Search

The rapid temporal evolution of convective rainfall poses a challenge for rainfall nowcasting models. With the growing potential of machine learning models for precipitation nowcasting to produce realistic-looking nowcasts for long lead times, it is important to investigate whether the nowcasts also produce realistic development for convective rainfall. Common verification metrics traditionally used to validate nowcasting models are often dominated by large-scale stratiform rainfall, and averaging the metrics across entire precipitation fields obscures how accurately the models replicate individual convective cells, which makes it difficult to distinguish the model skill for the growth and decay of convective rainfall. In this study, we present a convective cell tracking-based framework to investigate how accurately nowcasting models reproduce the development of convective rainfall. The framework consists of first identifying and tracking the convective cells in the input observation rainfall fields, and then identifying and tracking the cells separately in the target observations and the nowcast rainfall fields by continuing the cell tracks identified in the observations. Features describing the cells and cell tracks, such as the cell volume rain rate and area, are then extracted. In addition to the errors in these feature values, the models' skill in reproducing the existence of convective cells is estimated by calculating several contingency-table metrics, such as the Critical Success Index. The results allow the analysis of how accurately the models reproduce the growth and decay of convective rainfall and quantify the differences between the models, for example, due to differences in how the models smooth the nowcasts, i.e., blurring. The framework also allows differentiation of the results based on the initial conditions of the cell tracks, demonstrated here by separating the tracks into decaying or growing cell tracks based on the cell status when the nowcast is created. The framework is demonstrated using four open-source advection-based models: the advection nowcast, S-PROG, and LINDA implemented in the pysteps library, and L-CNN, with data from the Swiss radar network. The results indicate that the L-CNN model reproduced the existence of convective cells best among the models and had smaller errors in the cell volume rain rate than LINDA and S-PROG. LINDA had the smallest underestimation in the cell mean rain rate, whereas S-PROG significantly overestimated the cell volume rain rate and area because of blurring. [ABSTRACT FROM AUTHOR]

Precipitation nowcasting (forecasting locally for 0–6 h) serves both public security and industries, facilitating the mitigation of losses incurred due to, e.g., flash floods and is usually done by predicting weather radar echoes, which provide better performance than numerical weather prediction (NWP) at that scale. Probabilistic nowcasts are especially useful as they provide a desirable framework for operational decision-making. Many extrapolation-based statistical nowcasting methods exist, but they all suffer from a limited ability to capture the nonlinear growth and decay of precipitation, leading to a recent paradigm shift towards deep-learning methods which are more capable of representing these patterns. Despite its potential advantages, the application of deep learning in probabilistic nowcasting has only recently started to be explored. Here we develop a novel probabilistic precipitation nowcasting method, based on Bayesian neural networks with variational inference and the U-Net architecture, named DEUCE. The method estimates the total predictive uncertainty in the precipitation by combining estimates of the epistemic (knowledge-related and reducible) and heteroscedastic aleatoric (data-dependent and irreducible) uncertainties, using them to produce an ensemble of development scenarios for the following 60 min. DEUCE is trained and verified using Finnish Meteorological Institute radar composites compared to established classical models. Our model is found to produce both skillful and reliable probabilistic nowcasts based on various evaluation criteria. It improves the receiver operating characteristic (ROC) area under the curve scores 1 %–5 % over STEPS and LINDA-P baselines and comes close to the best-performer STEPS on a continuous ranked probability score (CRPS) metric. The reliability of DEUCE is demonstrated with, e.g., having the lowest expected calibration error at 20 and 25 dBZ reflectivity thresholds and coming second at 35 dBZ. On the other hand, the deterministic performance of ensemble means is found to be worse than that of extrapolation and LINDA-D baselines. Last, the composition of the predictive uncertainty is analyzed and described, with the conclusion that aleatoric uncertainty is more significant and informative than epistemic uncertainty in the DEUCE model. [ABSTRACT FROM AUTHOR]

