Prospective direction of geostatistic analysis and stochastic modeling taking into account economic costs precise agriculture

Abstract

Th e article discusses the current approaches to the assessment of the heterogeneity of soil properties. On the example of the experimental plot (106 ha) of the RUNE “Grodno Zonal Institute of Plant Industry of the National Academy of Sciences of Belarus” it is shown that ignoring the intra-fi eld heterogeneity leads to even more signifi cant diff erences. On the experimental fi eld, using satellite images and multi-temporal indices EVI, microplots with an area of 50 to 200 m2 with а signifi cantly reduced level of eff ective fertility, up to the death of plants, were identifi ed, on the basis of which а sampling grid was developed. During the fi eld work, 214 pHKCl samples were taken with а sampling step of 70 m and 53 additional samples in the “problem areas”. Using geostatistical methods, data on the spatial heterogeneity of pHKCl were obtained, the degree of spatial dependence and the accuracy of the results obtained were estimated for various numbers of samples (107, 216, and 268 points). Further use of stochastic modeling made it possible to construct high-precision cartograms of acidity with а pixel resolution of 4 m and 12 m. We also showed that the use of cell sizes less than the processing device is not useful for solving applied issues of precision farming. Probabilistic-statistical models were obtained, represented by cartograms of mean square errors and the probability of the phenomenon. Th e resulting cartograms showed the presence of ultra-acidic soils in the “problem zones”; such а reaction also determines the presence of toxic aluminum compounds and extremely low calcium and magnesium contents, which jointly lead to the absence of seedlings of cultivated vegetation. А detailed accounting of soil acidity in comparison with а classical agrochemical survey provided savings of $ 2807 by the application of СаС03, which made it possible to fully compensate for the additional costs of taking and analyzing а larger number of samples.