarchical levels such as elementary damaged areas, windthrow, and storm events. The database includes 102 747, 700, and 486 objects for each level, respectively. The total area of the spatial features is equal to 2966.1 km2. It is of note that we cannot determine whether the trees were felled or broken by the wind based on satellite images that even have very high resolution. Therefore, we use the single term “windthrow” for all types of wind-induced forest damage.

The overwhelming majority of stand-replacing windthrows found in ER, namely 97.4 % of the events and 95.3 % of wind-damaged area, are associated with convective storms and tornadoes (Table 7). More than half of all windthrow events are tornado-induced with, however, a relatively small damaged area (less than 13 % of the total wind-damaged area). Non-convective storms and snowstorms are responsible for less than 5 % of the area of stand-replacing windthrow in ER. This is somewhat in contrast to Western and Central Europe where most forest damage is induced by non-convective wind events, namely winter storms, caused by strong extratropical cyclones (Gardiner et al., 2010; Gregow et al., 2017). Indeed, winter windstorms affect Eastern Europe less compared to Western and Central Europe (Haylock, 2011).

Among 486 storm events that caused windthrow, 381 yielded only one windthrow area (Fig. 9), primarily tornado-induced. The rest of the 105 storms resulted in a smaller number of windthrow events (319) but larger damaged area – 2276.6 km2, namely 76.8 % of all damaged area. Most of these storms induced two or three successive or parallely located windthrow areas, and only 14 storms caused 5 or more windthrow events. We found a maximum of 17 separate windthrow areas that related to one storm. We found 71 storm events resulting in two or more successive windthrow areas, while 12 storm events led to the formation of two or more parallel windthrow areas, and 22 storm events include a family of both parallel and successive ones (Fig. 3). The maximum distances between the two nearest successive and two parallel windthrow areas amount to 150 and 26 km, respectively.

It should be noted that a single storm may cause both tornado- and non-tornado-induced windthrow, e.g., a supercell can lead to the formation of a tornado and a rear-flank downdraft (Karstens et al., 2013), both causing forest damage. In total, we found 30 storms that resulted in the formation of two types of windthrow.

We managed to match several storm events with storm reports at weather stations; in particular, the database contains 89 such cases. Among these 89 station reports, we found eight reports with wind gusts ≥ 30 m s-1, 14 reports with wind gusts 25–29 m s-1, and 30 reports with wind gusts 20–24 m s-1. This information has been included in the database and can be used in further studies to estimate the critical wind speed causing windthrow and to analyze the role of other accompanying weather phenomena, e.g., with snow, heavy rainfall, large hail, etc.

Windthrow events occur in the entire forest zone of ER (Fig. 10). However, the highest density is observed near 60∘ N and somewhat coincides with the highest percentage of forest-covered area (see Fig. 1). It is of note that two windthrow areas are located north of 66∘ N and one of them is even north of the Arctic Circle. The dominant direction of both tornado-induced and other windthrows is SW–NE (Fig. 15b), which is in line with previous studies on tornado climatology in northern Eurasia (Shikhov and Chernokulsky, 2018; Chernokulsky et al., 2020).

Three regions where windthrows have affected more than 0.75 % of forests can be highlighted (Fig. 11a). Two of them are related to the catastrophic storms which occurred on 27 June and 29 July 2010. In total, these two storms have damaged 1140 km2 of forests, which is 38.4 % of the total area of stand-replacing windthrow in ER in 1986–2017. The third area is located on the western slope of the northern Urals and coincides with the largest extent of dark coniferous forests in ER (Pakhuchiy, 1997). The most important windthrow events occurred here in June 1993, July 2012, and October 2016. The latter was induced by a snowstorm. The relatively high frequency of windthrow in this region was emphasized previously (Lassig and Mocalov, 2000; Shikhov and Chernokulsky, 2018; Shikhov et al., 2019b). It was hypothesized that it may be related to the combination of several factors, namely widespread old-growth forests, a high annual precipitation rate (up to 1000 mm yr-1), and large soil wetness, which all contribute to the forest's wind susceptibility (Dobbertin, 2002).