Precipitation nowcasting (forecasting locally for 0-6h) serves both public security and industries, facilitating the mitigation of losses incurred due to e.g. flash floods, and is usually done by predicting weather radar echoes, which provides better performance than NWP at that scale. Probabilistic nowcasts are especially useful as they provide a desirable framework for operational decision-making. Many extrapolation-based statistical nowcasting methods exist, but they all suffer from a limited ability to capture the nonlinear growth and decay of precipitation, leading to a recent paradigm shift towards deep learning methods, more capable of representing these patterns. Despite of its potential advantages, the application of deep learning in probabilistic nowcasting has only recently started to be explored. Here we develop a novel probabilistic precipitation nowcasting method, based on Bayesian neural networks with variational inference and the U-Net architecture, named DEUCE. The method estimates the total predictive uncertainty of precipitation by combining estimates of the epistemic (knowledge-related, reducible) and heteroscedastic aleatoric (datadependent, irreducible) uncertainties, and produces an ensemble of development scenarios for the following 60 minutes. DEUCE is trained and verified using Finnish Meteorological Institute radar composites against established classical models. Our model is found to produce both skillful and reliable probabilistic nowcasts based on various evaluation criteria. It improves ROC Area Under the Curve scores 1-5% over STEPS and LINDA-P baselines, and comes close to the best-performer STEPS on a CRPS metric. The reliability of DEUCE is demonstrated with, e.g., having the lowest Expected Calibration Error at 20 and 25 dBZ reflectivity thresholds, and coming second at 35 dBZ. On the other hand, deterministic performance of ensemble means is found to be worse than that of extrapolation and LINDA-D baselines. Lastly, the composition of the predictive uncertainty is analysed and described, with the conclusion that aleatoric uncertainty is more significant and informative than epistemic uncertainty in the DEUCE model. [ABSTRACT FROM AUTHOR]

Short-range forecasts (nowcasts) of rainfall facilitate providing early warning of severe rainfall and flooding, which is particularly important in densely populated urban areas. Nowcasts are conventionally obtained by the extrapolation of radar echoes from a constant altitude plan position indicator (CAPPI) or lowest-angle plan position indicator (PPI). Lacking a model for growth or decay, this approach has a limited ability to forecast the summertime convective storms. In a previous attempt to address this shortcoming, called RadVil, the predicted surface rain rate is obtained from mass balance equations of vertically integrated liquid (VIL) retrieved from volumetric reflectivity measurements. Predicting the growth and decay has also been attempted by using an autoregressive (AR) model, which led to the development of spectral prognosis (S-PROG). A novel combination of these two methodologies, called ANVIL, is proposed. In this approach, the growth and decay of VIL are modeled by an autoregressive integrated (ARI) process. It is shown that the predictability of growth and decay is scale-dependent. Thus, the key idea of ANVIL is to decompose the VIL into multiple spatial scales and apply a separate ARI model to each scale. The operational feasibility of ANVIL is evaluated using the Next-Generation Radar (NEXRAD)/WSR-88D radar that covers the Dallas–Fort Worth metropolitan area. The evaluation is done using ten convective events in 2018 and 2019. Using several verification metrics, it is shown that ANVIL has up to 25% improved skill compared to conventional nowcasting techniques. The improvement is consistent for a wide range of spatial scales (1–11 km) and rain rate thresholds (5–20 mmh−1). [ABSTRACT FROM AUTHOR]

Pysteps is an open-source and community-driven Python library for probabilistic precipitation nowcasting, that is, very-short-range forecasting (0–6 h). The aim of pysteps is to serve two different needs. The first is to provide a modular and well-documented framework for researchers interested in developing new methods for nowcasting and stochastic space–time simulation of precipitation. The second aim is to offer a highly configurable and easily accessible platform for practitioners ranging from weather forecasters to hydrologists. In this sense, pysteps has the potential to become an important component for integrated early warning systems for severe weather. The pysteps library supports various input/output file formats and implements several optical flow methods as well as advanced stochastic generators to produce ensemble nowcasts. In addition, it includes tools for visualizing and post-processing the nowcasts and methods for deterministic, probabilistic and neighborhood forecast verification. The pysteps library is described and its potential is demonstrated using radar composite images from Finland, Switzerland, the United States and Australia. Finally, scientific experiments are carried out to help the reader to understand the pysteps framework and sensitivity to model parameters. [ABSTRACT FROM AUTHOR]

