FORECASTING NATIONAL CROP YIELD DURING THE GROWING SEASON

• US 20160290918A1
• Filed: 04/01/2015
• Published: 10/06/2016
• Est. Priority Date: 09/12/2014
• Status: Active Grant

• Current Assignee: Climate Pty., Ltd.
• Original Assignee: The Climate Corporation (Bayer AG)
• Inventors: Xu, Ying, Xu, Lijuan
• Granted Patent: US 10,564,316 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: A01B 76/00 Parts, details or accessories of agricultural machines or implements, not provided for in groups A01B51/00 - A01B75/00
• CPC Class Codes: A01B 79/005 Precision agriculture
• CPC Class Codes: E02D 1/04 Sampling of soil E02D1/025 takes precedence
• CPC Class Codes: G01N 2021/1793 Remote sensing
• CPC Class Codes: G01N 2021/3155 Measuring in two spectral ranges, e.g. UV and visible
• CPC Class Codes: G01N 2033/245 for agricultural purposes
• CPC Class Codes: G01N 21/17 Systems in which incident light is modified in accordance with the properties of the material investigated where the material investigated is optically excited causing a change in wavelength of the incident light G01N21/63
• CPC Class Codes: G01N 21/31 Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry G01N21/72 takes precedence
• CPC Class Codes: G01N 21/359 using near infrared light
• CPC Class Codes: G01N 2201/0616 Ambient light is used
• CPC Class Codes: G01N 2201/129 Using chemometrical methods
• CPC Class Codes: G01N 33/0098 Plants or trees wood G01N33/46
• CPC Class Codes: G01N 33/24 Earth materials G01N33/42 takes precedence
• CPC Class Codes: G01V 20/00 Geomodelling in general
• CPC Class Codes: G01W 1/10 Devices for predicting weather conditions computers per se G06; display devices G09
• CPC Class Codes: G06F 2111/06 Multi-objective optimisation, e.g. Pareto optimisation using simulated annealing [SA], ant colony algorithms or genetic algorithms [GA]
• CPC Class Codes: G06F 30/27 using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
• CPC Class Codes: G06Q 10/04 Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem" market predictions or forecasting for commercial activities G06Q30/0202
• CPC Class Codes: G16C 20/20 Identification of molecular entities, parts thereof or of chemical compositions
• CPC Class Codes: G16C 20/30 Prediction of properties of chemical compounds, compositions or mixtures
• CPC Class Codes: G16C 99/00 : Subject matter not provided for in other groups of this subclass
• CPC Class Codes: Y02A 90/10 : Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation

Abstract

A method for determining national crop yields during the growing season using regional agricultural data is provided. In an embodiment, determining national crop yields during the growing season may be accomplished using a server computer system that receives, via a network, agricultural data records that are used to forecast a national crop yield for a particular year. Within the server computer system an agricultural time series module receives one or more agricultural data records that represent a type of covariate data value related to plants at a specific geo-location at a specific time. The agricultural time series module then aggregates the agricultural data records to create one or more geo-specific time series that represent a specific geo-location over a specified time. The agricultural time series module creates one or more aggregated time series that represent geographic areas from a subset of the one or more geo-specific time series. A crop yield estimating module selects a representative feature from the one or more aggregated time series and creates a covariate matrix for each specific geographic area in computer memory of the server computer system. The crop yield estimating module determines a specific state crop yield for a specific year by using a linear regression module to calculate the specific state crop yield from the covariate matrix that represents the specific state for that specific year. The crop estimation module determines a national crop yield for the specific year by using the distribution generation module to calculate the national crop yield for a specific year from the sum of the specific state crop yields for the specific year nationally adjusted using a national yield adjustment module.

