Numerical Weather Prediction Models Research Articles

Impact of Saharan dust outbreaks on short‐range weather forecast errors in Europe

AbstractMineral dust, the most abundant atmospheric aerosol by mass, interacts with radiation directly and alters cloud properties indirectly. Many operational numerical weather prediction models account for aerosol direct effects by using climatological mean concentrations and neglect indirect effects. This simplification may lead to shortcomings in model forecasts during outbreaks of Saharan dust towards Europe, when climatological mean dust concentrations deviate strongly from actual concentrations. This study investigates errors in model analyses and short‐range forecasts during such events. We investigate a pronounced dust event in March 2021 using the pre‐operational ICOsahedral Nonhydrostatic weather and climate model with Aerosols and Reactive Trace gases (ICON‐ART) with prognostic calculation of dust and the operational European Centre for Medium‐Range Weather Forecasts (ECMWF) Integrated Forecasting System (IFS) model, which deploys a dust climatology. We compare model analysis and forecast with measurements from satellite and in situ instruments. We find that inclusion of prognostic aerosol and direct radiative effects from dust improves forecasts of surface radiation during clear‐sky conditions. However, dust‐induced cirrus clouds are strongly underestimated, highlighting the importance of representing indirect effects adequately. These findings are corroborated by systematic quantification of forecast errors against satellite measurements. For this we construct an event catalogue with 49 dust days over Central Europe between January 2018 and March 2022. We classify model cells by simulated and observed cloudiness and simulated dustiness in the total atmospheric column. We find significant overestimations of brightness temperature for cases with dust compared with cases without dust. For surface shortwave radiation, we find median overestimations of 6.2% during cloudy conditions with dust optical depth greater than 0.1, however these are not significant compared with cloudy conditions without dust. Our findings show that the pre‐operational ICON‐ART and the operational IFS model still do not reproduce cloudiness adequately during events with Saharan dust over Central Europe. Missing implementations of prognostic dust, particularly of indirect effects on cloud formation, lead to significant underestimations of cloudiness and potentially overestimations of surface radiation.

Univariate Flux Partition Functions for Planetary Boundary Layer Schemes at Gray Zone Resolutions

Abstract When the horizontal grid spacing of a numerical weather prediction model approaches kilometer scale, the so-called gray zone range, turbulent fluxes in the convective boundary layer (CBL) are partially resolved and partially subgrid scale (SGS). Knowledge of the partition between resolved and SGS turbulent fluxes is key to building scale-adaptive planetary boundary layer (PBL) schemes that are capable of regulating the SGS fluxes with varying grid spacing. However, flux partition depends not only on horizontal grid spacing, but also on local height, bulk stability of the boundary layer, and the particular turbulent flux. Such multivariate functions are difficult to construct analytically, so their implementations in scale-adaptive PBL schemes always involve certain levels of approximation that can lead to inaccuracies. This study introduces a physically based perspective for the flux partition functions that greatly simplifies their implementation with high accuracy. By introducing an appropriate scaling length λ that accounts for both height and bulk stability dependencies, the dimensionality of the partition functions is reduced to a single dimensionless group. Based on the analysis of a comprehensive large-eddy simulation dataset of the CBL, it is further shown that λ’s height and bulk stability dependencies can be separately represented by a similarity length scale and a stability coefficient. The resulting univariate partition functions are incorporated into a traditional first-order PBL scheme as a proof of concept. Our results show that the augmented scheme well-reproduces the SGS fluxes at gray zone resolutions. Significance Statement Flux partition functions are a key component in most scale-adaptive planetary boundary layer (PBL) schemes developed for kilometer- and subkilometer-resolution numerical weather prediction models. They regulate the parameterized turbulent fluxes as a function of horizontal grid spacing, while they also depend on height and atmospheric stability. Such multivariate dependencies forbid simple analytical expressions, and as a result, partition functions implemented in scale-adaptive PBL schemes are generally simplified at the cost of accuracy in previous works. This study investigates the possibility of constructing partition functions that are both accurate and easy to parameterize. Utilizing a physically based length scale, univariate partition functions are built, evaluated, and put into a conventional PBL scheme to improve the gray zone turbulence parameterization.

