scholarly journals Nowcasting of surface wind speed using probabilistic, explainable Deep Learning

2021 ◽  
Author(s):  
Francesco Zanetta ◽  
Daniele Nerini
Keyword(s):  
Deep Learning ◽  
Wind Speed ◽  
Real Time ◽  
Weather Prediction ◽  
Surface Wind ◽  
Similarity Index ◽  
Surface Wind Speed ◽  
Second Phase ◽  
Post Processing ◽  
Wide Range

<div> <div> <div> <p>Surface wind is an extremely difficult parameter to predict, particularly in the complex topography of the Alps. Due to several important processes happening at sub-kilometer scale, even high resolution Numerical Weather Prediction models such as COSMO-1 still present substantial biases. To address this, a wide range of statistical post-processing methods are used. Recently, methods based on Deep Learning have emerged as a new solution and are now actively developed at many weather services, including MeteoSwiss. At the same time, efforts are made to obtain accurate representations of surface wind speed up to a few hours ahead by integrating all available information in real-time, an approach known as nowcasting.</p> <p>With the aim of seamlessly combining nowcasting and post-processing approaches for surface wind speed predictions, we developed a Deep Learning probabilistic post-processing model that is also able to integrate real time observations, and developed a new metric, the Similarity Index, for this purpose. The Similarity Index is a way to estimate the correlation of surface wind speed between two locations, based on their position and geomorphological setting, and can be used to choose the best available observation to be used at any point in space at any given time, and weigh that observation in a way that mimics geostatistical interpolation methods. The proposed methodology yields improved forecasts of wind speed where both systematic and random errors are reduced, thanks to the post-processing and nowcasting components respectively. In a second phase, we implemented a state- of-the-art explainability framework for machine learning, SHAP, and presented how it can be used to get insights into the model and build trust in the results.</p> </div> </div> </div>

An Investigation of the Stability Dependence of SST-Induced Vertical Mixing over the Ocean in the Operational Met Office Model

2017 ◽  
Vol 30 (1) ◽  
pp. 91-107 ◽  
Author(s):  
Qingtao Song ◽  
Dudley B. Chelton ◽  
Steven K. Esbensen ◽  
Andrew R. Brown
Keyword(s):  
Wind Speed ◽  
Weather Prediction ◽  
Surface Wind ◽  
Vertical Mixing ◽  
Surface Wind Speed ◽  
Satellite Observations ◽  
Sea Surface ◽  
Stable Conditions ◽  
Wide Range ◽  
The Impact

This study presents an assessment of the impact of a March 2006 change in the Met Office operational global numerical weather prediction model through the introduction of a nonlocal momentum mixing scheme. From comparisons with satellite observations of surface wind speed and sea surface temperature (SST), it is concluded that the new parameterization had a relatively minor impact on SST-induced changes in sea surface wind speed in the Met Office model in the September and October 2007 monthly averages over the Agulhas Return Current region considered here. The performance of the new parameterization of vertical mixing was evaluated near the surface layer and further through comparisons with results obtained using a wide range of sensitivity of mixing parameterization to stability in the Weather Research and Forecasting (WRF) Model, which is easily adapted to such sensitivity studies. While the new parameterization of vertical mixing improves the Met Office model response to SST in highly unstable (convective) conditions, it is concluded that significantly enhanced vertical mixing in the neutral to moderately unstable conditions (nondimensional stability [Formula: see text] between 0 and −2) typically found over the ocean is required in order for the model surface wind response to SST to match the satellite observations. Likewise, the reduced mixing in stable conditions in the new parameterization is also relatively small; for the range of the gradient Richardson number typically found over the ocean, the mixing was reduced by a maximum of only 10%, which is too small by more than an order of magnitude to be consistent with the satellite observations.

scholarly journals Wind Speed Forecast Based on Post-Processing of Numerical Weather Predictions Using a Gradient Boosting Decision Tree Algorithm

Atmosphere ◽  
2020 ◽  
Vol 11 (7) ◽  
pp. 738 ◽  
Author(s):  
Wenqing Xu ◽  
Like Ning ◽  
Yong Luo
Keyword(s):  
Wind Speed ◽  
Decision Tree ◽  
Wind Power ◽  
Surface Wind ◽  
Surface Wind Speed ◽  
Gradient Boosting ◽  
Post Processing ◽  
Near Surface ◽  
Wind Power Forecasting ◽  
Numerical Weather

