The assessment of precipitation variability and uncertainty is crucial in a variety of applications, such as flood risk forecasting, water resource assessments, evaluation of the hydrological impacts of climate change, determination of design floods, and hydrological modelling in general. Within this framework, this session aims to gather contributions on research, advanced applications, and future needs in the understanding and modelling of precipitation variability, and its sources of uncertainty.
Specifically, contributions focusing on one or more of the following issues are particularly welcome:
- Novel studies aimed at the assessment and representation of different sources of uncertainty versus natural variability of precipitation.
- Methods to account for different accuracy in precipitation time series, e.g. due to change and improvement of observation networks.
- Uncertainty and variability in spatially and temporally heterogeneous multi-source precipitation products.
- Estimation of precipitation variability and uncertainty at ungauged sites.
- Precipitation data assimilation.
- Process conceptualization and modelling approaches at different spatial and temporal scales, including model parameter identification and calibration, and sensitivity analyses to parameterization and scales of process representation.
- Modelling approaches based on ensemble simulations and methods for synthetic representation of precipitation variability and uncertainty.
- Scaling and scale invariance properties of precipitation fields in space and/or in time.
- Physically and statistically based approaches to downscale information from meteorological and climate models to spatial and temporal scales useful for hydrological modelling and applications.

Co-organized by AS1/CL2/NH1/NP3
Convener: Simone Fatichi | Co-conveners: Alin Andrei Carsteanu, Roberto Deidda, Andreas Langousis, Chris Onof
| Attendance Fri, 08 May, 14:00–15:45 (CEST)

Files for download

Download all presentations (87MB)

Chat time: Friday, 8 May 2020, 14:00–15:45

Chairperson: A. Langousis, C. Onof
D209 |
Ashish Sharma, Ze Jiang, and Fiona Johnson

As we write this abstract, Australia is experiencing widespread forest fires, Sydney has declared significant water restriction measures curtailing demand, and the entire country is experiencing a drought that is amongst the worst on record. Formulating a stable and practical approach for predicting drought into the future is being realised as an important need, as we enter an era of warmer climates that complicate this problem to an even greater extent. This study presents a novel basis for forecasting drought into the future. Use is made of a recently developed wavelets based methodology for transforming predictor variables so as to force greater consistency in spectral attributes with the response being modelled. Using a commonly adopted drought index, we demonstrate how the wavelets transformed predictor variables can be used to model the response with greater accuracy than otherwise. These transformed predictor variables are then used in conjunction with CMIP5 decadal climate forecasts to demonstrate the accuracy attainable at longer lead times than is currently possible. While our application focusses on the Australian mainland, the method is generic and can be adopted anywhere.

How to cite: Sharma, A., Jiang, Z., and Johnson, F.: Forecasting drought revisited – the importance of spectral transformations to dominant atmospheric predictor variables, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-12334, https://doi.org/10.5194/egusphere-egu2020-12334, 2020.

D210 |
Ioannis Sofokleous, Adriana Bruggeman, Corrado Camera, and George Zittis

The reconstruction of detailed past weather and climate conditions, such as precipitation, is an essential part of hydrometeorological impact studies. Although this can be achieved through dynamical downscaling of reanalysis datasets, different model setup options can result in significantly different simulated fields. To select an efficient ensemble of the WRF atmospheric model for the simulation of precipitation at high resolution, suitable for hydrological studies at catchment scale, a series of simulation experiments is performed. The model experiments center on Cyprus, in the Eastern Mediterranean, a small domain with an area of 225×145 km2 with complex topography. The simulations are made for the hydrologic year 2011-2012. Initial and boundary conditions are provided by the ERA5 reanalysis dataset. A stepwise approach is followed for the evaluation of monthly simulations for an ensemble comprised of 18 combinations of various model physics parameterizations. In the first step, the model ensemble is evaluated for three domain setups with different extends and nested downscaling steps, i.e. 19·105 km2 with 12-, 4- and 1-km grids (12-4-1), 19·105 km2 with 6- and 1-km grids (6-1a) and 7.28 ·105 km2 with 6- and 1-km grids (6-1b). The ensemble performance is then investigated for two initialization frequencies, 30 and 5 days, both with 6-hour spin-up. In the last step, the performance of the individual ensemble members is evaluated and the five best performing members are selected. A gridded precipitation dataset for the area over Cyprus is developed for the evaluation of the simulated precipitation. The statistical indicators used are bias, mean absolute error (MAE), Nash-Sutcliffe efficiency and Kling-Gupta efficiency. The four indicators are scaled and combined in a single composite metric score (CMS), ranging from 0 to 1.

The best overall performance was achieved with the 12-4-1 domain setup. This setup resulted in the lowest bias of accumulated precipitation of the 18-member ensemble, i.e. 1%, compared to 8% for 6-1a and 10% for 6-1b, for the wet month of January. The 12-4-1 setup was also found to add value, in terms of computational time, to the least computationally demanding 6-1b setup by reducing the monthly bias by 47 mm per 1000 cpu hours. The statistical metrics for the ensemble with 5-day initialization exhibited very small variation from the metrics for the monthly initialization, with less than 4% difference in the MAE of the accumulated precipitation. The added value of the 5-day initialization, relative to the monthly initialization, was found to be negative for all four metrics in January and for two of the metrics in May. Despite the variable performance of individual ensemble members in different months, the combined metric showed that the overall highest (lowest) ranked members, with a CMS value of 0.63 (0.43), were those using the Ferrier and WRF-Double-Moment-6th-class (WRF-Single-Moment-6th-class) microphysical schemes. The proposed stepwise evaluation approach allows the identification of a reduced number of ensemble members, out of the initial ensemble, with a model setup that can simulate precipitation at high resolution and under different atmospheric conditions.