On the Importance of Regime-Specific Evaluations for Numerical Weather Prediction Models as Demonstrated using the High Resolution Rapid Refresh (HRRR) Model

Abstract The scientific literature has many studies evaluating numerical weather prediction (NWP) models. However, many of those studies averaged across a myriad of different atmospheric conditions and surface forcings which can obfuscate the atmospheric conditions when NWP models perform well versus when they perform inadequately. To help isolate these different scenarios, we used observations from the U.S. Climate Reference Network (USCRN) obtained between 1 January and 31 December 2021 to distinguish among different near-surface atmospheric conditions (i.e., different near-surface heating rates (), incoming shortwave radiation (SWd) regimes, and 5-cm soil moisture (SM05)) to evaluate the High-Resolution Rapid Refresh (HRRR) model, which is a 3-km model used for operational weather forecasting in the U.S. On days with small (large) , we found afternoon T biases of about 2°C (−1°C) and afternoon SWd biases of up to 170 W m−2 (100 W m−2), but negligible impacts on SM05 biases. On days with small (large) SWd, we found daytime temperature biases of about 3°C (−2.5°C) and daytime SWd biases of up to 190 W m−2 (80 W m−2). Whereas different SM05 had little impact on T and SWd biases, dry (wet) conditions had positive (negative) SM05 biases. We argue that the proper evaluation of weather forecasting models requires careful consideration of different near-surface atmospheric conditions and is critical to better identifying model deficiencies which supports improvements to the parameterization schemes used therein. A similar, regime-specific model verification approach may also be used to help evaluate other geophysical models.

Operational Aviation Icing Forecast Algorithm for the Korea Meteorological Administration

Abstract In this study, we developed and evaluated the Korean Forecast Icing Potential (K-FIP), an in-flight icing forecast system for the Korea Meteorological Administration (KMA) based on the simplified forecast icing potential (SFIP) algorithm. The SFIP is an algorithm used to postprocess numerical weather prediction (NWP) model forecasts for predicting potential areas of icing based on the fuzzy logic formulations of four membership functions: temperature, relative humidity, vertical velocity, and cloud liquid water content. In this study, we optimized the original version of the SFIP for the global NWP model of the KMA through three important updates using 34 months of pilot reports for icing as follows: using total cloud condensates, reconstructing membership functions, and determining the best weight combination for input variables. The use of all cloud condensates and the reconstruction of these membership functions resulted in a significant improvement in the algorithm compared with the original. The weight combinations for the KMA’s global model were determined based on the performance scores. While several sets of weights performed equally well, this process identified the most effective weight combination for the KMA model, which is referred to as the K-FIP. The K-FIP demonstrated the ability to successfully predict icing over the Korean Peninsula using observations made by research aircraft from the National Institute of Meteorological Sciences of the KMA. Eventually, the K-FIP icing forecasts will provide better forecasts of icing potentials for safe and efficient aviation operations in South Korea. Significance Statement In-flight aircraft icing has posed a threat to safe flights for decades. With advances in computing resources and an improvement in the spatiotemporal resolutions of numerical weather prediction (NWP) models, icing algorithms have been developed using NWP model outputs associated with supercooled liquid water. This study evaluated and optimized the simplified forecast icing potential, an NWP model–based icing algorithm, for the global model of the Korean Meteorological Administration (KMA) using a long-term observational dataset to improve its prediction skills. The improvements shown in this study and the SFIP implemented in the KMA will provide more informative predictions for safe and efficient air travel.

