Basic statistics of the yield gain surface.

Agricultural experimentation requires careful selection of the experimental design and model for analyzing treatment data. However, even with rigorous experimental control, the discrepancies between treatments are so subtle that traditional statistical models fail to highlight statistically significant differences that occur in field practice. The incorporation of geotechnologies offers the ability to map agricultural variability, but a gap still exists in the availability of tools designed to map and evaluate the effectiveness of agricultural experiments. To overcome this limitation and promote the wider application of Geographic Information Systems in agriculture, the scope of this study focuses on the development of a resource in QGIS software, aimed at evaluating agricultural experiments using a randomized block design with up to five treatments. The resource developed incorporates spatial interpolation techniques using geostatistical kriging, map generation, and statistics. The study used yield samples from six different crops to identify quantitative and spatial differences between two-treatment experiments in terms of yield gain. The performance evaluation included statistical measures such as Root Mean Square Error (RMSE) and Pearson’s correlation coefficient to validate the accuracy of interpolated surfaces. The results consisted of two surfaces representing the study area treated with each of the treatments, as well as a surface reflecting the yield gain of the reference treatment in relation to the control treatment, accompanied by relevant descriptive statistics measures on this gain surface. The experiments showed that RMSE varied between 4.6% and 69.71%, depending on the crop and treatment, while Pearson’s correlation ranged from −0.16 to 0.86, indicating varying degrees of agreement between interpolated and observed data. The tool successfully generated yield gain maps, allowing spatial visualization of treatment differences, with an accuracy of up to 95.40% in detecting spatial and numerical variations between treatments. The tool, called GEOTrat - Points, offers the flexibility to evaluate agricultural experiments of various designs, encompassing different crops and different quantities of samples, providing both numerical and spatial analysis.