Short-Term Explicit Prediction

The Short-Term Explicit Prediction (STEP) Program is a multi-NCAR Laboratory activity with the overarching goal to improve the short-term (0-36 hours) forecasting of high-impact weather events such as severe thunderstorms (heavy rain, tornados, downburst, flash flood, lightning and hail), winter storms (snow, freezing rain and drizzle), and hurricanes. The STEP program emphasizes several research areas that are crucial for advancing the science and application of the short-term prediction of high-impact weather, through collaborative effort incorporating national and international scientists, engineers, and operational personnel from universities, government institutions and the private sector. Most of the forecasting/nowcasting systems and analysis tools developed under STEP are available to the communities for the support of research and real-time operations.

In FY18, RAL’s STEP effort continued to emphasize the research and real-time demonstration of the integrated Hydromet Prediction System (Figure 1) (STEP-Hydromet hereafter). The overarching objective of this effort is to advance the prediction of heavy rainfall, flash floods and streamflow through the integration of state-of-the-art rainfall estimation, precipitation forecasting/nowcasting, and hydrology modeling techniques into one seamless system. The major objectives this year were to conduct the analysis and evaluation of the real-time data collected during the 2017 STEP Hydroment Experiment and the third real-time Hydromet Experiment from 1 June – 15 August 2018 summer along the Colorado Front Range. Additionally, RAL continued to lead the STEP research theme on the improvement of WRF microphysics parameterization scheme as in previous years and STEP staff participated in analyses of the NSF-sponsored PECAN (see http://pecan15.org) experiment datasets and predictability of nocturnal, elevated convection initiation.

Figure 1.   Flowchart for the STEP Hydromet Prediction System.
Figure 1.   Flowchart for the STEP Hydromet Prediction System.

1. Development and Demonstration of STEP-Hydromet

The emphasis of the STEP-Hydromet is to provide prediction on the time scale of minutes out to one day, with particular emphasis on 0-12 hour forecasts and 0-1 hour nowcasts on very high-resolution spatial grids (from 100 m – 3 km in resolution). In addition to active research activities that support the development of the hydromet system, the integrated system has been demonstrated in real-time using the Colorado Front Range as a testbed since 2014. All components of STEP-Hydromet are run in real-time in an integrated fashion. The components included in the fully integrated system are:  1) radar based quantitative precipitation estimation (QPE) and rain gauge QPE, 2) quantitative precipitation nowcasting (QPN) from 10 min to 1 h from the heuristic-based Autonowcaster/Trident system. High-resolution winds and buoyancy analyses from VDRAS are also produced. 3) quantitative precipitation forecasts (QPF) from the WRF 3DVar NWP models with radar data assimilation and frequent update cycles, 4) streamflow prediction on a spatially-continuous 100 m resolution grid, from the WRF-Hydro hydrology model, and 5) near-real time performance evaluation of the QPE and QPF fields using a set of statistical metrics and techniques.

The Numerical Weather Prediction (NWP) forecast models and the WRF-Hydro model ran on NCAR’s Cheyenne supercomputer, while the AutoNowcaster, Trident and VDRAS nowcasting systems, and the EOL QPE mosaics ran on workstations located in RAL and EOL.  A dedicated web page

(https://ral.ucar.edu/projects/step_hydromet)

was set up for real-time viewing of precipitation accumulation fields, forecast and nowcast products, and streamflow prediction.  The JAZZ interactive java-based display system was also set up for viewing all the real-time observations (radar, satellite, surface station data), STEP nowcasts and NWP model forecast fields and was easily accessible on any workstation by clicking on a link on the dedicated STEP web site. Real-time streamflow, soil moisture and overland flow depth information was displayed on a dynamic web mapping service called HydroInspector, also developed as part of the STEP project. Upgrades and enhancements were made to the components of the Hydromet system prior to the start of the demonstration based on research conducted throughout the year. Ongoing research efforts towards improving the different components of this system are discussed below.

