HS7.7

Various natural phenomena, such as precipitation, generally exhibit spatial extremal dependence at short distances only, while the dependence usually fades away as the distance between sites increases arbitrarily. However, the available models proposed in the literature for spatial extremes, which are based on max-stable or Pareto processes or comparatively less computationally demanding "sub-asymptotic" models based on Gaussian location and/or scale mixtures, generally assume that spatial extremal dependence persists across the entire spatial domain. This is a clear limitation when modeling extremes over large geographical domains, but surprisingly, it has been mostly overlooked in the literature. In this paper, we develop a more realistic Bayesian framework based on a novel Gaussian scale mixture model, where the Gaussian process component is defined by a stochastic partial differential equation that yields a sparse precision matrix, and the random scale component is modeled as a low-rank Pareto-tailed or Weibull-tailed spatial process determined by compactly supported basis functions. We show that our proposed model is approximately tail-stationary despite its non-stationary construction in terms of basis functions, and we demonstrate that it can capture a wide range of extremal dependence structures as a function of distance. Furthermore, the inherently sparse structure of our spatial model allows fast Bayesian computations, even in high spatial dimensions, based on a customized Markov chain Monte Carlo algorithm, which prioritize calibration in the tail. In our application, we fit our model to analyze heavy monsoon rainfall data in Bangladesh. Our study indicates that the proposed model outperforms some natural alternatives, and that the model fits precipitation extremes satisfactorily well. Finally, we use the fitted model to draw inferences on long-term return levels for marginal precipitation at each site, and for spatial aggregates.

How to cite: Huser, R., Hazra, A., and Bolin, D.: Realistic and Fast Modeling of Spatial Extremes over Large Geographical Domains, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-6595, https://doi.org/10.5194/egusphere-egu22-6595, 2022.

Recent developments in Extreme Value Theory have led to the adoption of multivariate methods for modeling extreme rainfall in a way that the spatial dependence between the different measuring stations is used to "borrow“ information. A commonly used method, Brown-Resnick Max-Stable processes, extends the geostatistical concept of the variogram to suit block maxima, allowing to explicitly model the spatial extremal dependence shown by the data. This extremal dependence usually stems from physical processes that generate rainfall in such a way that several stations are affected simultaneously by the same extreme event, such as convective storms or frontal events. Depending on the region, this dependence can change in time, as different meteorological processes dominate the rainfall generation process for different seasons.

In this study, we analyze in the Berlin-Brandenburg region the change in extremal dependence for annual block maxima. We consider two different seasons – winter and summer – to investigate the effects of two different rainfall generating processes: frontal rainfall is more likely to occur in winter, while convection is dominating in summer. Furthermore, we investigate how this extremal dependence affects the accuracy of the estimation of return levels by using a Brown-Resnick Max-Stable process and comparing the estimated return levels to the results of a covariates model assuming spatial independence. We obtain the uncertainty of our estimates within a Bayesian modeling framework. The bivariate extremal coefficient shows a notable difference in the extremal dependence for summer and winter. Moreover, we observe a difference in the skill of the model when comparing the two seasons, suggesting that the difference in the extremal dependence has an impact on the marginal estimates from the model.

How to cite: Jurado, O. E. and Rust, H. W.: Estimating the impact of seasonal extremal dependence with a Max-Stable process for modeling extreme precipitation events over Berlin-Brandenburg, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-3660, https://doi.org/10.5194/egusphere-egu22-3660, 2022.

Re-gridding considerably alters precipitation statistics. Despite this fact, regridding precipitation datasets is commonly performed for coupling or comparing different models/datasets. In general, several studies have highlighted the effects of regridding at regional scale. In this study, the effects of re-gridding precipitation are emphasized at a global scale using different regridding methods, size of the shifts and resolutions of the dataset. Substantial differences are noted at high quantiles and precipitation dry (or wet-dry frequency) is altered to a great extent. Specifically, a difference of 46 mm in high (0.95) quantiles and a reduction of 30% wet-dry frequency is noted. The differences increase with the size of the grid shift at higher quantiles and vice versa for low quantiles. As the grid resolution increases, the difference between original and regridded data declines, yet the shift size dominates for high quantiles for which the differences are higher. Spatially, large differences at high quantiles in tropical land regions, and at low quantiles in polar regions are noted. These impacts are approximately same for the three different (first order conservative, bilinear, and distance weighted averaging) regridding methods considered in this study. Overall, re-gridding should be performed with caution as it can alter the statistical properties of precipitation to a great extent and adds uncertainty to further analysis of using in any models or in combined precipitation products.

How to cite: Rajulapati, C. R., Papalexiou, S. M., Clark, M. P., and Pomeroy, J. W.: Precipitation regridding – Impacts at global scale, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-5970, https://doi.org/10.5194/egusphere-egu22-5970, 2022.

