Two tributaries of the Upper Noguera Pallaresa are studied: the Romadriu River (in the left-hand margin) and the Flamisell River (in the right-hand margin) (Figure 1 and Table 1). In the Romadriu subcatchment, the Portainé and Ramiosa streams (tributaries of the Romadriu) were sampled between March 2014 and March and September 2015, while between the end of May and the beginning of June 2016, other areas close to the riverbanks of the Romadriu River itself were also sampled. The lower section of the Portainé stream, which forms an elongated cone of alluvial debris at its confluence with the Romadriu, has been intensively studied and the results obtained were published by Génova, et al. (2018) and Furdada, et al. (2020). The cone is covered by a very diverse hardwood forest, dominated by species such as Populus tremula L., Fraxinus excelsior L., Populus nigra L., Prunus avium L. and Betula pendula Roth, along with scattered specimens of Tilia platyphyllos, Quercus petraea (Matt.) Liebl., Juglans regia L., Acer campestre L., and Salix caprea L.). Around the Ramiosa stream, which in the same area flows into the Romadriu River, there is a similar forest, although here Quercus petraea and Fraxinus excelsior are more frequent. On the other hand, in the sampled areas of the Romadriu riverbank Populus tremula and Populus nigra dominate. The presence of a ski resort in Portainé from 2006-2008 resulted in the different behavior of flood and debris flow dynamics (Furdada, et al., 2020), leading to the sedimentation of the current cone at the confluence of the Portainé stream with the Romadriu River. In the Ramiosa catchment, the channel flow is incised in the bedrock.

In October 2018, a small area of ~500 m2 in the Flamisell subcatchment, was sampled in a dense riparian forest growing on a large gravel bar, consisting mainly of Alnus glutinosa (L.) Gaertn., together with scattered Populus nigra trees. The area is situated approximately 1 km south of the town of Senterada and downstream of the small dam or weir of Senterada (0.12 Hm3, 26.5 m long and 3.7 m high above the channel, built between 1919 and 1920; Galí, 1922). This sampling area, located immediately downstream of the weir, was chosen because a large number of trees with damage and scars were found, and also to explore the effect of this small hydraulic structure.

The dendrochronological sampling is focused mainly on trees that presented clear signs of damage (Stoffel & Bollschweiler, 2008; Díez-Herrero, et al., 2013; Stoffel & Corona, 2014; Génova, et al., 2015, 2018), potentially caused by the impact of boulders and sediment particles or floating material carried by the floods (mainly logs or large pieces of wood). The main indicators of macroscopic damage identified consisted of scars, tree crown loss or loss of the main stem, tilting trunks, and dead trees (Figure 2). Other nearby trees were also sampled, because although they did not show any external signs of damage, they may either have been wounded at the base covered by dragged material or perhaps had sustained old internal wounds covered by callus (Génova, et al., 2015, 2018).

Additional data were collected for each tree selected, including general dendrometric data (perimeter, height, dominance) and the size of the wounds. Furthermore, sketches of the sampled trees and the environment were made, their precise location was determined with differential GNSS, their detailed geomorphological position (elements or facets; such as river bank, channel bar, floodplain), and photographs were also taken.

The extraction of the dendrochronological samples (cylindrical cores) was carried out using a 400 mm long Pressler borer with an internal diameter of 5 mm (Grissino-Mayer, 2003). At least 2 samples were extracted from each selected specimen representing the trunk growth, one in the flow direction and the other in the opposite direction. Other samples close to the scars or the callus were likewise extracted to complete the information (Stoffel & Bollschweiler, 2008; Génova, et al., 2018), both the trunks and replacement branches of decapitated trees (Figure 3). The extracted samples were placed on wooden supports for conservation and drying and, once dry, the usual preparation procedures were followed before measurement. In addition, some wedges with parts of the callus and complete sections of the trunk of dead or badly damaged trees were also obtained (Génova, et al., 2018).

We used the LINTAB measurement table with a precision of 0.01 mm and the associated computer program TSAP-Win to measure the tree rings and compile temporal tree-ring growth data series. The tree-ring series were dated by assigning dates to each value of the series using synchronization techniques to perform qualitative and quantitative, visual and statistical analyses (Cook & Kairiukstis, 1990), as well as using the statistical functions (mainly Standard t-value and synchronicity) of TSAP-Win, and results were verified using the software Cofecha (Grissino-Mayer, 2001). First, we synchronized the tree-ring series from each tree, which was then averaged and synchronized with those from the other trees of the same species.