Quantitative Precipitation Nowcasting (QPN)

Nowcasting skill for summer 2017

The skill of 0-6 h precipitation nowcasting along the Colorado Front range has been monitored routinely during the STEP-Hydromet program for the last several years. Results for the summer of 2017 showed that radar data assimilation for the HRRR and 3DVAR provided greater improvement in skill than in previous years. This was particularly true for average precipitation amounts for nowcast periods between 1 and 3 hours and for larger areas (see Fig. 2c).  During the first hour the numerical model forecast were generally no better than persistence forecasts and had lower performance scores than simple extrapolation.  The NWP models and nowcast techniques are compared in Fig. 2.   These include the NCAR 3DVAR (3DVAR, red line), the NCAR 3DVAR with data assimilation (DA, green line), the High Resolution Rapid Refresh model (HRRR, blue line), persistence (same rainfall as the last hour, pres, purple line), one hour radar echo extrapolation nowcast (red star) and one hour growth and decay nowcast (Trident nowcasting system, black star). Figures 2 a-c show the skill (correlation coefficient) for each of the above techniques for the average rainfall over 230, 450 and 11,300 sq mi areas.

Figures 2 a-c.  Plots of correlation coefficient scores of average rainfall compared to nowcast time length. a) for the eastern Denver metropolitan region; b) for Boulder county; and c) within 60 km radius of the Denver KFTG radar.
Figures 2 a-c.  Plots of correlation coefficient scores of average rainfall compared to nowcast time length. a) for the eastern Denver metropolitan region; b) for Boulder county; and c) within 60 km radius of the Denver KFTG radar.

During 2018, the weights within the Trident and AutoNowcaster fuzzy-logic nowcasting systems were modified to improve its performance over pure extrapolation. 

Modifications to real-time Nowcasting systems

One of the activities planned for 2018, was to continue the real-time demonstration of the STEP Hydromet Heavy Rainfall Prediction System along the Colorado Front Range during the summer, incorporating and testing modifications and upgrades to the overall system (QPE, QPN, QPF and Streamflow prediction).   Following a change in focus of the overall STEP program direction, a scaled-back version of the real-time demonstration was conducted focusing primarily on the Trident and AutoNowcaster systems and using only the HRRR numerical model.   The primary objective for the scaled-back demonstration was to improve the very short-term nowcasting of convection initiation and heavy rainfall events.  Upgrades to the system included 1) ingest of the high temporal and spatial resolution of the new GOES-16 satellite imagery, 2) inclusion of two convergence fields from the COLIDE automated convergence boundary detection algorithm and the VDRAS boundary layer 4D-Var system and 3) inclusion of the HRRR numerical model short term forecasts.  Several engineering efforts were involved in preparation for the 2018 demonstration.  These efforts included 1) writing converters to ingest the high resolution GOES-16 spatial and temporal imagery and GLM data from the NOAA port into our system, 2) setup and porting of the Hydromet Heavy Rainfall system to a new workstation to support running all of the algorithms and processes and 3) setting up the VDRAS system to  use the HRRR numerical model as a background field.

The high resolution GOES-16 satellite visible and infrared imagery were examined throughout the summer to assess how to use the additional information more advantageously within the Trident/ANC system; particularly to improve the prediction and lead time of convection initiation.  Special attention was given to the two new water vapor channels, to monitor the variability in moisture and moisture advection at the different altitudes.  The HRRR model forecast fields were examined throughout the summer to assess its performance in providing accurate spatial and temporal forecasts of heavy rainfall.  Substantial time was spent in evaluating and comparing the performance of the COLIDE and VDRAS convergence fields and determining how best to optimize both of these fields for improvement in short-term nowcasting.  Based on the past several years of research, convergence boundaries not only play an important role in convection initiation, but also in the timing and location of heavy rainfall and flash flood events along the Front Range.  A comparison of the COLIDE and VDRAS fields are shown in Fig. 3.  Occasionally both techniques did a good job in detecting convergence boundaries such as shown in Fig.3.  However the majority of the time COLIDE did a better job of detecting the small spatial scale features associated with gust fronts, while VDRAS was more useful in representing broader convergence zones associated with larger scale features such as cold fronts.  A blending of capabilities from these two systems to produce a robust convergence detection capability within the Trident/AutoNowcaster system is warranted.

