A hand watering two golden sinks filled with produce.
CASE STUDY
How we helped 16 CA farms prevent invasive species.
Read more

Farms we've helped

Underwood Family Farms • Urban Edge Farm • California Endive Farms • Braga Fresh Family Farm • Frog Hollow Farm • Dutch Hollow Farms • Apricot Lane Farms • Simonian Farms • Victoria Island Farms • Windsor Family Farm • Tanaka Farms • Mountain Farm • Silo Farm Organics • Soil Born Farms • Cloverfield Organic Farm • Durst Organic Growers
Underwood Family Farms • Urban Edge Farm • California Endive Farms • Braga Fresh Family Farm • Frog Hollow Farm • Dutch Hollow Farms • Apricot Lane Farms • Simonian Farms • Victoria Island Farms • Windsor Family Farm • Tanaka Farms • Mountain Farm • Silo Farm Organics • Soil Born Farms • Cloverfield Organic Farm • Durst Organic Growers
SE + Central CA
LOCATION
Specialty crops
TYPE
Tumbleweed, fruit flies
FOCUS
Overview
Problem 1: Colorado Desert

This story is one of two sides: On one hand, over 10,900 farms in California's six southeastern counties grapple with the rapid spread of tumbleweeds (Salsola tragus), which clog irrigation canals, damage fences, and obstruct roads, leading to increased labor costs and reduced efficiency in farm operations. Tumbleweed clearance can consume up to 15% of a farm's annual maintenance budget, while their interference with irrigation systems can reduce water efficiency by up to 20%, exacerbating the region's existing water scarcity issues.

Problem 2: Central Valley

In the Central Valley, fruit growers face the Mediterranean fruit fly, which infests crops like oranges and peaches, causing up to 30% yield losses and risking quarantine restrictions that halt market access. These pests increase pesticide use and associated costs by an average of 25%. Agsight's innovative solutions could mitigate these challenges using near real time monitoring and predictive machine learning (ML) algorithms to optimize pest control measures, thereby reducing labor costs and enhancing crop protection.

A black farmer balancing a peach on her bucket hat.
01
Decomposing the problem.

Tumbleweed


In California's Colorado Desert farms, farmers reported that clearing tumbleweed occupies up to 15% of their annual maintenance budget, with some spending 300+ hours annually on manual removal. Field studies conducted across 8 farms revealed that 75% of those farms experienced irrigation system blockages, leading to an average 17.4% reduction in water efficiency and 7.9% decrease in yield.

Fruit flies

In California's Central Valley, fruit flies can lead to significant crop losses, with up to 30% of fruit yields being compromised in heavily infested areas. Field studies conducted across 6 fruit farms indicated that 50% of farmers reported increased costs for pest management, amounting to an average of $200 per acre annually and up to 18 hours per week on manual pest inspections during peak seasons. These studies used pheromone trap counts, crop damage assessments, and farmer surveys to gather data.

02
Translating findings into insights.
A miniature cartoon figure standing below a tomato stem whose fruit is missing.
A miniature cartoon delivery driver fulfilling a mobile app order on a scooter carrying half an avocado.
A miniature cartoon figure tapping "confirm" on a giant propped phone screen surrounded by leafy green vegetables.
Detection and prevention
Field studies showed that early detection  is critical to minimizing impact. Current methods like visual inspections and pheromone traps are labor-intensive and unpredictable.
Localized intervention
For tumbleweeds, this might involve targeted herbicide application based on wind patterns and growth stages. For fruit flies, farmers could deliver doses of biopesticides only to affected areas.
Data-supported decisions
Farmers should be able to optimize timing and dosage of interventions with actionable and personalized recommendations using pest population dynamics and crop phenology data.
How do you currently manage irrigation?
How do you deal with high soil salinity levels?
How do you receive weather forecasts and integrate them into irrigation planning?
What features do you find useful in your current irrigation tools?
How do you interpret data from sensors?
How easy is it for you to understand and act on salinity data?
How do you currently manage irrigation?
How do you deal with high soil salinity levels?
How do you receive weather forecasts and integrate them into irrigation planning?
What features do you find useful in your current irrigation tools?
How do you interpret data from sensors?
How easy is it for you to understand and act on salinity data?
What kind of personalized recommendations would you find most helpful?
How do you currently monitor salinity levels?
What information do you need to better manage soil salinity?
What methods have you tried to mitigate salinity in your soil?
How do you determine when to apply soil amendments to combat salinity?
How do you test your soil?
What kind of personalized recommendations would you find most helpful?
How do you currently monitor salinity levels?
What information do you need to better manage soil salinity?
What methods have you tried to mitigate salinity in your soil?
How do you determine when to apply soil amendments to combat salinity?
How do you test your soil?
03
Crafting something truly novel.
A black farmer smiling while balancing a peach on her bucket hat and holding two other peaches in each of her hands/
Addressing tumbleweed takeovers

