scholarly journals Skilful precipitation nowcasting using deep generative models of radar

Nature ◽  
2021 ◽  
Vol 597 (7878) ◽  
pp. 672-677
Author(s):  
Suman Ravuri ◽  
Karel Lenc ◽  
Matthew Willson ◽  
Dmitry Kangin ◽  
Remi Lam ◽  
...  
Keyword(s):  
Poor Performance ◽  
Heavy Rain ◽  
Generative Models ◽  
Generative Model ◽  
Alternative Methods ◽  
Systematic Evaluation ◽  
Lead Times ◽  
Physical Constraints ◽  
Rain Events ◽  
Rain Rates

AbstractPrecipitation nowcasting, the high-resolution forecasting of precipitation up to two hours ahead, supports the real-world socioeconomic needs of many sectors reliant on weather-dependent decision-making1,2. State-of-the-art operational nowcasting methods typically advect precipitation fields with radar-based wind estimates, and struggle to capture important non-linear events such as convective initiations3,4. Recently introduced deep learning methods use radar to directly predict future rain rates, free of physical constraints5,6. While they accurately predict low-intensity rainfall, their operational utility is limited because their lack of constraints produces blurry nowcasts at longer lead times, yielding poor performance on rarer medium-to-heavy rain events. Here we present a deep generative model for the probabilistic nowcasting of precipitation from radar that addresses these challenges. Using statistical, economic and cognitive measures, we show that our method provides improved forecast quality, forecast consistency and forecast value. Our model produces realistic and spatiotemporally consistent predictions over regions up to 1,536 km × 1,280 km and with lead times from 5–90 min ahead. Using a systematic evaluation by more than 50 expert meteorologists, we show that our generative model ranked first for its accuracy and usefulness in 89% of cases against two competitive methods. When verified quantitatively, these nowcasts are skillful without resorting to blurring. We show that generative nowcasting can provide probabilistic predictions that improve forecast value and support operational utility, and at resolutions and lead times where alternative methods struggle.

scholarly journals Comparisons of Instantaneous TRMM Ground Validation and Satellite Rain-Rate Estimates at Different Spatial Scales

2008 ◽  
Vol 47 (8) ◽  
pp. 2215-2237 ◽  
Author(s):  
David B. Wolff ◽  
Brad L. Fisher
Keyword(s):  
Rain Rate ◽  
Spatial Scales ◽  
Tropical Rainfall Measuring Mission ◽  
Heavy Rain ◽  
Global Ocean ◽  
Ground Validation ◽  
Microwave Imager ◽  
Rain Events ◽  
Trmm Microwave Imager ◽  
Rain Rates

Abstract This study provides a comprehensive intercomparison of instantaneous rain rates observed by the two rain sensors aboard the Tropical Rainfall Measuring Mission (TRMM) satellite with ground data from two regional sites established for long-term ground validation: Kwajalein Atoll and Melbourne, Florida. The satellite rain algorithms utilize remote observations of precipitation collected by the TRMM Microwave Imager (TMI) and the Precipitation Radar (PR) aboard the TRMM satellite. Three standard level II rain products are generated from operational applications of the TMI, PR, and combined (COM) rain algorithms using rain information collected from the TMI and the PR along the orbital track of the TRMM satellite. In the first part of the study, 0.5° × 0.5° instantaneous rain rates obtained from the TRMM 3G68 product were analyzed and compared to instantaneous Ground Validation (GV) program rain rates gridded at a scale of 0.5° × 0.5°. In the second part of the study, TMI, PR, COM, and GV rain rates were spatiotemporally matched and averaged at the scale of the TMI footprint (∼150 km2). This study covered a 6-yr period (1999–2004) and consisted of over 50 000 footprints for each GV site. In the first analysis, the results showed that all of the respective rain-rate estimates agree well, with some exceptions. The more salient differences were associated with heavy rain events in which one or more of the algorithms failed to properly retrieve these extreme events. Also, it appears that there is a preferred mode of precipitation for TMI rain rates at or near 2 mm h−1 over the ocean. This mode was noted over ocean areas of Kwajalein and Melbourne and has been observed in TRMM tropical–global ocean areas as well.

