Multicollinearity Analysis of Phenology Metrics From Modis Satellite Imagery for Maize and Soybean in Croatia

This study analyzed the phenology metrics of maize and soybean in Croatia across the years 2021 and 2022 based on Moderate Resolution Imaging Spectroradiometer (MODIS) satellite data, focusing on median dates of occurrence and their variability. The total number of filtered parcels from the Paying Agency for Agriculture, Fisheries and Rural Development database with area larger than 25 ha was 108 and 463 for maize, as well as 25 and 163 parcels for soybean in the years 2021 and 2022, respectively. Median maturity and peak growth stages occurred later in 2021 than in 2022 for both crops, suggesting potentially broader climatic impacts on growth duration. The correlation matrices for the study indicated strong relationships between maturity, senescence, and peak phases but not dormancy which had low levels of correlation with other measures. The presence of high and significant multicollinearity was underlined in the study by the analysis of variance inflation factors used when analyzing phenological parameters, especially with limited samplesizes. These findings suggest that dormancy and peak phenology can be useful for a characterization of crop yield and cropland conditions.