With the large-scale development of wind energy, wind power forecasting plays a key role in power dispatching in the electric power grid, as well as in the operation and maintenance of wind farms. The most important technology for wind power forecasting is forecasting wind speed. The current mainstream methods for wind speed forecasting involve the combination of mesoscale numerical meteorological models with a post-processing system. Our work uses the WRF model to obtain the numerical weather forecast and the gradient boosting decision tree (GBDT) algorithm to improve the near-surface wind speed post-processing results of the numerical weather model. We calculate the feature importance of GBDT in order to find out which feature most affects the post-processing wind speed results. The results show that, after using about 300 features at different height and pressure layers, the GBDT algorithm can output more accurate wind speed forecasts than the original WRF results and other post-processing models like decision tree regression (DTR) and multi-layer perceptron regression (MLPR). Using GBDT, the root mean square error (RMSE) of wind speed can be reduced from 2.7–3.5 m/s in the original WRF result by 1–1.5 m/s, which is better than DTR and MLPR. While the index of agreement (IA) can be improved by 0.10–0.20, correlation coefficient be improved by 0.10–0.18, Nash–Sutcliffe efficiency coefficient (NSE) be improved by −0.06–0.6. It also can be found that the feature which most affects the GBDT results is the near-surface wind speed. Other variables, such as forecast month, forecast time, and temperature, also affect the GBDT results.

scholarly journals Improved near real time surface wind resolution over the Mediterranean Sea

Ocean Science ◽  
2007 ◽  
Vol 3 (2) ◽  
pp. 259-271 ◽  
Author(s):  
A. Bentamy ◽  
H.-L. Ayina ◽  
P. Queffeulou ◽  
D. Croize-Fillon ◽  
V. Kerbaol
Keyword(s):  
Wind Speed ◽  
Mediterranean Sea ◽  
Real Time ◽  
Surface Wind ◽  
Surface Wind Speed ◽  
Remotely Sensed ◽  
Wind Fields ◽  
Remotely Sensed Data ◽  
Space And Time ◽  
The Mediterranean

Abstract. Several scientific programs, including the Mediterranean Forecasting System Toward Environmental Predictions (MFSTEP project), request high space and time resolutions of surface wind speed and direction. The purpose of this paper is to focus on surface wind improvements over the global Mediterranean Sea, based on the blending near real time remotely sensed wind observations and ECMWF wind analysis. Ocean surface wind observations are retrieved from QuikSCAT scatterometer and from SSM/I radiometers available at near real time at Météo-France. Using synchronous satellite data, the number of remotely sensed data available for each analysis epoch (00:00 h; 06:00 h; 12:00 h; 18:00 h) is not uniformly distributed as a function of space and time. On average two satellite wind observations are available for each analysis time period. The analysis is performed by optimum interpolation (OI) based on the kriging approach. The needed covariance matrixes are estimated from the satellite wind speed, zonal and meridional component observations. The quality of the 6-hourly resulting blended wind fields on 0.25° grid are investigated trough comparisons with the remotely sensed observations as well as with moored buoy wind averaged wind estimates. The blended wind data and remotely wind observations, occurring within 3 h and 0.25° from the analysis estimates, compare well over the global basin as well as over the sub-basins. The correlation coefficients exceed 0.95 while the rms difference values are less than 0.30 m/s. Using measurements from moored buoys, the high-resolution wind fields are found to have similar accuracy as satellite wind retrievals. Blended wind estimates exhibit better comparisons with buoy moored in open sea than near shore.

scholarly journals Evaluation of Spaceborne GNSS-R Retrieved Ocean Surface Wind Speed with Multiple Datasets

2019 ◽  
Vol 11 (23) ◽  
pp. 2747 ◽  
Author(s):  
Zhounan Dong ◽  
Shuanggen Jin
Keyword(s):  
Wind Speed ◽  
Weather Prediction ◽  
Surface Wind ◽  
Ocean Surface ◽  
Surface Wind Speed ◽  
Retrieval Algorithm ◽  
Navigation Satellite ◽  
Ocean Surface Wind ◽  
Global Navigation ◽  
Global Navigation Satellite

