Refining horizontal grid spacing for storm-scale forecasts: beneficial or detrimental?

With the increase in computational power, the horizontal grid-spacing in numerical models continues to decrease. However, current research has demonstrated no significant improvement in forecasting skill when the horizontal grid-spacing decreases beyond 4km. To guide the design of future forecasting systems, it is important to understand the reasons for the lack of improvement. A quantitative assessment of the sensitivity to grid spacing relative to other model parameters is also desirable.

One of the reasons why increasing horizontal resolution may not result in more skillful forecasts is the lack of high-resolution initial conditions. In addition, the error-doubling time for convective-scale forecasts, on the order of 1 h, is much shorter than for coarser resolution simulations. Furthermore, initial condition (IC) errors at small scales can grow and propagate upscale to affect mesoscale predictability (Hohenegger; Schar 2007; Zhang et al. 2002; Zhang et al. 2006). As the IC becomes more inappropriate with increasing horizontal resolution, and that small scale IC errors grow rapidly in high resolution models, it is pertinent to ask if there are expected benefits for high resolution forecasts.

To answer this question and to contribute to the understanding of the effect of horizontal grid-spacing on the predictability of precipitation, a series of sensitivity experiments using the Weather Research and Forecasting – Advanced Research version (WRF-ARW) are performed and analyzed. Three different aspects of the sensitivity to horizontal grid-spacing will be examined. First, the impact of using coarse resolution ICs to initialize high-resolution models will be investigated in a semi-idealized setting, where the initial conditions will be provided by a previous very high-resolution run sampled at different spatial increments. Second, the effect of grid-spacing on the simulation of real precipitation cases will be assessed though a validation against radar observations. Finally, the relative importance of horizontal grid spacing compared to other model parameters will be determined. The cases analyzed herein cover several precipitation systems over the Continental US, characterized by different synoptic environments, in order to highlight the importance of the forcing for mesoscale predictability.

References

Hohenegger, C., and C. Schar, 2007: Atmospheric predictability at synoptic versus cloud-resolving scales. B Am Meteorol Soc, 88, 1783-+.

Zhang, F. Q., C. Snyder, and R. Rotunno, 2002: Mesoscale predictability of the "surprise'' snowstorm of 24-25 January 2000. Monthly Weather Review, 130, 1617-1632.

Zhang, F. Q., A. M. Odins, and J. W. Nielsen-Gammon, 2006: Mesoscale predictability of an extreme warm-season precipitation event. Weather and Forecasting, 21, 149-166.