How to cite: Sofokleous, I., Bruggeman, A., Camera, C., and Zittis, G.: High-resolution ensemble precipitation simulations over a small domain with complex topography, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-390, https://doi.org/10.5194/egusphere-egu2020-390, 2020.

D211 |
| solicited
Ethan Gutmann, Roy Rasmussen, and Jeffrey Arnold

When is good enough, good enough? The spatio-temporal variability of precipitation makes measurements extremely challenging, particularly in the mountains.  Simultaneously, the improvements in physical realism of atmospheric models makes them increasingly valuable for fields such as hydrology, particularly in the mountains.  However, the computational cost of such models renders them impractical for many applications, in or out of the mountains.  Here we describe an intermediate complexity atmospheric model (ICAR) capable of capturing around 90% of the variability in orographic precipitation for 1% of the computational cost of a state of the science non-hydrostatic atmospheric simulation.  ICAR uses an analytical solution for flow perturbations created by topography and simulates the core processes responsible for orographic precipitation (e.g. orographic lifting, advection, cloud microphysical processes). We show that key aspects of orographic precipitation spatial patterns are well simulated in ICAR, including some that gridded observation based products are missing. We then show some early results when using ICAR to simulate regional climate changes forced by global models at higher spatial resolutions than it is currently practical to run traditional regional climate models. These simulations quantify plausible shifts in precipitation resulting in the transition from snow to rain, as well as elevation dependent warming caused by the snow albedo feedback.  Further, the computational efficiency of ICAR permits us to run these simulations with many different physics configurations to better explore the sensitivity of these changes to assumptions in the microphysics and land surface model components. 

How to cite: Gutmann, E., Rasmussen, R., and Arnold, J.: A Fast Intermediate Complexity Atmospheric Model for Precipitation Modeling, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-10983, https://doi.org/10.5194/egusphere-egu2020-10983, 2020.

D212 |
Emma D. Thomassen, Elisabeth Kendon, Hjalte J. D. Sørup, Steven Chan, Peter L. Langen, Ole B. Christensen, and Karsten Arnbjerg-Nielsen

Convection Permitting Models (CPM) are believed to improve the representation of precipitation extremes at sub-daily scale compared to coarser spatial scale Regional Climate Models (RCM). This study seeks to compare how the spatio-temporal characteristics of precipitation extremes differ between a 2.2km CPM and a 12km RCM from the UK Met Office with a pan-European domain.

Storm data have been re-gridded to a common 12km grid and all events in the period from 1999-2008 are tracked with the DYMECS tracking algorithm. A peak-over-threshold method is used to sample extreme events within a northern European case area. Maximum intensity and maximum area of extremes are sampled based on the maximum intensity and maximum size reached within their lifetime. Evolution in size and intensity, track patterns, and seasonal occurrence of extreme events are compared between the two models.

For the top 1000 extreme events with the highest maximum intensities, the two models show disagreement in movement direction and spatial and temporal occurrence. While the CPM data are dominated by south-north moving events occurring in summer over central Europe, the RCM data are dominated by west-east moving events occurring over UK and more uniformly distribution over the year. The CPM and RCM however show good agreement in these variables for extreme events instead selected based on largest spatial area. A comparison with the COSMO REA6 reanalysis model continuously nudged towards observations indicates a similar spatial and seasonal distribution of extreme events sampled by maximum intensity as in the CPM. Analysis of the evolution of storms over their lifetime shows on average higher intensities and spatial areas of the most intense storms in the RCM data compared to the most intense storms in the CPM data. Sampling of maximum intensity extreme events in each of the four seasons show larger disagreement between the two models in the evolution in intensity and size in autumn (SON) and winter (DJF) than in spring (MAM) and summer (JJA).

How to cite: Thomassen, E. D., Kendon, E., Sørup, H. J. D., Chan, S., Langen, P. L., Christensen, O. B., and Arnbjerg-Nielsen, K.: Comparison of spatio-temporal evolution of extreme precipitation events between two high-resolution models in a northern Europe case study, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-18467, https://doi.org/10.5194/egusphere-egu2020-18467, 2020.

D213 |
Jefferson Wong, Fuad Yassin, and James Famiglietti

Obtaining reliable precipitation measurements and accurate spatiotemporal distribution of precipitation remains as a challenging task for driving Hydrologic-Land Surface Models (H-LSMs) and better hydrological simulations and predictions. To further improve the accuracy of precipitation estimation for hydrological applications, the idea of generating a hybrid dataset by combining existing precipitation products has become a more appealing approach in recent years. The reliability of the hybrid dataset is evaluated against in-situ climate stations and error characteristics are calculated to compare to the existing products. However, the robustness of the hybrid dataset in representing spatial details could be problematic when evaluated only using a sparse network of in-situ observations at regional or basin scales. This study aims to develop a methodological framework that combines multiple precipitation products based on evaluation against not only climate stations but also streamflow stations that are spatially representative across large river basin. The framework is illustrated using a Canadian H-LSM named MESH (Modélisation Environmentale communautaire - Surface Hydrology) in the Saskatchewan River basin, Canada over the period of 2002 to 2012. Five existing precipitation datasets are considered as the candidates for generating the hybrid dataset. The framework consists of three components. The first component evaluates each precipitation candidate against the local gauge data for benchmarking, runs each candidate through MESH with 10 km spatial resolution and default parameterization, and calculates the overall streamflow performance in each sub-basins with equal weighting of three evaluation metrics. The second component generates the hybrid dataset by combining the best performing candidates (annual or seasonal) at sub-basin scale. The third component assesses the performance of the hybrid dataset at downstream gauge stations along the mainstream as a validation mechanism for comparison with the performance of the candidate datasets. Results shows that the hybrid dataset is able to perform equally well with the existing precipitation products in the headwater while improve the streamflow performance downstream. The successful application of the framework in this river basin could build the foundation and the confidence in applying the combination method to data-limited river basins in northern Canada.