Pysteps is an open-source and community-driven Python library for probabilistic precipitation nowcasting - that is to say, very-short range forecasting (0-6 h). The aim of pysteps is to serve two different needs. The first is to provide a modular and well-documented framework for researchers interested in developing new methods for nowcasting and stochastic space-time simulation of precipitation. The second aim is to offer a highly configurable and easily accessible platform for practitioners ranging from weather forecasters to hydrologists. In this sense, pysteps has the potential to become an important component for integrated early warning systems for severe weather. The pysteps library supports standard input/output file formats and implements several optical flow methods as well as advanced stochastic generators to produce ensemble nowcasts. In addition, it includes tools for visualizing and post-processing the nowcasts and methods for deterministic, probabilistic, and neighbourhood forecast verification. The pysteps library is described and its potential is demonstrated using radar composite images from Finland, Switzerland, United States, and Australia. Finally, scientific experiments are carried out to help the reader to understand the pysteps framework and sensitivity to model parameters. [ABSTRACT FROM AUTHOR]

Nowcasts (short-term forecasts) of heavy rainfall causing flash floods are highly valuable in densely populated urban areas. In the Collaborative Adaptive Sensing of the Atmosphere (CASA) project, a high-resolution X-band radar network was deployed in the Dallas–Fort Worth (DFW) metroplex. The Dynamic and Adaptive Radar Tracking of Storms (DARTS) method was developed as a part of the CASA nowcasting system. In this method, the advection field is determined in the spectral domain using the discrete Fourier transform. DARTS was recently extended to include a filtering scheme for suppressing small-scale precipitation features that have low predictability. Building on the earlier work, Stochastic DARTS (S-DARTS), a probabilistic extension of DARTS, is developed and tested using the CASA DFW radar network. In this method, the nowcasts are stochastically perturbed in order to simulate uncertainties. Two novel features are introduced in S-DARTS. First, the scale filtering and perturbation based on an autoregressive model are done in the spectral domain in order to achieve high computational efficiency. Second, this methodology is extended to modeling the temporal evolution of the advection field. The performance and forecast skill of S-DARTS are evaluated with different precipitation intensity thresholds and ensemble sizes. It is shown that S-DARTS can produce reliable probabilistic nowcasts in the CASA DFW domain with 250-m spatial resolution up to 45 min for lower precipitation intensities (below 2 mm h−1). For higher intensities (above 5 mm h−1), adequate skill can be obtained up to 15 min. [ABSTRACT FROM AUTHOR]

The quality of quantitative precipitation estimation (QPE) is degraded by considerable discrepancies between radar and ground measurements, which are common due to inherent uncertainties between these two kinds of sensor systems. The causes include measurement errors and differences in sampling schemes. Nevertheless, the remaining discrepancies can be statistically modeled. A model describing detection probabilities of ground rainfall, systematic biases as well as the variance of residual discrepancies between radar and rain gauges is developed. These are modeled by means of multiple explanatory variables such as rain rate and distance from radar. The model is implemented by using nonparametric kernel methods and spatiotemporal Kriging interpolation. A key feature of the model is that for a given radar-derived rainfall field and explanatory variables, it determines probability distributions for the corresponding ground rainfall. Unbiased estimates for ground rainfall can be obtained from the expected values of the distributions. From such distributions, one can also obtain uncertainty estimates and exceedance probabilities that are important for hydrological applications. Performance of the model is assessed by cross-validation using hourly rainfall accumulations measured by the Finnish rain gauges and C-band dual polarization radars. [ABSTRACT FROM AUTHOR]