We selected and analyzed marks and other evidence in the dated tree-ring series that could provide information on torrential floods (Shroder, 1980; Stoffel and Bollschweiler, 2008; Díez-Herrero, et al., 2013; Ruiz-Villanueva, et al., 2013; Stoffel & Corona, 2014; Génova, et al., 2018), mainly as follows:

Abrupt growth changes (suppressions and releases) in tree-ring growth series are responses to different possible inputs. Therefore, they were only taken into account as complementary data when they were correlated with other indications and affected more than 10 % of the trees. These indicators were analyzed using the LRM software (List Ring Measurement; Holmes, 1999), which identifies the significant years in growth change according to their representativeness in groups of trees. In the most recent study carried out in the Flamisell subcatchment, where the trees were young, this analysis was complemented with a study of growth changes according to the proposal put forward by Nowacki & Abrams (1997) and previously used in Dendrogeomorphology (for example in Sorg, et al., 2010 and Malik, et al., 2021). The percentage change in the growth of each tree from the date of the possible event (GC) was determined using the growth mean of 2 years before (M1) and the growth mean of 2 years after (M2), according to the formula:

considering the threshold for recognizing the change in each tree to be ≥25%.

On completion of the analysis of the different marks and signals registered in the tree rings, we follow the proposal by Génova, et al., (2018) to date the recognized evidence of damage. The events are therefore dated with a biannual denomination, the dendrogeomorphological year, which indicates that they may have first started at the end of the formation of the previous ring; that is, during the latency period of the trees until the end of the year’s growth.

Documentary data on recent floods and human activities were obtained from different technical reports (IGC, 2008; ICGC, 2010a,b, 2011 and 2013a,b; IGC et al., 2013). Historical data on the main rainfall and past flooding were obtained from local witnesses, archives (mainly from the Arxiu Comarcal del Pallars Sobirà), newspaper libraries, and the work by Balasch, et al. 2008. The PREDIFLOOD (Barriendos, et al., 2014) and INUNGAMA flood databases (Llasat, et al., 2013), handled and managed by M. Barriendos and C. Llasat (Universitat de Barcelona), respectively, were also consulted. Images provided by the company in charge of the Mal Pas hydroelectric infrastructure (https://ca.wikipedia.org/wiki/Central_hidroel%C3%A8ctrica_de_Mal_Pas) helped us to analyze the dynamics of the torrential flows that affected the Portainé stream in 2008 and 2010 (Garcia-Oteyza, et al., 2015; Génova, et al., 2018; Victoriano, et al., 2018b). The collection of photographs belonging to the La Pobla de Segur Fire station, showing the impacts of the 1982 floods, were used to corroborate the high energy of the event and its destructive capacity.

The detailed geomorphological mapping, together with the positions of the affected trees at the mouth of the Portainé stream to the Noguera Pallaresa River, were presented and analyzed by Victoriano, et al., (2018b), who related the different geomorphologic units with the relative position of the damaged trees and the hydraulics of the floods. The results obtained have been completed and compared with those of the Flamisell study area. Vertical aerial photographs and multi-temporal orthophotographs (period 1946-2018) were collected in both the Flamisell and Romadriu study areas. Photointerpretation was carried out to detect changes due to fluvio-torrential dynamics using criteria consisting of tree mass density, coloration, and texture (indicative of sediment transported and deposited shortly before the photograph was taken, etc.). The 1982 vertical aerial photographs, obtained one month after the flood, provide only limited information due to the shadows in the narrower and deeper valleys. Although it was not possible to interpret the entire valley floor of the Romadriu River, the photointerpretation of the shadow-free sections provided relevant information regarding the effects of the flood (see Results section).

