Hierarchical zoning for spatial sampling and digital mapping of soil attributes
Full text
| |
| Author(s): |
Derlei Dias Melo
Total Authors: 1
|
| Document type: | Master's Dissertation |
| Press: | Campinas, SP. |
| Institution: | Universidade Estadual de Campinas (UNICAMP). Faculdade de Engenharia Agrícola |
| Defense date: | 2024-07-30 |
| Examining board members: |
Lucas Rios do Amaral;
Guilherme Martineli Sanches;
André Freitas Colaço
|
| Advisor: | Lucas Rios do Amaral |
| Abstract | |
Adequate soil sampling plays a crucial role in obtaining reliable information for the development of soil attribute maps. However, due to financial constraints, sampling density is often limited, resulting in samples that may not accurately reflect the spatial variability of the soil. Therefore, it is essential to identify key sampling locations to produce more accurate maps. The objective of this research was to evaluate whether delineating agricultural areas into regions of macro and micro-variability allows for targeting sampling points to better capture the variability of soil attributes at the surface, and thus produce better digital soil maps compared to regular grid sampling. In this study, we proposed two densities for mapping. First, a limited sampling density of one sample every 2.5 hectares, referred to here as a sparse grid, which is commonly used by precision agriculture service providers. The second density was a grid with one sample per hectare, which tends to deliver good results for soil attribute mapping in many areas. Both sampling densities were configured in three sampling arrangements: a completely regular grid and two grids combining regular points and targeted points, comprising 25% and 50% targeted points. Initially, we defined zones of macro-homogeneity based on the apparent magnetic susceptibility of the soil (SM), assigning a number of targeted points to each zone considering its variability and area size. We then defined micro-homogeneity based on plant vigor, using the Enhanced Vegetation Index (EVI). Microzones of homogeneity were defined by directing each sampling point to the pixel with the highest Fuzzy membership value. Our results indicate that the sampling configuration with 50% of the sampling points directed improves interpolation quality for multiple soil attributes, both at higher densities and in sparse sampling grids. Although there is a predictive improvement in mapping with a density of one sample per hectare, the gains are not robust enough to replace the regular grid with the proposed method. However, when using the sparse grid, the predictive gains justify adopting the method. Nonetheless, the proposed approach requires prior knowledge of the behavior of attributes in the study area, as well as their relationship with the covariates used in delineating the macro and micro-zones. If such prior knowledge is not available, the proposed approach can be used as a two-stage mapping strategy, where a percentage of points from the regular grid is initially collected to understand the relationship with environmental variables. This would allow the analyst to understand the behavior of the attributes in the area and subsequently apply the most suitable covariates for targeting based on the proposed methodology (AU) | |
| FAPESP's process: | 23/02592-4 - Sampling optimization and multivariate predictions for mapping soil fertility and soil quality |
| Grantee: | Derlei Dias Melo |
| Support Opportunities: | Scholarships in Brazil - Master |