Investigation of precipitation extremes is traditionally based on point observations. Such rain gauges networks  often have an insufficient network density to  correctly capture the spatial extent of extreme events. An alternative is to use weather radar data which provide a spatially distributed rainfall field but these observations are prone to errors. To reduce the errors in the radar observations, a copula-based merging procedure is applied to combine radar and station observations with high temporal resolution. From this product, the spatial extent of extremes in investigated. This is done by extracting the connected rainfall areas from every rainfall field for several precipitation thresholds and temporal aggregations. The location, size, station data whithin these areas, areal mean precipitation value, and the areal maximum precipitation value are gathered and investigated.

This procedure was applied to the area covered by the German Weather Service (DWD) radar in Hannover with a 5 minutes temporal resolution and for the period 2000-2019. The first results of this investigation shows that station observations underestimate the true areal maxima in most of the cases. Moreover, the connected areas are categorized based on their size and the areal mean precipitation values are compared. It was found that with increasing area size the corresponding areal mean increased. This was observed until a certain area size is reached after which the areal mean almost stabilizes. A clustering of the continuous areas revealed that the occurrence of the areas is independent of the location and that extreme observation can occur anywhere within the study region.

How to cite: El Hachem, A., Bárdossy, A., Seidel, J., Goshtasbpour, G., and Haberlandt, U.: Investigating the Spatial Extent of Extreme Precipitation, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-9725, https://doi.org/10.5194/egusphere-egu22-9725, 2022.

Changes in the frequency and intensity of extreme precipitation resulting from climate change are responsible for natural disasters such as severe floods and have been a major study focus during the last decades. Previous studies have mainly focused on the trends of annual maxima precipitation at global and regional scales. However, little is known about how extreme precipitation trends change among different climate types. This study offers a global analysis of extreme precipitation changes in terms of climate type by using over 8500 gauge-based records. We focus on the period 1964 to 2013 when global warming was accelerating. A climate type is assigned to each station based on the Köppen Geiger (KG) climate classification, resulting in 30 KG climate subtypes. Mann-Kendall test and Sen’s slope estimator are applied to each time series, measuring the magnitude and significance of trends. The heaviness of the tail for each station is assessed based on the shape parameter of the Generalized Extreme Value distribution. Our results indicate a decreasing trend for the majority of stations associated with some of the arid, temperate, and continental subtypes (i.e., hot semi-arid (BSh); hot-summer temperate (Csa); warm-summer temperate (Csb); and warm, dry-summer continental (Dsb)). An increasing trend is observed for the stations associated with the remaining KG subtypes, especially stations associated with dry-summer subarctic (Dsc) and monsoon-influenced extremely cold subarctic (Dwd). A significant increasing trend is estimated for 9.7% of stations located in the eastern USA, Asia, and northern Europe. However, only 2% of stations, mainly in eastern Australia and the central USA have a significant decreasing trend. The heaviness of the tail is the largest in the Polar major climate type (E), followed by Tropical (A), Dry (B), Continental (D), and Temperate (C). For the climate subtypes, large heavy-tailed extremes are observed in extremely cold subarctic (Dfd), polar tundra (ET), and tropical monsoon (Am), while only light-tailed extremes were observed in subpolar oceanic (Cfc). This study reveals the relationship of extreme precipitation characteristics (e.g., tail heaviness and trend) with the climate types at the global scale.

How to cite: Hobbi, S., Nerantzaki, S., Papalexiou, S. M., and Rajulapati, C. R.: Global analysis of extreme precipitation changes in the Köppen-Geiger climate classification, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-10335, https://doi.org/10.5194/egusphere-egu22-10335, 2022.

Evaluating historical extreme flood events is fundamental due to their socioeconomic impacts. In this context, the spatial distribution of the event has a key role and the univariate approach, based on the analysis of local flood frequency on a single site, is not the proper one.

For this reason, in the recent past, an increasing amount of research has focused on the regional characterization of flood events, trying to describe their temporal and spatial distribution. The main objective of this work is the comparison of two different methods widely used for the selection and characterization of spatially distributed flood events for risk assessment purposes. Both methods were applied to the Italian territory and compared in terms of parameters used, results obtained, and technical analogies and differences. 

The two methodologies reveal similar results, comparable with a list of extreme events produced as a collection of historical flood reports. The results show that floods co-occurring in several basins are unevenly distributed, with a higher number of selected events occurred in Northern and Central Italy, where the largest Italian basins are located.

How to cite: Borre, A., Viglione, A., Gabellani, S., and Ghizzoni, T.: Methodologies for the characterisation of spatially distributed hydrological events: the Italian case study., EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-2384, https://doi.org/10.5194/egusphere-egu22-2384, 2022.

Prolonged heat periods have become a recurring feature of the European climate. Recent events like the 2003 heatwave in France, the 2010 Russian heatwave, and the 2019 European heatwave have caused considerable economic losses due to crop failure, imposed substantial stress on the health system, and caused thousands of heat-related deaths. Due to climate change, an increase in length and frequency of heatwaves has been observed since 1950 in most regions worldwide. However, until now, little knowledge is available on the generalized patterns of heatwaves since most studies focus on the analysis of single historical heatwave events.

