Systems and methods for quantifying agroecosystem variables through multi-tier scaling from ground data, to mobile platforms, and to satellite observations

Patent metadata

Field	Value
Publication number	US-12374104-B2
Application number	US-202118005022-A
Country	US
Kind code	B2
Filing date	Jul 9, 2021
Priority date	Jul 10, 2020
Publication date	Jul 29, 2025
Grant date	Jul 29, 2025

Abstract

The ability to scale data can provide numerous advantages, especially with regard to agricultural information. For example, agroecosystems include land and data associated with the land, such as physical traits and information. This can include, for example, information related to the soil, crops, other vegetation, and other information related to the land. In order to be able to quickly and accurately know such information and traits, ground truth data can be scaled using aerial and/or satellite imagery. Models and other machine learning can utilize ground truth data to scale limited field area data (e.g., 0.1-1 km) and accurately apply the same to large swaths of land (e.g., >100 km 2 ) with accuracy for the field traits and/or characteristics.

First claim

The invention claimed is:

1. A non-transitory computer-readable storage medium comprising executable instructions that, when executed by a processing system including a processor, perform operations, the operations comprising:
obtaining a first-tier dataset including ground truth values of agroecosystem variables, wherein the first-tier dataset has a first-tier geographic footprint, and wherein the ground truth values have been collected via field measurements;
obtaining a second-tier dataset including second-tier values of the agroecosystem variables, wherein the second-tier dataset has been collected from one or more mobile systems, wherein the second-tier dataset has a second-tier geographic footprint, wherein the second-tier geographic footprint is larger than the first-tier geographic footprint, and wherein the second-tier geographic footprint at least partially overlaps with the first-tier geographic footprint in a first tier overlapped area;
obtaining a third-tier dataset including third-tier values of the agroecosystem variables, wherein the third-tier dataset has been collected from one or more satellite systems, wherein the third-tier dataset has a third-tier geographic footprint, wherein the third-tier geographic footprint is larger than the second-tier geographic footprint, and wherein the third-tier geographic footprint at least partially overlaps with the second-tier geographic footprint in a second tier overlapped area;
generating estimates of values of the agroecosystem variables by applying at least a first-tier model across at least a part of the second-tier geographic footprint, wherein the at least the first-tier model uses the first-tier dataset for first-tier labels, and wherein the at least the first-tier model uses the second-tier dataset for first-tier inputs; and
generating estimates of secondary values of the agroecosystem variables by applying at least a second-tier model across at least a part of the third-tier geographic footprint, wherein the at least the second-tier model uses the second-tier dataset for second-tier labels, and wherein the at least the second-tier model uses the third-tier dataset for second-tier inputs.

2. The non-transitory computer-readable storage medium of claim 1 , wherein the operations further comprise sending an output comprising the estimates of the secondary values of the agroecosystem variables, to a display screen, to a printer, to a computer file, or to any combination thereof.

3. The non-transitory computer-readable storage medium of claim 2 , wherein the operations further comprise mapping the output of the third-tier values of the agroecosystem variables on a geographic map.

4. The non-transitory computer-readable storage medium of claim 1 , wherein the applying the at least the first-tier model across at least part of the second-tier geographic footprint comprises:
applying the at least the first-tier model across all of the second-tier geographic footprint; and
the applying the at least the second-tier model across at least part of the third-tier geographic footprint comprises applying the at least the second-tier model across all of the third-tier geographic footprint.

5. The non-transitory computer-readable storage medium of claim 4 , wherein the field measurements have been collected from the ground; and
the ground truth values obtained through direct measurements or observations without inference information include one or more agroecosystem variables based upon: a leaf sample analysis, a soil sample analysis, a field-level condition analysis, a plant gas-exchange experiment, a tractor or robot-carried sensor derived soil/plant/management information, a leaf spectra derived leaf trait, LiDAR-derived crop/soil/management information, radar-derived crop/soil/management information, gamma-ray sensor derived crop/soil/management information, ground-photo derived plant/soil/management information, or any combination thereof.

6. The non-transitory computer-readable storage medium of claim 1 , wherein the agroecosystem variables comprise: one or more crop traits; one or more soil traits; one or more agricultural management practices; or any combination thereof.

7. The non-transitory computer-readable storage medium of claim 6 , wherein the one or more crop traits comprise:
one or more biochemical crop traits;
one or more biophysical crop traits; or
any combination thereof.

8. The non-transitory computer-readable storage medium of claim 7 , wherein the one or more soil traits comprise:
one or more biochemical soil traits;
one or more biophysical soil traits; or
any combination thereof.

9. The non-transitory computer-readable storage medium of claim 8 , wherein the one or more crop traits comprise:
one or more pigments;
one or more biophysical properties;
one or more biochemical properties;
one or more crop functional properties;
one or more crop stress conditions;
or any combination thereof.

10. The non-transitory computer-readable storage medium of claim 9 , wherein the one or more soil traits comprise:
one or more physical properties;
one or more chemical properties;
one or more soil functional properties;
one or more amounts of soil organic carbon;
or any combination thereof.

11. The non-transitory computer-readable storage medium of claim 1 , wherein the operations further comprise training a classifier of the first tier model using the first tier dataset that is overlapped with the second-tier geographic footprint that at least partially overlaps with the first-tier geographic footprint in a first tier overlapped area.

12. The non-transitory computer-readable storage medium of claim 1 , wherein the operations further comprise training a classifier of the second tier model using the second tier dataset that is overlapped with the third-tier geographic footprint that at least partially overlaps with the second-tier geographic footprint in a second tier overlapped area.

13. The non-transitory computer-readable storage medium of claim 1 , wherein the operations further comprise outputting a mapped area with agroecosystem variables on a display.

14. The non-transitory computer-readable storage medium of claim 13 , wherein the mapped area comprises the third tier geographic footprint overlayed with the agroecosystem variables.

15. A device comprising:
a processing system including a processor; and
a memory that stores executable instructions that, when executed by the processing system, perform operations, the operations comprising:
obtaining at least one ground truth dataset of first-tier values of agroecosystem variables, wherein the at least one ground truth dataset has a first-tier geographic footprint;
obtaining at least one mobile dataset collected from at least one mobile system, wherein the at least one mobile dataset has a second-tier geographic footprint, and wherein the second-tier geographic footprint at least partially overlaps with the first-tier geographic footprint in a first tier overlapping area;
obtaining at least one satellite remote sensing dataset collected from at least one satellite system, wherein the at least one satellite remote sensing dataset has a third-tier geographic footprint, and wherein the third-tier geographic footprint at least partially overlaps with the second-tier geographic footprint in a second tier overlapping area;
configuring, for the first tier overlapping area, one or more first-tier models which use the at least one ground truth dataset as one or more first-tier labels and which use the at least one mobile dataset as one or more first-tier inputs;
generating predictions

Assignees

• Univ Illinois

Inventors

• Wang Sheng
• Guan Kaiyu
• Peng Bin
• Jiang Chongya
• Wang Sibo

Classifications

• G06V20/13

  Satellite images · CPC title

• G06F18/214

  Generating training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title

• G06F18/21

  Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation · CPC title

• A01B79/005

  Precision agriculture · CPC title

• G06V20/188 Primary

  Vegetation · CPC title