Browse

You are looking at 21 - 30 of 2,812 items for :

  • Weather and Forecasting x
  • Refine by Access: All Content x
Clear All
Kosuke Ono

Abstract

This study extends Bayesian model averaging (BMA) to a form suitable for time series forecasts. BMA is applied to a three-member ensemble for temperature forecasts with a 1-h interval time series at specific stations. The results of such an application typically have a problematic characteristic. BMA weights assigned to ensemble members fluctuate widely within a few hours because BMA optimizations are independent at each lead time, which is incompatible with the spatiotemporal continuity of meteorological phenomena. To ameliorate this issue, a degree of correlation among different lead times is introduced by the extension of latent variables to lead times adjacent to the target lead time for the calculation of BMA weights and variances. This extension approach stabilizes the BMA weights, improving the performance of deterministic and probabilistic forecasts. Also, an investigation of the effects of this extension technique on the shapes of forecasted probability density functions showed that the extension approach offers advantages in bimodal cases. This extension technique may show promise in other applications to improve the performance of forecasts by BMA.

Restricted access
Aaron J. Hill and Russ S. Schumacher

Abstract

Approximately seven years of daily initializations from the convection-allowing National Severe Storms Laboratory Weather Research and Forecasting model are used as inputs to train random forest (RF) machine learning models to probabilistically predict instances of excessive rainfall. Unlike other hazards, excessive rainfall does not have an accepted definition, so multiple definitions of excessive rainfall and flash flooding – including flash flood reports and 24-hr average recurrence intervals (ARIs) – are used to explore RF configuration forecast sensitivities. RF forecasts are analogous to operational Weather Prediction Center (WPC) day-1 Excessive Rainfall Outlooks (EROs) and their resolution, reliability, and skill are strongly influenced by rainfall definitions and how inputs are assembled for training. Models trained with 1-y ARI exceedances defined by the Stage-IV (ST4) precipitation analysis perform poorly in the northern Great Plains and southwest U.S., in part due to a high bias in the number of training events in these regions. Increasing the ARI threshold to 2 years or removing ST4 data from training, optimizing forecast skill geographically, and spatially averaging meteorological inputs for training generally results in improved CONUS-wide RF forecast skill. Both EROs and RF forecasts have seasonal skill – poor forecasts in the late fall and winter and skillful forecasts in the summer and early fall. However, the EROs are consistently and significantly better than their RF counterparts, regardless of RF configuration, particularly in the summer months. The results suggest careful consideration should be made when developing ML-based probabilistic precipitation forecasts with convection-allowing model inputs, and further development is necessary to consider these forecast products for operational implementation.

Restricted access
Makenzie J. Krocak and Harold E. Brooks

Abstract

While many studies have looked at the quality of forecast products, few have attempted to understand the relationship between them. We begin to consider whether or not such an influence exists by analyzing storm-based tornado warning product metrics with respect to whether they occurred within a severe weather watch and, if so, what type of watch they occurred within.

The probability of detection, false alarm ratio, and lead time all show a general improvement with increasing watch severity. In fact, the probability of detection increased more as a function of watch-type severity than the change in probability of detection during the time period of analysis. False alarm ratio decreased as watch type increased in severity, but with a much smaller magnitude than the difference in probability of detection. Lead time also improved with an increase in watch-type severity. Warnings outside of any watch had a mean lead time of 5.5 minutes, while those inside of a particularly dangerous situation tornado watch had a mean lead time of 15.1 minutes. These results indicate that the existence and type of severe weather watch may have an influence on the quality of tornado warnings. However, it is impossible to separate the influence of weather watches from possible differences in warning strategy or differences in environmental characteristics that make it more or less challenging to warn for tornadoes. Future studies should attempt to disentangle these numerous influences to assess how much influence intermediate products have on downstream products.

Restricted access
Soo-Hyun Kim, Hye-Yeong Chun, Dan-Bi Lee, Jung-Hoon Kim, and Robert D. Sharman

Abstract

Based on a convective gravity wave drag parameterization scheme in a Numerical Weather Prediction (NWP) model, previously proposed near-cloud turbulence (NCT) diagnostics for better detecting turbulence near convection are tested and evaluated by using global in situ flight data and outputs from operational global NWP model of the Korea Meteorological Administration for one year (from December 2016 to November 2017). For comparison, eleven widely used clear air turbulence (CAT) diagnostics currently used in operational NWP-based aviation turbulence forecasting systems are separately computed. For selected cases, NCT diagnostics predict more accurately localized turbulence events over convective regions with better intensity, which is clearly distinguished from the turbulence areas diagnosed by conventional CAT diagnostics that they mostly failed to forecast with broad areas and low magnitudes. Although overall performance of NCT diagnostics for whole one year is lower than conventional CAT diagnostics due to the fact that NCT diagnostics exclusively focus on the isolated NCT events, adding the NCT diagnostics to CAT diagnostics improves the performance of aviation turbulence forecasting. Especially in the summertime, performance in terms of an area under the curve (AUC) based on probability of detection statistics is the best (AUC = 0.837 with a 4% increase, compared to conventional CAT forecasts) when the mean of all CAT and NCT diagnostics is used, while performance in terms of root mean square error is the best when the maximum among combined CAT and single NCT diagnostic is used. This implies that including NCT diagnostics to currently used NWP-based aviation turbulence forecasting systems should be beneficial for safety of air travel.

