ALBERT

Description: Topographic rainfall induced by westbound tropical cyclones past an island mountain is investigated using an idealized Weather Research and Forecasting (WRF) Model. Idealized simulations with varying vortex core size R (100–250 km), vortex intensity Vmax (20–35 m s−1), and steering wind speed U (4–10 m s−1) are conducted. The results show that the geometric distributions of major rainfall over the island are not greatly sensitive to cloud microphysics schemes using either single momentum or double momentum. Major rainfall is produced over northeastern and southwestern slopes of the mountain for smaller U. As U is doubled, the rainfall, however, is considerably weakened or is present only over southwestern slopes. For smaller U, a bifurcation of island rainfall with a sudden change in intensity or geometric shifting exists within a tiny range of R or Vmax. When the bifurcation occurs with small track deviations, geometric distributions of major rainfall are also more sensitive to cloud microphysics schemes. Such formation of bifurcation or double-peak rainfall, however, is significantly reduced when the terrain size is doubled. Systematic experiments are conducted to relate the topographical rainfalls over the northern half, southern half, and the whole of the mountain slopes to varying R, Vmax, and U. Larger U tends to produce much larger southern rainfall than northern rainfall. The average and maximum rainfalls generally increase with increased Vmax, except for large R. The decrease of average rainfall and maximum rainfall with increased U is more evident for smaller R, while not necessarily true for larger R.

Description: The representation of model error in ensemble prediction systems (EPSs) can be limited by the assumptions within parameterization schemes. Stochastic perturbed parameterization tendencies (SPPT) is one representation of model error that randomly perturbs parameterized physical tendencies using a spatially and temporally correlated red-noise field. This research investigates the sensitivity of ensemble rainfall forecasts produced by the Weather Research and Forecasting (WRF) Model to the configuration of SPPT and independent SPPT (iSPPT) for three meso–synoptic-scale heavy rainfall events over the United States and Taiwan, primarily focusing on the ensemble mean and standard deviation as well as forecast skill. Thirty-two 20-member ensembles, which represent a combination of eight configurations of the stochastic perturbation time scale, length scale, and amplitude scale, and four perturbed parameterization schemes, as well as an unperturbed control simulation, are examined for each event. In each case, rainfall standard deviation is most sensitive to the perturbation time scale and amplitude scale. Moreover, microphysics tendency perturbations are associated with the largest standard deviation in two of the three events, followed by perturbations to the total (nonmicrophysics), turbulent mixing, and radiation parameterized tendencies. Additionally, microphysics tendency perturbations are associated with an increase in the areal coverage of heavy rainfall compared to the control forecast, regardless of whether the control forecast over or underrepresents the observed rainfall distribution.

Description: Most ensembles suffer from underdispersion and systematic biases. One way to correct for these shortcomings is via machine learning (ML), which is advantageous due to its ability to identify and correct nonlinear biases. This study uses a single random forest (RF) to calibrate next-day (i.e., 12–36-h lead time) probabilistic precipitation forecasts over the contiguous United States (CONUS) from the Short-Range Ensemble Forecast System (SREF) with 16-km grid spacing and the High-Resolution Ensemble Forecast version 2 (HREFv2) with 3-km grid spacing. Random forest forecast probabilities (RFFPs) from each ensemble are compared against raw ensemble probabilities over 496 days from April 2017 to November 2018 using 16-fold cross validation. RFFPs are also compared against spatially smoothed ensemble probabilities since the raw SREF and HREFv2 probabilities are overconfident and undersample the true forecast probability density function. Probabilistic precipitation forecasts are evaluated at four precipitation thresholds ranging from 0.1 to 3 in. In general, RFFPs are found to have better forecast reliability and resolution, fewer spatial biases, and significantly greater Brier skill scores and areas under the relative operating characteristic curve compared to corresponding raw and spatially smoothed ensemble probabilities. The RFFPs perform best at the lower thresholds, which have a greater observed climatological frequency. Additionally, the RF-based postprocessing technique benefits the SREF more than the HREFv2, likely because the raw SREF forecasts contain more systematic biases than those from the raw HREFv2. It is concluded that the RFFPs provide a convenient, skillful summary of calibrated ensemble output and are computationally feasible to implement in real time. Advantages and disadvantages of ML-based postprocessing techniques are discussed.

