Premium
The convective adjustment time‐scale as indicator of predictability of convective precipitation
Author(s) -
Keil Christian,
Heinlein Florian,
Craig George C.
Publication year - 2013
Publication title -
quarterly journal of the royal meteorological society
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.744
H-Index - 143
eISSN - 1477-870X
pISSN - 0035-9009
DOI - 10.1002/qj.2143
Subject(s) - predictability , forcing (mathematics) , parametrization (atmospheric modeling) , precipitation , convection , climatology , environmental science , forecast skill , quantitative precipitation forecast , atmospheric sciences , meteorology , mathematics , geology , physics , statistics , quantum mechanics , radiative transfer
Predictability of convective precipitation depends on the interaction between synoptic forcing and local‐scale flow characteristics. In order to assess different predictability levels it is desirable to objectively determine the dominant process in a given meteorological situation. Such a measure is given by the convective adjustment time‐scale τ , a physically based quantity that distinguishes between strong and weak synoptically forced precipitation regimes. By employing the convective adjustment time‐scale diagnostic, forecasts of the convection‐permitting COSMO‐DE ensemble prediction system available for a total of 88 days in summer 2009 are examined. Based on the normalized ensemble spread of hourly precipitation rates, it is shown that the practical predictability of total precipitation is higher during strong large‐scale forcing than during weak forcing. Likewise, the forecast skill, determined using two deterministic scores, is higher during strong than during weak forcing conditions. Different predictability levels of convective precipitation can be revealed by examining distinct sub‐ensembles depending on their source of uncertainty. The impact of variations in the boundary conditions of the driving global models used in the ensemble system is quite insensitive to the prevailing flow regime, while the impact of physics perturbations representing the model error is clearly weather regime dependent, exhibiting a strong contribution only during weakly forced conditions. Then convective precipitation turns out to be especially sensitive to variations in the physics parametrization even at forecast lead times of 12 to 18 hours during the main convective period in the afternoon. Two case‐studies exemplifying the strong and weak forcing regimes are shown, to illustrate how forecast skill varies and the different ensemble members cluster as the precipitation event evolves.