Getting to the core of winter frosts in central South America through decision trees
- 1Complutense University of Madrid, Faculty of Physical Sciences, Physics of the Earth and Astrophysics, Madrid, Spain (scollazo@ucm.es)
- 2University of Buenos Aires, Faculty of Exact and Natural Sciences, Department of Atmospheric and Ocean Sciences, Buenos Aires, Argentina
- 3Institute of Geosciences (IGEO), Spanish National Research Council – Complutense University of Madrid (CSIC–UCM), Madrid, Spain
Agricultural production in Central South America (CSA) is highly sensitive to frost events. This study aims to provide a detailed characterisation and quantification of the physical processes leading to frost conditions in CSA between 1979 and 2022, using gridded observational data. To accomplish this objective, we focus on three innovative elements: a thorough examination of regional frost properties, a novel multi-parametric method for the description of upper-level jets, and the quantification of the underlying mechanisms through decision trees. We selected this methodology due to its high interpretability, particularly advantageous for users within the agricultural sector, owing to its visually intuitive representation.
After regionalisation and statistical description of the frosts, we have been able to identify five homogeneous regions in CSA. Regarding the synoptic mechanisms that promote the occurrence of extensive frosts, we have found that the physical processes are similar across all regions. These processes are triggered by the passage of a cold front and the establishment of a migratory post-frontal anticyclone, which is reinforced by the subsidence generated on the southern flank of the entrance region of the Subtropical Jet (STJ). We have computed six parameters to characterise the STJ and seven for the Polar Front Jet (PFJ). They consider the latitude, intensity, height, tilting, longitudinal extent, and branches of the jets. By employing decision trees, we have effectively discerned a subset of jet parameters that are closely linked to frost events. The most relevant of these are the latitude of the STJ, the longitudinal extent of the Atlantic branch of the STJ, and the tilt of the Atlantic branch. In general, when the STJ is positioned north of 31°S, the likelihood of frost in the CSA rise. However, in more northern areas, the probability of frost occurrence further increases with a slight northward shift in its location. In addition, extensions of the Atlantic STJ ranging from 25 to 55°, coupled with a negative tilt, are two characteristics often linked to frost occurrences. These features are especially accentuated in events occurring in the northernmost regions. The application of decision trees has allowed us to summarise this information for every region of CSA in a user-friendly way, helpful for decision-makers in the agricultural sector.
Acknowledge: The SAFETE project, which has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 847635
How to cite: Collazo, S. and García-Herrera, R.: Getting to the core of winter frosts in central South America through decision trees, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-277, https://doi.org/10.5194/ems2024-277, 2024.