How to cite: Wong, J., Yassin, F., and Famiglietti, J.: A Methodological Framework to Combine Multiple Precipitation Datasets for Improving Streamflow Simulations: A test study in the Saskatchewan River basin, Canada, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-12547, https://doi.org/10.5194/egusphere-egu2020-12547, 2020.

D214 |
Hannes Müller-Thomy, Korbinian Breinl, David Lun, and Günter Blöschl

Precipitation is a key input variable for precipitation-runoff models. For catchments without precipitation observations generating rainfall fields is a possibility to enable precipitation-runoff simulations. These synthetic precipitation fields have to reproduce the spatial precipitation distribution adequately, especially at large catchment scales. Since the spatial precipitation coherence in ungauged catchments is unknown, it has to be transferred from an existing observational network. Ideally, the meteorological regime of the area of the observational network should be similar to that of the ungauged catchment in terms of the processes and factors controlling the spatial precipitation coherence.

This study identifies these processes and conceptualises them for rainfall modelling. We analyse precipitation time series of 1200 stations in the Greater Alpine Region (including Austria and Southern Germany, ~300,000 km²). Precipitation data subsets are constructed based on space-dependent (including climate zone, land use, altitude, slope, exposition) and time-dependent factors (seasons, circulation patterns, temperature). The analyses are carried out for different temporal resolutions (1, 12 and 24 hours) to unravel possible time-dependencies. The spatial precipitation coherence is represented by bivariate characteristics (Pearson’s correlation coefficient, continuity ratio, probability of occurrence) as a function of station separation distance. Uncertainty and variability of the spatial coherence are quantified via function spaces. Self-organizing maps are applied to translate the multi- dimensional results into low-dimensional maps.

In the low lands of the study domain, time-dependent factors are expected to influence the spatial precipitation coherence stronger than space-dependent factors, while in the mountainous regions the space-dependent factors will have a stronger influence due to the air movement being forced by the topography.

How to cite: Müller-Thomy, H., Breinl, K., Lun, D., and Blöschl, G.: Unravelling the process controls of the spatial coherence of precipitation, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-8554, https://doi.org/10.5194/egusphere-egu2020-8554, 2020.

D215 |
Monica Estebanez Camarena, Nick van de Giesen, Marie-Claire ten Veldhuis, and Sandra de Vries

West Africa’s economy is mainly sustained on agriculture and over 70% of crops are rain-fed. Economic growth and food security in this region is therefore highly dependent on the knowledge of rainfall patterns. According to the IPCC, the Global South will seriously suffer from climate change. As traditional rainfall patterns shift, accurate rainfall information becomes crucial for farmers to optimize food production.

The scarce rain gauge distribution and data transmission challenges make rainfall analysis difficult in these regions. Satellites could offer a solution to this problem, but present satellite products do not account for local characteristics and perform poorly in West Africa. For example, comparing the widely used TAMSAT and CHIRPS satellite rainfall products with ground data in our pilot area in the Northern Region of Ghana, we found a very poor correlation with TAMSAT and CHIRPS grossly overestimating the number of rainy days, while underestimating the amount of rainfall per event.

The RainRunner rainfall retrieval algorithm, developed within the Schools and Satellites (SaS) project, aims to overcome the lack of ground data and good rainfall satellite products through Earth Observation and advanced Machine Learning (ML). SaS is being funded by the European Space Agency as one of the pilot projects of CSEOL (Citizen Science and Earth Observation Lab). It is being developed in a cooperation between TU Delft, PULSAQUA, TAHMO Ghana, Smartphones4Water and the Ghana Meteorological Agency (GMet).

Research suggests that local characteristics are the reason for traditional rainfall retrieval algorithms to perform poorly in West Africa, where the land surface temperature and the concentration of atmospheric aerosols are higher than in other regions in the world. Hence, RainRunner will utilize information relevant to the rain process other than the traditionally used cloud top temperature, namely, cloud amount, atmospheric aerosols, soil moisture and land surface temperature. These data are derived from diverse sensors onboard ESA’s Sentinel satellites (S1, S2, S3 and S5P), as well as MSG’s Aviris. The satellite products, together with a Digital Elevation Model, will be pre-processed into datacubes to be fed to a Convolutional Neural Network (CNN) to estimate precipitation for a certain geographic point.

CNNs have shown to achieve better results when modelling complex natural processes than other ML algorithms, when provided with big amounts of data and well-designed architectures that represent the physical process knowledge. Furthermore, they have the main advantages of computing efficiency and the ability to represent processes beyond numerical simulations. The latter is essential for understanding the complex interactions between variables, therefore resulting in not only improving rainfall estimates but also in increasing our understanding of processes in poorly measured regions.