The highest density of tornado-induced windthrows is found between 59 and 62∘ N, 48 and 56∘ E (Fig. 11b), which is in good agreement with previous estimates (Shikhov and Chernokulsky, 2018). However, the ratio of the tornado-damaged area to the total forested area is higher in the western part of ER (Fig. 11b). It is of note that higher values of so-called convective instability indices are also observed in this region (Taszarek et al., 2018).

The species composition and age of forest stands have substantial influence on the spatial distribution of windthrow (Dobbertin, 2002, Suvanto et al., 2016; Gregow et al., 2017). Using the presented dataset, estimates of the relationship between windthrow area and forest stands characteristics can be carried out in future studies at a regional scale.

We successfully determined the year of occurrence for all windthrow events and the month of occurrence for 263 (67.9 %) tornado-induced and 224 (71.5 %) non-tornado-induced windthrow events. We established the dates of occurrence for 339 windthrow events, including 149 (39.2 %) tornado-induced and 187 (59.7 %) non-tornado-induced ones. It is of note that the dates of the most impactful large-scale windthrows with a damage area > 10 km2 were determined for 44 out of 49 cases (90 %). Windthrow events with known dates have a total area of 2599 km2, i.e., 87.7 % of the total wind-damaged area.

The storm-damaged area has a relatively high interannual variability (Fig. 12). The largest area of windthrow, i.e., > 1200 km2, is found in 2010, when two exceptional storm events occurred. An extremely high number of tornado-induced windthrow events occurred in 2009 and 2017. Storm events causing windthrows have been observed every year and range from 2 to 36, with the maximum in 2012 and minimum in 2001. In general, the annual number of windthrows and storm events was lower before 2001, when the EEFCC data were used to identify windthrow, and higher after 2001, when the GFC data were utilized. The annual number of windthrow events for these periods amounts to 12.1 and 30.5, respectively; in its turn, the annual number of storm events amounts to 8.3 and 20.9. This temporal inhomogeneity, related to the different initial data used, should be taken into account when interannual variability is analyzed. More details on the dataset limitations are provided in Sect. 6.

Windthrow events occur in ER from May to October (Fig. 13). The seasonal maximum of the number of windthrow events is found in June – both for tornadoes and for other storm events. This is in concordance with the previous estimates on tornado climatology (Shikhov and Chernokulsky, 2018; Chernokulsky et al., 2020). The maximum frequency of the occurrence of storm events causing windthrows is also observed in June. Moreover, more than 90 % of storm events with known dates occur in summer. It is important to note that we failed to establish the month of appearance for 127 tornado-induced windthrow areas and 98 non-tornado-induced ones, which have a total area of 245 km2.

Sometimes, two or more storm events causing windthrows occurred in ER on the same day. In total, we found 7 outbreaks with more than 10 windthrow areas per day. The most remarkable outbreaks occurred on 18 July 2012 when 9 storms resulted in 25 windthrow areas and on 7 June 2009 when 5 storms resulted in 24 windthrow areas. However, the largest forest damage is associated with a single storm, namely the long-lived convective storm “Asta” (Suvanto et al., 2016). This storm passed over the northwestern part of ER and Finland on 29 July 2010 and damaged 639 km2 of forests in Russia alone.

No winter windthrow events were found. It is of note that both GFC and EEFCC Landsat-based products reveal stand-replacing windthrow areas regardless of the season of their appearance. In particular, if windthrow happened in winter, it would be clearly seen in images taken in subsequent vegetation periods because of a rather slow forest recovery process. Therefore, the lack of winter windthrow revealed is feasibly due to the climatic conditions of the study area and is not associated with data limitations. In particular, winter storms from Western Europe reach the territory of Russia that is already weakened (Haylock, 2011), In addition, in ER and Northern Europe, low temperatures and soil freezing also prevent trees from falling because of windstorms during the winter season (Suvanto et al., 2016). According to Suvanto et al. (2016), winter windthrows are not typical for Finland either.

