A Diagnostic Method for Detecting Frontal Boundaries and Structures

As the National Oceanic and Atmospheric Administration (NOAA) transitions from the operational Global Forecasting System (GFS) Global Spectral Model (GSM) to the Finite Volume Cubed Sphere dynamical core (FV3), detailed analyses of how well the new dynamical core handles the larger scale dynamics under different combinations of physics will be essential in crafting the the FV3 into a model that consistently outperforms the operational GFS. Recent evaluations of the geopotential height Anomaly Correlation Coefficient at 500-mb from the Model Evaluation Group place the FV3 performance at nearly the same level as the operational GFS, both for short-term and medium-range forecasting (e.g. GFS-0.871 and FV3-0.876 at day 5 between March and June of 2018). Although viewed as reassuring, more rigorous and careful analyses will be essential in order for the FV3 to consistently outperform the GFS and become widely accepted as the new forecasting standard. Some approaches that could lead to the improvement of the FV3 include, but are not limited to, modifications in the data assimilation system, the evaluation of physics through the single column model utility, quick-view utilities comparing models and model performance for different physics combinations, and the creation of diagnostic tools that peer into the characteristics and structure of different weather and meteorological phenomenon. The work that is presented here features a front detection algorithm created with Python and demonstrates a case highlighting 1) the 3-D structure of a front, 2) the relationship between a cold front and low-level jet (LLJ) ahead of the front, 3) an analysis of moisture transport about the LLJ, and 4) regions of forced convection ahead of the front. The intent of this work is to show the utility of this tool as a way to evaluate the spatiotemporal characteristics of a front (whether a cold front or dryline), and to show the potential of its use in evaluating the formation of clouds and precipitation for diagnostic purposes.

The front detection algorithm, which has been adapted for detecting both cold fronts and drylines over CONUS, utilizes both the vertically integrated potential temperature gradient and integrated water vapor (cold front) or dew point temperature gradient (dryline) at each latitude as a way to determine the location of a front. From there a proximity analysis is conducted to determine the highest concentration of points, or where there is a consistently large thermal gradient being reported during the execution of the algorithm. Spatial limitations of the front are included by evaluating which point has the lowest report of sea-level pressure according the model (northern limit), and an arbitrary cut-off point of 30N at the southern extent. For instances where more than one front is present over CONUS, the algorithm determines the number of geographical points where the thermal gradient is largest at a particular latitude, identifies the central location of those groupings, and then performs an area delimited recalculation of the thermal gradient to pick out the number of identifiable fronts present over CONUS. The delimited area is dependent upon the separation distance between neighboring fronts, where a final optimization is made to ensure that only neighboring points along the frontal axis are being selected. Calculations of the angle perpendicular to the frontal axis are made where transects are drawn through the front in order to create cross sections for enhanced visualizations.

The example presented uses a combination of fields such as specific humidity, vertical and ageostrophic wind, potential temperature, the zonal and meridional wind, the 0-degC isotherm, sea-level pressure, precipitable water, and the cloud mixing ratio. Evaluations of various transects along the front show the importance of the ageostrophic circulation as a feedback mechanism in reinforcing the thermal structure of the front, namely the tilting/torqueing of the frontal boundary; the response of low-level winds as a result of a strengthening gradient northward along the axis of the front; the convergence moisture and moist convection ahead of the front; and localized convective episodes along the LLJ axis. Other interesting features related to moist transport are evident behind the surface front with transport occurring above the 0-degC line, and the local modification of the thermal structure as a result of both the presence of moisture and the transition of phase from water to ice. A final analysis is conducted along the LLJ axis with the axis automatically selected by identifying where the strongest low-level winds are located ahead of the front. Results along the LLJ show a substantial drop in sea-level pressure, with localized pressure minima corresponding to regions of enhanced vertical motion, localized convection, and large cloud water mixing ratios extending toward the upper troposphere. It is suspected that variations in the ageostropic circulation are in part responsible to the convective variability ahead of the front as it acts to modify the vertical behavior of the front and accelerate the near surface front. Future analyses with this diagnostic algorithm include the evaluation of potential vorticity along the frontal boundary, the evolution of fronts using different combinations of physics and how that impacts the spatiotemporal evolution, as well as the inclusion of severe weather metrics to understand how the potential for severe convective episodes are either enhanced or suppressed.