The Proof-of-Concept algorithm will be trained and validated with TAHMO and GMet ground measurements. Eventually, the training and validation dataset will incorporate data acquired by a rainfall observation network combining low-cost sensors and Citizen Science data collected by schoolchildren in Ghana.

Once operative, the RainRunner will guide agricultural extension agents, support crop insurance and ultimately contribute to economic growth and food security in the Global South.

How to cite: Estebanez Camarena, M., van de Giesen, N., ten Veldhuis, M.-C., and de Vries, S.: RainRunner - Machine Learning and Earth observation for reliable rainfall information in West Africa , EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-22073, https://doi.org/10.5194/egusphere-egu2020-22073, 2020.

D216 |
Anastasios Perdios and Andreas Langousis

Over the years, several studies have been carried out to investigate how the statistics of peak annual discharges vary with the size of basins, with diverse findings regarding the observed type of scaling (i.e. simple scaling vs multiscaling), especially in cases where the data originated from regions with significantly different hydroclimatic characteristics. In this context, two important questions arise: a) how rainfall climatology affects the scaling of peak annual discharges, and b) how one can effectively conclude on an approximate type of statistical scaling of annual discharge maxima with respect to the basin size. The present study aims at addressing these two questions, using daily discharges from 805 catchments located in different parts of the United Kingdom, with at least 30 years of recordings. In doing so, we isolate the effects of the catchment area and the local rainfall climatology, and examine how the statistics of the standardized discharge maxima vary with the basin scale. The obtained results show that: a) the local rainfall climatology is an important contributor to the observed statistics of annual peak discharges, and b) when the effects of the local rainfall climatology are properly isolated, the scaling of the standardized annual discharge maxima with the area of the catchment closely follows that of the underlying rainfall process, deviating significantly from the simple scaling rule. The aforementioned findings explain to a large extent the diverse results obtained by previous studies in the absence of rainfall information, shedding light to the approximate type of scaling of peak annual discharges with the basin size.

How to cite: Perdios, A. and Langousis, A.: Revisiting the statistical scaling of peak annual discharges with respect to the basin size in the light of rainfall climatology, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-227, https://doi.org/10.5194/egusphere-egu2020-227, 2020.

D217 |
Brunella Bonaccorso, Giuseppina Brigandì, and Giuseppe Tito Aronica

Depth (or intensity)-duration-frequency (DDF or IDF) curves are commonly applied in hydrology to derive storms of fixed duration and return period for hydraulic infrastructures design and risk assessment. Usually, annual maxima rainfall (AMR) data from 1 to 24-hour duration are used to develop DDF or IDF curves. However, design of urban drainage systems or flood risk assessment in small catchments often requires knowledge of very short-duration rainfall events (less than 1 hour), whose data are often unavailable or too scarce for estimating reliable quantile values. Regularities in the temporal pattern exhibited by storm records, known as scaling properties, could help in characterizing extreme storms at partially gauged sites better than the application of traditional statistical techniques. In this work, a scaling approach for estimating the distribution of sub-hourly extreme rainfall in Sicily (Italy) is presented based on data from high-resolution rain gauges with a short functioning period and from low-resolution rain gauges with longer samples. First, simple scaling assumption versus multiple scaling one is verified for annual maxima rainfall (AMR) data from 10 minute to 24-hour duration, revealing that the simple scaling regime holds from 20 to 60 minutes for most of the stations. Three scaling homogeneous regions are classified based on the scaling exponent values. In each region, this parameter is regionalized by means of power law relationships with the maximum 1 hour AMR data. Then, regional DDF curves are developed by combining the scale-invariant framework with the generalized extreme value (GEV) probability distribution, in order to estimate T-year sub-hourly extreme rainfalls at sites where only rainfall data for longer durations (≥ 1 hour) are available. The regional GEV simple scaling DDF model is validated against sub-hourly historical observations at five rain gauges. Results indicate that the proposed model provides reliable sub-hourly estimates.

How to cite: Bonaccorso, B., Brigandì, G., and Aronica, G. T.: A regional scale invariant depth-duration-frequency model for sub-hourly extreme rainfall estimation in Sicily, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-5994, https://doi.org/10.5194/egusphere-egu2020-5994, 2020.

D218 |
Jana Ulrich, Madlen Peter, Oscar E. Jurado, and Henning W. Rust

Intensity-Duration-Frequency (IDF) Curves are a popular tool in Hydrology for estimating the properties of extreme precipitation events. They describe the relationship between rainfall intensity and duration for a given non-exceedance probability (or frequency). For a site where precipitation measurements are available, these curves can be estimated consistently over durations using a duration-dependent GEV (d-GEV, after Koutsoyiannis et al. 1998). In this approach, the probability distributions are modeled simultaneously for all durations.

Additionally, we integrate covariates to describe the spatial variability of the d-GEV parameters so that we can model the distribution of extreme precipitation for a range of durations and locations in one step. Thus IDF Curves can be estimated even at ungauged sites. Further advantages are parameter reduction and more efficient use of the available data. We use the Quantile Skill Score to investigate under which conditions this method leads to an improved estimate compared to the single-site approach and to evaluate the performance at ungauged sites.

How to cite: Ulrich, J., Peter, M., Jurado, O. E., and Rust, H. W.: Estimating IDF-Relations consistently using a duration-dependent GEV with spatial covariates, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-18724, https://doi.org/10.5194/egusphere-egu2020-18724, 2020.

