Understanding Mesoscale Error Growth and Predictability

Technological advances have made atmospheric mesoscale modeling at very fine resolutions readily available to a great number of organizations. Though initial operational results show some skill with respect to synoptic scale forecasts, many of the problems associated with mesoscale error growth and predictability have been ignored. Understanding mesoscale error is critical to accurately interpreting mesoscale model results and output from tactical decision aids (TDA's). This study examines mesoscale error growth and predictability through controlled numerical model experiments. A known "true" atmosphere is created through the use of the US Navy's Coupled Oceanographic/Atmospheric Mesoscale Prediction System (COAMPS). Virtual observations are randomly sampled from this atmosphere to provide data for ingest into forecasts using the NCARIPenn State MM5 mesoscale model. Forecast results for ten cases are compared against the "true" atmospheric solution and error statistics are calculated for wind speed and geopotential height fields. Results show how error growth and predictability are affected by different variables such as boundary conditions, weather regime, sample size and sample distribution. A scale separation of error is also performed in order to assess the impact of synoptic scale error on mesoscale error.