**1. Introduction**

Extreme rainfall resulting in flood events is a common phenomenon in different environments, including both mountain ranges and lowlands [1,2]. The systematic monitoring of flow stages at gauging stations, including precipitation and flow discharge prediction, is currently well applied and documented in many medium- and large-sized rivers [3–5]. In contrast, data from mountain steep headwater catchments and, in general, first-order catchments [6] are still poor, due to the insufficient network of stream gauges and the sporadic amount of processed documentary evidence [7]. Not only do mountain headwater streams generate sediment-laden flows with aftermaths within and along fans [8,9], but also streams and gullies of first-order catchments (up to 10 km2) in moderate relief can be responsible for local damage to infrastructure. Ozturk et al. [10] analysed extraordinary flash flood events (140 mm per 2 h) in a small catchment (6 km2) that resulted in damage to infrastructure and a high amount of suspended sediments (t/km2) due to intense hillslope–channel coupling. In addition, Terti et al. [11] pointed to a short response time of small catchments to flash floods, thereby increasing the probability of trapping people during outdoor activities.

Despite the lack of gauging records from forested first-order catchments, several approaches exist to describe the hydrogeomorphic impacts of extremely high flow stages. The dating of flood scars on riparian vegetation using dendrogeomorphic methods [12,13] and recording their position and height above the channel bottom with a combination of hydraulic models is a well-established approach for the estimation of peak flood discharge, flow velocity, and unit stream power [14]. Using flood scars as a palaeostage indicator (PSI; maximum height of the scar above the channel bottom) allowed the interpretation of past flood events in medium- and large-scale rivers (catchment areas larger than 40 km2), for example, in the Western Mediterranean [15], North America [16,17], the Carpathians [18], and the Himalayas [19]. The dendrogeomorphic response of flash floods in small catchments is generally considered lower [20] but may increase due to the presence of erodible sediments amplifying lateral bank erosion. In such conditions, while small streams may not generate flood waves as large as those of large rivers, the presence of exposed and scarred tree roots [21] may complete a relatively low number of scarred tree stems.

The alluvial streams draining first-order catchments often do not display clear relationships between channel geometry, bed substrate, unit stream power and drainage area, and their resulting form and evolution trajectory are unpredictable unless local conditions (e.g., bedrock resistance, intensity of hillslope-channel coupling processes, land use history, and presence of large instream wood) are constrained [22–26]. These streams are characterized by a more or less developed stepped-bed morphology with a wide range of sediment size, where individual steps controlling channel bed stability consist of interlocked boulders, bedrock outcrops or large wood pieces [27]. This implies the relative stability of their channel beds under relatively high discharges (up to floods of 20–50-year recurrence intervals) and thus only limited adjustments of the channel morphology and geometry to lower (e.g., bankfull) flows owing to the presence of generally shallow flows, particle-size interactions (hide/protrusion effect) and additional bed form resistance [25,28–31]. The correlations between the unit stream power, sediment calibre and prevailing fluvial process may exist at the reach scale when spatially limited depositional reaches can be accompanied by local bed sediment fining and an abrupt decrease in the unit stream power of a high-magnitude flood [23]. Nevertheless, our knowledge of the direct relationships between the transport capacity of a particular flood event, bed stability and the resulting fluvial processes in first-order catchments is still somewhat limited by the lack of detailed post-flood field surveys of geomorphic consequences.

As mentioned, two-dimensional hydraulic models have been successfully applied to peak discharge reconstructions in ungauged or poorly gauged catchments of various sizes and environments [18,32–35]. This peak discharge is mostly defined not as the single deterministic value but rather as a range of values due to uncertainties inherent in the process of the estimation [36]. Discharges predicted in such a way could be used in flood frequency analyses [18], where they serve as outliers to the measured discharges. Consequently, the predicted discharges could serve as a basis for the evaluation and mitigation of flood risk [34] or as information about the flood magnitude in the historical period [32]. Moreover, the results of hydraulic simulations can be used for the estimation of stream transport capacity and channel stability during a particular flood event at a very detailed scale. In this sense, the bed shear stress and unit stream power are relevant parameters for calculating the incipient motion of coarse bed particles and thus evaluating the stability of stepped-bed channels consisting of relatively stable cobble to boulder steps [30,37].

For our purposes, we selected a first-order catchment (1.8 km2) in the moderate relief of the Eastern Sudetes (NE Czechia), where, during the last 15 years, the occurrence of several flash flood events caused substantial geomorphic imprints [38]. In particular, the last intense precipitation event (27 May 2014) resulted in a moderate flood risk in the case of medium-sized rivers (2-year recurrence interval), but a discharge of an approximately 100-year recurrence interval was estimated at ungauged small streams. The short-lived storm, with a total rainfall amount of 40 mm (locally up to 80 mm), had an intensity of between 40 and 60 mm/h. Its hydrogeomorphic response was primarily due to

unprecedented rainfall intensity and large antecedent precipitation. The financial costs of the flash flood aftermaths in the affected region were calculated as approximately EUR 200,000 [39].

Our aims were to (i) create the chronology of the flash flood events in this small catchment using dendrogeomorphic approaches, (ii) estimate the parameters (peak flow discharge, flow velocity, bed shear stress, and unit stream power) of the last 2014 flash flood event using the combination of PSI and 2D hydraulic modelling, and (iii) describe the stream transport capacity and channel stability during the 2014 flash flood event based on the hydraulic simulation data and sedimentological parameters. Such a comprehensive approach may help to better quantify the flash flood parameters of ungauged streams and thus contribute to more reliable management of small streams in future extreme climate events.