Claims

1. A method comprising:

using an aggregated time series module in a server computer system, receiving one or more agricultural data records that represent a type of covariate data value for plants at a specific geo-location at a specific time, wherein the type of covariate data value includes at least one of a remotely sensed spectral property of plant records at a particular spectral bandwidth range, and a soil moisture record;

using the aggregated time series module, aggregating the one or more agricultural data records to create one or more geo-specific time series, wherein each geo-specific time series represents a specific geo-location over a specified time;

using the aggregated time series module, creating one or more aggregated time series that represent specific geographic areas, from a subset of the one or more geo-specific time series;

using a crop yield estimating module in a server computer system, selecting a representative feature from the one or more aggregated time series and creating for each specific geographic area a covariate matrix in computer memory comprising the representative features selected from the one or more aggregate time series;

using the crop yield estimating module, determining a specific state crop yield for a specific year by using a linear regression module to calculate the specific state crop yield from the covariate matrix that represents the specific state for that specific year, wherein one or more regression coefficients in the linear regression module for the specific state are calculated by using a distribution generation module and wherein an error term in the linear regression module for the specific state is calculated by using the distribution generation module where a mean parameter is zero and a variance parameter is a state specific bias coefficient;

using the crop yield estimating module, determining a national crop yield for the specific year by using the distribution generation module to calculate the national crop yield for the specific year from a sum of the specific state crop yields for the specific year nationally adjusted using a national yield adjustment module.

2. The method of claim 1, wherein the national yield adjustment module comprises:

calculating a mean parameter for the distribution generation module wherein the mean parameter comprises a sum of each specific state crop yield multiplied by a specific state weighted value, multiplied by a first national bias coefficient, and added to a second national bias coefficient; and

calculating a variance parameter for the distribution generation module wherein the variance parameter comprises a sum of each specific state bias coefficient squared multiplied by the specific state weighted value, multiplied by the first national bias coefficient squared, and added to a third national bias coefficient squared.

3. The method of claim 1, wherein aggregating the one or more agricultural data records to create one or more geo-specific time series further comprises filtering one or more agricultural data records that correspond to known geo-locations that are not used as agricultural growth areas.

4. The method of claim 1, wherein aggregating the one or more agricultural data records to create one or more geo-specific time series further comprises selecting favorable data values from the one or more agricultural data records based upon a vegetation index, wherein a vegetation index is a combination of one or more wavelength ranges of remotely sensed spectral properties.

5. The method of claim 1, wherein aggregating the one or more agricultural data records to create one or more geo-specific time series comprises aggregating soil moisture data records that correspond to a specific geo-location at a specific point in time.

6. The method of claim 1, wherein creating one or more aggregated time series comprises applying locally weighted scatterplot smoothing to the subset of the one or more geo-specific time series and selecting median values from for each time interval to create the one or more aggregated time series.

7. The method of claim 1, wherein selecting a representative feature from the one or more aggregated time series comprises selecting one of the following:

a maximum value of the aggregated time series, an integral over a specified time period of the aggregated time series, an integral over a specified time period of the aggregated time series divided by a mean of the aggregated time series provided the mean is above a defined value threshold, an aggregated time series value at a period of time before the maximum value of the aggregated time series, or an aggregated time series value at a period of time after the maximum value of the aggregated time series.

8. The method of claim 1, wherein the aggregated time series represents a specific state in a country.

9. The method of claim 1, wherein determining the specific state crop yield for the specific year by applying a normal distribution function comprises using independent and identically distributed random variables.

10. The method of claim 1, wherein determining the national crop yield for the specific year by applying a normal distribution function comprises using independent and identically distributed random variables.

11. The method of claim 1, wherein determining a national crop yield for the specific year further comprises determining a prediction interval associated with the national crop yield, wherein the prediction interval is range of values that measures a level of certainty associated with the national crop yield.

12. The method of claim 1, wherein determining a national crop yield for the specific year further comprises determining a distribution set associated with the national crop yield, wherein the distribution set measures a level of certainty associated with the national crop yield.

14. The one or more non-transitory storage media of claim 11, wherein the national yield adjustment module comprises:

calculating a mean parameter for the distribution generation module wherein the mean parameter comprises a sum of each specific state crop yield multiplied by a specific state weighted value, multiplied by a first national bias coefficient, and added to a second national bias coefficient; and

calculating a variance parameter for the distribution generation module wherein the variance parameter comprises a sum of each specific state bias coefficient squared multiplied by the specific state weighted value, multiplied by the first national bias coefficient squared, and added to a third national bias coefficient squared.