Description: Statistical postprocessing models can be used to correct bias and dispersion errors in raw precipitation forecasts from numerical weather prediction models. In this study, we conducted experiments to investigate four factors that influence the performance of regression-based postprocessing models with normalization transformations for short-term precipitation forecasts. The factors are 1) normalization transformations, 2) incorporation of ensemble spread as a predictor in the model, 3) objective function for parameter inference, and 4) two postprocessing schemes, including distributional regression and joint probability models. The experiments on the first three factors are based on variants of a censored regression model with conditional heteroscedasticity (CRCH). For the fourth factor, we compared CRCH as an example of the distributional regression with a joint probability model. The results show that the CRCH with normal quantile transformation (NQT) or power transformation performs better than the CRCH with log–sinh transformation for most of the subbasins in Huai River basin with a subhumid climate. The incorporation of ensemble spread as a predictor in CRCH models can improve forecast skill in our research region at short lead times. The influence of different objective functions (minimum continuous ranked probability score or maximum likelihood) on postprocessed results is limited to a few relatively dry subbasins in the research region. Both the distributional regression and the joint probability models have their advantages, and they are both able to achieve reliable and skillful forecasts.

Description: This study compares ensemble precipitation forecasts from 10-member, 3-km grid-spacing, CONUS domain single- and multicore ensembles that were a part of the 2016 Community Leveraged Unified Ensemble (CLUE) that was run for the 2016 NOAA Hazardous Weather Testbed Spring Forecasting Experiment. The main results are that a 10-member ARW ensemble was significantly more skillful than a 10-member NMMB ensemble, and a 10-member MIX ensemble (5 ARW and 5 NMMB members) performed about the same as the 10-member ARW ensemble. Skill was measured by area under the relative operating characteristic curve (AUC) and fractions skill score (FSS). Rank histograms in the ARW ensemble were flatter than the NMMB ensemble indicating that the envelope of ensemble members better encompassed observations (i.e., better reliability) in the ARW. Rank histograms in the MIX ensemble were similar to the ARW ensemble. In the context of NOAA’s plans for a Unified Forecast System featuring a CAM ensemble with a single core, the results are positive and indicate that it should be possible to develop a single-core system that performs as well as or better than the current operational CAM ensemble, which is known as the High-Resolution Ensemble Forecast System (HREF). However, as new modeling applications are developed and incremental changes that move HREF toward a single-core system are made possible, more thorough testing and evaluation should be conducted.

Description: Hurricane Joaquin (2015) was characterized by high track forecast uncertainty when it approached the Bahamas from 29 September 2015 to 1 October 2015, with 5-day track predictions ranging from landfall in the United States to east of Bermuda. The source of large track spread in Joaquin forecasts is investigated using an ensemble prediction system (EPS) based on the Hurricane Weather Research and Forecasting (HWRF) Model. For the first time, a high-resolution analysis of an HWRF-based EPS is performed to isolate the factors that control tropical cyclone (TC) track uncertainty. Differences in the synoptic-scale environment, the TC vortex structure, and the TC location are evaluated to understand the source of track forecast uncertainty associated with Joaquin, especially at later lead times when U.S. landfall was possible. EPS members that correctly propagated Joaquin into the central North Atlantic are compared with members that incorrectly predicted U.S. landfall. Joaquin track forecasts were highly dependent on the evolution of the environment, including weak atmospheric steering flow near the Bahamas and three synoptic-scale systems: a trough over North America, a ridge to the northeast of Joaquin, and an upper-tropospheric trough to the east of Joaquin. Differences in the steering flow were associated with perturbations of the synoptic-scale environment at the model initialization time. Ultimately, members that produced a more progressive midlatitude synoptic-scale pattern had reduced track errors. Joaquin track forecast uncertainty was not sensitive to the TC vortex structure or the initial TC position.

