Statistical approach to identify storm deposits and cryptic event layers from grain-size data, Mechelinki, Poland, Baltic Sea

Identifying storm-induced coastal flooding deposits involves multiple laboratory techniques, with one of the fundamental methods being grain size and shape analysis. This study presents a new statistical approach for analysing grain size and shape data, enabling granulometric characterisation of storm deposits. The set of statistical methods used to objectively identify storm-related deposits includes multivariate techniques such as principal component analysis (PCA), factor analysis (FA), cluster analysis (CLA), linear discriminant analysis (LDA), and end-member modelling analysis (EMMA). The selected statistical analyses have proven valuable in identifying the storm origins of deposits. Moreover, they have enabled the depiction of cryptic event layers, referring to those not associated with discrete, macroscopically detectable sand layers. The statistical approach to granulometric data proved to be particularly useful in cases where the other lines of evidence, such as i.e. geochemical or micropalaeontological evidence are weak. Identifying cryptic event layers, the term defined in the article for the first time, improves paleoclimatic interpretations by revealing previously unexplored past storms and periods of storminess.