RFJ Spatial variability in fertigated coffee yields and plant nutrients in soil saturation extracts

The spatial distribution and levels of available plant nutrients (elements) in the soil can limit coffee yield and must be evaluated for effective crop management. Therefore, we analyzed spatial variability in yield and plant nutrients in the saturation extract of a clayey Oxisol cropped with fertigated coffee. The experiment was carried out on 14 hectares of coffee in Monte Carmelo, Minas Gerais, Brazil. Soil samples were collected (0 - 0.2m layer) at 61 regular grid points (spaced 50x50m) and used to determine plant nutrients in the saturation extract. Coffee yield was also determined at these points. Descriptive statistics were calculated for each variable and geostatistics were used to build a spatial variability model representing the physical attributes of the soil. Variographic analysis was performed using semivariograms. These showed that yield and soil chemistry varied throughout the study site. Thus, the maps generated from geostatistics can be useful tools for soil management in fertigated coffee crops.

MÉTODOS DE INTERPOLAÇÕES DE DADOS OBTIDOS POR SENSORES DE AGRICULTURA DE PRECISÃO

Os sensores que medem as características do solo em campo são importantes ferramentas para o manejo da agricultura de precisão, entre eles, destaca-se o sensor de contato, que mede a condutividade elétrica (CE), matéria orgânica (MO) e potencial hidrogeniônico (pH) do solo. Objetivou-se avaliar os erros dos métodos de interpolação por krigagem, inverso da distância e normal da distância a partir de dados de CE, MO e pH do solo. O experimento foi realizado no município de Candói – PR, onde foram amostradas duas áreas, os dados foram coletados por sensor de contato, o qual foi conFigurado para uma coleta de 150 pontos por hectare para a condutividade elétrica e a matéria orgânica, e para potencial hidrogeniônico a frequência de coleta foi de 15 pontos por hectare, o equipamento foi acoplado em um trator operado a uma velocidade de 8 km h-1 com passadas paralelas de 20 m. Realizaram-se análises variográficas, validação cruzada e elaboração de mapas. Os menores erros de interpolação ou “jack knifing” para CE, MO e pH foram apresentados pelo método de interpolação inverso da distância, para o talhão T2, e no talhão T1 o método da Krigagem obteve os menores erros para o pH. Concluiu-se que distância das amostragens foi adequada e a krigagem e o inverso da distância foram mais eficientes que o normal da distância. Verificou-se que quanto maior a potência de elevação, tanto para o método do inverso da distância quanto para normal da distância, os erros aumentam e também o grau de contagiosidade.Palavras-chave: geoestatística, sensor Veris, variabilidade espacial. METHODS OF DATA INTERPOLATIONS OBTAINED BY PRECISION AGRICULTURE SENSORS ABSTRACT:The sensors that measure soil characteristics in the field are important tools for the management of precision agriculture, among them the contact sensor, which measures the electrical conductivity (EC), organic matter (OM) and hydrogenation potential (pH) of the soil. The objective of this study was to evaluate the errors of the interpolation methods by kriging, inverse distance and normal distance from the data of EC, MO and soil pH. The experiment was carried out in the city of Candói - PR, where two areas were sampled, the data were collected by contact sensor, which was configured for a collection of 150 points per hectare for electrical conductivity and organic matter, and for potential the collection frequency was 15 points per hectare, the equipment was coupled in a tractor operated at a speed of 8 km h-1 with parallel passes of 20 m. Variographic analysis, cross-validation and mapping were performed. The smallest interpolation errors or jack knifing for CE, MO and pH were presented by the inverse distance interpolation method for the T2 field, and in T1 field the Kriging method obtained the lowest errors for pH. It was concluded that distance from the samplings was adequate and the kriging and the inverse of the distance were more efficient than the normal distance. It was verified that the higher the elevation power, both for the inverse distance and the normal distance method, the errors increase and also the degree of contagiousness.Keywords: geostatistics, Veris sensor, spatial variability.