Description: Differences between forecasts and observations at eight atmospheric river observatories (AROs) in the western United States during winter 2015/16 are analyzed. NOAA’s operational RAP and HRRR 3-h forecasts of wind, integrated water vapor (IWV), integrated water vapor flux (IWV flux), and precipitation from the grid points nearest the AROs were paired with ARO observations presented in the NOAA/Physical Sciences Division’s water vapor flux tool (WVFT). The focus of this paper is to characterize and quantify the differences in the WVFT observations and forecasts. We used traditional forecast evaluation methods since they were compatible with the design of the tool: a near-real-time visual depiction of hourly observed and forecasted variables at a single location. Forecast root-mean-squared errors (RMSEs) and unbiased RMSEs, standard deviations of the observed and forecasted variables, and frequency bias scores (FBS) for all of the fields, plus equitable threat scores for precipitation, are presented. Both models forecasted IWV at all AROs and the winds that drive orographic precipitation at most AROs within a reasonable range of the observations as indicated by comparisons of the standard deviations and RMSEs of the forecasts with the standard deviations of the observations and FBS. These results indicated that forecasted advection of moisture to the stations was adequate for generating precipitation. At most stations and most hourly precipitation rates, the HRRR underpredicted precipitation. At several AROs the RAP precipitation forecasts more closely matched the observations at smaller ( 2 mm h−1.

Description: On 25 December 2016, a 984-hPa cyclone departed Colorado and moved onto the northern plains, drawing a nearby Arctic front into the circulation and wrapping it cyclonically around the equatorward side of the cyclone. A 130-km-wide and 850-km-long swath of surface winds exceeding 25 m s−1 originated underneath the comma head of the lee cyclone and followed the track of the Arctic front from Colorado to Minnesota. These strong winds formed in association with a downslope windstorm and mountain wave over Colorado and Wyoming, producing an elevated jet of strong winds. Central to the distribution of winds in this case is the Arctic air mass, which both shielded the elevated winds from surface friction behind the front and facilitated the mixing of the elevated jet down to the surface just behind the Arctic front, due to steep lapse rates associated with cold-air advection. The intense circulation south of the cyclone center transported the Arctic front and the elevated jet away from the mountains and out across Great Plains. This case is compared to an otherwise similar cyclone that occurred on 28–29 February 2012 in which a downslope windstorm occurred, but no surface mesoscale wind maximum formed due to the absence of a well-defined Arctic front and postfrontal stable layer. Despite the superficial similarities of this surface wind maximum to a sting jet (e.g., origin in the midtroposphere within the comma head of the cyclone, descent evaporating the comma head, acceleration to the top of the boundary layer, and an existence separate from the cold conveyor belt), this swath of winds was not caused by a sting jet.

Description: HarmonEPS is the limited-area, short-range, convection-permitting ensemble prediction system developed and maintained by the HIRLAM consortium as part of the shared ALADIN–HIRLAM system. HarmonEPS is the ensemble realization of HARMONIE–AROME, used for operational short-range forecasting in HIRLAM countries. HarmonEPS contains a range of perturbation methodologies to account for uncertainties in the initial conditions, forecast model, surface, and lateral boundary conditions. This paper describes the state of the system at the version labeled cycle 40 and highlights some directions for further development. The different perturbation methods available are evaluated and compared where appropriate. Several institutes have operational or preoperational implementations of HarmonEPS, such as MEPS (Finland, Norway, and Sweden), COMEPS (Denmark), IREPS (Ireland), KEPS (the Netherlands), AEMET-γSREPS (Spain), and RMI-EPS (Belgium), and these systems are briefly described and compared with the ensemble prediction system (IFS ENS) from the European Centre for Medium-Range Weather Forecasts (ECMWF).