CR 25:265-274 (2004)  -  doi:10.3354/cr025265

ABSTRACT: We compare 2 different approaches applied in the evaluation of heat-related mortality: a Œtraditional¹ one, based on the analysis of relationships between mortality and individual meteorological variables and/or indices (such as heat index), and a synoptic one, which links mortality to objectively determined air masses, taking into account the entire weather situation rather than single elements. Various methods of cluster analysis and of final forming of air masses were tested in the air-mass classification procedure; all classifications comprise an Œoffensive¹ air mass, which is associated with a pronounced rise in mortality of 20 to 30 deaths d^-1 (7 to 10% relative increase). It is characterized by elevated temperatures, low cloud cover and a relatively strong flow with a southern component. Regression analysis reveals a slightly stronger correlation of mortality with heat index than temperature, enhanced response in females over males, and the strongest relationship for unlagged variables. Positive correlations hold for lags of 0 to 3 d only, while at lags of 4 to 25 d, the link is negative, which demonstrates the mortality displacement effect and its time extent. Significantly increased mortality is observed at daily maximum (average, minimum) temperatures reaching or exceeding 26°C (20°C, 14°C) and their anomalies from mean seasonal courses of 2°C or larger. The results demonstrate that the air-mass-based approach can be applied to central European conditions and is a suitable alternative to the more commonly used methods in assessing impacts of heat stress on mortality in this area.

KEY WORDS: Mortality · Heat stress · Air masses · Cluster analysis · Czech Republic