Advances in technology make soil sampling quick and more cost-efficient.

Technological advancements in agriculture continue to shape the industry, bringing new and often better options and ways of doing things to farmers. While lagging some compared to other ag technologies, soil testing is quickly catching up.

Sensor technology, combined with new artificial intelligence (A.I.), builds a digital twin of the soil in the cloud. This form of machine learning technology makes soil samples available more quickly and cost-efficiently. It also provides whole-field analysis, instead of traditional grid sampling for nutrients, compaction, and carbon.

A skeptic, though, might ask how exactly does machine learning work. To break it down:

• Sampling Plan Development: A sampling plan is developed to account for varied topography, soil type, land use and other factors, and to ensure that a sufficient number of samples are collected.
• Data Collection: Soil samples are collected from prescribed locations and the fields are scanned using proximal sensor technology. EarthOptics ground truths their scanned data with soil samples, which calibrate sensor measurements.
• Data Preprocessing: The collected data is preprocessed to ensure its quality and prepare it for analysis. This involves cleaning the data, removing any outliers or errors, handling missing values, and normalizing or standardizing the data if necessary.
• Training Data Preparation: The dataset is split into training and testing sets. The training set is used to teach the machine learning model to recognize patterns and relationships between the features and soil properties. It is important to have a diverse and representative training set to ensure the model's generalization capabilities.
• Model Selection: Various machine learning algorithms that can be used for soil property prediction. The choice of algorithm depends on the specific problem, available data, and desired accuracy.
• Model Training: The selected machine learning model is trained using the labeled training data. The model learns from the patterns and relationships in the data to create a function that can map the input features to the desired output (e.g., predicting soil pH based on geographic coordinates, climate data, and land use).
• Model Evaluation: The trained model is evaluated using the testing set – the dense samples. This allows EarthOptics to assess the model's performance on unseen data and measure its accuracy or other relevant metrics.
• Model Deployment: Once the model is satisfactory, it can be used to predict soil properties at new or unexplored locations.

EarthOptics’ sensor technology has measured more than a million acres with a high level of accuracy. On an 1,100-acre crop field, the C-Mapper™ tool had a mean error rate of 0.062 and a bulk-density map with an error rate of 0.006. Soil sampling technology is the wave of the future: higher accuracy, less labor-intensive and more cost-efficient.