Although there are numerous meteorological stations around the Flamisell and the Romadriu subcatchments, none of them are located within the Flamisell area and only those of Salòria and Portainé (in operation since 2011) are situated within the Romadriu subcatchment. Furthermore, many of them have short time series or considerable gaps. Figure 4 shows the stations from which data have been used in this work. Original data have been used (both from https://www.meteo.cat/observacions/xema, with point-selective daily data from June 2009 to the present, and from MeteoPirineu) and data cited in the works by Palau, et al., (2017), Trapero, et al., (2013a) and Bech, et al., (2011). Palau, et al., (2017) take precipitation values from various meteorological stations in a detailed study of the 21/Aug./2015 event. Bech, et al., (2011) defined isohyets from real data of 24 h accumulated precipitation in the regional 1-2/Nov./2008 event, while Trapero, et al. (2013a) performed simulations of the same event using the data from Bech, et al., (2011), considering 24 h of accumulated precipitation. From these two works, we selected the ranges of precipitation registered in the set of stations analyzed. In addition, monthly and annual daily maximum rainfall data have been obtained from the Anuaris de dades meteorològiques (Yearbooks of meteorological data) of the Catalan Meteorological Service (https://www.meteo.cat/wpweb/climatologia/serveis-i-dades-climatiques/anuaris-de-dades-meteorologiques/).

Meteorological data for the Portainé stream have been collected and analyzed by Victoriano, (2018) and Furdada, et al. (2020). The most significant data were selected for the present work and are presented in the following sections. Moreover, reports on “Singular weather events” (https://www.meteo.cat/wpweb/climatologia/butlletins-i-episodis-meteorologics/episodis-meteorologics/) were consulted and cross-checked with rainfall and gauging data whenever necessary or possible (due to short series, which do not cover the periods of interest or contain data gaps). For example, for the Flamisell, when no “Singular weather events” were found that could explain the damage to the vegetation (e.g. in the dendrogeomorphological year 2009-10, see section 4.3) the dates of rainfall events of interest (see section 4.3) were explored in the meteorological yearbooks. For these dates, using the Automatic Stations Map, the precipitation in each one of the stations was selected and compared. It should be noted that the significant rainfall events addressed by Bech, et al., (2011) and Trapero, et al., (2013a), and also described in the “Singular weather events”, are indeed well known, but their hydrological and derived geomorphological effects have not been studied before.

There are no gauging stations in the Romadriu subcatchment, so there are no discharge data for this subcatchment. In the Flamisell subcatchment, there are 3 gauging stations (Table 2).

We use data from station 9267 - Capdella (in operation since 1989) - as it provides the longest discharge series and is the station that contains the periods of the establishment of the riparian trees and the dendrogeomorphological years with detected damage on these trees (see Results section), as well as being relatively close to the sampling area (~18 km upstream). For this station, data are available on daily average discharges and monthly maximum instantaneous discharges (m³/s). Although no data exist for the following periods: 26/Nov./1990 - 18/Jan./1991; 19/Jul./2000 - 19/Aug./2000; 31/Jul/2002 - 05/Sep./2002; 01/Oct./2006 - 30/Sep./2007; 01/Oct./2008 - 28/Mar./2011 and while these data do not reflect the total daily discharges, they nevertheless indicate the days on which significant flooding occurred.

At station 9181 - La Pobla de Segur (in operation from 1965 to 1991) - approx. 9 km below the sampling area, at the confluence with the Noguera Pallaresa, there are no data from 01/Nov./1982 to 01/Oct./1983 due to the destruction of the gauging station caused by the November 1982 flood. The series ends when the period of establishment of riparian trees begins, so no useful data was available for interpreting the damage to the vegetation. In any case, the data of the monthly contributions in hm³ of the Flamisell were retrieved to obtain an order of magnitude of the contributions of this subcatchment. These data were compared with the 1982 Flamisell flood discharge reconstruction obtained by Balasch, et al., (2008). The 9274 - Sallente P.P. station - provided no useful data for the present work.

All the data obtained, both documentary and geomorphological, were unfortunately incomplete, and the rainfall and discharge data series have gaps or are short. Faced with these limitations, the methodology used is based on the compilation and chronological ordering of each type of data and, subsequently, on their comparison and joint analysis. Geomorphological and forest density changes in the sampled areas were analyzed from the middle of the 20^th century to the beginning of the 21^st using aerial photographs and orthophotos from different periods. They were then compared with documentary data on floods (using historical and testimonial data and photographs of the events) and cross-checked with the dendrogeomorphological years with the detected damage to trees. The few available instrumental series (meteorological and hydrological) were also compared to the establishment of the riparian forest, the documented torrential flood events, and the dated damage to the trees. All these analyses contribute to the understanding of the dynamics of subcatchments of different dimensions facing floods of different magnitudes. As Kondolf & Piegay (2003) suggest, working with the convergence of evidence between the various data sources and comparing multiple zones increases knowledge and takes the complexity of the systems into account. In our work, the areas under study consist of different spatial scales, have undergone their own geomorphological and vegetation colonization evolution, and present a particular response to torrential flooding.