scholarly journals Thailand Daily Rainfall and Comparison with TRMM Products

2008 ◽  
Vol 9 (2) ◽  
pp. 256-266 ◽  
Author(s):  
Roongroj Chokngamwong ◽  
Long S. Chiu
Keyword(s):  
Daily Rainfall ◽  
Tropical Rainfall Measuring Mission ◽  
Heavy Rain ◽  
Absolute Difference ◽  
Rain Event ◽  
Critical Success ◽  
Global Precipitation Climatology Centre ◽  
Temporal And Spatial ◽  
Rain Events ◽  
Rain Rates

Abstract Daily rainfall data collected from more than 100 gauges over Thailand for the period 1993–2002 are used to study the climatology and spatial and temporal characteristics of Thailand rainfall variations. Comparison of the Thailand gauge (TG) data binned at 1° × 1° with the Global Precipitation Climatology Centre (GPCC) monitoring product shows a small bias (1.11%), and the differences can be reconciled in terms of the increased number of stations in the TG dataset. Comparison of daily TG with Tropical Rainfall Measuring Mission (TRMM) version 6 (V6) 3B42 rain estimates shows improvements over version 5 (V5) in terms of bias and mean absolute difference (MAD). The V5 is computed from the adjusted Geostationary Operational Environmental Satellite (GOES) precipitation index (AGPI) and V6 is computed using the TRMM Multisatellite Precipitation Analysis (TMPA) algorithm. The V6 histogram is much closer to that of TG than V5 in terms of rain fraction and conditional rain rates. Scatterplots show that both versions of the satellite products are deficient in capturing heavy rain events. In terms of detecting rain events, a critical success index (CSI) shows no difference between V6 and V5 3B42. The CSI for V6 is higher for the rainy season than the dry season. These results are generally insensitive to rain-rate threshold and averaging periods. The temporal and spatial autocorrelation of daily rain rates for TG, V6, and V5 3B42 are computed. Autocorrelation function analyses show improved temporal and spatial autocorrelations for V6 compared to TG over V5 with e-folding times of 1, 1, and 2 days, and isotropic spatial decorrelation distances of 1.14°, 1.87°, and 3.61° for TG, V6, and V5, respectively. Rain event statistics show that the V6 3B42 overestimates the rain event durations and underestimates the rain event separations and the event conditional rain rates when compared to TG. This study points to the need to further improve the 3B42 algorithm to lower the false detection rate and improve the estimation of heavy rainfall events.

scholarly journals Incorporating the Effects of Moisture into a Dynamical Parameter: Moist Vorticity and Moist Divergence

2015 ◽  
Vol 30 (6) ◽  
pp. 1411-1428 ◽  
Author(s):  
Weihong Qian ◽  
Jun Du ◽  
Xiaolong Shan ◽  
Ning Jiang
Keyword(s):  
Potential Vorticity ◽  
Eastern China ◽  
Heavy Rain ◽  
Systematic Evaluation ◽  
Rain Event ◽  
Dynamical Parameter ◽  
Rain Area ◽  
Moisture Effects ◽  
Moist Potential Vorticity ◽  
Rain Events

Abstract Properly including moisture effects into a dynamical parameter can significantly increase the parameter’s ability to diagnose heavy rain locations. The relative humidity–based weighting approach used to extend the moist potential vorticity (MPV) to the generalized moist potential vorticity (GMPV) is analyzed and demonstrates such an improvement. Following the same approach, two new diagnostic parameters, moist vorticity (MV) and moist divergence (MD), have been proposed in this study by incorporating moisture effects into the traditional vorticity and divergence. A regional heavy rain event that occurred along the Yangtze River on 1 July 1991 is used as a case study, and 41 daily regional heavy rain events during the notorious flooding year of 1998 in eastern China are used for a systematic evaluation. Results show that after the moisture effects were properly incorporated, the improved ability of all three parameters to capture a heavy rain area is significant (statistically at the 99% confidence level): the GMPV is improved over the MPV by 194%, the MD over the divergence by 60%, and the MV over the vorticity by 34% in terms of the threat score (TS). The average TS is 0.270 for the MD, 0.262 for the MV, and 0.188 for the GMPV. Application of the MV and MD to assess heavy rain potential is not intended to replace a complete, multiscale forecasting methodology; however, the results from this study suggest that the MV and MD could be used to postprocess a model forecast to potentially improve heavy rain location predictions.

