LEÃO, Armindo Bezerra.; http://lattes.cnpq.br/0248891519767458; LEÃO, Armindo Bezerra.
Resumen:
The soil atributes may present a great variability, due to formation and management
factores, such as an inadequate irrigation that coutj produce salinity problems. The
classical statistics consider that soil variability occurs in a randomized form, however
several studies have shown that the attributes present an strong spatial dependence,
nedding thus geoestatical analises. The objective of this work was to study the variables
that characterize the soil salinity by classical statistics and geostatistics and also to evaluate the spatial variability, the spatial dependence amongst samples, to verify the crossed validation of the interpolation method (Kriging) and to build maps of isolines of the ali variables. The present work was conducted at the Engenheiro Arcoverde irrigated
perimeter located at Condado, Paraíba, aiming to measure the soil parameters that define
the soil salinity, to study the spatial variability of them throughout geoestatical analyses, to
study the liability of the Kriging interpolation method and to construct salinity and sodicity
maps, furnishing subsides for an adequate soil management which could allows their
reclamation. For this, 159 soil samples were collected at 0-20, 20-40 and 40-60 cm depth
intervals, on an irregular grid with the sampling points separated at approximately 100 m.
To evaluate the spatial variability initially it was used classical statistics to verify position
and dispersion measures and afterwards geoestatistics to analise the semivariograms,
crossed validation and to obtain the best adjusted models. After interpolation by kriging,
isoline maps were constructed showing the salinity and sodicity situation of the area. The
results obtained allowed to observe a great spatial variability of eletrical conductivity and
exchange sodium percentage. The pH variable showed a low spatial variability. The spatial
dependence of the studied attributes permitted their mapping using kriging techniques. The crossed validation offered an excellent precision to estimate data. The compartments
identification, sub-regions of the studied area, was excellent to characterize the soil
salinity. Small variations on the soil relief, on the soil formation and management
conditioned the variability found for the studied attributes. Finally, the geoestatistics
showed the possibility to obtain a greater volume of information from a small volume of
data, which means efficience, saving of time and resource