Spatiotemporal Convolutional Approach for the Short-Term Forecast of Hourly Heavy Rainfall Probability Integrating Numerical Weather Predictions and Surface Observations

Abstract The accurate prediction of short-term rainfall, and in particular the forecast of hourly heavy rainfall (HHR) probability, remains challenging for numerical weather prediction (NWP) models. Here, we introduce a deep learning (DL) model, PredRNNv2-AWS, a convolutional recurrent neural network designed for deterministic short-term rainfall forecasting. This model integrates surface rainfall observations and atmospheric variables simulated by the Precision Weather Analysis and Forecasting System (PWAFS). Our DL model produces realistic hourly rainfall forecasts for the next 13 h. Quantitative evaluations show that the use of surface rainfall observations as one of the predictors achieves higher performance (threat score) with 263% and 186% relative improvements over NWP simulations for the first 3 h and the entire forecast hours, respectively, at a threshold of 5 mm h−1. Noting that the optical-flow method also performs well in the initial hours, its predictions quickly worsen in the final hours compared to other experiments. The machine learning model, LightGBM, is then integrated to classify HHR from the predicted hourly rainfall of PredRNNv2-AWS. The results show that PredRNNv2-AWS can better reflect actual HHR conditions compared with PredRNNv2 and PWAFS. A representative case demonstrates the superiority of PredRNNv2-AWS in predicting the evolution of the rainy system, which substantially improves the accuracy of the HHR prediction. A test case involving the extreme flood event in Zhengzhou exemplifies the generalizability of our proposed model. Our model offers a reliable framework to predict target variables that can be obtained from numerical simulations and observations, e.g., visibility, wind power, solar energy, and air pollution.

Improvement of Albedo and Snow-Cover Simulation during Snow Events over the Tibetan Plateau

Abstract The snow albedo is a vital component of land–atmosphere coupling models. It plays a critical role in regulating land surface energy exchange by controlling incoming solar radiation absorbed by the land surface and influencing the timing and rate of snowmelt. Accurate snow albedo simulation is essential to obtain surface energy balance and snow-cover estimates. Here, the simulation of albedo and snow cover using the Weather Research and Forecasting Model and an improved snow albedo scheme is verified against satellite-retrieved products during and immediately following eight snowfall events over the Tibetan Plateau. The improved model successfully characterizes the spatial pattern and inverted U-shaped temporal pattern of albedo over the entire Tibetan Plateau. This is attributed to the local optimization of snow-age parameters and explicit consideration of snow depth in the improved scheme. Compared with the previous model, the model proposed herein greatly decreases the overestimated albedo (by 0.13–0.27), yielding a bias range of ±0.08, mean relative bias decrease of 70%, and significant increase in the spatial correlation coefficient of 0.03–0.39 (mean: 0.13). The significant improvements of albedo estimates appear in deep snow-covered regions, largely attributed to parameter optimization related to snow albedo decay, while less improvements appear over the shallow snow-covered regions. Accurate reproduction of the spatiotemporal variation in albedo alleviated snow-cover overestimation by small amounts. For snow-cover estimates, the improved model consistently decreases the false-alarm rate by 0.03, and increases the overall accuracy and equitable threat score by 0.04 and 0.03, respectively. Moreover, the improved scheme shows an equivalent improvement of albedo estimates at both 1- and 5-km grid spacing over the eastern Tibetan Plateau; this is also true for snow-cover estimates. Significance Statement Snow albedo schemes in widely used numerical weather prediction models show notable shortcomings in complex mountainous regions, hindering accurate surface energy balance and snow-cover prediction. The purpose of this study is to better understand the role of snow albedo on snow-cover estimates and reveal the application potential of an improved snow albedo scheme across the Tibetan Plateau. This is important because snow albedo influences the timing and rate of snowmelt, and in turn snow-cover estimates, through regulating the surface energy budget. Our results highlight the strong application potential of our improved scheme in reducing snow simulation errors, confirm the importance of snow depth on snow albedo, and provide a new perspective for improving the accuracy of snow forecast over the topographically high Tibetan Plateau.

