Caractérisation et cartographie des propriétés physico-chimiques des sols de la région de Ghardaïa (Cas de la palmeraie de ZELFANA) Algérie.

Abstract

Oasis agrosystems play an important role in the development of arid lands. Water is the limiting factor for the improvement of these areas. However, if development under irrigation contributes to the increase of agricultural production, its extension is accompanied by serious threats to the preservation of the environment, especially the quality of the soil, which is already weakened by extreme climatic conditions. In this topic, the spatial variability of the properties of irrigated soils is a means of knowing the state of the latter. This study aims to determine the spatial variability of organic carbon, salinity, PHWATER, pHKCL, total and active limestone in the soil using a geostatistical approach. The present study was carried out in the region of Zelfana (Ghardaïa) located in the Algerian central Sahara. We opted for a random sampling method, where we carried out 15 profiles. The edaphic study was done on three depths (0-30 cm, 30-60 cm, 60-120 cm). The main results obtained show that the irrigation water coming from the Albian aquifer has a medium to poor quality (C3 S1). The granulometric results show that the study area is characterised by a predominantly sandy to sandy-clay texture. The analytical results show that the soils studied are generally very poor in organic carbon (CO < 1%) in all three levels. The soil is low to very saline, with EC values at 25°C ranging from 0.64 to 3.87 dS/m with an ascending saline profile. The soil is slightly to highly calcareous, with total limestone values ranging from 2.08 to 30.94% 15.62%. The soil is alkaline with a pH ranging from 8.25 to 8.31.The results of the spatial distribution of the studied parameters show a high to very high variation of organic carbon, moderate to very high for limestone, and high to moderate for salinity. On the other hand, the spatial variability of pHWATER and pHKCL is low. The nugget effect is very low for all parameters studied. The results of the cross-validation led to the selection of the most reliable variogram models, namely the circular, spherical, gaussian and exponential, which were used to produce the spatial variability maps by ordinary kriging.