110 trees of 11 different species (Acer campestre, Alnus glutinosa, Fraxinus excelsior, Juglans regia, Pinus sylvestris, Populus nigra, Populus tremula, Prunus avium, Quercus petraea, Salix caprea, and Tilia platyphyllos) are analyzed, including; a) 81 trees in the Romadriu subcatchment (9 trees in the Ramiosa stream, 15 trees in the Romadriu river, and the 57 trees in the Portainé stream already studied in our previous work; and b) 29 trees in the Flamisell subcatchment. Almost all of them are angiosperms belonging to very different families and with different types of xylem, from ring pore (e.g.: Quercus petraea, Fraxinus excelsior) to diffuse pore (e.g.: Alnus glutinosa, Tilia platyphyllos) or with intermediate characteristics (e.g.: Populus spp., Prunus avium). Among the sampled species, in which 4 or more trees are analyzed (Table 3), poplars (Populus nigra) and oaks (Quercus petraea) stand out as the thickest. It should be noted that, since the oaks have a slightly lower average perimeter, they can reach twice the age of the poplars. Quaking aspen (Populus tremula) almost doubled the perimeter and the tree-ring width mean of the common ash (Fraxinus excelsior), despite having the same estimated mean age, while the alders studied (Alnus glutinosa), which present perimeters similar to common ash, rarely exceeded twenty years of age (Table 3).

In the Flamisell subcatchment, both poplars and alders have the greatest average ring width (although the fact that these are comparatively younger trees must be taken into account), while in the Romadriu subcatchment, the poplar is also the species with the largest tree-ring width average.

Large differences are identified between the average and maximum ages obtained in the Romadriu subcatchment, compared to those obtained in the Flamisell (Table 3), and therefore also in terms of the dates of forest establishment. In the Romadriu, most of the trees belonging to the most predominant species (Populus tremula, Fraxinus excelsior, Populus nigra) that comprise the riparian forest were established in the 1950s or thereafter. Other trees were older, establishing themselves between the early 20^th century and the 1920s: oak (Quercus petraea), lime (Tilia platyphyllos), and walnut (Juglans regia); while only a few others, such as cherry trees (Prunus avium), were much younger. In contrast, the alders and poplars on the Flamisell riverbanks were much more recent, their establishment being estimated as in the 1990s.

Three tables summarizing significant meteorological data selected from different sources and with different criteria are presented below. Tables 5 and 6 present selected meteorological data recorded in the dendrochronological years 2007-08 and 2009-10 (when there was damage to the vegetation in both Romadriu and Flamisell subcatchments), and in the dendrochronological year 2014- 2015 (when damage to the vegetation of the Flamisell subcatchment was verified). The data correspond to the dates of rainfall events of interest, both because of the damage caused to the vegetation (known from technical reports), and because these events are considered significant on a regional scale as well as in the “Singular weather events” of METEOCAT, and consequently may have caused damage to the vegetation.

Table 5 includes selected rainfall data and its classification as local or regional events of the rainfall events of interest described above. Table 6 presents the rainfall comparison of the events of interest described above with the monthly and annual rainfall maximums of the network (Figure 4); it specifically shows: a) the maximum monthly rainfall in 24 h in the months of the dates of rainfall events of interest defined above; b) the maximum precipitation in 24 h of the year, indicating the month and day of occurrence; and c) the total accumulated precipitation in each one of the selected rainfall events. When the month in which the maximum precipitation in 24 h occurs coincides (or fails to coincide) with the dates of the rainfall events of interest, it is highlighted in dark blue or light blue, respectively.

Tables 5 and 6 highlight the large variability of rainfall in the different stations for each episode. Table 6 shows that only in the regional episode of November 1^st and 2^nd, 2008, was the maximum annual 24 h precipitation recorded at all stations (except Sort, where there was no record). Considerable variability is found in the other two regional events since while some stations register their maximum annual rainfall, others do not even register a monthly maximum. It is also evident that, in the four local events, half of the stations did not even register a maximum monthly precipitation in 24 h. This underlines the large variability of precipitation in these mountainous areas.

