The ERICA IOP 5 Storm. Part II: Sensitivity Tests and Further Diagnosis Based on Model Output

This paper continues the study of the ERICA IOP 5 storm begun in a companion paper. The latter documented the storm development, utilizing both conventional and special observations, and presented the results of a successful simulation of the storm by the Pennsylvania State University-NCAR Mesoscale Model MM4. At 24 h into the simulation, the MM4 predicted a central pressure of 984 mb, close to the observed value, whereas the Nested Grid Model (NGM) of the National Meteorological Center forecasted a depth of only 997 mb for the same hour. Here the results of experiments designed to test the sensitivity of the development to latent heating, surface energy fluxes, Gulf Stream position, and grid size are first presented. A high sensitivity to latent heating and a moderate sensitivity to the other parameters are found. A comparison with other cases in the literature reveals that the sensitivity to latent heating, and to the fluxes, was unusually large. In view of this finding, further diagnosis is made of the behavior of a number of moisture-sensitive parameters in the model, namely, the potential vorticity (PV), the stability of the storm environment to vertical and slantwise ascent, and the surface energy fluxes. The diagnosis revealed that (i) large diabatically produced PV, capable of sub-stantially impacting the storm intensity, appeared in the lower portion of the warm-frontal cloud mass, (ii) the storm environment was neutral or oven unstable to vertical ascent (near and ahead of the cold front) and to slantwise ascent (in and above the warm-frontal zone), and (iii) the movement of the storm near and parallel to the Gulf Stream allowed heated and moistened air to be continuously ingested into the storm. In seeking clues to the cause of the superior performance of the MM4, additional experiments are carried out in which a Kuo-type convection scheme, such as employed in the NGM, replaces the Grell scheme used in the previous MM4 simulations. It is concluded that approximately 60% of the difference between the MM4 and NGM predictions can be accounted for by the utilization of the Grell scheme and a finer grid (30 km vs 85 km) and that the effects of grid size and convective parameterization are highly coupled in this case. The remaining difference is attributed to other elements of the predictions that are not further investigated. An experiment conducted on a slowly deepening ERICA storm (IOP 7) demonstrates that, in this case at least, the MM4 shows no tendency to produce excessive deepening of ocean storms.