Deep Learning for Postprocessing Global Probabilistic Forecasts on Subseasonal Time Scales

Abstract Subseasonal weather forecasts are becoming increasingly important for a range of socioeconomic activities. However, the predictive ability of physical weather models is very limited on these time scales. We propose four postprocessing methods based on convolutional neural networks to improve subseasonal forecasts by correcting systematic errors of numerical weather prediction models. Our postprocessing models operate directly on spatial input fields and are therefore able to retain spatial relationships and to generate spatially homogeneous predictions. They produce global probabilistic tercile forecasts for biweekly aggregates of temperature and precipitation for weeks 3–4 and 5–6. In a case study based on a public forecasting challenge organized by the World Meteorological Organization, our postprocessing models outperform the bias-corrected forecasts from the European Centre for Medium-Range Weather Forecasts (ECMWF), and achieve improvements over climatological forecasts for all considered variables and lead times. We compare several model architectures and training modes and demonstrate that all approaches lead to skillful and well-calibrated probabilistic forecasts. The good calibration of the postprocessed forecasts emphasizes that our postprocessing models reliably quantify the forecast uncertainty based on deterministic input information in the form of ECMWF ensemble mean forecast fields only.

Multi‐Platform Observations of Severe Typhoon Koinu

AbstractSevere Typhoon Koinu passed south of Hong Kong on 8 and 9 October 2023, triggering the issuance of the Increasing Gale or Storm Signal No. 9, the second highest tropical cyclone (TC) warning signal in Hong Kong. Koinu was a difficult case for TC warning service due to its compact size and rather erratic movement over the coastal waters of Guangdong. To monitor Koinu's movement and wind structure, the Hong Kong Observatory utilized various observational platforms, including meteorological aircraft, ocean radar, and synthetic aperture radar on polar orbiting satellites. The paper presents major observations derived from these measurements. The aircraft probe and dropsonde data suggested boundary layer inflow, warm core structure, eyewall updraft, and high turbulence in the eyewall of the typhoon. The weather radar observations indicated the occurrence of a waterspout in the vicinity of the typhoon. Additionally, the study highlights the forecasting performance of the AI‐based Pangu‐Weather model, which could outperform the conventional global numerical weather prediction models in forecasting TC track in the region. The documentation of these observations aims to provide valuable references for weather forecasters and stimulate further research on forecasting this type of tropical cyclone.

Ice Crystal Alignment: A Concept for High-Temporal-Resolution Monitoring Using Polarimetric Phased Array Radar

Abstract This article describes a concept whereby future operational polarimetric phased array radars (PPAR) routinely monitor ice crystal alignment regions caused by thundercloud electric fields with volume scan updates (∼12 min−1) sufficient to resolve the temporal variation due to lightning and subsequent rapid electric field regeneration in nonsevere thunderstorms. Routine observations of crystal alignment regions may enhance thunderstorm nowcasting through comparison of their temporal and spatial structure with other polarimetric signatures, integration with lightning detection data, and assimilation into convection resolving numerical weather prediction models. If crystal alignment observations indicate strong electrification well in advance of the first lightning strike and likewise reliably indicate the decay of strong electric fields at the end of a storm, this capability may improve warning for lightning-sensitive activities such as airport ramp operations and space launch. Experimental observations of crystal alignment volumes in central Oklahoma severe storms and their relation to those storms’ structures are presented and used to motivate discussion of possible PPAR architectures. In one case—a tornadic supercell—these observations illustrate an important limitation. Even the hypothesized 12 min−1 volume scan update rate would not resolve the temporal variation of the crystal alignment regions in such storms, suggesting that special, adaptive scanning methods may be appropriate for such storms. We describe how future operational phased array radars could support a crystal alignment measurement mode via parallel, time-multiplexed processing and discuss potential impacts on the radar’s primary weather observation mission. We conclude by discussing research needed to better understand technical challenges and operational benefits.

