Spatio-temporal decision uncertainty of selected soil physical parameters can enhance variable rate irrigation

Abstract

The effectiveness of crop irrigation via sprinkler systems could be improved by environmental variability inherent to field conditions, thus leading to the sub-optimal irrigation of certain sections. To rectify this discrepancy, in-situ soil characteristics were methodically correlated with the region's hydrological circumstances and root zone, assigning a distinct level of uncertainty to each decision point. A stochastic geodatabase was then generated, offering prospective applications in precision agriculture. The experimental agricultural field in Nyírbátor, Hungary, served as the reference point, with the constraints posed by variability being surmounted through a dual-layer iteration of random sampling structures employing the sequential Gaussian simulation (SGS) method. For this purpose, 25 physical, 9 chemical, and 11 soil microelements were examined from samples extracted from 105 boreholes in an 85-hectare cornfield while adopting a regular sampling scheme within a 100 × 100 m grid. Each soil parameter estimation underwent the following process: 1. Organization of data and application of exploratory statistics for outlier identification; 2. Normal score transformation; 3. Exploratory variography; 4. Sequential Gaussian simulations, leading to the construction of a series of plausible, equally probable realizations; 5. Computation of medians and the 95% confidence intervals. These methodologies were deployed concerning the soil characteristics, with porosity being selected as the representative soil parameter for the Nyírbátor cornfield. Porosity was our focus physical parameter because the micro and macro soil structures greatly influence the hydraulic characteristics of the soil such as water infiltration, hydraulic conductivity and moisture retention. Comparative assessments of the Hydrus 3D hydrological models of kriged and sequential Gaussian simulation surfaces were conducted. Results highlighted the efficacy of sequential Gaussian simulation in encapsulating the field's heterogeneity, and the accompanying uncertainty served as a decision-making tool in the diversified water application across the field. The results were validated using field data observations of soil moisture in the corn field from 2020 and 2021 respectively and nonetheless, the uncertainty divergence between the Hydrus outputs unveiled the knowledge deficit concerning actual spatial patterns of soil porosity. The established workflow offers a cost-efficient dynamic methodology for water resource management, potentially curtailing overall irrigation expenditure by variably applying water to parcels based on uncertainty estimates.