No abstract

No abstract

Abstract. Quantitative precipitation nowcasting (QPN) has become an essential technique in various application contexts, such as early warning or urban sewage control. A common heuristic prediction approach is to track the motion of precipitation features from a sequence of weather radar images and then to displace the precipitation field to the imminent future (minutes to hours) based on that motion, assuming that the intensity of the features remains constant (“Lagrangian persistence”). In that context, “optical flow” has become one of the most popular tracking techniques. Yet the present landscape of computational QPN models still struggles with producing open software implementations. Focusing on this gap, we have developed and extensively benchmarked a stack of models based on different optical flow algorithms for the tracking step and a set of parsimonious extrapolation procedures based on image warping and advection. We demonstrate that these models provide skillful predictions comparable with or even superior to state-of-the-art operational software. Our software library (“rainymotion”) for precipitation nowcasting is written in the Python programming language and openly available at GitHub (https://github.com/hydrogo/rainymotion, Ayzel et al., 2019). That way, the library may serve as a tool for providing fast, free, and transparent solutions that could serve as a benchmark for further model development and hypothesis testing – a benchmark that is far more advanced than the conventional benchmark of Eulerian persistence commonly used in QPN verification experiments.

Abstract. Climate change is likely to impact the seasonality and generation processes of floods in the Nordic countries, which has direct implications for flood risk assessment, design flood estimation, and hydropower production management. Using a multi-model/multi-parameter approach to simulate daily discharge for a reference ) and a future (2071-2099) period, we analysed the projected changes in flood seasonality and generation processes in six catchments with mixed snowmelt/rainfall regimes under the current climate in Norway. The multi-model/multi-parameter ensemble consists of (i) eight combinations of global and regional climate models, (ii) two methods for adjusting the climate model output to the catchment scale, and (iii) one conceptual hydrological model with 25 calibrated parameter sets. Results indicate that autumn/winter events become more frequent in all catchments considered, which leads to an intensification of the current autumn/winter flood regime for the coastal catchments, a reduction of the dominance of spring/summer flood regimes in a high-mountain catchment, and a possible systematic shift in the current flood regimes from spring/summer to autumn/winter in the two catchments located in northern and south-eastern Norway. The changes in flood regimes result from increasing event magnitudes or frequencies, or a combination of both during autumn and winter. Changes towards more dominant autumn/winter events correspond to an increasing relevance of rainfall as a flood generating process (FGP) which is most pronounced in those catchments with the largest shifts in flood seasonality. Here, rainfall replaces snowmelt as the dominant FGP primarily due to increasing temperature. We further analysed the ensemble components in contributing to overall uncertainty in the projected changes and found that the climate projections and the methods for downscaling or bias correction tend to be the largest contributors. The relative role of hydrological parameter uncertainty, however, is highest for those catchments showing the largest changes in flood seasonality, which confirms the lack of robustness in hydrological model parameterization for simulations under transient hydrometeorological conditions.

Abstract. The potential of weather radar observations for hydrological and meteorological research and applications is undisputed, particularly with increasing world-wide radar coverage. However, several barriers impede the use of weather radar data. These barriers are of both scientific and technical nature. The former refers to inherent measurement errors and artefacts, the latter to aspects such as reading specific data formats, geo-referencing, visualisation. The radar processing library wradlib is intended to lower these barriers by providing a free and open source tool for the most important steps in processing weather radar data for hydro-meteorological and hydrological applications. Moreover, the community-based development approach of wradlib allows scientists to share their knowledge about efficient processing algorithms and to make this knowledge available to the weather radar community in a transparent, structured and well-documented way.

Abstract. In this study, we present RainNet, a deep convolutional neural network for radar-based precipitation nowcasting. Its design was inspired by the U-Net and SegNet families of deep learning models, which were originally designed for binary segmentation tasks. RainNet was trained to predict continuous precipitation intensities at a lead time of 5 min, using several years of quality-controlled weather radar composites provided by the German Weather Service (DWD). That data set covers Germany with a spatial domain of 900 km×900 km and has a resolution of 1 km in space and 5 min in time. Independent verification experiments were carried out on 11 summer precipitation events from 2016 to 2017. In order to achieve a lead time of 1 h, a recursive approach was implemented by using RainNet predictions at 5 min lead times as model inputs for longer lead times. In the verification experiments, trivial Eulerian persistence and a conventional model based on optical flow served as benchmarks. The latter is available in the rainymotion library and had previously been shown to outperform DWD's operational nowcasting model for the same set of verification events. RainNet significantly outperforms the benchmark models at all lead times up to 60 min for the routine verification metrics mean absolute error (MAE) and the critical success index (CSI) at intensity thresholds of 0.125, 1, and 5 mm h−1. However, rainymotion turned out to be superior in predicting the exceedance of higher intensity thresholds (here 10 and 15 mm h−1). The limited ability of RainNet to predict heavy rainfall intensities is an undesirable property which we attribute to a high level of spatial smoothing introduced by the model. At a lead time of 5 min, an analysis of power spectral density confirmed a significant loss of spectral power at length scales of 16 km and below. Obviously, RainNet had learned an optimal level of smoothing to produce a nowcast at 5 min lead time. In that sense, the loss of spectral power at small scales is informative, too, as it reflects the limits of predictability as a function of spatial scale. Beyond the lead time of 5 min, however, the increasing level of smoothing is a mere artifact – an analogue to numerical diffusion – that is not a property of RainNet itself but of its recursive application. In the context of early warning, the smoothing is particularly unfavorable since pronounced features of intense precipitation tend to get lost over longer lead times. Hence, we propose several options to address this issue in prospective research, including an adjustment of the loss function for model training, model training for longer lead times, and the prediction of threshold exceedance in terms of a binary segmentation task. Furthermore, we suggest additional input data that could help to better identify situations with imminent precipitation dynamics. The model code, pretrained weights, and training data are provided in open repositories as an input for such future studies.

[1] Hydrologic modelers often need to know which method of quantitative precipitation estimation (QPE) is best suited for a particular catchment. Traditionally, QPE methods are verified and benchmarked against independent rain gauge observations. However, the lack of spatial representativeness limits the value of such a procedure. Alternatively, one could drive a hydrological model with different QPE products and choose the one which best reproduces observed runoff. Unfortunately, the calibration of conceptual model parameters might conceal actual differences between the QPEs. To avoid such effects, we abandoned the idea of determining optimum parameter sets for all QPE being compared. Instead, we carry out a large number of runoff simulations, confronting each QPE with a common set of random parameters. By evaluating the goodness-of-fit of all simulations, we obtain information on whether the quality of competing QPE methods is significantly different. This knowledge is inferred exactly at the scale of interest-the catchment scale. We use synthetic data to investigate the ability of this procedure to distinguish a truly superior QPE from an inferior one. We find that the procedure is prone to failure in the case of linear systems. However, we show evidence that in realistic (nonlinear) settings, the method can provide useful results even in the presence of moderate errors in model structure and streamflow observations. In a real-world case study on a small mountainous catchment, we demonstrate the ability of the verification procedure to reveal additional insights as compared to a conventional cross validation approach.Citation: Heistermann, M., and D. Kneis (2011), Benchmarking quantitative precipitation estimation by conceptual rainfall-runoff modeling, Water Resour. Res., 47, W06514,

No abstract