Arva Intelligence Corp.

John A McEntire

Thomas A Dye

Elijah Benjamin Hoffman

Leland David Bernstein

Christopher William Hardin Fedor

A system and method for determining optimal sampling parameters is described. The method gathers soil data from a soil data source, which is associated with a soil organic carbon (SOC) project area. The crop prediction engine then determines that the soil data is less than optimal, but that the soil data is sufficient to generate an optimal sampling plan. The method completes a Monte Carlo simulation, which generates an empirical sampling distribution. The optimal sampling plan is determined by defining a margin of error, which provides a deviation from a predictive analysis of measured soil chemistry for a plurality of collected soil samples, and performing Monte Carlo simulations that include one Monte Carlo simulation having a lowest sampling density that satisfies the margin of error. The optimal sampling plan has the lowest sampling density and includes one or more sampling locations.

Optimal sampling with soil stratification

A system and method for interpolating soil chemistry variables to different plots of land is described. A first interpolation training model includes a machine learning model that receives soil composition information. A distance field training model generates spatial predictors that are applied to the machine learning model. The first interpolation training model prioritizes spatial smoothing over accuracy. A second interpolation training model is also applied that includes a distance weighting training model that more greatly weighs interpolated soil composition information closer to a point of interpolation than interpolated soil composition information that is further away to the point of interpolation. The second interpolation training model prioritizes accuracy over spatial smoothing. The illustrative crop prediction engine estimates soil chemistry values at different locations with the first interpolation training model and the second interpolation training model.

Estimating soil chemistry at different crop field locations

A system and method for visualizing one or more crop response surfaces. The system includes a geospatial database associated with a crop prediction engine. The geospatial database receives soil composition information for plots of land. The crop prediction engine identifies covariates from the soil composition information, which has a feature matrix. The crop prediction engine generates a multi-dimensional covariate training data set from the covariates. The crop prediction engine then applies the multi-dimensional covariate training data set to a machine learning training model to generate at least one predictive crop-yield predictive model. The crop prediction engine ranks covariates having feature set interactions. Subsequently, the crop prediction engine determines a dominant crop-yield feature set interaction from the ranked covariates having feature set interactions. The crop prediction engine generates a crop response surface from the dominant crop-yield feature set interaction. The crop prediction engine then visualizes the crop response surface.

Determining crop-yield drivers with multi-dimensional response surfaces

A system and method for identifying ground types from one or more interpolated covariates. The method proceeds by accessing soil composition information for plots of land, in which the soil composition information includes measured soil sample results, environmental results, soil conductivity results or any combination thereof. The method continues by identifying covariates from the soil composition information. Subsequently, the method interpolates covariates associated with different locations with an interpolation training model. Voxels are generated that are each associated with interpolated covariates having a corresponding geographical location. The method trains a random forest training model with the interpolated covariates. The voxels traverse the trained random forest model to identify clusters of voxels that are co-associated. The method identifies a ground type by combining the co-associated clusters. Each ground type is associated with a crop zone, a soil fertility, or a farm management recommendation.

Identifying ground types from interpolated covariates

A system and method for predicting a crop yield for a type of seed in a location is described. The method includes receiving, at a client device, seasonal crop data for the type of seed, soil data associated with the location, and mapping data associated with the location. The soil data includes soil variables, and the location is represented by voxels. The seasonal crop data, the soil data and the mapping data are uploaded to a geospatial database associated with a crop prediction engine. A random forest prediction model is applied to the seasonal crop data, the soil data and mapping data in the geospatial database by the crop prediction engine, which then ranks covariates to determine one or more significant covariates. The crop prediction engine then re-applies the significant covariates to the random forest prediction model to predict the crop yield for the type of seed at the location.

Growing community of inventors

Park City, UT, United States of America

John A McEntire