On the Changes in MCS Cold Pool Characteristics due to Simultaneous Changes in Horizontal and Vertical Grid Spacing in WRF Runs

Questions remain regarding the optimal resolution for operational CAMS, which most appropriately resolves convection for forecasting purposes, but is computationally feasible. While previous research shows that horizontal grid spacing (∆x) should be decreased to 100 m to fully resolve convection, numerous works have shown that adequate resolving of convective storms may occur with dx as coarse as 1 km. Still, convection and turbulent mixing are 3D processes, and little work has been done to demonstrate the benefits of refining model resolution by simultaneously decreasing dx and vertical grid spacing (∆z). The current work evaluates simulated MCS cold pool characteristics in WRF, using a 1-way nested modeling frame work, with 3-, 1-, and 0.333-km ∆x. For all domains, ∆z is varied in four ways: (1) using the default 50 levels (Z1), (2) doubling ∆z to 100 levels throughout the troposphere (Z2), (3) using 50 vertical levels with half of the levels concentrated below 700 mb to better resolve vertical low-level hydrometeor tendencies and evaporation/condensation processes (Z3), and the same as Z3 but with 100 levels (Z4). This modeling framework and subsequent subset of resolution sensitivity tests are first performed on an observed case in order to evaluate the sensitivity in MCS cold pool solutions (with respect to varying ∆x and ∆z) from an operational perspective. The case selected will involve an MCS which traversed Oklahoma, allowing for surface cold pool verification to be performed using either the raw Oklahoma Mesonet data, or NCEP’s Unrestricted Mesoscale Analysis surface data (URMA, which ingests Oklahoma Mesonet data). It is hypothesized that 1-km ∆x will adequately resolve the intensity, magnitude distribution, and areal coverage of the simulated surface cold pool (compared to observations) when enhanced ∆z (Z4) is applied, minimizing computational costs, as 3D hydrometeor mixing and convective turbulent processes will, in turn, be adequately resolved. 0.333-km solutions however, are expected to show similar results to 1-km, as microphysical and boundary layer processes will remain parameterized, inhibiting the further resolving of convective features, and thus not worth the computational cost in an operational environment.

To provide a physical explanation as to why 1-km ∆x with enhanced ∆z (Z4) grid spacing adequately resolves MCS cold pools, quasi-idealized WRF runs, compositing the initial conditions of 35 observed MCS cases, will be employed, using the earlier mentioned 1-way nested framework with varying ∆x and ∆z grid spacings. The idealized nature of the runs will simplify the analyses to a single case, while attempting to preserve some heterogeneity introduced by reality via the initial conditions. In this manner, the aforementioned hypothesis is expected to be validated with the most realistic depiction (based on previous works) of 3D rainwater, cloudwater and turbulent mixing processes, as well as the evaporation and condensation of rainwater and cloudwater.