RN-Net: A Deep Learning Approach to 0–2 h Rainfall Nowcasting Based on Radar and Automatic Weather Station Data

Precipitation has an important impact on people’s daily life and disaster prevention and mitigation. However, it is difficult to provide more accurate results for rainfall nowcasting due to spin-up problems in numerical weather prediction models. Furthermore, existing rainfall nowcasting methods based on machine learning and deep learning cannot provide large-area rainfall nowcasting with high spatiotemporal resolution. This paper proposes a dual-input dual-encoder recurrent neural network, namely Rainfall Nowcasting Network (RN-Net), to solve this problem. It takes the past grid rainfall data interpolated by automatic weather stations and doppler radar mosaic data as input data, and then forecasts the grid rainfall data for the next 2 h. We conduct experiments on the Southeastern China dataset. With a threshold of 0.25 mm, the RN-Net’s rainfall nowcasting threat scores have reached 0.523, 0.503, and 0.435 within 0.5 h, 1 h, and 2 h. Compared with the Weather Research and Forecasting model rainfall nowcasting, the threat scores have been increased by nearly four times, three times, and three times, respectively.

Download Full-text

Numerical Prediction of Submesoscale Flow in the Nocturnal Stable Boundary Layer over Complex Terrain

Monthly Weather Review ◽

10.1175/mwr-d-11-00061.1 ◽

2012 ◽

Vol 140 (3) ◽

pp. 956-977 ◽

Cited By ~ 24

Author(s):

Nelson L. Seaman ◽

Brian J. Gaudet ◽

David R. Stauffer ◽

Larry Mahrt ◽

Scott J. Richardson ◽

...

Keyword(s):

Prediction Models ◽

Weather Prediction ◽

Weather Research And Forecasting ◽

Stable Boundary Layers ◽

Stable Boundary ◽

Central Pennsylvania ◽

Stable Conditions ◽

Improved Model ◽

Mean Square Errors ◽

Numerical Weather Prediction Models

Abstract Numerical weather prediction models often perform poorly for weakly forced, highly variable winds in nocturnal stable boundary layers (SBLs). When used as input to air-quality and dispersion models, these wind errors can lead to large errors in subsequent plume forecasts. Finer grid resolution and improved model numerics and physics can help reduce these errors. The Advanced Research Weather Research and Forecasting model (ARW-WRF) has higher-order numerics that may improve predictions of finescale winds (scales <~20 km) in nocturnal SBLs. However, better understanding of the physics controlling SBL flow is needed to take optimal advantage of advanced modeling capabilities. To facilitate ARW-WRF evaluations, a small network of instrumented towers was deployed in the ridge-and-valley topography of central Pennsylvania (PA). Time series of local observations and model forecasts on 1.333- and 0.444-km grids were filtered to isolate deterministic lower-frequency wind components. The time-filtered SBL winds have substantially reduced root-mean-square errors and biases, compared to raw data. Subkilometer horizontal and very fine vertical resolutions are found to be important for reducing model speed and direction errors. Nonturbulent fluctuations in unfiltered, very finescale winds, parts of which may be resolvable by ARW-WRF, are shown to generate horizontal meandering in stable weakly forced conditions. These submesoscale motions include gravity waves, primarily horizontal 2D motions, and other complex signatures. Vertical structure and low-level biases of SBL variables are shown to be sensitive to parameter settings defining minimum “background” mixing in very stable conditions in two representative turbulence schemes.

Download Full-text

On the effective resolution of WRF simulations at microscale grid resolution.

10.5194/egusphere-egu21-229 ◽

2021 ◽

Author(s):

Pedro Bolgiani ◽

Javier Díaz-Fernández ◽

Lara Quitián-Hernández ◽

Mariano Sastre ◽

Daniel Santos-Muñoz ◽

...

Keyword(s):

Prediction Models ◽

Weather Prediction ◽

Wrf Model ◽

Research Community ◽

Weather Research And Forecasting ◽

Large Uncertainty ◽

Partial Explanation ◽

Synoptic Conditions ◽

Large Effort ◽

Numerical Weather Prediction Models

<p>As the computational capacity has been largely improved in the last decades, the grid configuration of numerical weather prediction models has stepped into microscale resolutions. Even if mesoscale models are not originally designed to reproduce fine scale phenomena, a large effort is being made by the research community to improve and adapt these systems. However, reasonable doubts exist regarding the ability of the models to forecast this type of events, due to the unfit parametrizations and the appearance of instabilities and lack of sensitivity in the variables. Here, the Weather Research and Forecasting (WRF) model effective resolution is evaluated for several situations and grid resolutions. This is achieved by assessing the curve of dissipation for the wind kinetic energy. Results show that the simulated energy spectrum responds to different synoptic conditions. Nevertheless, when the model is forced into microscale grid resolutions the dissipation curves present an unrealistic atmospheric energy. This may be a partial explanation to the aforementioned issues and imposes a large uncertainty to forecasting at these resolutions.</p>

Download Full-text

Adaptive Kalman Filtering for Postprocessing Ensemble Numerical Weather Predictions

Monthly Weather Review ◽

10.1175/mwr-d-17-0084.1 ◽

2017 ◽

Vol 145 (12) ◽

pp. 4837-4854 ◽

Cited By ~ 13

Author(s):

Anna Pelosi ◽

Hanoi Medina ◽

Joris Van den Bergh ◽

Stéphane Vannitsem ◽

Giovanni Battista Chirico

Keyword(s):

