From Field to Snack: Building the BRITO Potato Yield Prediction System That Boosts Annual Yields by 10% and Cuts Costs by 30%

The recipient of the Fifth Hundred Young Farmers award, Cheng Chia-Hsiang, is an innovative farmer who masterfully bridges the gap between agriculture and brand practice. Managing his own potato fields and laboratory, he refrigerates and preserves his harvest every March and April, while also sourcing potatoes from other young farmers through contract farming. Operating on a single annual harvest, he ensures a year-round supply. Concurrently, he runs his self-owned brand, "BRITO," transforming potatoes into snack foods. His vision is that every bag of potatoes should represent not just delicious flavor, but also a deep love for the land and a shared lifestyle.

Under this operational model, "yield prediction" has become the linchpin for the brand's stable supply. If the yield falls short, it could lead to a failure to fulfill agreements with processing plants or retail channels, potentially resulting in breach of contract. Conversely, a surplus would lead to insufficient cold storage space and soaring overhead costs. To address this, Chia-Hsiang collaborated with AgriWeather to implement a customized potato yield prediction system. This system integrates field weather data, historical records, and real-time sensor information to serve as the baseline for annual inventory stockpiling and shipping cadences.

For more fascinating stories about how BRITO potatoes are grown, head straight to the 👉 BRITO Facebook page to learn more!



・・・



1. Unstable Climate and Quality Control Difficulties: With only one harvest per year, climate conditions have a massive impact. In particular, high temperatures, strong winds, and sudden shifts in rainfall can expose potatoes above the ground, damaging their quality and even hurting yields.
2. Insufficient Seed Potato and Management Information: Potato farming relies heavily on healthy seed potatoes and precision management; however, Taiwan's local seed potato supply chain remains incomplete.
3. High Post-Harvest Storage Pressure: Because harvests are concentrated within a short window, a spike in yield driven by inaccurate predictions will create intense pressure on cold storage and logistics.
4. Mismatched Supply and Demand Leading to Breach of Contract: If the volume delivered by contract farmers fails to meet the needs of downstream clients (such as processing plants and retail channels), it will result in a loss of brand trust or even financial damages.



・・・



AgriWeather assisted Chia-Hsiang in implementing a rolling yield prediction system. The key tools include:

1. Weather Stations and Soil Sensor 3-in-1s
   • These tools monitor real-time changes in field temperature, humidity, soil moisture, rainfall, and wind speed. For example, when the soil moisture drops below 20%, an alert is automatically sent to trigger irrigation adjustments.
2. Historical Data Archiving and Comparison
   • By compiling local climate and yield data from recent years, a baseline predictive model was constructed. This data includes actual harvest timelines and corresponding weather conditions, allowing the model to more closely align with localized field realities and risk variations.
3. Predictive Modeling for Yield Forecasting
   • Estimates are calculated based on both real-time and historical data, with yield predictions updated every two weeks. As the harvest period approaches, the prediction accuracy becomes increasingly precise.



・・・



1. Improved Cash Flow: Saving NT$7 Million per 100 Hectares

Armed with yield predictions, if a surplus is anticipated, we can adjust our sales strategy ahead of time. This allows us to ship a specific grade of potatoes—approximately 10,000 kg per hectare—directly to processing plants.

Although these potatoes are still priced at NT$13/kg, skipping stages like warehousing, grading, and packaging saves roughly NT$7/kg in cash expenditures.

Calculated across 100 hectares, this saves a total of NT$7,000,000 in cash outflows, which is immensely helpful for the brand's overall capital deployment.

2. Increased Yield and Revenue: +NT$6 Million

By fine-tuning management and harvest strategies based on predictive data, the yield is estimated to increase by 10%. This equates to an extra 300 tons of potatoes (+3 tons/hectare × 100 hectares).

Priced at NT$20/kg, this translates to a revenue boost of up to NT$6,000,000.

3. Precision Labor Deployment, Reducing Personnel Waste

The weather forecasting tool integrates rainfall probability and precipitation levels, which helps determine the optimal time to harvest. This avoids the scenario of "wasted labor deployment" (workers showing up with no harvest possible), saving approximately 300 worker-days in personnel expenses.

4. Smart Greenhouse Management: 30% Cost Savings, 20% Increase in Value

Within the 0.1-hectare (1-fen) seedling greenhouse, sensors are used to regulate moisture, electrical conductivity (EC), temperature, and humidity, while automatically controlling drip irrigation and cooling systems:

• Saves 30% in costs during the cultivation period (labor, water, fertilizer, and electricity).
• Boosts output value by 20% while effectively curbing energy waste from "ineffective cooling."



・・・



Q1: Can this system be applied to other crops?
A1: Yes. Models can be customized based on the specific crop. Currently, AgriWeather also assists with forecasting for fruit trees, rice, and pest/disease outbreaks, adjusting the models according to each crop's unique characteristics.

Q2: Will the predictions be inaccurate? Weather changes so drastically nowadays.
A2: The system updates its predictions every two weeks. The data sources include localized, real-time sensors and long-term historical weather records, allowing the accuracy to continually improve as the harvest date approaches.

Q3: Will this kind of system be very expensive? Can farmers afford it?
A3: Pricing is currently flexible and based on field scale and crop type. Smallholder farmers can choose to implement the system in phases and may also apply for government subsidies.

Q4: Do I need to provide past yield and weather data myself?
A4: Past yield data is definitely required in order to construct a tailored model and make precise estimates. However, for weather data, the system can directly utilize official open data provided by the Central Weather Administration (CWA), so farmers do not need to provide their own—AgriWeather will handle the data integration.



・・・



If you are a farmer, brand owner, processing plant, or regional revitalization team curious about this smart agriculture system, please feel free to contact us!

We are always excited to share our experience and co-create collaborative models, ensuring that every piece of land can utilize the most tailored climate data and predictive tools.