|
1000
|
Abstract/Summary
|
-
Delineation of management zones (MZ) based on soil mineral nitrogen (SMN) dynamics can enhance site-specific management, reduce nitrate leaching, and improve nutrient efficiency. We tested proximal sensing as an alternative to standard laboratory methods to capture the spatial variability of SMN, nitrate (NO3−), and soil moisture (SM) and combined these data with topographic and remote sensing inputs to delineate MZ using data fusion and k-means clustering. Two conventionally managed fields with winter oilseed rape (Brassica napus L.) and winter barley (Hordeum vulgare L.) were chosen for Field-A and Field-B. Fresh soil samples were analyzed in the laboratory using KCl extraction, while global positioning system-labeled data from a proximal soil sensor (FarmLab) were accessed via cloud storage. FarmLab estimated NO3− and SMN were higher than laboratory results (p < 0.05), whereas SM showed no significant difference between the two methods. Bland–Altman analysis, which assesses the limit of agreement between methods to ensure consistency, revealed significant discrepancies in NO₃⁻ estimated by both methods, particularly in Field-B, with limits of agreement ranging from −17.40 to 29.66 mg kg−1. Results of k-means clustering, a method for grouping data into similar categories, were evaluated using 11 feature sets, which combine data from multiple sources (laboratory and FarmLab data, satellites, and topographic data) to create a comprehensive dataset for analysis at different time points in autumn and spring. The results showed that the optimal clustering result varied depending on the field and date. Feature sets with topographic variables performed well in Field-A, while feature sets with remote sensing, topography, and FarmLab data improved MZ in Field-B. This study demonstrates how the FarmLab device can capture within-field SMN variability and examines the similarities and differences between both methods (laboratory and FarmLab). Despite discrepancies between methods, FarmLab showed the potential of integrating in-season NO3− and SMN data with topographic and remote sensing data to delineate MZ. This approach can be scaled up to farm and landscape scale, allowing farmers to leverage proximal and remote sensing data for in-season SMN monitoring, which enables efficient nutrient management and promotes sustainable farming practices with economic and environmental benefits.
|
