Densidades amostrais combinadas com pontos adicionais na variabilidade espacial de atributos químicos do solo

Abstract

Precision agriculture comprises tools and technologies for the rational use of agricultural inputs. In order to evaluate the spatial variability of soil fertility attributes, such as CEC (cation exchange capacity), V% (base saturation), K (potassium) and P (phosphorus), the georeferenced soil sampling technique guided by remote sensing products, such as NDVI, which has a dependence relationship with the productivity of soybean and corn crops. Thus, the goal was to evaluate regular sampling meshes and the allocation of additional collection points based on the transition regions of the NDVI maps of the previous culture in determining the spatial variability of attributes, phosphorus, potassium, cation exchange capacity and base saturation, in addition to the recommendations of fertilization and correction and its correlation with the soybean NDVI in the previous harvest. The experiment was carried out in a plot of 224 ha of a commercial farm located in the city of Maracaju - MS, the soil was oxissol. The sampling arrangements were generated by a regular grid of one sample per hectare (1:1), with the addition of up to 50% of additional points allocated by the NDVI map, thus composing 336 sample points and the other grids were composed by the withdrawal of 10 % of additional points up to 0% and likewise for the grid of 1 sample every 4 hectares (1:4). P (mg dm-3), K (Cmolc dm-3), CEC (Cmolc dm-3) and V (%) were evaluated, submitted to descriptive statistics, geostatistics, ordinary kriging, evaluation of maps by relative deviation coefficients (CRD) and kappa. The NDVI of the 2021/2022 soybean crop was also evaluated based on available images from the Sentinel 2 satellite and its correlation with soil attributes. Phosphorus (P) even with appropriate to high values in the area allows separating grids of different sampling intensities, with denser grids suitable for assessing the variability of this attribute. It appears that the grids with a sample per hectare already achieve satisfactory results, but with 30% of additional points the best representation of the variability occurs. Potassium (K) showed high values in the whole area, which made it impossible to distinguish the sample grids. For CEC, the sample grids studied were equivalent in estimating spatial variability. Base saturation (V%) had better identification of spatial variability with the use of grids of at least 1:1 (one sample per hectare). For limestone recommendation purposes, the grids with a sample per hectare (1:1) are more efficient in representing the variability than those with a sample for 4 ha (1:4) and the measure that increases the number of additional points improves the estimate in the 1:1 grid. The correlation between P and NDVI is very weak, weak with K and moderate with CEC and V%.