East Asia Atmospheric River Forecast With a Deep Learning Method: GAN‐UNet

AbstractAccurate forecasting of atmospheric rivers (ARs) holds significance in preventing losses from extreme precipitation. However, traditional numerical weather prediction (NWP) models are computationally expensive and can be limited in accuracy due to inaccurate physical parameter settings. To overcome these limitations, we propose a deep learning (DL) model, called GAN‐UNet, to forecast the AR occurrence, position, and intensity in East Asia. GAN‐UNet can capture the complex nonlinear relationship between the inputs at the past moment, including the vertically integrated water vapor transport (IVT), zonal wind at 850 hPa (U850), and meridional wind at 850 hPa (V850), and the forecast output (IVT, U850, or V850), whose results are comparable to NWP models. In addition, the average model (AM) by integrating the results generated by GAN‐UNet and European Centre for Medium‐Range Weather Forecasts (ECMWF) outperforms all the NWP models selected in this study, demonstrating its potential to improve the performance of NWP through the DL method. Specifically, the 5‐day average F1 scores of the AM are 0.777 and 0.845, whose values are significantly better than those obtained by ECMWF (0.712 and 0.794) in the two key regions of East Asia; The AM 5‐day average intersection over unions are 0.706 and 0.688 while the values of ECMWF are 0.675 and 0.64; in terms of intensity forecast, GAN‐UNet and AM exhibited lower differences in most of the intensity bins, except for the final bin with IVT more than 825 kg m−1 s−1. With this thorough analysis, GAN‐UNet is shown as an effective model to forecast ARs.

Multimodel comparison and assessment of short to medium range precipitation and temperature forecasts over India: Implications towards forecasting of meteorological indices in India

AbstractA comprehensive assessment of the forecast skill of various meteorological indices over the Indian region from different numerical weather prediction (NWP) models is lacking in the literature. In this study the performance of four NWP models, namely Global Ensemble Forecast System (GEFSv12), European Center for Medium Range Forecasting (ECMWF), Climate Forecast System (CFSv2) and Indian Institute of Tropical Meteorology (IITM) towards forecasting of precipitation, temperature and associated meteorological indices, is evaluated at short to medium timescales across the Indian region. Further, the effect of ocean atmospheric (OA) oscillations on the precipitation/temperature forecast skill from the different NWP models is also assessed. Results show that the NWP models are better in predicting the meteorological indices than the quantitative forecasts. The ECMWF model was found to be the best for PCP forecasting with CFSv2 performing poorly. For temperature the GEFSv12 model performance was the lowest, compared to rest of the models. The models show poor skill in forecasting monsoon season precipitation compared to non‐monsoon and the temperature forecasts from the NWP models are particularly poor for the northern basins. Skilful temperature forecasts are observed in the Northwestern, Indo‐Gangetic and Central basins for the CFSv2 and ECMWF models. The forecast skill of precipitation indices are higher in the northwestern, central and Indo‐Gangetic basins compared to the rest. The skill of precipitation indices, namely rainy days, extreme rainy days and consecutive wet days, is higher during the monsoon seasons while the prediction skill of consecutive dry days is higher during the non‐monsoon season. OA analysis revealed that the ENSO phases have dominant effect on the forecast skill of the precipitation only. The temperature and meteorological indices forecasts are not affected significantly by the OA phases. The outcomes of this study have implications towards irrigation scheduling and water resources management decision making in India.

