Delineation of management zones using mobile measurements of soil apparent elec...

• Author(s): Moral, F. J. ; Terron, J. M. ; Marques da Silva, José Rafael
• Date: 2010
• Persistent ID: http://hdl.handle.net/10174/3145
• Origin: Repositório Científico da Universidade de Évora
• Subject(s): soil apparent electrical conductivity; Site-specific crop management; Principal component analysis; Regression kriging; Fuzzy c-means
• Document Type: Article
• Language: English

Description

Site-specific management promotes the identification and management of areas within the field, which represent subfield regions with homogeneous characteristics (management zones). However, determination of subfield areas is difficult because of the complex combination of factors which could affect crop yield. One possibility to capture yield variability is the use of soil physical properties to define the management zones as they are related to plant available water. With the aim of characterizing the spatial variability of the main soil physical variables and using this information to determine potential management zones, soil samples were taken from 70 locations in a 33-ha field in Badajoz, southwestern Spain. Firstly, accurate spatial distribution maps of the soil attributes were generated by using regression kriging as the most suitable algorithm in which exhaustive secondary information on soil apparent electrical conductivity (ECa) was incorporated. ECa measurements were carried out with a Veris 3100 operating in both shallow (0–30 cm), ECs, and deep (0–90 cm), ECd,mode. Clay, coarse sand and fine sand were the soil physical properties which exhibited higher correlation with ECa (positively correlated with the finer texture component, clay, and negatively correlated with the coarser ones, coarse and fine sands), particularly with ECs. Thus, this was the secondary variable used to obtain the kriged maps. Later, principal component analysis and fuzzy cluster classification were performed to delineate management zones, resulting in two subfields to be managed separately. This number of subfields was determined using the fuzzy performance index and normalized classification entropy as the way to optimize the classification algorithm.