D219 |
Dongkyun Kim and Christian Onof

We introduce a stochastic model reproducing various rainfall characteristics at timescales between 5 minutes and one decade. The model is composed of three moduels as follow: First, the model generates the fine-scale rainfall data based on a type of Bartlett-Lewis rectangular pulse model; Second, sequence of the generated rainstorms are shuffled so that their correlation structure can be preserved; Third, the time series is rearranged at the monthly timescale to reflect the coarse scale correlation structure. The method was tested based on the 69 years of 5-minute rainfall data of Bochum, Germany. The mean, variance, covariance, skewness, and proportion of wet/dry periods were well reproduced at the timescales from 5 minutes to a decade. The extreme values were also successfully reproduced at the timescales between 5 minutes and 3 days. The antecedent moisture condition before an extreme rainfall event was reproduced well too.

How to cite: Kim, D. and Onof, C.: A stochastic rainfall model that can reproduce important rainfall properties across the timescales from several minutes to a decade, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-4072, https://doi.org/10.5194/egusphere-egu2020-4072, 2020.

D220 |
Ross Pidoto and Uwe Haberlandt

Climate impact studies regarding hydrology require long precipitation time series of high spatial and temporal resolution. Global climate models (GCMs) provide global predictions of future climates, however they are a poor choice to accurately represent future surface precipitation conditions, especially at high resolutions. Instead, statistical downscaling from a GCM to a stochastic precipitation model is one common method to provide unbiased time series of arbitrary length for use within climate impact studies.

This study considers an alternating renewal stochastic rainfall model conditioned on fuzzy-rule based climate classes. The key research question for this study is whether stationarity of the climate classes can be assumed, meaning that changes to future rainfall can be explained by changes in climate class frequency alone. If stationarity of the climate classes cannot be assumed, what further steps, for example a delta-change approach, are required to adequately account for this non-stationarity.

An event based alternating renewal rainfall model has been conditioned on a fuzzy-rule based climate classification, using re-analysis climate data as input for the classification. The classification is created via an automated objective optimisation procedure that derives climate classes of non-mean (either dry or wet) rainfall behaviour.

The study area is the northern German federal state of Lower Saxony. ERA5 re-analysis climate data was used as input for the fuzzy-based classification. Previous studies using this classification method used atmospheric pressure data only, whereas this study also incorporates additional climate variables such as wind, temperature, humidity etc. 18 high-resolution rainfall gauges with a time series length of at least 15 years were used as observations for the rainfall model. A regional climate model (RCM) will be used as a reference for both past and future rainfall conditions in order to test the stated hypothesis. The climate classes derived from the re-analysis data will be reproduced for future climates using simulation results from a GCM.

Initial results indicate that the conditioning on climate classes using additional climate variables improves the single site performance of the rainfall model, particularly regarding extremes. The climate classes themselves were also shown to be more robust and diverse in terms of their rainfall behaviour when compared to classes generated from atmospheric pressure data alone. It is also hypothesised that the climate conditioned model will show improvements in predicting future precipitation conditions compared to previous studies.

How to cite: Pidoto, R. and Haberlandt, U.: Can a statistically downscaled stochastic rainfall model conditioned on climate variables sufficiently represent future rainfall scenarios?, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-9974, https://doi.org/10.5194/egusphere-egu2020-9974, 2020.

D221 |
Barbara Haese, Sebastian Hörning, Maximilian Graf, Adam Eshel, Christian Chwala, and Harald Kunstmann

Precipitation is one of the crucial variables within the hydrological system, and accordingly one of the main drivers for terrestrial hydrological processes. The quality of many hydrological applications such as climate prediction, water resource management, and flood forecasting, depends on the correct reproduction of its spatiotemporal distribution. However, the global network of precipitation observations is relatively sparse in large areas of the world. Compared to these observation network, inhabited areas typically have a relative dense network of Commercial Microwave Links (CMLs). These CMLs can be used to calculate path-averaged rain rates, derived from their attenuation. One challenge when using path-averaged rain rates is the construction of spatial precipitation fields. To address these challenges, we apply Random Mixing Whittaker-Shannon (RMWSPy) to stochastically simulate precipitation fields. Therefore, we generate precipitation fields as a linear combination of unconditional spatial random fields, where the spatial dependence structure is described by copulas. The weights of the linear combination are optimized in such a way that the observations and the spatial structure of the precipitation observations are reproduced. Within this method the path-averaged rain rates are used as non-linear constrains. One big advantage when using RMWSPy is the ability to simulate precipitation field ensembles of any size, where each ensemble member is in concordance with the underlying observations. The spread of such an ensemble enables an uncertainty estimation of the simulated fields. In particular, it reflects the precipitation variability along the CML path and the uncertainty between the observation locations. We demonstrate RMWSPy using CML observations within various areas of Germany with a different density of observations. We show, that the reconstructed precipitation fields reproduce the observed spatial precipitation pattern in a comparable good quality as the RADOLAN weather radar data set provided by the German Weather Service (DWD).

How to cite: Haese, B., Hörning, S., Graf, M., Eshel, A., Chwala, C., and Kunstmann, H.: Using Commercial Microwave Links for a stochastic reconstruction of precipitation field ensembles, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-9080, https://doi.org/10.5194/egusphere-egu2020-9080, 2020.