Kalman Filtering ◽

Prediction Models ◽

Southern Italy ◽

Weather Prediction ◽

Adaptive Method ◽

Ensemble Forecasts ◽

Numerical Weather ◽

Weather Stations ◽

Meteorological Observations ◽

Numerical Weather Prediction Models

Forecasts from numerical weather prediction models suffer from systematic and nonsystematic errors, which originate from various sources such as subgrid-scale variability affecting large scales. Statistical postprocessing techniques can partly remove such errors. Adaptive MOS techniques based on Kalman filters (here called AMOS), are used to sequentially postprocess the forecasts, by continuously updating the correction parameters as new ground observations become available. These techniques, originally proposed for deterministic forecasts, are valuable when long training datasets do not exist. Here, a new adaptive postprocessing technique for ensemble predictions (called AEMOS) is introduced. The proposed method implements a Kalman filtering approach that fully exploits the information content of the ensemble for updating the parameters of the postprocessing equation. A verification study for the region of Campania in southern Italy is performed. Two years (2014–15) of daily meteorological observations of 10-m wind speed and 2-m temperature from 18 ground-based automatic weather stations are used, comparing them with the corresponding COSMO-LEPS ensemble forecasts. It is shown that the proposed adaptive method outperforms the AMOS method, while it shows comparable results to the member-by-member batch postprocessing approach.

Download Full-text

The Latest Update of JMA Numerical Weather Prediction Models and its Solar Power Forecasting Errors

IEEJ Transactions on Power and Energy ◽

10.1541/ieejpes.138.881 ◽

2018 ◽

Vol 138 (11) ◽

pp. 881-892 ◽

Cited By ~ 1

Author(s):

Hideaki Ohtake ◽

Fumichika Uno ◽

Takashi Oozeki ◽

Yoshinori Yamada

Keyword(s):

Numerical Weather Prediction ◽

Solar Power ◽

Prediction Models ◽

Weather Prediction ◽

Numerical Weather ◽

Solar Power Forecasting ◽

Forecasting Errors ◽

Numerical Weather Prediction Models ◽

Power Forecasting

Download Full-text

Forecasting of Tornadoes and Downbursts

The Oxford Handbook of Non-Synoptic Wind Storms ◽

10.1093/oxfordhb/9780190670252.013.3 ◽

2020 ◽

Author(s):

Djordje Romanic

Keyword(s):

Chaos Theory ◽

Prediction Models ◽

Weather Prediction ◽

Small Scale ◽

Weather Forecasts ◽

Wind Speeds ◽

The Past ◽

Extreme Wind ◽

Extreme Wind Speeds ◽

Numerical Weather Prediction Models

Tornadoes and downbursts cause extreme wind speeds that often present a threat to human safety, structures, and the environment. While the accuracy of weather forecasts has increased manifold over the past several decades, the current numerical weather prediction models are still not capable of explicitly resolving tornadoes and small-scale downbursts in their operational applications. This chapter describes some of the physical (e.g., tornadogenesis and downburst formation), mathematical (e.g., chaos theory), and computational (e.g., grid resolution) challenges that meteorologists currently face in tornado and downburst forecasting.

Download Full-text

Machine learning based algorithms for uncertainty quantification in numerical weather prediction models

Journal of Computational Science ◽

10.1016/j.jocs.2020.101295 ◽

2021 ◽

Vol 50 ◽

pp. 101295

Author(s):

Azam Moosavi ◽

Vishwas Rao ◽

Adrian Sandu

Keyword(s):

Machine Learning ◽

Uncertainty Quantification ◽

Numerical Weather Prediction ◽

Prediction Models ◽

Weather Prediction ◽

Numerical Weather ◽

Numerical Weather Prediction Models

Download Full-text

Parametrizations of Liquid and Ice Clouds’ Optical Properties in Operational Numerical Weather Prediction Models

Atmosphere ◽

10.3390/atmos12010089 ◽

2021 ◽

Vol 12 (1) ◽

pp. 89

Author(s):

Harel. B. Muskatel ◽

Ulrich Blahak ◽

Pavel Khain ◽

Yoav Levi ◽

Qiang Fu

Keyword(s):

Optical Properties ◽

Numerical Weather Prediction ◽

Radiation Transfer ◽

Prediction Models ◽

Weather Prediction ◽

Effective Size ◽

Asymmetry Factor ◽

Numerical Weather ◽

Numerical Weather Prediction Models ◽

Spectral Averaging

Parametrization of radiation transfer through clouds is an important factor in the ability of Numerical Weather Prediction models to correctly describe the weather evolution. Here we present a practical parameterization of both liquid droplets and ice optical properties in the longwave and shortwave radiation. An advanced spectral averaging method is used to calculate the extinction coefficient, single scattering albedo, forward scattered fraction and asymmetry factor (bext, v, f, g), taking into account the nonlinear effects of light attenuation in the spectral averaging. An ensemble of particle size distributions was used for the ice optical properties calculations, which enables the effective size range to be extended up to 570 μm and thus be applicable for larger hydrometeor categories such as snow, graupel, and rain. The new parameterization was applied both in the COSMO limited-area model and in ICON global model and was evaluated by using the COSMO model to simulate stratiform ice and water clouds. Numerical weather prediction models usually determine the asymmetry factor as a function of effective size. For the first time in an operational numerical weather prediction (NWP) model, the asymmetry factor is parametrized as a function of aspect ratio. The method is generalized and is available on-line to be readily applied to any optical properties dataset and spectral intervals of a wide range of radiation transfer models and applications.

Download Full-text