Using Agsight, we helped these farms resolve these issues via satellite imagery analysis, meteorological data, and drone surveys, followed by image preprocessing (noise reduction, normalization, and segmentation) and classification by a convolutional neural network (CNN) to distinguish tumbleweeds from native vegetation based on visual features like leaf morphology, color, and texture to generate spatial and temporal maps that delineate its extent and severity.

Using a machine learning algorithm to analyze wind patterns and predict the likely spread of tumbleweeds by correlating this data with topographic information, the app simulates potential growth and spread scenarios of tumbleweed infestations, identifying high-risk areas before infestations occur. Using this pipeline, Agsight recommends specific grass species or cover crops that can outcompete tumbleweeds for resources and soil.

Addressing fly flurries

Using Agsight, we helped Central Valley fruit farms tackle the Mediterranean fruit fly issue through an approach combining satellite imagery analysis, real-time meteorological data, and pheromone trap data collected via drone surveys. We used this data to improve the performance of our existing pest, disease, and stress diagnosis feature.

Here, we integrated a CNN to classify and differentiate fruit fly infestations to classify and differentiate fruit fly infestations from other types of damage based on distinct visual markers like fruit discoloration, size, and spotting patterns. By correlating this data with historical infestation records and crop susceptibility profiles, Agsight provided farmers with real-time alerts and intervention strategies.

Terminating tumbleweed.

Field trials were conducted on 8 farms, where the model's predictions were tested against real-world scenarios. High-risk areas identified by the model were monitored, and the actual spread of tumbleweeds was recorded. The model's predictions were found to be accurate within a margin of 89%, confirming its reliability.

A hand picking an apple off its branch.
More than just swatting flies.

Field trials across Central Valley fruit farms confirmed the accuracy and reliability of the monitoring system, with over a 91% detection rate for early signs of fruit fly infestations compared to traditional methods and an 88% accuracy for predictive analytics, which forecasted potential outbreaks. Farmers reported a 50% decrease in crop losses as a result.

04
Impact.
A peach tree in an orchard.
A peach dangling from a branch.
Tumbleweed

For tumbleweed, we tracked and validated changes in infestation severity and spread across affected farms using satellite imagery and drone surveys. Our system accurately mapped tumbleweed clusters and predicted their growth patterns, enabling proactive management strategies that reduced labor costs associated with manual removal and minimized infrastructure damage.

Fruit flies

For fruit flies, we evaluated impact by monitoring pest populations, crop damage rates, and pesticide usage across participating fruit farms in the Central Valley. The integrated monitoring system provided early detection of fruit fly activity, allowing farmers to implement timely interventions and reduce pest population growth by applying targeted pesticides safely.

A corn farmer holding a corn on the cob in his corn field.
Two corns on the cob.
69
%
decrease in labor hours dedicated to tumbleweed
A farmer tossing a corn on the cob in the air.
58
%
decrease in crop losses attributed to fruit flies
63
%
fewer incidents of infrastructural damage
45
%
reduction in indiscriminate pesticide use
Two miniature cartoon figures with party poppers shooting produce.

RELEASE

Broader release scheduled for 2025

This feature is currently in beta testing and undergoing a larger scope of validation to ensure it's perfect when released!

NEED A HAND?

Need help transforming your farm?
We'd love to talk!
Let's talk!