Restricted access
Laurel L. DeHaan, Andrew C. Martin, Rachel R. Weihs, Luca Delle Monache, and F. Martin Ralph

Abstract

Accurate forecasts of atmospheric rivers (ARs) provide advance warning of flood and landslide hazards, as well as greatly aid effective water management. It is therefore critical to evaluate the skill of AR forecasts in numerical weather prediction (NWP) models. A new verification framework is proposed leveraging freely available software and metrics previously used for different applications. Specifically, AR detection and statistics are computed for the first time using the Method for Object-based Diagnostic Evaluation (MODE). In addition, the measure of effectiveness (MoE) is introduced as a new metric for understanding AR forecast skill in terms of size and location. The MoE provides a quantitative measure of the position of an entire forecasted AR compared to observation, regardless of whether the AR is making landfall. In addition, the MoE can provide qualitative information about the evolution of a forecast by lead time with implications about the predictability of an AR. We analyze AR forecast verification and skill using 11 years of cold season forecasts from two NWP models, one global and one regional. Four different thresholds of integrated vapor transport (IVT) are used in the verification revealing differences in forecast skill based on the strength of an AR. In addition to MoE, AR forecast skill is also addressed in terms of intensity error, landfall position error, and contingency table metrics.

Restricted access
Makenzie J. Krocak, Jinan N. Allan, Joseph T. Ripberger, Carol L. Silva, and Hank C. Jenkins-Smith

Abstract

Nocturnal tornadoes are challenging to forecast and even more challenging to communicate. Numerous studies have evaluated the forecasting challenges, but fewer have investigated when and where these events pose the greatest communication challenges. This study seeks to evaluate variation in confidence among US residents in receiving and responding to tornado warnings by hour-of-day. Survey experiment data comes from the Severe Weather and Society Survey, an annual survey of US adults. Results indicate that respondents are less confident about receiving warnings overnight, specifically in the early morning hours (12 AM to 4 AM local time). We then use the survey results to inform an analysis of hourly tornado climatology data. We evaluate where nocturnal tornadoes are most likely to occur during the time frame when residents are least confident in their ability to receive tornado warnings. Results show that the Southeast experiences the highest number of nocturnal tornadoes during the time period of lowest confidence, as well as the largest proportion of tornadoes in that time frame. Finally, we estimate and assess two multiple linear regression models to identify individual characteristics that may influence a respondent’s confidence in receiving a tornado between 12 AM and 4 AM. These results indicate that age, race, weather awareness, weather sources, and the proportion of nocturnal tornadoes in the local area relate to warning reception confidence. The results of this study should help inform policymakers and practitioners about the populations at greatest risk for challenges associated with nocturnal tornadoes. Discussion focuses on developing more effective communication strategies, particularly for diverse and vulnerable populations.

Restricted access
YANSHUANG XIE, SHAOPING SHANG, JINQUAN CHEN, FENG ZHANG, ZHIGAN HE, GUOMEI WEI, JINYU WU, BENLU ZHU, and YINDONG ZENG

Abstract

Accurate storm surge forecasts provided rapidly could support timely decision-making with consideration of tropical cyclone (TC) forecasting error. This study developed a fast storm surge ensemble prediction method based on TC track probability forecasting and searching optimization of a numerical scenario database (SONSD). In a case study of the Fujian Province coast (China), a storm surge scenario database was established using numerical simulations generated by 93,150 hypothetical TCs. In a GIS-based visualization system, a single surge forecast representing 2562 distinct typhoon tracks and the occurrence probability of overflow of seawalls along the coast could be achieved in 1–2 min. Application to the cases of Typhoon Soudelor (2015) and Typhoon Maria (2018) demonstrated that the proposed method is feasible and effective. Storm surge calculated by SONSD had excellent agreement with numerical model results (i.e., mean MAE/RMSE: 7.1/10.7 cm, correlation coefficient: >0.9). Tide prediction also performed well with MAE/RMSE of 9.7/11.6 cm versus the harmonic tide, and MAE/RMSE of phase prediction for all high waters of 0.25/0.31 h versus observations. The predicted high-water level was satisfactory (MAE of 10.8 cm versus observations) when the forecasted and actual positions of the typhoon were close. When the forecasted typhoon position error was large, the ensemble surge prediction effectively reduced prediction error (i.e., the negative bias of −58.5 cm reduced to −5.2 cm versus observations), which helped avoid missed alert warnings. The proposed method could be applied in other regions to provide rapid and accurate decision-making support for government departments.