Spaceborne Global Navigation Satellite Systems-Reflectometry (GNSS-R) can estimate the geophysical parameters by receiving Earth’s surface reflected signals. The CYclone Global Navigation Satellite System (CYGNSS) mission with eight microsatellites launched by NASA in December 2016, which provides an unprecedented opportunity to rapidly acquire ocean surface wind speed globally. In this paper, a refined spaceborne GNSS-R sea surface wind speed retrieval algorithm is presented and validated with the ground surface reference wind speed from numerical weather prediction (NWP) and cross-calibrated multi-platform ocean surface wind vector analysis product (CCMP), respectively. The results show that when the wind speed was less than 20 m/s, the RMS of the GNSS-R retrieved wind could achieve 1.84 m/s in the case where the NWP winds were used as the ground truth winds, while the result was better than the NWP-based retrieved wind speed with an RMS of 1.68 m/s when the CCMP winds were used. The two sets of inversion results were further evaluated by the buoy winds, and the uncertainties from the NWP-derived and CCMP-derived model prediction wind speed were 1.91 m/s and 1.87 m/s, respectively. The accuracy of inversed wind speeds for different GNSS pseudo-random noise (PRN) satellites and types was also analyzed and presented, which showed similar for different PRN satellites and different types of satellites.

scholarly journals Validation and Calibration of Significant Wave Height and Wind Speed Retrievals from HY2B Altimeter Based on Deep Learning

2020 ◽  
Vol 12 (17) ◽  
pp. 2858
Author(s):  
Jiuke Wang ◽  
Lotfi Aouf ◽  
Yongjun Jia ◽  
Youguang Zhang
Keyword(s):  
Deep Learning ◽  
Wind Speed ◽  
Wave Height ◽  
Significant Wave Height ◽  
Surface Wind ◽  
Surface Wind Speed ◽  
Significant Wave ◽  
Sea Surface Wind Speed ◽  
Sea Surface Wind ◽  
Key Parameter

HY2B is now the latest altimetry mission that provides global nadir significant wave height (SWH) and sea surface wind speed. The validation and calibration of HY2B are carried out against National Data Buoy Center (NDBC) buoy observations from April 2019 to April 2020. In general, the HY2B altimeter measurements agree well with buoy observation, with scatter index of 9.4% for SWH, and 15.1% for wind speed. However, we observed a significant bias of 0.14 m for SWH and −0.42 m/s for wind speed. A deep learning technique is novelly applied for the calibration of HY2B SWH and wind speed. Deep neural network (DNN) is built and trained to correct SWH and wind speed by using input from parameters provided by the altimeter such as sigma0, sigma0 standard deviation (STD). The results based on DNN show a significant reduction of the bias, root mean square error (RMSE), and scatter index (SI) for both SWH and wind speed. Several DNN schemes based on different combination of input parameters have been examined in order to obtain the best model for the calibration. The analysis reveals that sigma0 STD is a key parameter for the calibration of HY2B SWH and wind speed.

scholarly journals Impact Assessment of Assimilating NASA’s RapidScat Surface Wind Retrievals in the NOAA Global Data Assimilation System

2018 ◽  
Vol 146 (4) ◽  
pp. 929-942
Author(s):  
Ling Liu ◽  
Kevin Garrett ◽  
Eric S. Maddy ◽  
Sid-Ahmed Boukabara
Keyword(s):  
Wind Speed ◽  
Weather Prediction ◽  
Surface Wind ◽  
Space Station ◽  
Surface Wind Speed ◽  
Background Field ◽  
Wind Vector ◽  
Forecast Models ◽  
Wind Retrievals ◽  
The Impact

The National Aeronautics and Space Administration (NASA) RapidScat scatterometer on board the International Space Station (ISS) provides observations of surface winds that can be assimilated into numerical weather prediction (NWP) forecast models. In this study, the authors assess the data quality of the RapidScat Level 2B surface wind vector retrievals and the impact of those observations on the National Oceanic and Atmospheric Administration (NOAA) Global Forecast System (GFS). The RapidScat is found to provide quality measurements of surface wind speed and direction in nonprecipitating conditions and to provide observations that add both information and robustness to the global satellite observing system used in NWP models. The authors find that with an assumed uncertainty in wind speed of around 2 m s−1, the RapidScat has neutral impact on the short-range forecast of surface wind vectors in the tropics but improves both the analysis and background field of surface wind vectors. However, the deployment of RapidScat on the ISS presents some challenges for use of these wind vector observations in operational NWP, including frequent maneuvers of the spacecraft that could alter instrument performance.