D222 |
Lianyi Guo

Four bias-correction methods, i.e. Gamma Cumulative Distribution Function (GamCDF), Quantile-Quantile Adjustment (QQadj), Equidistant CDF Matching (EDCDF) and Transform CDF (CDF-t), were applied to five daily precipitation datasets over China produced by LMDZ4-regional that was nested into five global climate models (GCMs), BCC-CSM1-1m, CNRM-CM5, FGOALS-g2, IPSL-CM5A-MR and MPI-ESM-MR, respectively. A unified mathematical framework can be used to define the four methods, which helps understanding their nature and essence in identifying the most reliable probability distributions of projected climate. CDF-t is shown to be the best bias-correction algorithm based on a comprehensive evaluation of different rainfall indices. Future precipitation projections corresponds to the global warming levels of 1.5°C and 2°C under RCP8.5 were obtained using the bias correction methods. The multi-algorithm and multi-model ensemble characteristics allow to explore the spreading of results, considered as a surrogate of climate projection uncertainty, and to attribute such uncertainties to different sources. It was found that the spread among bias-correction methods is smaller than that among dynamical downscaling simulations. The four bias-correction methods with CDF-t at the top all reduce the spread among the downscaled results. Future projection using CDF-t is thus considered having higher credibility.

How to cite: Guo, L.: Projected precipitation changes over China for global warming levels at 1.5°C and 2°C in an ensemble of regional climate simulations: Impact of bias-correction algorithms, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-4811, https://doi.org/10.5194/egusphere-egu2020-4811, 2020.

D223 |
Masoud Mehrvand and András Bárdossy

Generating synthetic precipitation for weather generators were always a challenging issue in hydro-climate simulations because of its high variability in time and space. We present a spectral method for generating the synthetic precipitation time series which is in accordance with the observed precipitation statistical characteristics not only for the observed points, but also for any desired location by interpolating the time series spectrum. In this regard, time series spectra derived from the observed signal converting from its time domain to the corresponding frequency domain using the Fourier transform.

The main problem for spectral interpolation of precipitation time series is highly occurrence of non-rainy days which can be even more inaccurate for the finer resolutions such as hourly and sub-hourly data. In order to overcome the highly frequent occurrence of non-rainy days, transformation between indicator and normal correlation has been taken into account.

This method enables us to generate synthetic time series with same statistical characteristics for the observed points and also for any point of interests rather than the observed points. The introduced so called spectral and spatial interpolation method applied for daily and hourly precipitation time series for the selected stations in state Baden-Württemberg, Germany.

How to cite: Mehrvand, M. and Bárdossy, A.: Spectral and spatial interpolation of precipitation for daily and hourly time series, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-20520, https://doi.org/10.5194/egusphere-egu2020-20520, 2020.

D224 |
Masaru Yamamoto

The present study investigates short-term (four-day) atmospheric response to regionally warm sea surface in Tsushima Strait for two periods (a sunny period, 19-22 August 2013 and a rainy period, 23-26 August 2013) using ensemble WRF simulations with initial condition altered in the presence and absence of an extremely warm SST core. In this presentation, the author focuses on the influence of regionally warm sea surface on moisture and extreme rainfall. The moisture response is quite different between the sunny and rainy periods. Ensemble averaged distribution of time-mean moisture variation induced by a regionally warm sea surface is well correlated with the SST increase during the sunny period. However, it is not clearly correlated with the SST increase during the rainy period when vapor fluctuated because of frequent rainfall. The high SST enhanced time-mean precipitation in the central area of the warm SST core. In the ensemble experiment, the warm SSTs do not always enhance hourly rainfall because the water-vapor concentrations are decreased by prior rainfall events in some members. In a simulation that well reproduces heavy rainfall at Izuhara located in Tsushima Strait in the presence of the warm SST core, high SSTs induced extreme precipitation (~50 mm/h) in the morning. Water vapor decreased after the morning heavy rainfall. The decreased moisture led to low precipitation in the afternoon. In contrast, a low-SST experiment with the warm-SST core removed shows that water-vapor concentrations were higher after weaker morning rainfall, compared to the high SST experiment with the warm core. Because of the high water-vapor concentrations, low SST led to greater precipitation in the afternoon. Thus, when responses of hourly precipitation to SST are investigated, we must consider the temporal water-vapor variation associated with prior rainfall event.

How to cite: Yamamoto, M.: Influence of regionally warm sea surface on moisture and extreme rainfall in Tsushima Strait during August 2013, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-6809, https://doi.org/10.5194/egusphere-egu2020-6809, 2020.

D225 |
Martina Lagasio, Agostino N. Meroni, Luca Pulvirenti, Giuseppe Squicciarino, Antonio Parodi, Alexia Tsouni, Haris Kontoes, and Nikos Bartsotas

