Intensity-duration-frequency (IDF) curves usefully quantify extreme precipitation over various durations and return periods for engineering design. Unfortunately, sparse, infrequent, or short observations hinder the creation of robust IDF curves in many locations. This paper presents the first global, multi-temporal (1-360 h) dataset of generalized extreme value (GEV) parameters at 31 km resolution dubbed PXR-2 (Parametrized eXtreme Rain). Using these data we generalize site-specific studies to show that that GEV parameters typically scale robustly with event duration (r 2 >0.88). Thus, we propose a universal IDF formula that allows estimates of rainfall intensity for a continuous range of durations (PXR-4). This parameter scaling property opens the door to estimating sub-daily IDF from daily records. We evaluate this characteristic for selected global cities and a high-density rain gauge network in the United Kingdom. We find that intensities estimated with PXR-4 are within ±20% of PXR-2 for durations ranging between 2 and 360 h. PXR is immediately usable by earth scientists studying global precipitation extremes and a promising proof-of-concept for engineers designing infrastructure in data-scarce regions.

Digital elevation models (DEM) are fundamental for hydrologic and hydraulic modelling. Many practitioners rely on open-access global data sets due to the cost and sparse coverage of sources of higher resolution. In 2016, the Japanese Aerospace Exploration Agency released the ALOS World 3D-30m (AW3D30), an open-access global elevation model at an horizontal resolution of 30 m. So far no published study has done an assessment of the flood modelling capabilities of this new product. The purpose of this investigation is to (a) assess the utility of the AW3D30 for flood modelling purposes and (b) compare its performance with regards to computed water levels and flood extent maps calculated using other freely available 30m DEM (e.g., SRTM and ASTER). For this comparison, the reference to reality is given by the maps computed using a light detection and ranging (LiDAR)-based digital terrain model. This study is carried out in two catchments with contrasting topographic gradients. Results show that AW3D30 performs better than the SRTM. In mountainous regions, the results derived with the AW3D30 are comparable in skill to those obtained with a LiDAR-derived digital surface model. This encouraging performance paves the way to more accurate modelling for both data-scarce regions and global flood models. K E Y W O R D S ALOS, AW3D30, digital elevation model, flood

Abstract. Worldwide, floods are acknowledged as one of the most destructive hazards. In human-dominated environments, their negative impacts are ascribed not only to the increase in frequency and intensity of floods but also to a strong feedback between the hydrological cycle and anthropogenic development. In order to advance a more comprehensive understanding of this complex interaction, this paper presents the development of a new open-source tool named "Itzï" that enables the 2-D numerical modelling of rainfall-runoff processes and surface flows integrated with the open-source geographic information system (GIS) software known as GRASS. Therefore, it takes advantage of the ability given by GIS environments to handle datasets with variations in both temporal and spatial resolutions. Furthermore, the presented numerical tool can handle datasets from different sources with varied spatial resolutions, facilitating the preparation and management of input and forcing data. This ability reduces the preprocessing time usually required by other models. Itzï uses a simplified form of the shallow water equations, the damped partial inertia equation, for the resolution of surface flows, and the Green-Ampt model for the infiltration. The source code is now publicly available online, along with complete documentation. The numerical model is verified against three different tests cases: firstly, a comparison with an analytic solution of the shallow water equations is introduced; secondly, a hypothetical flooding event in an urban area is implemented, where results are compared to those from an established model using a similar approach; and lastly, the reproduction of a real inundation event that occurred in the city of Kingston upon Hull, UK, in June 2007, is presented. The numerical approach proved its ability at reproducing the analytic and synthetic test cases. Moreover, simulation results of the real flood event showed its suitability at identifying areas affected by flooding, which were verified against those recorded after the event by local authorities.