Êtes-vous un étudiant de l'EPFL à la recherche d'un projet de semestre?
Travaillez avec nous sur des projets en science des données et en visualisation, et déployez votre projet sous forme d'application sur Graph Search.
This work analyses the temporal and spatial characteristics of bioclimatic conditions in the Lower Silesia region. The daily time values (12UTC) of meteorological variables in the period 1966–2017 from seven synoptic stations of the Institute of Meteorology and Water Management (IMGW) (Jelenia Góra, Kłodzko, Legnica, Leszno, Wrocław, Opole, Śnieżka) were used as the basic data to assess the thermal stress index UTCI (Universal Thermal Climate Index). The UTCI can be interpreted by ten different thermal classes, representing the bulk of these bioclimatic conditions. Stochastic autoregressive moving-average modelling (ARMA) was used for the statistical analysis and modelling of the UTCI as well as separately for all meteorological components. This made it possible to test differences in predicting UTCI as a full index or reconstructing it from single meteorological variables. The results show an annual and seasonal variability of UTCI for the Lower Silesia region. Strong significant spatial correlations in UTCI were also found in all stations of the region. “No thermal stress” is the most commonly occurring thermal class in this region (about 38%). Thermal conditions related to cold stress classes occurred more frequently (all cold classes at about 47%) than those of heat stress classes (all heat classes at about 15%). Over the available 52-year period, the occurrence of “extreme heat stress” conditions was not detected. Autoregressive analysis, although successful in predicting UTCI, was nonetheless unsuccessful in reconstructing the wind speed, which showed a persistent temporal correlation possibly due to its vectorial origin. We conclude thereby that reconstructing UTCI using linear autoregressive methods is more suitable when working directly on the UTCI as a whole rather than reconstructing it from single variables.
Victor Panaretos, Laya Ghodrati
Julia Schmale, Jakob Boyd Pernov, Jules Gros-Daillon