Subseasonal Prediction of Summer Temperature in West Africa Using Artificial Intelligence: A Case Study of Senegal

Despite the rapid growth of machine learning (ML) and its far-reaching applications in various fields such as healthcare, finance, and urban heat management, there are still some unresolved challenges in the field of climate change. Reliable subseasonal forecasts of summer temperatures would be a great benefit to society. Although numerical weather prediction (NWP) models are better at capturing relevant sources of predictability, such as temperatures, land, and sea surface conditions, the subseasonal potential is not fully exploited. One such challenge is accurate subseasonal temperature forecasting using cutting-edge ML technology. This study aims to assess and predict the changes in subseasonal temperature during the summer season (from March to June) in Senegal on 2-weeks time scales. Six ML techniques, including linear regression (LR), decision tree (DT), support vector machine (SVM), artificial neural network (ANN), long short-term memory (LSTM), and gated recurrent units (GRU), are used. The experiments utilize a multivariate approach by incorporating variables of the ERA-5 dataset from 1981 to 2022. The results compared all the performances of the methods to assess their overall effectiveness in forecasting air temperature (t2m) values over 2 weeks. Our analysis demonstrates that the GRU model outperforms the other ML models, achieving a Nash–Sutcliffe efficiency (NSE) score of 74.68% and a mean absolute percentage error (MAPE) of 2.51%. The GRU model effectively captures long-term dependencies and exhibits superior performance in temperature forecasting. Furthermore, a comparison between the observed and predicted values confirms the accuracy of the GRU model in aligning with actual temperature trends. Overall, this study contributes an impactful deep learning model to the field of subseasonal temperature forecasting in West Africa (Senegal), which offers local authorities the capability to anticipate climatic events and enact preventive measures accordingly.

Deep learning model for heavy rainfall nowcasting in South Korea

Accurate nowcasting is critical for preemptive action in response to heavy rainfall events (HREs). However, operational numerical weather prediction models have difficulty predicting HREs in the short term, especially for rapidly and sporadically developing cases. Here, we present multi-year evaluation statistics showing that deep-learning-based HRE nowcasting, trained with radar images and ground measurements, outperforms short-term numerical weather prediction at lead times of up to 6 h. The deep learning nowcasting shows an improved accuracy of 162%–31% over numerical prediction, at the 1-h to 6-h lead times, for predicting HREs in South Korea during the Asian summer monsoon. The spatial distribution and diurnal cycle of HREs are also well predicted. Isolated HRE predictions in the late afternoon to early evening which mostly result from convective processes associated with surface heating are particularly useful. This result suggests that the deep learning algorithm may be available for HRE nowcasting, potentially serving as an alternative to the operational numerical weather prediction model.

Velocity estimation of thunderstorm movement and dealiasing of single Doppler radar during convective events

AbstractThe use of meteorological radars in monitoring current weather conditions is crucial regarding the observation of the evolution and dissipation of thunderstorms. Thus, Doppler velocities being measured in each radar scan and velocity vectors derived from Numerical Weather Prediction (NWP) models—that are usually not as highly resolving as radar scans—are combined, as a monitoring utility during the evolution of a convective weather event. The objective of this study is to develop a new method that allows the implementation of a thunderstorm movement velocity estimation technique combining block matching and optical flow techniques. This new method constitutes a nowcasting (NWC) application that enables the use of a single Doppler radar without the need of using NWPs. The method relies on the estimation of the thunderstorm movement vector velocities (Doppler velocity) for each constant altitude plan position indicator (CAPPI) and through correction for aliasing errors to obtain 3D vector velocity fields for convective systems. The performance of the method is evaluated for selected case studies of convective thunderstorms under different synoptic scale conditions over Greece, a geographical area with challenges in forecasting due to its sharp relief and the need for optimization of the use of radar products.

Probabilistic post-processing of short to mediumrange temperature forecasts: Implications for heatwave prediction in India.