Table 7 presents the accumulated rainfall at Meteocat stations for three selected dates in 2010, when no “Singular weather events” were found that could explain the damage to the Flamisell vegetation in this dendrogeomorphological year. As explained in the methodology section, the days on which monthly maximum precipitation occurred in each of the weather stations in Table 6 were explored in the meteorological yearbooks and the precipitation in each of the stations presented in Table 7 was obtained from the Automatic Stations Map. This allowed us to discard the rainfall event of 02/Nov./2010 and to consider 09/June/2010 as the one that most likely caused the flood and the damage.

The existing discharge data (average daily discharge and maximum instantaneous monthly discharge peaks) at gauging station 9267-Capdella (Flamisell subcatchment) are presented below. Figure 11 shows the comparison of these data with the establishment riparian forest period and with the years of damage to trees, and the dates of some notable peak flows were highlighted. There is no clear, univocal relationship among the peak discharges or some of the notable historical flows and the damage caused.

The monthly contributions in hm³ of the Flamisell at station 9181 - La Pobla de Segur (1965-1991) were obtained (Table 8). The contribution of the 1982 flood (16.4 hm³) was greater than the average monthly contribution, reinforcing the evidence of the great destructive power of this flood. On the other hand, as mentioned before, the weir of Senterada has a capacity of 0,12 hm³, which corresponds to a storage capacity of barely 0.24% of the maximum monthly contribution and is less than the minimum monthly contribution.

The first objective of a dendrogeomorphological study is to adequately select the sampling areas and species, a process that entails a series of limitations. In the search for areas where trees can provide a record of past floods, forested areas near riverbanks and streams are selected if they include externally damaged trees affected by floods. Various factors act on the evolution and development of riparian forests, but large flood events can destroy or partially eliminate them, with the loss of evidence of previous events and the subsequent establishment of a new forest of uniform age or “cohort” (Oliver & Larson, 1996). It can be especially significant in narrow and steep channels in mountainous areas. In fact, in these contexts, the flood events that are best recorded as damage to trees are those of intermediate magnitude, since the larger ones may destroy all the vegetation and the smaller ones leave little evidence (Ruiz-Villanueva, et al., 2010; Génova, et al., 2015).

In this sense, the estimated dates of the establishment of the trees in the different areas studied in the upper catchment of the Noguera Pallaresa River present great differences. In the Romadriu subcatchment and, more specifically, in the riparian forest of the Portainé stream, it was determined that most of the trees had progressively established themselves in the alluvial cone since the early 1950s (Génova, et al., 2018). During the previous decades, the western margin of the stream probably underwent intensive cultivation (remains of the ancient presence of fruit trees are still found today, García-Oteyza, et al., 2015). When agricultural activity ceased, the area was colonized by natural vegetation, as evidenced by the analysis of orthoimages from different years (Génova, et al., 2018). On the other hand, the maximum ages estimated in this work for the trees of the Flamisell subcatchment are much more recent, which indicates an almost simultaneous establishment in the 1990s (Table 3), and is, therefore, a clear indication of the great devastation caused by the flood of 1982.

The large flood of 1982 was the last of the three exceptionally destructive events that occurred in the Noguera Pallaresa during the 20^th century, in 1907, 1937, and 1982 (Balasch, et al., 2008), although no dendrogeomorphological evidence of the first two has been identified. The extensive devastation caused by the 1982 event has been documented, for example, in the Servei Geològic de Catalunya, (1983), in Corominas & Alonso, (1990), and in Balasch, et al., (2008). Documentary information indicates that the analyzed area of the Flamisell subcatchment was deforested when the weir was built in the 1920s (Figure 7); likewise, with a subsequent aerial photograph taken in 1956, and especially with further photographs of the 1982 flood up until the late 1980s (Figure 8), from when the establishment of the riparian forest started again. However, even though during this event the transport of materials in the Romadriu River was enormous, producing a ~10 m high dam that blocked its mouth in the Noguera Pallaresa (Figure 6), scars on trees established earlier in this subcatchment have not been dated. Most of these trees were over ten years old at that time (Table 3) and already constituted a more or less homogeneous forest in the Portainé stream, and they did show other evidence (suppressions and releases) affected growth from the following year in more than half of the trees analyzed (Génova, et al., 2018), but not in the other studied areas of the Romadriu basin.

In the dendrogeomorphological dating of flood events, we use the proposal put forward by Génova, et al. (2018); i.e. a biannual denomination that corresponds to the dendrogeomorphological year. This biannual dating shows that the event, and therefore the damage caused, may have occurred from the beginning of the dormant period of the trees, which in these latitudes approximately coincides with the end of the summer period, and up until the end of the annual growth. Furthermore, this establishes a coincidence with the hydrological year in Spain, which covers the period that runs from October 1^st to September 30^th of the following year.

