The goal of this project is to determine what electromagnetic radiation wavelengths correlate with the changes in the nitrogen status of three cranberry cultivars. For this project, isolated propagation containers or “mini-bogs” were evenly split between three cultivars: Early Black, Stevens, and Mullica Queen. Each “mini-bog” was planted with 98 cranberry plugs. Each cultivar was subsequently split into four even groups with each group assigned one of four fertilizer regimes: 10% optimal, 50% optimal, 100% optimal, and 150% optimal. Over that growing season, physiochemical and spectrographic data was collated from 48 sets of cranberry plants grown separately in “mini-bogs”. Canopy and contact level spectrographic data was collected using the ASD FieldSpec 4 field spectrometer and leaf clip attachment. Clippings from each “mini-bog” were collected after spectrographic data collection and sent for wet digestion total nitrogen laboratory analysis. Using the Automated Radiative Transfer Models Operator (ARTMO) package within MATLABs and ARTMO’s Machine Learning Regression Algorithms (MLRAs) toolbox and Spectral Indices (SI) toolboxes, we were able to examine 298 datasets collected during the 2024 growing season. MLRA results show strong correlation between the changes in the nitrogen concentration and the spectrographic readings. The MLRA produced correlation results for 30 machine learning regression algorithms, including gaussian processes, kernel ridge process, random forest processes, linear regressions, and neural network processes. Early Black had a correlation up to 98.41%, Stevens up to 91.43% correlation, and Mullica Queens had up to 99.98% correlation. Applying the strongest correlation of the MLA functions to the band analysis tool within MLRA, we identified the top 20 bands out of 2151 bands that strongly correlate with the changes in the nitrogen concentration. This study showed that over 20 electromagnetic radiation wavelengths correlated strongly with the changes in the nitrogen status of our cultivars. Combining these wavelengths with reference wavelength in a spectral index is the next step to finding a combination that can accurately and precisely derive the nitrogen status of cranberry vegetation.