Finding a low-dimensional embedding of a graph ofnnodes inis an essential task in many applications. For instance, maximum variance unfolding (MVU) is a well-known dimensionality reduction method that involves solving this problem. The standard approach is to formulate the embedding problem as a semidefinite program (SDP). However, the SDP approach does not scale well to large graphs. In this paper, we exploit the fact that many graphs have an intrinsically low dimension, and thus the optimal matrix resulting from the solution of the SDP has a low rank. This observation leads to a quadratic reformulation of the SDP that has far fewer variables, but on the other hand, is a difficult convex maximization problem. We propose an approach for obtaining a solution to the SDP by solving a sequence of smaller quadratic problems with increasing dimension. Utilizing an augmented Lagrangian and an interior-point method for solving the quadratic problems, we demonstrate with numerical experiments on MVU problems that the proposed approach scales well to very large graphs. [ABSTRACT FROM AUTHOR]

The latest established generation of weather radars provides polarimetric measurements of a wide variety of meteorological and nonmeteorological targets. While the classification of different precipitation types based on polarimetric data has been studied extensively, nonmeteorological targets have garnered relatively less attention beyond an effort to detect them for removal from meteorological products. In this paper we present a supervised learning classification system developed in the Finnish Meteorological Institute (FMI) that uses Bayesian inference with empirical probability density distributions to assign individual range gate samples into 7 meteorological and 12 nonmeteorological classes, belonging to five top-level categories of hydrometeors, terrain, zoogenic, anthropogenic, and immaterial. We demonstrate how the accuracy of the class probability estimates provided by a basic naive Bayes classifier can be further improved by introducing synthetic channels created through limited neighborhood filtering, by properly managing partial moment nonresponse, and by considering spatial correlation of class membership of adjacent range gates. The choice of Bayesian classification provides well-substantiated quality estimates for all meteorological products, a feature that is being increasingly requested by users of weather radar products. The availability of comprehensive, fine-grained classification of nonmeteorological targets also enables a large array of emerging applications, utilizing nonprecipitation echo types and demonstrating the need to move from a single, universal quality metric of radar observations to one that depends on the application, the measured target type, and the specificity of the customers' requirements. Significance Statement: In addition to meteorological echoes, weather radars observe a wide variety of nonmeteorological phenomena including birds, insects, and human-made objects like ships and aircraft. Conventionally, these data have been rejected as undesirable disturbance, but lately their value for applications like aeroecological monitoring of bird and insect migration has been understood. The utilization of these data, however, has been hampered by a lack of comprehensive classification of nonmeteorological echoes. In this paper we present a comprehensive, fine-grained, probabilistic classifier for all common types of nonmeteorological echoes which enables the implementation of a wide range of novel weather radar applications. [ABSTRACT FROM AUTHOR]

Last years, the policies to reduce the impacts caused by severe storms and heavy precipitation include adaptive risk management approaches (e.g. the Flood Risk Management Plans promoted by the EU Flood Directive, which identified the importance of Early Warning Systems). In this context, the development of tools capitalizing on the recent advances in real-time systems (including observations, improved capacity to anticipate the phenomena and their impacts, and characterization of the uncertainty) is crucial to support decision and response.This project, lasting the years 2018-2019, develops tools for assessing and forecasting pan-European hazards induced by severe storms and flash floods. In particular, the project will focus on (i) probabilistic nowcasting of convective storms that cause hazards in urban areas, (ii) providing a continuous hazard forecasting horizon (from 1 hour to week) by seamlessly integrating the nowcasts of precipitation with the NWP outputs of ECMWF, and (iii) developing tools to characterize the uncertainty affecting the different components of the hazard forecasting algorithms.This supposes an extension of the tools developed in the EC Civil Protection projects HAREN, EDHIT and ERICHA. Similarly, the developments made in the project are targeted to be integrated in the very-short term forecasting layer of the European Flood Awareness System (EFAS), which is the reference system for flood monitoring and forecasting used by agents in charge of the emergency management and response. Specifically, the following development will be performed: Operational real-time, quality-controlled, multi-source, pan-European fields of heavy rainfall will be piloted by integrating rainfall estimates from the data of the EUMETNET project OPERA, rain gauges, EUMETSAT product CRR and lightning location (GLD 360) into a truly pan-European composite. From them probabilistic, rainfall nowcasts for lead times 0-6 hours are generated that are seamlessly blended (in the lead time range of 3-8 h) with the ECMWF ensemble predictions of rainfall. Algorithms that can at best generate reasonable seamless blending of the multi-source probabilistic nowcasts are so far commonly lacking. In topic (i) European-scale object-oriented cell tracking -based probabilistic nowcasts of convective heavy rainfall will be compared to nowcasts obtained with the original high time-space resolution of radar data.The SMUFF products will be integrated into new layers on the ERICHA/SMUFF platform, including probabilistic flood hazard products based on the respective rainfall ensemble predictions. Optimal characterization of probabilities and uncertainties in the products from the end user perspective as well as real-time evaluation of the rainfall and flooding hazard assessment products and user interfaces is performed in collaboration with Civil Protection Agencies. The presentation in EGU will highlight the SMUFF project, its mid-term results and final objectives. [ABSTRACT FROM AUTHOR]