scholarly journals A Robust Error-Based Rain Estimation Method for Polarimetric Radar. Part II: Case Study

2012 ◽  
Vol 51 (9) ◽  
pp. 1702-1713 ◽  
Author(s):  
Acacia S. Pepler ◽  
Peter T. May
Keyword(s):  
Estimation Method ◽  
Heavy Rain ◽  
Rain Gauge ◽  
Polarimetric Radar ◽  
Composite Methods ◽  
A New Technique ◽  
Rain Events ◽  
Radar Part ◽  
Rain Rates

AbstractRainfall estimation using polarimetric radar involves the combination of a number of estimators with differing error characteristics to optimize rainfall estimates at all rain rates. In Part I of this paper, a new technique for such combinations was proposed that weights algorithms by the inverse of their theoretical errors. In this paper, the derived algorithms are validated using the “CP2” polarimetric radar in Queensland, Australia, and a collocated rain gauge network for two heavy-rain events during November 2008 and a larger statistical analysis that is based on data from between 2007 and 2009. Use of a weighted combination of polarimetric algorithms offers some improvement over composite methods that are based on decision-tree logic, particularly at moderate to high rain rates and during severe-thunderstorm events.

scholarly journals Estimation of Rain Intensity Spectra over the Continental United States Using Ground Radar–Gauge Measurements

2012 ◽  
Vol 25 (6) ◽  
pp. 1901-1915 ◽  
Author(s):  
Xin Lin ◽  
Arthur Y. Hou
Keyword(s):  
United States ◽  
Heavy Rain ◽  
Rain Intensity ◽  
Light Rain ◽  
Geographical Variations ◽  
Satellite Sensors ◽  
Rain Events ◽  
Rain Rates ◽  
The Impact ◽  
Microwave Sensors

Abstract A high-resolution surface rainfall product is used to estimate rain characteristics over the continental United States as a function of rain intensity. By defining data at 4-km horizontal resolutions and 1-h temporal resolutions as an individual precipitating or nonprecipitating sample, statistics of rain occurrence and rain volume including their geographical and seasonal variations are documented. Quantitative estimations are also conducted to evaluate the impact of missing light rain events due to satellite sensors’ detection capabilities. It is found that statistics of rain characteristics have large seasonal and geographical variations across the continental United States. Although heavy rain events (>10 mm h−1) only occupy 2.6% of total rain occurrence, they may contribute to 27% of total rain volume. Light rain events (<1.0 mm h−1), occurring much more frequently (65%) than heavy rain events, can also make important contributions (15%) to the total rain volume. For minimum detectable rain rates setting at 0.5 and 0.2 mm h−1, which are close to sensitivities of the current and future spaceborne precipitation radars, there are about 43% and 11% of total rain occurrence below these thresholds, and they respectively represent 7% and 0.8% of total rain volume. For passive microwave sensors with their rain pixel sizes ranging from 14 to 16 km and the minimum detectable rain rates around 1 mm h−1, the missed light rain events may account for 70% of rain occurrence and 16% of rain volume. Statistics of rain characteristics are also examined on domains with different temporal and spatial resolutions. Current issues in estimates of rain characteristics from satellite measurements and model outputs are discussed.

