Diagnosing MCS Forecast Errors in the Warn-on-Forecast System

Mesoscale convective systems (MCSs) develop as individual thunderstorms merge and grow upscale into a contiguous precipitation area >100 km across. MCSs occur within the U.S. primarily during the warm season and can be associated with severe weather hazards that include tornadoes, damaging winds, hail, and flash floods. The experimental Warn-on-Forecast System (WoFS) is a rapidly updating convection-allowing ensemble system used for the short-term (0–6 h) probabilistic prediction of severe weather hazards. Britt et al.’s (2023) systematic verification of WoFS forecasts for 63 MCSs that occurred in 2017–2021 showed that WoFS predicts MCSs particularly well when the system is present in the forecast initial conditions, but that the northern portion of the forecasted MCSs have a northwest position bias compared to observations that increases with lead time. The aim of this presentation is to test the hypothesis that WoFS is producing cyclonic mesoscale convective vortices that are stronger and deeper than observed, contributing to the northwestward position bias at 3–6 h forecast leads.

Potential vorticity (PV) diagnosis is used to examine the development of mesoscale convective vortices and line-end vortices in observed and forecasted MCSs for 1584 WoFS forecasts in 2019-2020. To do this, PV, vorticity, and potential temperature from WoFS forecasts are compared against the 0-h analyses available from the WoFS data assimilation system. Area averages from within the stratiform (defined as 1 km reflectivity < 45 dBZ) and deep convection (at least 45 dBZ) regions are computed for comparison of forecasts against 0-h analyses. Statistical significance of forecast errors is tested using bootstrap resampling without replacement.

Results confirm that of the 1584 6-h forecasts, 49% (784) exhibited a northwest position bias and 8% (124) exhibited an opposite southeast position bias. For all 1584 WoFS forecasts, a statistically significant positive PV bias in the 900-700 hPa layer and significant low PV bias in the 650-250 hPa layer in the stratiform region was apparent. In the lower troposphere, the positive PV bias is accompanied by a significant positive vorticity bias maximized at 900 hPa and significant positive potential temperature bias maximized at 700 hPa. For the upper-troposphere, the negative PV bias is accompanied by a negative vorticity bias maximized at 275 hPa and a positive potential temperature bias maximized at 225 hPa. A similar vertical structure of the PV bias was present for the subset of WoFS forecasts with a northwest position bias. The PV bias was apparent in the lower troposphere for the southeast bias forecasts, except in a shallower layer compared to the northwest position bias forecasts. A significant PV bias was not apparent in the upper troposphere for the southeast position bias forecasts.

The question remains on the cause of the low- to mid-level positive PV bias in the stratiform region of the forecasted MCSs. Examination of WoFS forecasts of the MCS on 17-18 July 2020 in the northern Great Plains (a northwest position bias case) shows that the stratiform region had a positive bias of 5-10 dBZ in derived reflectivity highlighting the possibility that the PV errors are related to condensational heating. Detailed examination of ensemble member 6 initialized at 0000 UTC 18 July reveals how a positive PV bias began to develop near 750 hPa 45 minutes into the forecast and persisted through 6 hours. The development of the PV bias coincided with the development of a positive vorticity bias in the midtroposphere that intensified and deepened after 2 hours. The vorticity originated along the leading convective line and rapidly amplified and moved rearward into the stratiform region. While the positive PV bias was preceded by a warm bias in the upper troposphere, a warm bias developed in the mid troposphere near 650 hPa in concert with the development with the PV bias. In all, this case study suggests that the positive PV bias in the stratiform region of MCSs in WoFS forecasts may be related to the overdone coupling of low-level vertical mass flux and cyclonic vorticity within the deep convection region. The resulting vorticity increases and deepens, evolving into an intense mesoscale line-end vortex. Results from additional MCS cases in WoFS and experiments with alternative microphysics and horizontal resolution will be discussed.