At N.O.A.H., our yield prediction system is designed to help farmers and insurers make informed decisions under normal conditions. Here's a deeper look into how our process works.
1. Data Collection from Diverse Sources
We collect data from a variety of sources, including:
- Field data: Soil moisture, agricultural indices, and other on-ground and soil metrics.
- Weather data: Current and forecasted weather conditions.
- Satellite data: We gather high-resolution satellite images from multiple sources, including Synthetic Aperture Radar (SAR) data, which is particularly useful in providing all-weather, day-and-night observations. Optical and SAR data allows us to detect minute changes in the field, including soil moisture levels and crop growth patterns, even under cloud cover. This adds a critical layer of precision that optical imagery alone cannot achieve.
All this information is fed into our machine learning (ML) algorithm.
2. Determining Maximum Potential Yield (MAX)
Once we have the relevant data, our ML model calculates the maximum potential yield (MAX) for the current crop in the specific field. The most influential factors in determining the yield include:
- Solar radiation
- Soil moisture
- Rainfall patterns
The result is an estimate of the optimal yield under perfect weather conditions. We consider this value to be 100% of the possible yield.
3. Historical Analysis
Next, we use the same modeling process to compute the MAX yield for previous years. We compare this with the actual historical harvest data to calculate a "farmer's efficiency index"—a measure of how close the actual yield was to the potential MAX yield. For instance, if a farm typically achieves 80% of the MAX, we use this figure to predict the current year’s yield.
4. Yield Prediction for the Current Year
By understanding past performance, we can make a reliable prediction for the current year. If the farm achieved 80% of MAX in previous years, we assume a similar outcome for this year unless drastic changes occur.
5. Fine-Tuning Crop Insurance Policies
Our goal is to use this yield prediction to refine crop insurance policies, particularly within the Multi-Peril Crop Insurance (MPCI) program. The system allows us to recommend coverage levels, such as advising that a farmer opts for 70% MPCI coverage while adding a natural catastrophe policy (natcat) for certain events. While we don't account for events like pest infestations or wildfires—which represent 15-20% of risk—our focus on weather-related factors addresses 80% of yield-impacting events.
6. Why Weather Matters Most
Weather remains the single most critical factor affecting crop yield. With our system’s ability to provide accurate, real-time insights using satellite data and machine learning, we can significantly mitigate the unpredictability of weather events and help fine-tune insurance decisions for both farmers and insurers.
The Power of SAR and Satellite Data
Our use of SAR data, in particular, sets us apart from traditional prediction methods that rely solely on optical imagery. The ability to monitor fields under any weather conditions, combined with historical data and advanced modeling, enables us to provide highly accurate yield predictions that empower more strategic crop insurance decisions.
By harnessing technology to better predict outcomes, N.O.A.H. helps optimize crop insurance, giving farmers and insurers the insights they need to manage risk effectively.
---
With the continuous advancement of satellite technology and data analysis, we believe our system will further improve its accuracy and expand its capabilities to offer even more value to the agricultural industry.
.png)
.png)
.png)
