Surface Formations Salinity Survey in an Estuarine Area of Northern Morocco, by Crossing Satellite Imagery, Discriminant Analysis, and Machine Learning - Archive ouverte HAL Accéder directement au contenu
Article Dans Une Revue Soil Systems Année : 2023

Surface Formations Salinity Survey in an Estuarine Area of Northern Morocco, by Crossing Satellite Imagery, Discriminant Analysis, and Machine Learning

Résumé

The salinity of estuarine areas in arid or semi-arid environments can reach high values, conditioning the distribution of vegetation and soil surface characteristics. While many studies focused on the prediction of soil salinity as a function of numerous parameters, few attempted to explain the role of salinity and its distribution within the soil profile in the pattern of landscape units. In a wadi estuary in northern Morocco, landscape units derived from satellite imagery and naturalistic environmental analysis are compared with a systematic survey of salinity by means of apparent electrical conductivity (Eca) measurements. The comparison is based on the allocation of measurement points to an area of the estuary from Eca measurements alone, using linear discriminant analysis and four machine learning methods. The results show that between 57 and 66% of the points are well-classified, highlighting that salinity is a major factor in the discrimination of estuary zones. The distribution of salinity is mainly the result of the interaction between capillary rise and flooding by the tides and the wadi. The location of the misclassified points is analysed and discussed, as well as the possible causes of the confusions.

Dates et versions

hal-04119965 , version 1 (07-06-2023)

Identifiants

Citer

Youssouf El Jarjini, Moad Morarech, Vincent Valles, Abdessamad Touiouine, Meryem Touzani, et al.. Surface Formations Salinity Survey in an Estuarine Area of Northern Morocco, by Crossing Satellite Imagery, Discriminant Analysis, and Machine Learning. Soil Systems, 2023, 7 (2), pp.33. ⟨10.3390/soilsystems7020033⟩. ⟨hal-04119965⟩
11 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More