scholarly journals A statistical model for sea surface diurnal warming driven by numerical weather prediction fluxes and winds

2010 ◽  
Vol 7 (4) ◽  
pp. 1497-1532 ◽  
Author(s):  
M. J. Filipiak ◽  
C. J. Merchant ◽  
H. Kettle ◽  
P. Le Borgne
Keyword(s):  
Heat Flux ◽  
Wind Speed ◽  
Statistical Model ◽  
Weather Prediction ◽  
Surface Wind ◽  
Surface Wind Speed ◽  
Surface Heat ◽  
Sea Surface ◽  
Diurnal Warming ◽  
Nwp Model

Abstract. A statistical model is derived relating the diurnal variation of sea surface temperature (SST) to the net surface heat flux and surface wind speed from a numerical weather prediction (NWP) model. The model is derived using fluxes and winds from the European Centre for Medium-Range Weather Forecasting (ECMWF) NWP model and SSTs from the Spinning Enhanced Visible and Infrared Imager (SEVIRI). In the model, diurnal warming has a linear dependence on the net surface heat flux integrated since (approximately) dawn and an inverse quadratic dependence on the maximum of the surface wind speed in the same period. The model coefficients are found by matching, for a given integrated heat flux, the frequency distributions of the maximum wind speed deceedance and the observed warming exceedance. Diurnal cooling, where it occurs, is modelled as proportional to the integrated heat efflux divided by the heat capacity of the seasonal mixed layer. The model reproduces the statistics (mean, standard deviation, and 95-percentile) of the diurnal variation of SST seen by SEVIRI and reproduces the geographical pattern of mean warming seen by the Advanced Microwave Scanning Radiometer (AMSR-E). We use the functional dependencies in the statistical model to test the behaviour of two physical model of diurnal warming that display contrasting systematic errors.

scholarly journals Improved near real time surface wind resolution over the Mediterranean Sea

2006 ◽  
Vol 3 (3) ◽  
pp. 435-470 ◽  
Author(s):  
A. Bentamy ◽  
H.-L. Ayina ◽  
P. Queffeulou ◽  
D. Croize-Fillon
Keyword(s):  
Wind Speed ◽  
Mediterranean Sea ◽  
Real Time ◽  
Surface Wind ◽  
Surface Wind Speed ◽  
Remotely Sensed ◽  
Wind Fields ◽  
Remotely Sensed Data ◽  
Space And Time ◽  
The Mediterranean

Abstract. Several scientific programs, including the Mediterranean Forecasting System Toward Environmental Predictions (MFSTEP project), request high space and time resolutions of surface wind speed and direction. The purpose of this paper is to focus on surface wind improvements over the global Mediterranean Sea, based on the blending near real time remotely sensed wind observations and ECMWF wind analysis. Ocean surface wind observations are retrieved from QuikSCAT scatterometer and from SSM/I radiometers available at near real time at Météo-France. Using synchronous satellite data, the number of remotely sensed data available for each analysis epoch (00:00 h; 06:00 h; 12:00 h; 18:00 h) is not uniformly distributed as a function of space and time. On average two satellite wind observations are available for each analysis time period. The analysis is performed by optimum interpolation (OI) based on the kriging approach. The needed covariance matrixes are estimated from the satellite wind speed, zonal and meridional component observations. The quality of the 6-hourly resulting blended wind fields on 0.25° grid are investigated trough comparisons with the remotely sensed observations as well as with moored buoy wind averaged wind estimates. The blended wind data and remotely wind observations, occurring within 3 h and 0.25° from the analysis estimates, compare well over the global basin as well as over the sub-basins. The correlation coefficients exceed 0.95 while the rms difference values are less than 0.30 m/s. Using measurements from moored buoys, the high-resolution wind fields are found to have similar accuracy as satellite wind retrievals. Blended wind estimates exhibit better comparisons with buoy moored in open sea than near shore.

Sign in / Sign up

Export Citation Format

Share Document