Multidimensional agroclimatic-ai platform for predicting yellow onion storability

Overview

This solution leverages multi-fusion artificial intelligence to predict the maturity and storability of yellow onions by integrating field-level agroclimatic data. Growers currently rely on subjective decision-making, leading to harvests at variable maturities and unpredictable storage outcomes. By combining soil characteristics, weather patterns, remote sensing imagery, and crop phenology into a unified predictive model, this platform enables precision harvest scheduling and improved resource management. The result is reduced uncertainty around storage duration and marketability, helping growers deliver consistent quality to consumer outlets.

Technical specifications

Key features:

• Integration of soil type, composition, moisture, temperature, and salinity data
• Incorporation of weather variables including relative humidity, air temperature, precipitation, solar radiation, and wind speed
• Spatiotemporal aerial remote sensing collected at 15-day intervals post peak growth stage
• Crop phenology and maturity ratings collected in tandem with environmental data
• Machine learning models that fuse multidimensional datasets to predict maturity and storage duration
• Post-harvest monitoring of bulb quality and texture at 7-day intervals under 4°C storage conditions
• Independent validation of AI models for reliability and accuracy

Prior validation highlights:

• Aerial imagery indices correlated strongly with maturity for select cultivars (p < 0.001)
• K-Nearest Neighbor machine learning model achieved R² = 0.71 and Normalized-RMSE of 24% in predicting maturity for peanut crops
• Identified potential for improved accuracy by incorporating weather and soil-specific inputs

Technology readiness level

The technology is currently at an early-to-mid stage of development. Prior work on a related underground crop (peanut) demonstrated proof-of-concept for AI-based maturity prediction using aerial imagery and machine learning, achieving moderate accuracy. The next phase of validation will extend this approach to yellow onions by fusing expanded agroclimatic datasets, including soil climate and field weather data, with remote sensing inputs. Harvested onions will undergo controlled storage trials to validate storability predictions. The solution is not yet commercially deployed but is positioned for field validation and refinement through collaborative research partnerships.