Numerous dendrogeomorphological indicators have made it possible to date important events in the Romadriu subcatchment (Table 4), to which must be added an event that occurred in 1969-70, when many trees in the Portainé stream may have been decapitated (Génova, et al., 2018). The new data presented here do not modify the dates of the evidence gathered from the Portainé stream but increase its amount and reliability, as in the case of the dendrogeomorphological year 1997-98 (Table 4). Among all these indicators, those dated 1997-98, 2007-08, and 2009-10 stand out, due to the greater abundance of recorded evidence, especially scars. While some coincident damages were identified in the Flamisell subcatchment, the greatest correspond to the dendrogeomorphological year 2014-15, when up to 72 % of the trees analyzed presented wounds and bark removal that in some cases reached up to a height of 2 m (Figure 2).

The coincidence of damage in the same dendrogeomorphological year in both subcatchments (Flamisell and Romadriu) does not necessarily imply that they were the consequence of the same precipitation episode that generated simultaneous floods, but may correspond to different events. As already indicated in the previous section, the dendrogeomorphological record of flood events entails limitations, since there does not necessarily have to be a direct correlation between the confirmed dates of the major events and the number and extent of dendrogeomorphological evidence (Ruiz-Villanueva, et al., 2010). Thus, for example, the effects of successive events can be masked when two floods occur in the same dendrogeomorphological year, or when the later event of greater magnitude destroys all the dendrogeomorphological evidence of a less intense previous event (Ballesteros-Cánovas, et al., 2013). Moreover, the most recent damages (which occurred one or two years before) are very difficult to date, because the callus still cannot be properly identified (Génova, et al., 2018).

Another limitation is related to the impossibility of developing reference chronologies to contrast the undisturbed and disturbed tree-ring series, due to the lack of sufficient undisturbed trees in the studied sites. This has been offset by with the contrast of the data on growth anomalies between different species with different ecological ranges. Therefore, whenever possible, it will be more explanatory as well to use temporally detailed documentary and instrumental data that can be related to dendrogeomorphological evidence.

When contrasting dendrogeomorphological evidence and data from instrumental records in mountainous areas, it is necessary to take into account that the rainfall in those areas is strongly conditioned by the orography, which generates high-intensity rainfall nuclei, while in neighboring areas the rainfall may be much lower. In the November 1982 event, the inhabitants of the Romadriu Valley explained that the storm stagnated and precipitated directly, hitting the south-facing slope (opposite to the slope where the Portainé and Ramiosa streams are located). This matches the work by Trapero, et al., (2013b), in which simulated rainfall at the head of the Romadriu was much higher than in the small streams of Portainé and Ramiosa. The extreme rainfall of 1982 at the head of the Romadriu subcatchment produced the erosion and transport of readily available sediment, which generated the aforementioned obstruction at the confluence with the Noguera Pallaresa. On the other hand, in the case of the Portainé and Ramiosa torrents, only local erosions on the forested banks may have occurred, as well as the possible uprooting and transport of some specimens, consequently giving rise to suppressions and releases, but without enough damage to enable detection in the current riparian forest.

The dendrogeomorphological evidence of 1997-1998 in Portainé (Génova, et al., 2018) and in other areas of the Romadriu subcatchment are correlated with significant regional rainfall and flooding. In the Flamisell subcatchment, gauging station 9267 - Capdella recorded a maximum daily instantaneous flow on 18/Dec./1997, although no dendrogeomorphological evidence of this exists. Curiously, it was during the establishment of poplars and alders in the sampled area (1992-1999) when the highest flow peaks of the series occurred, including that of 1997 (Figure 11). Given their youth, shortness, and flexibility, it is likely that the trees leaned against possible floaters, so it is not possible for damage or other evidence to be identified.

