Spatial variability of selected soil agrochemical parameters and its relation to soil sampling strategy

In paper is presented estimation of spatial variability of selected soil agrochemical parameters (pH/KCl, available phosphorus and potassium) through basic geostatistical methods and some consequences in relation to soil sampling strategy. Problem was solved on example two fields within agricultural trade-producing farm Kocin, district Piestany (western part of the Slovak Republic). Soil samples were taken in regular grid (within field Chmelnica with acreage 8 ha in grid 25 x 25 m and within field Zahumnie with acreage 17 ha in grid 50 x 50 m) from topsoil layer. Soil on both field is classified as loamy Haplic Fluvisol on loess, or loess loams and polygenetic loams, respectively. Composite soil sample was created from three sub-samples that were taken approximately from area with diameter 3 m by soil auger. Point co-ordinates were obtained from vectored field maps in scale 1 : 5000. Semivariograms estimation from measured data was made with help of point kriging method and usual types of models (spherical, exponential, power function, and gaussian model). For semivariance estimation were the best suitable spherical and exponential models. From statistical treatment of measured data follows that range for selected agrochemical parameters (pH/KCl, available phosphorus and potassium) fluctuated between 68.9 and 155.8 m. Range values are not related with classical variability indicators (e.g. Coefficient of variation). Crossvalidation characteristics were satisfactory in all three agrochemical parameters on field Chmelnica and partly at pH/KCl and available phosphorus on field Zahumnie. Less favourable results in the case of available potassium are probably caused by occurrence of greater number of high values among the relatively low. In principle, range values indicate density of soil sampling under that soil parameters are spatially dependent. Recommendation of many authors to organise soil sampling in grid 40 - 100 x 40 - 100 m was in coherence with sampling density within two selected fields. In comparison with present practice within agrochemical soil testing, soil sampling in grid 100 x 100 m or one sample per hectare seems problematic and economical demanding from view-point of amount of soil samples and number of analysis. Recommended decreasing for laboratory analysis costs is not related with decrease samples soils number but rather with the following purposive composing and thus lowering the analysis number. The precision of range parameter estimation depends on density of soil sampling and length of investigated area. Insufficient number of soil samples and pairs number may skew estimation of semivariogram and spatial distribution of selected parameters within field especially on fields with small acreage, irregular/extreme shape, while sampling grid will be not condensed. Economical effect of differentiated fertilization within field is possible assume in the case of different soil texture or different soil fertility. Our results confirm that identification of spatial variability of soil agrochemical parameters does not always serve the next utilization at fertilizer application (e.g. liming on field Chmelnica is not actual)