Low-cost visible-light imaging for canola flowering phenology and site-specific crop management

Overview

Canola flowering timing and intensity are key indicators for site-specific crop management decisions, particularly the targeted application of fungicides and other crop protection inputs. This solution demonstrates that low-cost visible-light cameras, when paired with intelligent image processing algorithms, can estimate canola flowering dynamics as effectively as expensive multispectral sensors. Because flowering changes are visually observable in the green and blue wavelengths, a well-designed index using only visible-light imagery captures canopy size and bloom progression with comparable accuracy. The result is a more accessible, scalable phenology monitoring tool for growers and agronomists.

Technical specifications

• Sensor approach: Low-cost RGB cameras (including consumer-grade options such as a Canon DSLR) are compared against multispectral cameras and hyperspectral radiometers.
• Imaging platforms: Ground-based and drone-based aerial imagery acquisition for field-scale coverage.
• Analytical method: Regression of image-derived indices onto flower counts and hand-harvested yield, with a proposed normalized difference index of green and blue light to estimate percent bloom, return bloom, and bloom duration.
• Experimental design: Trials across multiple water and nitrogen environments to create a wide range in leaf area and flowering intensity, with subplot flower counts and yield measurements for ground truth.
• Statistical analysis: ANOVA to assess interactions between drainage and soil nitrogen on flowering intensity and duration.
• Key finding: Correlations between sensor data and yield or flower counts were not significantly different across low-cost visible-light, multispectral, and hyperspectral sensors.

Technology readiness level

The approach has been validated in a one-year (2014) canola field experiment near Adams, Oregon, across three water environments with nitrogen randomization. The next stage involves on-farm trials in three fields with variation in drainage and nitrogen randomization within each drainage regime, with one-meter blocks subsampled for flower counts. The technology is at an early-to-mid readiness stage, with demonstrated proof-of-concept performance and a clear path to broader field validation through drone-based imagery collection.