The forecast of snow events at airports is a major challenge in airport operation during winter time. In contrast to rain, snow has to be actively removed from aircraft or operation areas. Short term forecasting - or nowcasting - of snow events is required for airport operation. Airport stakeholders need to know the start and duration of snow events, but they need also an estimate of the snow accumulation during the event. Extrapolation techniques based upon weather radar observations are used to generate a probabilistic nowcast for airports. These techniques assume a linear propagation of the precipitation fields. In case of intensifying, decaying or non-linear propagation the forecast quality is considerably reduced. In this paper we show that lake effects along the coastlines or flow within the proximity of mountains degrade the forecast quality and the lead time for reliable nowcasts is shorter than for situations which are not affected by heterogeneous terrain. [ABSTRACT FROM AUTHOR]

Rainfall-runoff simulations in urban environments require meteorological input data with high temporal and spatial resolutions. The availability of precipitation data is constantly increasing due to the shift towards more open data sharing. However, the applicability of such data for urban runoff assessments is often unknown. Here, the feasibility of Finnish Meteorological Institute’s open rain gauge and open weather radar data as input sources was studied by conducting Storm Water Management Model simulations at a very small (33.5 ha) urban catchment in Helsinki, Finland. In addition to the open data sources, data were also available from two research gauges, one of them located on-site, and from a research radar. The results confirmed the importance of local precipitation measurements for urban rainfall-runoff simulations, implying the suitability of open gauge data to be largely dictated by the gauge’s distance from the catchment. Performance of open radar data with 5 min and 1 km 2 resolution was acceptable in terms of runoff reproduction, albeit peak flows were constantly and flow volumes often underestimated. Gauge adjustment and advection interpolation were found to improve the quality of the radar data, and at least gauge adjustment should be performed when open radar data are used. Finally, utilizing dual-polarization capabilities of radars has a potential to improve rainfall estimates for high intensity storms although more research is still needed. [ABSTRACT FROM AUTHOR]

The Finnish educational system is known for its equality. However, in many key areas in national and international assessments, Swedish-language schools in Finland have lagged behind their Finnish-language counterparts. So far there is little research into the underlying reasons for this discrepancy. In this article, in order to illuminate the phenomenon, we take a closer look at a number of school-level indices based on the Finnish OECD Programme for International Student Assessment (PISA) 2009 data and compare school principals’ responses across these two school categories. Our findings reveal differences in school policies and practices, in school climate, and in school resources between the two parallel language-specific sectors of basic education in Finland. [ABSTRACT FROM PUBLISHER]

We describe the intermediate results of a SJU/EU -funded research activity, called Probabilistic Nowcasting of Winter Weather for Airports (PNOWWA), developing methods to support the Air Traffic Management (ATM) challenged by winter weather. Our methodology is based on probabilistic nowcasting of winter weather, which will enable the estimation of winter weather conditions affecting the ground part of air traffic 4D trajectories. This kind of ATM methods and tools are called for, because the uncertainties during flight, departure and arrival at airports create a need to effectively utilize probability forecasts, both in the local operational user environment and en-route. [ABSTRACT FROM AUTHOR]