In the framework of the E-SHAPE “EuroGEO Showcase: Applications Powered by Europe” project, the Pilot 2 application of the Disasters Resilience Showcase concerns the disasters in urban environment. Starting from the results and methodologies analyzed in the framework of the STEAM project, the E-SHAPE pilot exploits the new capacities for designing and delivering innovative services for extreme-scale fire/hydro-meteorological modelling chain assimilating Copernicus data and core services directly ingested through the Copernicus Open Access Hub APIS, and the DIAS platform, as well as citizen scientists data, to enable more precise predictions and decision-making support for high impact events in urban and peri urban environment. Contributing to the Disaster Resilience SBA, one of the main activities listed in the GEO Space and Security Community Activity is to get maximum benefit from the use of large and heterogeneous datasets to potentially fill in the observational and capability gaps at EU decision making level. To this end, the application proposes also the integration of the datasets and tools made available in the frame of the pilot application (weather, citizen science, hydrological and fire models included in CIMA’s Platforms Dewetra and RASOR and NOA’s BEYOND Systems FireHub and FloodHub) for the impact assessment of natural hazards over areas of interest with regard to human security issues. An example of innovative service is the ingestion of high-resolution Copernicus remote sensing products in Numerical Weather Prediction (NWP) models. The rationale is that NWP models are presently able to produce forecasts with a spatial resolution in the order of 1 km, but unreliable surface information or poor knowledge of the initial state of the atmosphere may imply an inaccurate simulation of the weather phenomena. It is expected that forecast inaccuracies could be reduced by ingesting high resolution Earth Observation products into the models. In this context, the Copernicus Sentinel satellites represent an important source of data, because they can provide a set of high-resolution observations of physical variables (e.g. soil moisture, land/sea surface temperature, wind over sea, columnar water vapor) used in NWP model runs. The possible availability of a spatially dense Personal Weather Stations network could also be exploited to allow NWP models to assimilate timely updated data such as temperature, humidity and pressure. In this work a preliminary experiment design and methodology will be presented.

How to cite: Lagasio, M., Meroni, A. N., Pulvirenti, L., Squicciarino, G., Parodi, A., Tsouni, A., Kontoes, H., and Bartsotas, N.: Sentinel products assimilation in a complete hydro/fire-meteorological chain: nearly operational experiments in the framework of the E-SHAPE project , EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-7168, https://doi.org/10.5194/egusphere-egu2020-7168, 2020.

D226 |
Harald Zandler, Isabell Haag, and Cyrus Samimi

Gridded precipitation data is of central importance for various geoscientific research applications and is often the only available resource to derive spatial and temporal rainfall quantities. Numerous studies exist that evaluate respective products using gauge measurements. However, many existing approaches ignore the impact of temporal changes in incorporated observation data, the location of the observations and the potential overlap of evaluation and dataset stations. Considering these issues, we quantitatively evaluated monthly precipitation values of frequently used precipitation raster datasets (GPCC Full Data Monthly Product Version 2018, GPCC Monitoring Product Version 6, CRU TS 4.03, GPCP Version 2.3, PERSIANN-CDR, TRMM 3B43, MERRA-2, MERRA-2 bias corrected, ERA5) in the peripheral Pamir mountains with a focus on the two periods 1980–1994 and 1998–2012 as they are characterized by considerable observation data changes. The coefficient of efficiency, a dimensionless hydroclimatic evaluation measure, showed that only three of the precipitation raster datasets (GPCC Full Data Monthly Product Version 2018, GPCC Monitoring Product Version 6, MERRA-2 bias corrected) are able to provide better surface precipitation values than the long-term station mean in this observation data poor region. Results of the gauge-based products also document a fourfold increase of errors during periods with low availability of station data compared to periods with higher observation data inputs. In conclusion, the study clearly illustrates that gridded precipitation products may be connected to major problems in peripheral mountain regions with limited measurement infrastructure as most datasets directly or indirectly depend on observation networks. Significant differences of errors related to incorporated observation data variations demonstrate the need for temporal and spatial evaluation approaches as a prerequisite for the scientific utilization of precipitation raster datasets.

How to cite: Zandler, H., Haag, I., and Samimi, C.: Evaluate before use – temporal performance differences of gridded precipitation products in complex terrain, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-7339, https://doi.org/10.5194/egusphere-egu2020-7339, 2020.

D227 |
Ke-Sin Yu, Jihn-Sung Lai, and Yi-Huan Hsieh

Under the impact of climate change, rainfall-induced flood disasters have become more frequent in some areas. The development of an hourly rainfall forecast with higher time and spatial accuracy under different rainfall patterns and the connection between meteorological forecast and hydrological flood simulation are urgent issues. In this study, eight flood cases in 2019 in Taipei city, a high-risk urban area with high economic and social resource density, caused by different rainfall patterns were chosen to be analyzed. To improve the accuracy of meteorological data, WRF base ensemble prediction system (WEPS), a quantitative precipitation forecast (QPF) produced by Central Weather Bureau (CWB) of Taiwan was selected as the main meteorological data source, and after processed by objective quantitative analysis methods, the data then be input into the drainage–inundation model. As a one-dimensional and two-dimensional flood simulation system, SOBEK was used to verify the depth and location of floods. Results indicated that the WEPS data would have better performance in drainage–inundation model among the cases in 2019. Combining meteorological forecast data and hydrological simulation can somehow improve the accuracy of flood early warning system in a small catchment.

How to cite: Yu, K.-S., Lai, J.-S., and Hsieh, Y.-H.: Simulation and analysis for flood early warning system in small catchments caused by rainfall-induced disaster, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-15493, https://doi.org/10.5194/egusphere-egu2020-15493, 2020.

D228 |
Citlali Cabrera Gutiérrez and Jean Christophe Jouhaud

Complex models calculations can be very expensive and time consuming. A surrogate model aims at producing results which are very close to the ones obtained using a complex model, but with largely reduced calculation times. Building a surrogate model requires only a few calculations with the real model. Once the surrogate model is built, further calculations can be quickly realized.

In this study, we propose to build surrogate models by combining Proper Orthogonal Decomposition (POD) and kriging (also known as Gaussian Process Regression) for immediate forecasts. More precisely, we create surrogate models for rainfall forecasts on short deadlines. Currently rainfall forecasts in France are calculated for 15 minutes time laps using the AROME-PI model developed by M ́et ́eo-France. In this work, we show that the results obtained with our surrogate models are not only close to the ones obtained by AROME-PI, but they also have a better time resolution (1 minute) and a reduced calculation time.