Accurate and reliable air temperature forecasts are necessary for predicting and responding to thermal disasters such as heat strokes. Forecasts from Numerical Weather Prediction(NWP) models contain biases which require post-processing. Studies assessing the skill of probabilistic post-processing techniques (PPTs) on temperature forecasts in India are lacking. This study aims to evaluate probabilistic post-processing approaches such as Nonhomogeneous Gaussian Regression (NGR) and Bayesian Model Averaging (BMA) for improving daily temperature forecasts from two NWP models, namely, the European Centre for Medium Range Weather Forecasts (ECMWF) and the Global Ensemble Forecast System (GEFS), across the Indian subcontinent. Apart from that, the effect of probabilistic PPT on heatwave prediction skills across India is also evaluated. Results show that probabilistic PPT comprehensively outperform traditional approaches in forecasting temperatures across India at all lead times. In the Himalayan regions where the forecast skill of raw forecasts is low, the probabilistic techniques are not able to produce skillful forecasts even though they perform much better than traditional techniques. The NGR method is found to be the best performing PPT across the Indian region. Post-processing Tmax forecasts using the NGR approach was found to considerably improve the heatwave prediction skill across highly heatwave prone regions in India. The outcomes of this study will be helpful in setting up improved heatwave prediction and early warning systems in India.

Real-Time Evaluation of the Uncertainty in Weather Forecasts Through Machine Learning-Based Models

Meteorological events have always been of great interest because they have influenced everyday activities in critical areas, such as water resource management systems. Weather forecasts are solved with numerical weather prediction models. However, it sometimes leads to unsatisfactory performance due to the inappropriate setting of the initial state. Precipitation forecasting is essential for water resource management in semi-arid climate and seasonal rainfall areas such as the Ebro basin. This research aims to improve the estimation of the uncertainty associated with real-time precipitation predictions presenting a machine learning-based method to evaluate the uncertainty of a weather forecast obtained by the Weather Research and Forecasting model. We use a model trained with ground-truth data from the Confederación Hidrográfica del Ebro, and WRF forecast results to compute uncertainty. Experimental results show that Decision Tree-based ensemble methods get the lowest generalization error. Prediction models studied have above 90% accuracy, and root mean square error has similar results compared to those obtained with the ground truth data. Random Forest presents a difference of -0.001 concerning the 0.535 obtained with the ground truth data. Generally, using the ML-based model offers good results with robust performance over more traditional forms for uncertainty calculation and an effective alternative for real-time computation.

A tolerant hydrologic technique for real-time selection of optimum QPFs from NWPMs for flood warning applications.

The most important information required to successfully issue a flood warning is the quantitative precipitation forecasts (QPFs). This is important to run subsequent rainfall-runoff simulations. A rainfall-runoff simulation derives its accuracy mainly from the accuracy of the input QPFs. The dynamically based global numerical weather prediction models (NWPMs) are strong candidate sources of QPFs. A main problem is the real-time selection of which NWPM should be used to provide the QPFs for flood warning simulations. This paper develops an automated technique to solve this problem. The technique performs real-time comparisons with measured rainfall fields using a novel 'tolerant' hydrologic approach. The 'tolerant' approach performs the comparison on the basin scale and allows for timing shifts in the forecasts. This is because QPFs can be good but only a few hours early or late. Two events are used for illustration, and the proposed real-time application in flood warning is presented. The developed technique, employing the tolerant approach, could eliminate the effects of the timing shifts and, accordingly, succeeded to select the QPFs to be used. A Python package was developed for automation. The developed technique is expected to also be useful for offline assessments of historical performances of NWPMs.

Evaluating pySTEPS optical flow algorithms for convection nowcasting over the Maritime Continent using satellite data

