Source Advances Harvest Forecasting for Tomatoes with Next Generation AI Model
Fundamental changes to how the model learns and handles grower input mean significantly less manual work and meaningfully better accuracy.
To get an accurate picture of plant development, growers need to look beyond joules, especially when using artificial light. By focusing on Photosynthetically Active Radiation (PAR), the specific light wavelengths used for photosynthesis, we can provide a much more reliable metric for daily growth and harvest modeling.
Today, we are pleased to announce the release of Approximated PAR Sum. This feature provides a high-fidelity estimate of the daily light reaching your crop, ensuring you have the data you need for precise management, even without a physical PAR sensor.
Standard weather stations measure total solar energy, much of which isn't used by the plant for growth. Because photosynthesis is driven by the 400–700 nm spectrum, tracking PAR allows our models to reflect the energy your crops actually consume. This transition from "joules" to "PAR Sum" leads to more realistic plant modeling and more predictable harvest cycles. Artificial lights emit light in a completely different spectrum from the sun. Converting that into joules is technically inconsistent, while daylight PAR and PAR from artificial lights can simply be added and reflect what the plant can use for photosynthesis.
The Approximated PAR Sum isn't a simple estimate; it’s a calculated reconstruction of your greenhouse environment. Every five minutes, our system synthesizes multiple data points to determine the light levels at the canopy:
Solar input: Real-time radiation data from your weather station.
Structural transmission: Adjustments for light lost through the greenhouse glass or structure.
Screen dynamics: Accounting for the position and type of screens.
Artificial lighting: Integrating the specific output (μmol/m²/s) of your HPS or LED systems.
For facilities that already utilize PAR sensors, this feature serves as a valuable cross-reference. Physical sensors are susceptible to dust, shading from structural elements, or calibration drift over time. By comparing your sensor readings against the Approximated PAR Sum, the system can flag discrepancies early, allowing you to maintain data integrity across your entire operation.
The launch of Approximated PAR Sum represents a move away from hardware dependency toward intelligent, software-driven plant science. By providing every grower with access to high-fidelity PAR metrics, Source.ag continues to lower the barrier to precision agriculture, ensuring that data integrity and biological accuracy are available to all, regardless of their greenhouse configuration.
Fundamental changes to how the model learns and handles grower input mean significantly less manual work and meaningfully better accuracy.
AMSTERDAM – Monday, 16 February – Source.ag, the Dutch-leading pioneer in AI-driven solutions for high-tech greenhouses, today announced the appointment of Alberto Carbajal Rodriguez as AI Solutions Specialist in Mexico. This strategic hire underscores the company’s commitment to empowering Mexican growers with the tools and expertise needed to optimize and scale sustainable food production.