15. The one or more non-transitory storage media of claim 11, wherein aggregating the one or more agricultural data records to create one or more geo-specific time series further comprises filtering one or more agricultural data records that correspond to known geo-locations that are not used as agricultural growth areas.

16. The one or more non-transitory storage media of claim 11, wherein aggregating the one or more agricultural data records to create one or more geo-specific time series further comprises selecting favorable data values from the one or more agricultural data records based upon a vegetation index, wherein a vegetation index is a combination of one or more wavelength ranges of remotely sensed spectral properties.

17. The one or more non-transitory storage media of claim 11, wherein aggregating the one or more agricultural data records to create one or more geo-specific time series comprises aggregating soil moisture data records that correspond to a specific geo-location at a specific point in time.

18. The one or more non-transitory storage media of claim 11, wherein creating one or more aggregated time series comprises applying locally weighted scatterplot smoothing to the subset of the one or more geo-specific time series and selecting median values from for each time interval to create the one or more aggregated time series.

19. The one or more non-transitory storage media of claim 11, wherein selecting a representative feature from the one or more aggregated time series comprises selecting one of the following:

a maximum value of the aggregated time series, an integral over a specified time period of the aggregated time series, an integral over a specified time period of the aggregated time series divided by a mean of the aggregated time series provided the mean is above a defined value threshold, an aggregated time series value at a period of time before the maximum value of the aggregated time series, or an aggregated time series value at a period of time after the maximum value of the aggregated time series.

20. The one or more non-transitory storage media of claim 11, wherein the aggregated time series represents a specific state in a country.

21. The one or more non-transitory storage media of claim 11, wherein determining the specific state crop yield for the specific year by applying a normal distribution function comprises using independent and identically distributed random variables.

22. The one or more non-transitory storage media of claim 11, wherein determining the national crop yield for the specific year by applying a normal distribution function comprises using independent and identically distributed random variables.

23. The one or more non-transitory storage media of claim 11, wherein determining a national crop yield for the specific year further comprises determining a prediction interval associated with the national crop yield, wherein the prediction interval is range of values that measures a level of certainty associated with the national crop yield.

24. The one or more non-transitory storage media of claim 11, wherein determining a national crop yield for the specific year further comprises determining a distribution set associated with the national crop yield, wherein the distribution set measures a level of certainty associated with the national crop yield.

13. One or more non-transitory storage media storing instructions which, when executed by one or more computing devices, cause performance of a method comprising the steps of:

using an aggregated time series module in a server computer system, receiving one or more agricultural data records that represent a type of covariate data value for plants at a specific geo-location at a specific time, wherein the type of covariate data value includes at least one of a remotely sensed spectral property of plant records at a particular spectral bandwidth range, and a soil moisture record;

using the aggregated time series module, aggregating the one or more agricultural data records to create one or more geo-specific time series, wherein each geo-specific time series represents a specific geo-location over a specified time;

using the aggregated time series module, creating one or more aggregated time series that represent specific geographic areas, from a subset of the one or more geo-specific time series;

using a crop yield estimating module in a server computer system, selecting a representative feature from the one or more aggregated time series and creating for each specific geographic area a covariate matrix in computer memory comprising the representative features selected from the one or more aggregate time series;

using the crop yield estimating module, determining a specific state crop yield for a specific year by using a linear regression module to calculate the specific state crop yield from the covariate matrix that represents the specific state for that specific year, wherein one or more regression coefficients in the linear regression module for the specific state are calculated by using a distribution generation module and wherein an error term in the linear regression module for the specific state is calculated by using the distribution generation module where a mean parameter is zero and a variance parameter is a state specific bias coefficient;

using the crop yield estimating module, determining a national crop yield for the specific year by using the distribution generation module to calculate the national crop yield for the specific year from a sum of the specific state crop yields for the specific year nationally adjusted using a national yield adjustment module.