Equivariant Adversarial Network for Image-to-image Translation

2021 ◽  
Vol 17 (2s) ◽  
pp. 1-14
Author(s):  
Masoumeh Zareapoor ◽  
Jie Yang
Keyword(s):  
State Of The Art ◽  
Generative Models ◽  
Generative Model ◽  
Target Domain ◽  
Adversarial Network ◽  
Proposed Model ◽  
Image Translation ◽  
Great Performance ◽  
Representative Model ◽  
The Ideal

Image-to-Image translation aims to learn an image from a source domain to a target domain. However, there are three main challenges, such as lack of paired datasets, multimodality, and diversity, that are associated with these problems and need to be dealt with. Convolutional neural networks (CNNs), despite of having great performance in many computer vision tasks, they fail to detect the hierarchy of spatial relationships between different parts of an object and thus do not form the ideal representative model we look for. This article presents a new variation of generative models that aims to remedy this problem. We use a trainable transformer, which explicitly allows the spatial manipulation of data within training. This differentiable module can be augmented into the convolutional layers in the generative model, and it allows to freely alter the generated distributions for image-to-image translation. To reap the benefits of proposed module into generative model, our architecture incorporates a new loss function to facilitate an effective end-to-end generative learning for image-to-image translation. The proposed model is evaluated through comprehensive experiments on image synthesizing and image-to-image translation, along with comparisons with several state-of-the-art algorithms.

scholarly journals Comparison of Radar-Based Hail Detection Using Single- and Dual-Polarization

2019 ◽  
Vol 11 (12) ◽  
pp. 1436 ◽  
Author(s):  
Skripniková ◽  
Řezáčová
Keyword(s):  
Comparative Analysis ◽  
Heavy Rain ◽  
Radar Data ◽  
Severe Weather ◽  
Detection Methods ◽  
Polarimetric Radar ◽  
Dual Polarization ◽  
Rain Events ◽  
Single Polarization ◽  
Strong Agreement

The comparative analysis of radar-based hail detection methods presented here, uses C-band polarimetric radar data from Czech territory for 5 stormy days in May and June 2016. The 27 hail events were selected from hail reports of the European Severe Weather Database (ESWD) along with 21 heavy rain events. The hail detection results compared in this study were obtained using a criterion, which is based on single-polarization radar data and a technique, which uses dual-polarization radar data. Both techniques successfully detected large hail events in a similar way and showed a strong agreement. The hail detection, as applied to heavy rain events, indicated a weak enhancement of the number of false detected hail pixels via the dual-polarization hydrometeor classification. We also examined the performance of hail size detection from radar data using both single- and dual-polarization methods. Both the methods recognized events with large hail but could not select the reported events with maximum hail size (diameter above 4 cm).

scholarly journals Off-the-shelf deep learning is not enough, and requires parsimony, Bayesianity, and causality

2021 ◽  
Vol 7 (1) ◽  
Author(s):  
Rama K. Vasudevan ◽  
Maxim Ziatdinov ◽  
Lukas Vlcek ◽  
Sergei V. Kalinin
Keyword(s):  
Neural Networks ◽  
Computer Vision ◽  
Deep Learning ◽  
Bayesian Methods ◽  
Deep Neural Networks ◽  
Applied Research ◽  
Modern Science ◽  
Generative Models ◽  
Knowledge Development ◽  
Physical Constraints

AbstractDeep neural networks (‘deep learning’) have emerged as a technology of choice to tackle problems in speech recognition, computer vision, finance, etc. However, adoption of deep learning in physical domains brings substantial challenges stemming from the correlative nature of deep learning methods compared to the causal, hypothesis driven nature of modern science. We argue that the broad adoption of Bayesian methods incorporating prior knowledge, development of solutions with incorporated physical constraints and parsimonious structural descriptors and generative models, and ultimately adoption of causal models, offers a path forward for fundamental and applied research.

