Analysis of spatial interpolation for optimising management of a salinized field cultivated with lettuce

Abstract

The lack of randomisation in irrigation experiments is usually a disadvantage. The introduction of spatial variable experimental design offers a convenient tool to help solving this problem. In order to understand the variation of some soil physical and chemical properties in an experimental block and its effect on lettuce ( Lactuca sativa L.) production, graphical interpretation of those soil properties was done with the use of geostatistics in a geographic information system (GIS). In this work three techniques of geostatistics were used for the creation of several maps of soil properties in an experimental plot cultivated with lettuce. Lettuces were evaluated for individual weight and diameter at the end of the cropping season. The soil properties studied were: total mineral nitrogen, phosphorus, potassium, pH, electric conductivity and saturated soil hydraulic conductivity. The techniques used were: ordinary kriging, inverse distance and Thiessen polygon. Cross validation used to compare the prediction performances of the three geostatistical interpolation algorithms determined that kriging was the best technique for each soil property. Prior to the creation of the maps, semivariograms were produced for each soil property. The maps resulting from the interpolation techniques were introduced in a GIS and their values reclassified. After that, spatial modelling was used to develop a final overlay map from all the information of the analysed soil properties simulating a “lettuce production capability map”. This final map was created with the objective to determine which areas in the plot had optimal conditions for lettuce development. It was concluded that the plot did not had an optimal area for lettuce production. Localized problems with soil properties were found that could be solved with simple geographically restricted amendment treatments. Final lettuce yield had high correlation ( r 2 = 0.83) with the lettuce capability map derived.