SPATIAL PREDICTION OF SOIL INFILTRATION USING FUNCTIONAL GEOSTATISTICS

  • Diego Leonardo Cortes-D Universidad Nacional de Colombia, Department of Civil Engineering and Agricultural, Colombia
  • Jesus Herman Camacho-Tamayo Universidad Nacional de Colombia, Department of Civil Engineering and Agricultural, Colombia
  • Ramon Giraldo Universidad Nacional de Colombia, Department of Civil Engineering and Agricultural, Colombia
Keywords: functional data; functional kriging; Kostiakov; Philip; spatial variability

Abstract

The infiltration of water into the soil is a necessary parameter for irrigation systems design. Characterizing its spatial behavior allows a site-specific management of water according to soil conditions and crop requirements. The aim of this study is to establish the spatial distribution of infiltration in an Andisol by means of two geostatistical approaches: on the one hand by means of functional kriging, taking as input infiltration curves (obtained after a smoothing stage), and on the other hand by using classical ordinary kriging on the parameters of the Kostiakov and Phillip models. The comparison between these methodologies is carried out taking as a criterion the sum of squared errors of a leave-one-out cross-validation analysis. The results show a high correlation between observations and predictions (R2 values around 99%), which indicates that the use of functional geostatistics in this context could be a good alternative. Moreover, from a descriptive point of view, we can point out that the contour maps of basic infiltration (BI), cumulative infiltration (Ci), saturated hydraulic conductivity (Ks), and sorptivity (S) obtained with theĀ  observed data, as well as the predictions by functional geostatistics, show a very similar behavior, which empirically validates the use of this methodology.

Published
2018-12-21
How to Cite
Cortes-D, D., Camacho-Tamayo, J., & Giraldo, R. (2018). SPATIAL PREDICTION OF SOIL INFILTRATION USING FUNCTIONAL GEOSTATISTICS. AUC Geographica, 53(2), 149-156. https://doi.org/10.14712/23361980.2018.15
Section
Original Articles