In the small Portainé subcatchment, documentary data made it possible to determine in detail the exact date of recent local floods (García-Oteyza, et al., 2015; Génova, et al., 2018; Furdada, et al., 2020). In 2008 and 2010, two floods occurred each year, which were recorded in the dendrogeomorphological years 2007-08 and 2009-10. In the Flamisell subcatchment, the documentary data (Figures 10 and 11) do not clearly coincide with a notable flow peak at the gauging station, although a certain increase in discharge was recorded in spring. Furthermore, the meteorological event of 1-2/Nov./2008 was of a regional nature (Bech, et al., 2011; Trapero, et al., 2013a) and may have produced a flood peak, although no discharge data is available. Therefore, the wounds in the Flamisell sampling area could have been caused by at least one of these two events or by both, and it is possible that the regional event affected both subcatchments (Romadriu and Flamisell). Damage was caused to the trees of Romadriu and Flamisell in the dendrogeomorphological year 2009-10. The meteorological events that affected Portainé were of a local nature, while the damage to the Flamisell vegetation could be explained mostly by the regional rainfall on 06/09/2010 (Table 7). From 2010 to 2015, the rainfalls recorded in the Romadriu subcatchment were not extraordinary. The local precipitations occurred in July 2011 and 2013 and in August 2014 and 2015 (less than a 10-year return period, although very intense (Furdada, et al., 2020), causing debris flows in the Portainé stream (ICGC, 2013a, b; IGC et al., 2013; Palau, et al., 2017), and some evidence of damage was observed in September 2015 (Génova, et al., 2018).

On 17-18/June/2013, there were significant rainfalls and snowmelt in the Val d’Aran and the headwaters of the Noguera Pallaresa and Flamisell, which caused significant flooding in the Val d’Aran (Victoriano, et al., 2016). At gauging station 9267 - Capdella, a significant flow peak was observed, higher than those of 2007, 2008, 2011, and 2015, although no flooding is documented for that occasion. In the sampling area (18 km downstream) there were no tree damages, possibly due to the attenuation of the flood. Conversely, in those other years, the precipitations may possibly have affected a significantly greater area of the subcatchment and the flow increased downstream, thereby causing damage to the trees.

The “Singular weather events” of 27/Nov.-01/Dec./2014 and 10-25/March/2015 could have caused the serious damage detected in the Flamisell forest (the dendrogeomorphological year 2014-15). The gauging station data indicate slight but unremarkable increases in flow. In both events, rainfall data presented maxima at the stations located to the north and west of the subcatchment (Tables 5 and 6). The contribution of flow from the Bòssia River (the tributary on the right bank of the Flamisell that drains ~20 % of the subcatchment; Figure 1) may have led to a significant increase in discharge but went unrecorded at the gauging station because it was located upstream from the confluence.

Despite the floods that caused wounds and other damage between 2008 and 2015 in the forest sampled in the Flamisell subcatchment, the vegetation has since then been densifying, which implies that: 1) the floods that caused the damage were of magnitudes less than that of 1982 when the riverside forest was devastated; 2) the weir has no regulatory role in or influence on these floods of intermediate magnitudes (nor, of course, in extreme ones), and its lamination capacity when a flood occurs is practically nil. In consequence, it does not prevent injuries or other evidence on trees.

Furthermore, it should be noted that, in general, the data from the 9267 - Capdella gauging station provide little or no useful data for interpreting the damage to the vegetation in the sampling area. None of the notable flow peaks indicated in Figure 11, except those of 25 and 27/May/2008, seem to correspond to the damage recorded. This is most probably because an extensive area of the subcatchment is downstream from the station, so the events that affect the headwaters are attenuated downstream and those that are regional and affect the entire subcatchment are underestimated. In the sampling zone, which is close to the mouth, dendrogeomorphological damage is recorded when extraordinary floods occur, but in the case of extreme floods, such as that of 1982, the evidence is destroyed. In addition, the information of a meteorological nature entails the limitation of the location of the stations. As commented above, the orography of the area gives rise to a highly heterogeneous rainfall pattern (Tables 5 and 6). Thus, the data from stations outside the catchment of interest provide only a general idea of the events of a regional nature (and not always), and if they are of a local nature they may not be recorded. If the catchment is very small, as is the case of the Portainé, the best indicator of its dynamics is the dendrogeomorphological data; a rain gauge will only provide really relevant data when it is located at the headwaters. If the subcatchment is larger, as is the case of Flamisell, the local events or those that affect only part of the subcatchment may generate attenuated flows downstream.

The wide and diverse set of data analyzed enables a more comprehensive knowledge of torrential flood dynamics in both subcatchments, which respond to a very inhomogeneous mountain rainfall events. Moreover, the documentary data, both scientific and testimonial, reinforce, complete, and outline the interpretation of the dendrogeomorphological results.