How to cite: Cabrera Gutiérrez, C. and Jouhaud, J. C.: Efficient POD-Kriging Surrogate Models for Rainfall Forecasting, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-21780, https://doi.org/10.5194/egusphere-egu2020-21780, 2020.

D229 |
Dario Ruggiu, Francesco Viola, and Andreas Langousis

In an effort to assess the accuracy of the normality assumption for annual rainfall totals (ART) in data-poor regions, we develop a non-parametric procedure based on the marginal statistics of daily rainfall. In doing so we start by using three goodness-of-fit metrics to conclude on the approximate convergence of the empirical ART distribution to a normal shape, and classify daily rainfall timeseries into Gaussian (G) and non-Gaussian (NG) groups. At a second step, we apply logistic regression analysis to identify the statistics of daily rainfall that are most descriptive of the G/NG classification. In the third and final step, we use a random-search algorithm to conclude on a set of constraints to classify ART samples based on the marginal statistics of daily rainrates. The analysis is conducted using 3007 daily rainfall timeseries from the NOAA-NCDC Global Historical Climatology Network (GHCN) database, and aims at developing a statistical tool towards informed decision making for water management purposes. The conducted analysis highlights that the Anderson-Darling (AD) test statistic is the most conservative one in determining approximate Gaussianity of ART samples (followed by Cramer-Von Mises and Kolmogorov-Smirnov), while daily rainfall timeseries with fraction of dry days in excess of 90% and skewness coefficient of positive rainrates that exceeds 5.92 deviate significantly from the normal shape. Further, our results indicate that continental climate exhibits the highest fraction of Gaussian distributed ART samples, followed by warm temperate, equatorial, polar, and arid climates.

How to cite: Ruggiu, D., Viola, F., and Langousis, A.: A non-parametric procedure to assess the accuracy of the normality assumption for annual rainfall totals, based on the marginal statistics of daily rainfall: An application to NOAA-NCDC rainfall database, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-10365, https://doi.org/10.5194/egusphere-egu2020-10365, 2020.

D230 |
Marc Schleiss and Venkat Roy

We present a dynamic state model estimation method for rainfall nowcasting in which we assume that the short term spatio-temporal evolution of rainfall can be approximated by a linear state model with stochastic perturbations.  We estimate the model parameters using radar reflectivity measurements for one-step as well as multiple-step ahead rainfall nowcasting. If the rainfall intensity at location x and time index t is given by ut(x), then the overall rainfall field intensity vector at any time t over N pixels (of the target area) can be represented by ut = [ut(xN),...ut(xN)]T. Following the aforementioned formalism, the spatio-temporal evolution of the rainfall field can be described by the following linear state model given by
ut = Htut-1 + qt
where Ht is an unknown time-varying state-transition matrix of dimensions NxN and qt is a stochastic process noise vector of length N. We present an iterative least squares based method to estimate Ht and explore simpler algebraic structures (e.g., scaled affine transformations) to reduce the numbers of unknown parameters during estimation. We evaluate the performances of the proposed model using simulations and radar reflectivity data from the Royal Netherlands Meteorological Institute (KNMI). We observe that the nowcasting performances strongly depend on the size of the target area (number of pixels N), the type of events as well as the parameterization of Ht. The key advantage of the proposed approach over classical nowcasting methods based on Lagrangian persistence is the possibility to incorporate prior information about future rainfall evolution from external sources of information such as satellites or numerical weather prediction models during the estimation of the parameters.

How to cite: Schleiss, M. and Roy, V.: A Dynamic and Flexible State Model for Rainfall Nowcasting, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-21400, https://doi.org/10.5194/egusphere-egu2020-21400, 2020.

D231 |
Yuanyuan Ma

Sudden turn from drought to flood (STDF) is a unique representation of intra-seasonal extreme events and occurs frequently. However, it is notoriously difficult to represent in climate simulations due to the accumulation of model errors. This study uses a regional climate model (RCM) with different initialization and nudging schemes to explore effective approaches for capturing a STDF event. Results show that the conventional continuous integration with single initialization cannot reproduce the STDF event, while nudging or re-initialization can. Furthermore, spectral nudging and re-initialization outperform the conventional continuous simulation in reproducing precipitation features, but grid nudging induces the largest biases for precipitation though it has the smallest biases for other meteorological elements. Scale separation analysis shows that the large-scale features of the conventional continuous simulation drift far from the actual fields and force erroneous small-scale features, whereas the nudging and re-initialization successfully prevent the model from drifting away from the forcing fields at large-scales. The different performance for simulating precipitation among spectral nudging, re-initialization and grid nudging can be attributed to that the former two methods generate their own small-scale information via the RCM, while grid nudging over-suppresses the small-scale information while retaining the large-scale features. The difference in small-scale features affects the simulation of different moisture fluxes and convergences, as well as clouds, and then results in diverse precipitation. These results illustrate that both the consistency with large-scale features and the local variability from small-scale features are both robust factors for reproducing precipitation features during extreme events using RCMs.

How to cite: Ma, Y.: How essential of the balance between large and small scale features to reproduce precipitation during a sudden sharp turn from drought to flood, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-12558, https://doi.org/10.5194/egusphere-egu2020-12558, 2020.