期刊名称：Science et changements planétaires / Sécheresse
印刷版ISSN：1147-7806
电子版ISSN：1777-5922
出版年度：2011
卷号：22
期号：3
页码：178-185
DOI：10.1684/sec.2011.0312
出版社：John Libbey Eurotext
摘要：Figures See all figures Authors Houria Dakak , Brahim Soudi , Aicha Ben Mohammadi , Ahmed Douaik , Mohamed Badraoui , Rachid Moussadek Institut national de la recherche agronomique (Inra) Avenue Mohamed Belarbi Alaoui BP 6356 10101 Rabat Maroc, IAV Hassan II Département des sciences du sol 10101 Rabat Maroc, Université Ibn Tofail Faculté des sciences Département de géologie UFR ST 11/DOC/K 14050 Kénitra Maroc, Institut national de la recherche agronomique (Inra) Avenue de la Victoire BP 415 10000 Rabat Maroc Key words: modelling, Morocco, salinisation, soil, space-time distribution, watershed DOI : 10.1684/sec.2011.0312 Page(s) : 178-85 Published in: 2011 The control of soil salinity based on knowledge of its spatial distribution and evolution in time, is becoming necessary in order to secure strategies for sustainable agricultural development. The conventional method of determining salinity by measuring the electrical conductivity of the extract of saturated soil paste (ECe) remains cumbersome due to the large spatial variability in soil salinity. As an alternative, electromagnetic induction can be used to measure apparent electrical conductivity (ECa) of soil in the field. It is fast procedure and provides more sampling in space and time. This study has the aim of assessing whether sampling density is sufficient for interpolating soil salinity reliably. In addition, it has the objective of optimizing this sampling density by seeking to define the maximal distance to respect between measurement points. It concerns a 2,060 hectare watershed located in the semiarid plain of Tadla, Morocco. A series of 112 measures of ECa were carried out on a regular grid of 500m x 500m with the Geonics EM38 instrument while 12 samples were collected for the determination of ECe. A simple linear regression model was used to convert values of ECa into ECe. The variogram was then used for describing and modeling soil salinity variability. Finally, kriging was used for prediction at non-sampled locations. A soil salinity map was produced on the basis of these predictions. The results show that the variogram is better fitted with a spherical model with a nugget effect with a range of 1,000 m and the random component representing 54% of the total variation in soil salinity. This is due mainly to the size of the sampling grid and errors related to calibration. This study shows clearly that extrapolation is difficult due to sparse density of measurement. A maximal distance of around 250 m between measurement points would be required in order to describe and reliably model spatial variability of soil salinity at a reasonable cost with the aim of predicting at non-sampled locations and delineating different levels of soil salinity.