Heavy rainfall in the northern coast of Ecuador in the aftermath of El Niño 2015/2016 and its predictability 

2021 ◽  
Author(s):  
Luis E. Pineda ◽  
Juan Changoluisa ◽  
Ángel G. Muñoz
Keyword(s):  
Heavy Rainfall ◽  
Heavy Rain ◽  
Rainfall Anomaly ◽  
Precipitation Event ◽  
Lead Times ◽  
Northern Coast ◽  
Atmospheric Conditions ◽  
The North ◽  
Ensemble Mean ◽  
Rainfall Anomalies

<p>In January 2016, a high precipitation event (HPE) affected the northern coast of Ecuador leading to devastating flooding in the Esmeraldas’ river basin. The HPE appeared in the aftermath of the 2015/2016 El Niño as an early onset of heavy rainfalls otherwise expected in the core rainy season (Mar-Apr). Using gauge data, satellite imagery and reanalysis we investigate the daily and ‘weather-within-climate’ characteristics of the HPE and its accompanying atmospheric conditions. The convective storms developed into a mesoscale convective complex (MCC) during nighttime on 24<sup>th</sup> January. The scale size of the heavy rainfall system was about 250 km with a lifecycle lasting 16 hours for the complete storm with 6 hours of convective showers contributing to the HPE. The genesis of the MCC was related to above-normal moisture and orographic lifting driving convective updrafts; the north-south mountain barrier acted as both a channel boosting upslope flow when it moves over hillslopes; and, as a heavy-rain divide for inner valleys. The above normal moisture conditions were favored by cross-time-scale interactions involving the very strong El Niño 2015/2016 event, an unusually persistent Madden–Julian oscillation (MJO) in phases 3 and 6, remotely forced by tropical synoptic scale disturbances. In the dissipation stage, a moderate low-level easterly shear with wind velocity of about 10 m/s moved away the unstable air and the convective pattern disappear on the shore of the Esmeraldas basin.</p><p> </p><p>We use ECMWF re-forecast from the Sub-seasonal to Seasonal (S2S) prediction project dataset and satellite observations to investigate the predictability of the HPE. Weekly ensemble-mean rainfall anomaly forecasts computed from raw (uncorrected) S2S reforecast initialized on 31st Dec 2015, 7th, 14th and 21st Jan 2016 are used to assess the occurrence of rainfall anomalies over the region. The reforecast represents consistently, over all lead times, the spatial pattern of the HPE. Also, the ensemble-mean forecast shows positive rainfall anomalies at times scales of 1-3 weeks (0-21 days) at nearly all initialization dates and lead times, predicting this way successfully the timing and amplitude of the highest HPE leading the 25th January flood.</p>

scholarly journals Spatial structures and directionalities in Monsoonal precipitation over South Asia

2010 ◽  
Vol 17 (5) ◽  
pp. 371-381 ◽  
Author(s):  
N. Malik ◽  
N. Marwan ◽  
J. Kurths
Keyword(s):  
Indian Subcontinent ◽  
Monsoon Season ◽  
Daily Rainfall ◽  
Heavy Rain ◽  
Data Sets ◽  
Spatial Structures ◽  
Break Phase ◽  
Rainfall Events ◽  
Monsoonal Precipitation ◽  
Rain Events

Abstract. Precipitation during the monsoon season over the Indian subcontinent occurs in form of enormously complex spatiotemporal patterns due to the underlying dynamics of atmospheric circulation and varying topography. Employing methods from nonlinear time series analysis, we study spatial structures of the rainfall field during the summer monsoon and identify principle regions where the dynamics of monsoonal rainfall is more coherent or homogenous. Moreover, we estimate the time delay patterns of rain events. Here we present an analysis of two separate high resolution gridded data sets of daily rainfall covering the Indian subcontinent. Using the method of event synchronization (ES), we estimate regions where heavy rain events during monsoon happen in some lag synchronised form. Further using the delay behaviour of rainfall events, we estimate the directionalities related to the progress of such type of rainfall events. The Active (break) phase of a monsoon is characterised by an increase(decrease) of rainfall over certain regions of the Indian subcontinent. We show that our method is able to identify regions of such coherent rainfall activity.

Sign in / Sign up

Export Citation Format

Share Document