Abstract. The Maritime Continent (MC) regularly experiences powerful convective storms that produce intense rainfall, flooding and landslides, which numerical weather prediction models struggle to forecast. Nowcasting uses observations to make more accurate predictions of convective activity over short timescales (∼ 0–6 h). Optical flow algorithms are effective nowcasting methods as they are able to accurately track clouds across observed image series and predict forward trajectories. Optical flow is generally applied to weather radar observations; however, the radar coverage network over the MC is not complete and the signal cannot penetrate the high mountainous regions. In this research, we apply optical flow algorithms from the pySTEPS nowcasting library to satellite imagery to generate both deterministic and probabilistic nowcasts over the MC. The deterministic algorithm shows skill up to 4 h on spatial scales of 10 km and coarser and outperforms a persistence nowcast for all lead times. Lowest skill is observed over the mountainous regions during the early afternoon, and highest skill is seen during the night over the sea. A key feature of the probabilistic algorithm is its attempt to reduce uncertainty in the lifetime of small-scale convection. Composite analysis of 3 h lead time nowcasts, initialised in the morning and afternoon, produces reliable ensembles but with an under-dispersive distribution and produces area under the curve scores (i.e. ratio of hit rate to false alarm rate across all probability thresholds) of 0.80 and 0.71 over the sea and land, respectively. When directly comparing the two approaches, the probabilistic nowcast shows greater skill at ≤ 60 km spatial scales, whereas the deterministic nowcast shows greater skill at larger spatial scales ∼ 200 km. Overall, the results show promise for the use of pySTEPS and satellite retrievals as an operational nowcasting tool over the MC.

A two-fold deep-learning strategy to correct and downscale winds over mountains

Abstract. Assessing wind fields at a local scale in mountainous terrain has long been a scientific challenge, partly because of the complex interaction between large-scale flows and local topography. Traditionally, the operational applications that require high-resolution wind forcings rely on downscaled outputs of numerical weather prediction systems. Downscaling models either proceed from a function that links large-scale wind fields to local observations (hence including a corrective step) or use operations that account for local-scale processes, through statistics or dynamical simulations and without prior knowledge of large-scale modeling errors. This work presents a strategy to first correct and then downscale the wind fields of the numerical weather prediction model AROME (Application of Research to Operations at Mesoscale) operating at 1300 m grid spacing by using a modular architecture composed of two artificial neural networks and the DEVINE downscaling model. We show that our method is able to first correct the wind direction and speed from the large-scale model (1300 m) and then accurately downscale it to a local scale (30 m) by using the DEVINE downscaling model. The innovative aspect of our method lies in its optimization scheme that accounts for the downscaling step in the computations of the corrections of the coarse-scale wind fields. This modular architecture yields competitive results without suppressing the versatility of the DEVINE downscaling model, which remains unbounded to any wind observations.

Relative economic value of global ensemble prediction system of NCMRWF, India, for extreme weather events

AbstractAn ensemble prediction system quantifies the uncertainty in the forecast by a numerical weather prediction model and helps in decision‐making. The global ensemble prediction system (NEPS‐G) of NCMRWF, India, has 1 control and 22 perturbed ensemble members. In the present study, the relative economic value of NEPS‐G forecast has been investigated based on a simple decision‐analytic model. Temperature at 850 hPa (T850), maximum surface temperature (Tmax) and precipitation (rf) over the Indian domain have been studied. Ensemble mean forecast offers a larger value to the users than the control forecast. For rare extreme events, the area under the curve (T‐AUC) of a relative operating characteristic (ROC) obtained using the trapezoidal approximation for 22‐member NEPS‐G (NEPS‐22) is significantly smaller than the area under the curve (Z‐AUC) of the full ROC obtained using binormal model and Z‐transformation. As a result, the discrimination ability and relative economic value of NEPS‐G are underestimated. A method (SUB‐EM) which subdivides the lowest probability category into several subcategories and uses ensemble mean forecast as the secondary decision variable to enhance T‐AUC and improve relative economic value of NEPS‐G has been investigated. Tmax > 45°C and rf > 65 mm·day−1 have been considered as extreme events. The impact of increasing the number of perturbed ensemble members from 22 to 33 (NEPS‐33) by running additional 11 perturbed members from 0000 UTC initial condition has also been investigated and compared with the results obtained using SUB‐EM. SUB‐EM offers larger economic value to the users with small cost–loss ratio (α) without engaging any extra computational resources. An experiment is also carried out to study the impact of using different ensemble summary statistics as a secondary decision variable. Ensemble Maximum provides maximum value and benefits the largest range of users.