HS7.8

Due to the limited length of locally available sequences of precipitation extremes, estimates of design rainstorms at a given location (i.e. point rainfall depth associated with given durations and non-exceedance probabilities) are traditionally obtained from regional frequency analysis. Several statistical regionalization methods proposed in the literature enable one to exploit sequences of precipitation extremes observed at a number of sites that supposedly share the same frequency regime of rainfall extremes with the site of interest (herein also referred to as a homogeneous pooling group of sites). Homogeneous pooling groups of sites can be identified by looking at specific climatic descriptors; for instance, some reliable authors successfully utilize Mean Annual Precipitation (MAP) as the sole proxy for locally characterizing the frequency regime of sub-daily rainfall extremes and for grouping sequences of rainfall extremes records. We aim at advancing this traditional approach (1) by relaxing the hypothesis of the existence of a homogeneous pooling group of sites characterized by a unique regional parent distribution and (2) by incorporating additional morphological and climatic information in the regional model. Our research focuses on a large study area in Northern Italy, counting more than 2350 Annual Maximum Series of rainfall depth for different time-aggregation intervals between 1 and 24 hours, that have been collected between 1928 and 2011 in the Italian Rainfall Extreme Dataset (I2-RED).  We refer to local MAP value as well as to several other morphologic descriptors (e.g. minimum distance to the coast, elevation of orographic barriers, aspect, terrain slope, etc.) for characterizing the frequency regime of sub-daily rainfall extremes. We train a probabilistic neural network that uses the descriptors cited above as input layers for modeling the local frequency regime of observed rainfall annual maxima. We resort to a Generalized Extreme Value (GEV) distribution whose parameters are data-driven functions of the local morphoclimatic descriptors as well as the time-aggregation interval. We then perform a series of cross-validation experiments targeted at assessing the accuracy of the developed data-driven regional frequency model relative to a simpler regional model in which GEV parameters are functions of MAP and time aggregation intervals.

Our results address the following research problems: (a) identification of the most descriptive morphological proxies for representing the frequency regime of sub-daily rainfall extremes, (b) assessment of potential, limitations, and robustness of data-driven multivariate regional frequency models of sub-daily rainfall extremes relative to simpler and more traditional regionalization schemes.

How to cite: Magnini, A., Lombardi, M., Valtancoli, E., and Castellarin, A.: Regional predictions of sub-daily rainfall extremes through data-driven blends of morphoclimatic descriptors, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-570, https://doi.org/10.5194/egusphere-egu22-570, 2022.

Precipitation extremes are space-time phenomena and traditionally the statistical analyses on such occurrences have treated them merely as point events. Many of the consequences of such events like floods are related to the water volume, hence the spatial aspect of them cannot be neglected. This work aims to bring the areal aspect of the extreme rainfall into play by introducing the area into the Extreme Value Analysis (EVA) and providing Area-Depth-Duration-Frequency (ADDF) curves. For this purpose, different spatial rainfall products have been used and compared with each other. Processed raw radar data, a product of conditional merging of the radar and station data as well as the RADKLIM data (a product of the German Weather Service designed for climate research) have been used for the EVA. Unexpected patterns have been observed in the ADDF curves based on the processed radar data which were not in agreement with the assumptions of the classical approach of areal reduction factor.  Usually, it is assumed that areal precipitation extremes increase with decreasing area, so in practice reduction factors are used to estimate areal precipitation extreme values from point observations. This behavior was observed as expected mostly for durations shorter than 2 hours in all the study locations whereas the opposite was present for longer durations, where the precipitation is increasing with increasing area, so that the ADDF curves, representing different areas, show crossings at these durations. Different hypotheses about the reason for the crossings like seasonality and spatial non-stationarity have been tested and did not explain the crossings. On the other hand, the ADDF curves of the merged rainfall product hardly showed such patterns and followed the classical assumptions. Therefore, the appearance of such crossings in the ADDF curves of a spatial rain product might be an indicator of artifacts in the radar rainfall product. It has to be investigated in further tests if these results hold and if these crossings could be used as an indicator for unplausible radar data.  

How to cite: Goshtsasbpour, G., Haberlandt, U., El Hachem, A., Seidel, J., and Bárdossy, A.: Statistical Analysis of Space-times Dynamics of Extreme Precipitation, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-2738, https://doi.org/10.5194/egusphere-egu22-2738, 2022.

Rainfall intensity-duration-frequency (IDF) curves are required for the design of several water systems and protection works. Typically, long (more than 40 years) station data are employed first to generate annual extremes (AMS) for different durations and then to fit a GEV probability distribution. Since station data are only point measurements, regionalization techniques are applied to estimate IDF curves at ungauged locations.  Prior results revealed that the best way to obtain IDF maps for Germany was kriging interpolation of parameters from very long stations with the parameters of the short stations acting as an external drift. However, how certain the obtained IDFs values are, and how to derive the uncertainty range at each location, remain still unanswered. Therefore, it is the objective of this study, to investigate the propagation of uncertainty in the regionalization of the IDF curves for Germany.

For this purpose, the available station data from the German Weather Service (DWD) for whole Germany are employed, which includes; 1100 sub-hourly (5min) recordings with observations period shorter than 20 years, and 89 sub-hourly (5min) recordings with 60-70 years of observations. Annual extremes are extracted at each location for different durations (from 5mins up to 7days), and local IDF curves are estimated according to Koutsoyiannis et al. (1998). The parameters of the obtained IDF functions are then interpolated using external drift kriging. Finally, quantiles are derived for each location, duration and given return period (Ta=2, 10, 20, 50 and 100 years). Through a non-parametric bootstrap, the uncertainty is estimated for three different components of the regionalization: i) local estimation of parameters, ii) variogram estimation and iii) spatial sampling distribution.  Simulated annealing is employed to ensure that the spatial resampling of locations represents the obtained variograms. The final uncertainty range is then considered as the 95% confidence interval of the obtained IDF curve for each location, duration and return period.      

The results of this study reveal how the uncertainty of annual rainfall extremes propagates from local estimation to the regionalization of IDF curves based on kriging theory. The comparison of the three components will shed light to the following questions: Which is the contribution of each component to the final uncertainty range of IDFs curve? How is the uncertainty range changing based on different durations and return periods? Are there any spatial trends in Germany regarding the uncertainty range of IDFs curves? 

How to cite: Shehu, B. and Haberlandt, U.: Uncertainty analysis of regionalized intensity-duration-frequency curves in Germany, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-987, https://doi.org/10.5194/egusphere-egu22-987, 2022.

Long term time series of meteorologic variables are generally lacking and is especially the case at sub-daily temporal resolutions. These time series are needed for applications such as hydrological modelling of catchments and derived flood frequency analyses. Stochastic weather generators are one such solution and are able to generate long synthetic time series of arbitrary length.

This study explores the coupling of an hourly space-time rainfall model with a non-parametric K-NN resampling approach for non-rainfall climate variables such as temperature, humidity and global radiation. A daily gridded observational climate dataset is used for the resampling. As a last step, a simple disaggregation technique is applied to the resampled non-rainfall climate variables to achieve an hourly timestep.

To study the effectiveness and performance of the hybrid weather generator, synthetic time series for 400 mesoscale catchments within Germany consisting of 700 rainfall stations were generated and compared to observations. Results show that the hybrid weather generator adequately reproduces observed statistics for rainfall and the non-rainfall climate variables in addition to maintaining cross correlations between the climate variables.

How to cite: Pidoto, R. and Haberlandt, U.: A multi-scale space-time hybrid weather generator, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-6720, https://doi.org/10.5194/egusphere-egu22-6720, 2022.