We restored the time of occurrence with 6 h accuracy for 216 windthrow events, 136 among them using weather station reports and 80 using other data sources. We found 122 windthrow events (56.4 %) occurring between 15:00 and 21:00 LT (local time), which coincides with the afternoon maximum of the development of a deep convection. However, several more impactful storms, including, for instance, the “Asta” storm, occurred around midnight local time. No windthrows were found between 06:00 and 10:00 LT during the morning minimum of the convection diurnal cycle. A similar diurnal cycle was found for tornado events in northern Eurasia (Chernokulsky et al., 2020).

The area of EDAs varies between 0.0018 and 30.9 km2. Most of the EDAs are less than 0.01 km2 (Fig. 14a), but their total area is less than 10 %. In turn, 1 % of the largest EDAs account for 36.8 % of the total area of windthrows. Using the Kolmogorov–Smirnov (K–S) test, we found that at the 0.01 significance level, we can reject the null hypothesis that two samples of AEDA within each pair of windthrow types are drawn from the same distribution (at the 0.01 level). Because of the small sample size of windthrow areas induced by non-convective storms, later in the article, we will not discuss the results of the K–S test to compare distributions of characteristics of this type with those of other types.

Tornado-induced windthrow areas contain fewer EDAs than windthrow areas induced by strong wind (Fig. 14b). In particular, most of the tornado-induced windthrow areas include 10–25 EDAs, and only 2.5 % of them consist of more than 100 EDAs. In contrast, about 43 % of non-tornado-induced windthrow areas include more than 100 EDAs, while 5.5 % of them consist of more than 1000 EDAs. Based on the K–S test, we found that samples of a number of EDAs in tornado- and convective-storm-induced windthrow areas are from different distributions.

A relatively small number of severe storm events are responsible for most of the area of windthrow (Fig. 15a). Indeed, the 10 most destructive storm events occurred in ER over 1986–2017 and damaged 1758 km2 of forests, namely 59.2 % of the total area of windthrows in the database. This peculiarity is less pronounced for tornado-induced windthrow areas since their area usually is less than 10 km2. In particular, 10 tornadoes with the largest areas damaged 96.6 km2 of forests – 25.5 % of the total tornado-damaged area. Thus, the distribution of tornado-damaged areas is less skewed to high values than the distribution of other windthrow areas. The K–S test shows that samples of A for tornado- and convective-storm-induced windthrow areas are from different distributions.

The length of windthrows ranges from 0.8 to 283.6 km (Fig. 15b). More than 44 % of tornado-induced windthrow areas have path length < 5 km, while path lengths of 5–15 km are most frequent for non-tornado-induced ones. Based on the K–S test, we found that samples of the number of L for tornado- and convective-storm-induced windthrow areas are from different distributions. The maximum length of a storm track, consisting of several subsequent windthrow areas, reaches 544 km. This damage track was caused by the storm on 27 June 2010. In addition, another nine storm tracks have a length exceeding 250 km – most of them are among the most destructive in terms of forest-damaged area. Such series of windthrows with exceptionally long path lengths were likely caused by derechos. Derechos are long-lived mesoscale convective systems producing widespread damaging winds and causing large-scale forest damage in the United States (Johns and Hirt, 1987; Peterson, 2000), Europe (Taszarek et al., 2019), and South America (Negrón-Juárez et al., 2010), although not a single derecho event has been reported previously in Russia. A more detailed further analysis of these storm events should be carried out to confirm their nature.

Most of the tornado-induced windthrow areas have Wmax and Wmean less than 200 m (Fig. 15c, d). Instead, the distribution of Wmax of non-tornado-induced windthrow areas shifted toward larger Wmax. In particular, 103 windthrow areas (32.9 %) have Wmax>1000 m. The K–S test shows that samples of both Wmax and Wmean for tornado- and convective-storm-induced windthrow areas are from different distributions. The width of storm tracks is several times higher than the width of windthrow areas. Moreover, WTRmax of non-tornadic storms is several times higher than their WTRmean. WTRmax exceeds 30 km for the three widest convective storms: two derechos occurred on 27 June and 29 July 2010, and one non-convective storm occurred on 7–8 August 1987.