Figure 3.  Comparison of COLIDE automated, gridded convergence boundary detections with the gridded convergence field from VDRAS. Denver KFTG radar reflectivity (upper left); Denver KFTG radial velocity (upper right); gridded COLIDE detections (lower left) where brighter colors indicated stronger signals; VDRAS convergence field (lower right) where brown and yellow shades represent convergence regions.
Figure 3.  Comparison of COLIDE automated, gridded convergence boundary detections with the gridded convergence field from VDRAS. Denver KFTG radar reflectivity (upper left); Denver KFTG radial velocity (upper right); gridded COLIDE detections (lower left) where brighter colors indicated stronger signals; VDRAS convergence field (lower right) where brown and yellow shades represent convergence regions.

Qualitative Precipitation Forecasting (QPF)

The model-based QPF effort had two focus areas in FY18. First was evaluation and verification of the real-time results from the 2017 STEP Hydromet Experiment; second was a preliminary study on convective-scale hybrid data assimilation using WRFDA and DART. In 2017, in addition to the demonstration of the impact of radar data assimilation, two parallel runs with a newly developed LSAC (large-scale analysis constraint) with and without radar observations were executed to test the impact of LSAC. The LSAC scheme constrains the WRFDA analysis using GFS forecast to ensure that the large-scale component of the 3DVar analysis does not deviate too much from that of the global forecast.

The evaluation and verification of the real-time results were performed on the entire 3km WRF model domain. The verification was done for both July and August of 2017 by comparing WRF 0-12h QPF with/without radars against MRMS gauge-corrected precipitation analysis. The Fraction Skill Scores for 1mm and 2.5mm are shown in Figure 4, each compares the skills for the following QPF runs:

  • CTRL: initialized by WRFDA 3DVar analysis with continuous 3-hourly update cycles,
  • assimilating only conventional observations;
  • CTRL_RD: partial cycled hourly radar data assimilation with CTRL as the first guess at the multiples of 3 hour;
  • HRRR: operational HRRR mapped to the same domain as the other runs;
  • LSAC: Same as CTRL but with the large-scale analysis constraint;
  • LSAC_RD: same as CTRL_RD but with LSAC.

 

The results in Figure 4 indicate that the runs with radar data assimilation (HRRR, CTRL_RD, and LSAC_RD) improve the skill over those without radar for the entire 12 hour forecast period. For July, WRFDA 3DVar-based radar data assimilation run CTRL_RD has higher skill than HRRR and the LSAC has positive impact. For August, however, the LSAC has no impact in general and WRFDA has lower skill than HRRR for the 1mm threshold and higher skill for the 5mm threshold.

Figure 4. Comparison of FSSs among the five experiments CTRL, CTRL_RD, LSAC, LSAC_RD, and HRRR for the hourly precipitation threshold of 1mm (left) and 5mm (right) for July (upper) and August (lower) 2018.
Figure 4. Comparison of FSSs among the five experiments CTRL, CTRL_RD, LSAC, LSAC_RD, and HRRR for the hourly precipitation threshold of 1mm (left) and 5mm (right) for July (upper) and August (lower) 2018.

Detailed examination revealed that the LSAC had positive impact on smaller precipitation systems but occasionally negative impact on larger precipitation systems. Research and development is currently been conducted to improve the LSAC scheme such that its effect will be determined by a flow-dependent measure of the deviation of WRF forecast from GFS forecast.

The effort to develop a hybrid convective-scale hybrid data assimilation scheme was performed using a convective case archived during the STEP Hydromet 2017. Figure 5 is a flow chart showing the hybrid data assimilation process. The hybrid data assimilation enables the use of a flow-dependent background error covariance generated by DART in a variational framework. Our preliminary study shows encouraging result. Figure 6 compares the hybrid data assimilation (one way and two ways) with 3DVar and it demonstrates that the two-ways hybrid scheme outperforms 3DVar for the high precipitation threshold (5mm).

Figure 5. Flow chart of WRFDA and DART hybrid data assimilation showing the data assimilation procedure.
Figure 5. Flow chart of WRFDA and DART hybrid data assimilation showing the data assimilation procedure.
 
Figure 6. FSS Comparison of three data assimilation experiments 3DVar, one-way hybrid and two-way hybrid.
Figure 6. FSS Comparison of three data assimilation experiments 3DVar, one-way hybrid and two-way hybrid.