This study aims to increase the general understanding of heatwaves by identifying and analyzing stable classes, i.e., recurring patterns, of heatwaves present in Europe. In this study, we use data from a regional climate model large ensemble (Canadian Regional Climate Model version 5, CRCM5-LE) consisting of 50 possible realizations of climate in the years 1981-2010 in the EUR-11 domain. We use the 95th percentile of three days' mean temperature as a threshold of heatwave occurrence. Those events are additionally filtered to at least one percent of the land area to ensure that the events have a considerable spatial extent. We repeatedly apply hierarchical agglomerative clustering to find a dozen stable heatwave patterns in Europe. Those results are in good correspondence with clustering on an observational dataset (E-OBS) and when comparing those to historical events. Therefore it is shown that the catastrophic historical events can be explained as an extreme manifestation of the same recurring pattern.

Moreover, we analyze the obtained typical patterns regarding a precipitation deficit present before or after the event. We find that, e.g., after a summer heatwave in South-East Europe, there is a high chance of having increased precipitation in autumn, while no such trend can be observed in Scandinavia. Moreover, the study serves as a blueprint for the analysis of other spatial extreme events (e.g., droughts). 

How to cite: Felsche, E., Böhnisch, A., and Ludwig, R.: Spatial classification of typical European heatwaves using clustering, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-8617, https://doi.org/10.5194/egusphere-egu22-8617, 2022.

The compound dry and hot event (CDHE) has been paying attention in recent decades due to its disastrous impacts on diverse sectors. The teleconnections between dry conditions and large-scale circulation patterns have been widely studied at different spatial and temporal scales. However, studies investigating the links between large-scale circulation patterns and CDHE using a multiscale approach is missing. Quantifying the external forcing of compound dry and hot extremes (CDHE) is tedious and demands in-depth understanding.  We introduce a novel method by integrating wavelets, entropy and complex networks to quantify the potential drivers linked with CDHE. Firstly, a standardized dry and hot Index (SDHI) is developed to model the combined effect of precipitation and temperature using a copula approach. Second, the SDHI and Sea Surface Temperature (SST) is decomposed using wavelets to comprehend multiscale dynamical processes across time scales. Next, entropy is employed to quantify the similarity between SDHI and SST across multiple timescales. The proposed method uses the wavelet energy distribution of CDHI at different time scales and compares it with the wavelet energy distribution of SST to quantify the similarity. From similarity, complex networks is constructed to bridge the links between CDHE and circulation patterns. To investigate the efficiency and reliability, the proposed method is explored to improve the understanding and quantify the potential drivers of CDHE at a regional scale during the summer monsoon in India.  The results show that an integrated approach combining wavelets, entropy and complex networks offers a fresh perspective in analyzing the teleconnections between the compound extremes and large scale circulation patterns.

How to cite: Guntu, R. K., Merz, B., and Agarwal, A.: Understanding and quantifying potential drivers of compound extremes: A complex networks based on multiscale entropy approach, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-215, https://doi.org/10.5194/egusphere-egu22-215, 2022.

How to cite: Richards, J., Huser, R., Bevacqua, E., and Zscheischler, J.: Partially interpretable neural networks for high-dimensional extreme quantile regression: With application to wildfires within the Mediterranean Basin, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-2179, https://doi.org/10.5194/egusphere-egu22-2179, 2022.

A wide literature dealing with the assessment of the critical time scale for basin hydrologic response exist worldwide. The time of concentration (t_C) is recognized as the most frequently used time parameter, followed by the lag time (t_L). However, despite the high sensitivity of design flood peaks to the estimated time parameter values, there is still no agreement on the conceptual and operational definitions of these two parameters, resulting in several different approaches and formulations available.

In our work, we suggest a conceptual approach to validate formulas of the basin time of concentration, with the aim of drawing some guidance in the choice of a robust formulation to be used in hydrological modelling and hydrograph design. To this end, 47 empirical and semi-empirical formulations to quantify t_C have been selected and their structure compared in dimensional terms, using the hydraulic Chezy formula as a litmus paper. Using the river network morphology of 197 watersheds in north-western Italy we have then examined and compared the variability of the estimated average flow velocity within the most hydraulically compatible formulas.

Mindful of recent outcomes on tracer studies (see Azizian, 2019), our results lead to justify some of the coefficients of just a few of the empirical expressions of the critical basin travel time and to further clarify the distinction between t_C and t_L, according to some theoretical justifications discussed in Beven (2020).

How to cite: Evangelista, G. and Claps, P.: Dimensional analysis and intercomparison of the basin time of concentration formulas, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-471, https://doi.org/10.5194/egusphere-egu22-471, 2022.