Restricted access
Jörg Steinert, Patrick Tracksdorf, and Dirk Heizenreder

Abstract

The analysis and forecast of precipitation characteristics is a key task for national meteorological services for providing high quality weather forecasts and warnings. Beside the precipitation amount, the precipitation type is essential to describe and evaluate the recent, ongoing and future weather situations.

This paper introduces a new surface-based hybrid hydrometeor classification algorithm. The presented method combines polarimetric radar observations at radar beam height from the C-band dual-polarization weather radar network of the German Weather Service (DWD, Deutscher Wetterdienst) with corrected thermodynamical profiles of numerical weather prediction (NWP) model output and extrapolates the hydrometeor classes at radar beam height to a height of 2 m above ground level (AGL). The implemented technique parametrizes the microphysical processes in the lower troposphere based on the appropriate thermodynamical profile the hydrometeors have to pass along their way from the radar beam height to the surface. Due to errors in NWP output, the NWP vertical profiles of temperature and humidity are adjusted by using several types of surface stations with high spatial and temporal resolution.

Verification results show considerable improvements in the hydrometeor classification near the ground compared to the radar sweep height. After an additional positive in-house evaluation, the presented method is integrated into DWD’s operational environment. The topic of this paper is to describe the processing steps for the computation of the near-surface precipitation type. In addition, example cases and a verification study complement the explanations.

Restricted access
Eva–Maria Walz, Marlon Maranan, Roderick van der Linden, Andreas H. Fink, and Peter Knippertz

Abstract

Current numerical weather prediction models show limited skill in predicting low-latitude precipitation. To aid future improvements, be it with better dynamical or statistical models, we propose a well-defined benchmark forecast. We use the arguably best currently high-resolution, gauge-calibrated, gridded precipitation product, the Integrated Multi-Satellite Retrievals for GPM (Global Precipitation Measurement) (IMERG) “final run” in a ± 15-day window around the date of interest to build an empirical climatological ensemble forecast. This window size is an optimal compromise between statistical robustness and flexibility to represent seasonal changes. We refer to this benchmark as Extended Probabilistic Climatology (EPC) and compute it on a 0.1°×0.1° grid for 40°S–40°N and the period 2001–2019. In order to reduce and standardize information, a mixed Bernoulli-Gamma distribution is fitted to the empirical EPC, which hardly affects predictive performance. The EPC is then compared to 1-day ensemble predictions from the European Centre for Medium-Range Weather Forecasts (ECMWF) using standard verification scores. With respect to rainfall amount, ECMWF performs only slightly better than EPS over most of the low latitudes and worse over high-mountain and dry oceanic areas as well as over tropical Africa, where the lack of skill is also evident in independent station data. For rainfall occurrence, EPC is superior over most oceanic, coastal, and mountain regions, although the better potential predictive ability of ECMWF indicates that this is mostly due to calibration problems. To encourage the use of the new benchmark, we provide the data, scripts, and an interactive webtool to the scientific community.

Restricted access
Hung Ming Cheung, Chang-Hoi Ho, Minhee Chang, Dasol Kim, Jinwon Kim, and Woosuk Choi

Abstract

Despite tremendous advancements in dynamical models for weather forecasting, statistical models continue to offer various possibilities for tropical cyclone (TC) track forecasting. Herein, a track-pattern-based approach was developed to predict a TC track for a lead time of 6–8 days over the western North Pacific (WNP), utilizing historical tracks in conjunction with dynamical forecasts. It is composed of four main steps: (1) clustering historical tracks similar to that of an operational five-day forecast in their early phase into track patterns, and calculating the daily mean environmental fields (500-hPa geopotential height and steering flow) associated with each track; (2) deriving the two environmental variables forecasted by dynamical models; (3) evaluating pattern correlation coefficients between the two environmental fields from step (1) and those from dynamical model for a lead times of 6–8 days; and (4) producing the final track forecast based on relative frequency maps obtained from the historical tracks in step (1) and the pattern correlation coefficients obtained from step (3). TCs that formed in the WNP and lasted for at least seven days, during the 9-year period 2011–2019 were selected to verify the resulting track-pattern-based forecasts. In addition to the performance comparable to dynamical models under certain conditions, the track-pattern-based model is inexpensive, and can consistently produce forecasts over large latitudinal or longitudinal ranges. Machine learning techniques can be implemented to incorporate non-linearity in the present model for improving medium-range track forecasts.

Restricted access