FY19 plans

  • Follow up and publication of recent work
  • Analysis of RELAMPAGO data
  • Development of Benchmarking toolkit and application to past STEP forecasts
  • Development of Precipitation Gauge and Lightning assimilation
  • Define requirements for ensemble hydrologic forecast

 

2. Evaluation and improvement of model microphysics parameterizations

Development of prototype multi-moment graupel/hail hybrid category in Thompson microphysics parameterization

Previous STEP work that explored the impacts of varying graupel density was the foundation for developing a multi-moment graupel/hail hybrid category in the Thompson microphysics scheme.  For this multi-moment graupel/hail category, number concentration and bulk volume mixing ratio are now predicted variables, in addition to the bulk mass mixing ratio. This allows for the graupel density to vary in space and time and be diagnosed from the new predicted variables (Fig. 7).  The major overhaul to the code for this new scheme was completed in FY17, and in FY18 it was tested by running 3D idealized simulations, as well as simulations of real convective cases including a PECAN case on 19-20 June 2015.  The NEXRAD radar data from the PECAN case was used to compare with the test simulations to assess if the new scheme was producing realistic results. Sensitivity studies were also run with the idealized cases to determine impacts of varying certain parameterization settings.  Based upon these tests, the code was refined and prepared for use in a real-time forecast model to be run during the RELAMPAGO field campaign.

Establish methods for baseline performance evaluation

In order to evaluate impacts on storm structure, QPF, and, in particular, storm evolution due to physics parameterizations, including updates made to the Thompson microphysics scheme, an object-based evaluation tool that tracks storms over time is needed.  The Method for Object-based Diagnostic Evaluation (MODE)-Time Domain (TD) is a tool developed at NCAR for such analyses, allowing users to set several parameters to evaluate storms with certain spatial scales and intensities.  Moreover, there are several parameters that users can set that impact how observed and modeled objects are matched for comparative evaluation metrics. In FY18, sensitivity tests were performed that varied the matching parameters to quantify the impact of these choices on the evaluation outcomes.  This effort will be continued to establish methods for how to apply MODE-TD to quantify impacts of various physics choices, with specific emphasis on the new multi-moment graupel/hail category, and in order to determine baseline performance metrics.

Figure 7. Graupel/hail category density as predicted in the 3D idealized (a) supercell and (b) squall line simulations.
Figure 7. Graupel/hail category density as predicted in the 3D idealized (a) supercell and (b) squall line simulations.

FY19 plans

  • Evaluate and improve the prototype multi-moment graupel/hail hybrid category in the Thompson microphysics scheme to improve forecasted convective storm structure, evolution, and QPF.
  • Run the new prototype microphysics scheme during RELAMPAGO field campaign and evaluate in cases observed during the campaign. Deploy hail pads for ground truth measurements of hail size during RELAMPAGO.
  • Use MODE-TD to evaluate the prototype microphysics scheme compared to the original Thompson scheme and other microphysics schemes and/or forecast ensemble members as available. 
  • Use evaluation results to inform further microphysics parameterization improvements.

 

3. Study on convective initiation with PECAN data

A case occurred during Plains Elevated Convection at Night (PECAN) experiment was studied using VDRAS to understand the convective triggering process. Trier et al (2017) found that the environment had substantial mesoscale lifting in the PECAN area, but the analysis using convectional observations could not reveal the convective triggering processes. Rapid updated 4DVar analysis with VDRAS showed that the initiation was partly contributed by a dissipating previous storm located southeast of the new initiation. Figure 8 shows the VDRAS wind analysis overlaid on S-pol reflectivity. The VDRAS wind convergence corresponds well with the convergence line indicated by the S-pol reflectivity observation. Trajectory analysis is being performed to trace the source of the air and the process that lifted the air up to result in the elevated convection.

Figure 8. S-pol reflectivity at 1o  elevation angle overlaid by VDRAS wind at four different times.
Figure 8. S-pol reflectivity at 1o  elevation angle overlaid by VDRAS wind at four different times.

An oral presentation on the study was given in a special PECAN symposium in 2017 AMS annual meeting.

FY19 Plans

Further analysis to understand the formation mechanism of the elevated convection