Method and system for providing site-specific fertilizer recommendations
By generating vegetation indices and adjusting crop nutrient status, the problem of insufficient quality of remote image data was solved, enabling accurate fertilizer recommendations under cloud coverage and revisit time constraints, thus improving the accuracy and efficiency of application decisions.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- YARA INTERNATIONAL ASA
- Filing Date
- 2022-08-31
- Publication Date
- 2026-06-16
Smart Images
Figure CN117881278B_ABST
Abstract
Description
Technical Field
[0001] This disclosure relates to a system and method for providing site-specific, variable fertilizer recommendations to crops in agricultural fields. Background Technology
[0002] Determining the appropriate amount of fertilizer for a crop is one of the most important decisions farmers face. Nitrogen deficiency reduces yield, while excess causes economic losses and environmental damage. Deficiencies in other nutrients can lead to crop defects and reduced crop quality. However, crops exhibit field variability due to differences in parameters affecting their growth and development (water, soil properties), making it challenging to establish location-specific fertilizer recommendations that are optimal for the overall crop.
[0003] Among the various methods for estimating crop nitrogen levels, the use of remote imagery generated by satellites or other equivalent unmanned aerial vehicles is becoming increasingly important because publicly available imagery is available from long-standing satellite platforms such as SENTINEL and LANDSAT. Remote sensing allows for the determination of crop nitrogen levels in remote fields without the need for field verification. However, using remote imagery has well-known drawbacks. Due to time constraints for revisiting the site, coupled with the presence of clouds, adequately high-quality images are often unavailable when fertilizer application should be performed. This situation has necessitated the development of various methods for cloud detection and compensation or for extrapolation. However, these methods for detecting and correcting for cloudy and blurred pixels require significant computational effort, as well as image preprocessing and post-processing to avoid degrading readings.
[0004] The purpose of this disclosure is to establish a method for providing variable, site-specific fertilizer recommendations that can compensate for time constraints due to revisit time and time intervals caused by cloud and haze, and overcome the aforementioned problems.
[0005] Existing technology
[0006] Two known methods related to cloud compensation in image data (as relevant to this invention) can be found in: US2020125844 (Climate Corp), which uses a machine learning system to train a system for cloud detection based on multiple images already classified as clouds and cloud shadows; and further WO2020 / 160641A1, which uses a multilayer perceptron to use another method to classify image segments to generate cloud masks and classify each pixel. Summary of the Invention
[0007] This invention aims to provide a solution to the aforementioned problems. Avoiding computationally intensive cloud compensation methods requires utilizing available data on cloudless days.
[0008] According to a first aspect of this disclosure, this and other objectives are achieved by a method for providing site-specific variable fertilizer recommendations for crops at a given point in time, the method comprising the steps of: identifying at least one agricultural field comprising at least one crop; scheduling fertilizer application for the at least one agricultural field; determining the crop nutrient status of the at least one agricultural field, wherein determining the crop nutrient status of the at least one agricultural field includes receiving remote data, the remote data including image data of the at least one agricultural field and a timestamp indicating when the image data was taken; and generating at least one vegetation index (SX) indicative of the crop nutrient status based on the image data. i,j tm The crop nutrient status is determined based on at least one vegetation index; the crop nutrient status is adjusted based on the time difference between image timestamps and scheduled fertilizer application; and a variable fertilizer recommendation is determined for at least one agricultural field based on the adjusted crop nutrient status.
[0009] Using this method, the current crop nutrient status can be obtained using old remote image data.
[0010] According to another embodiment, determining the crop nutrient status of at least one agricultural field further includes: receiving remote data comprising multiple image data of at least one agricultural field on multiple dates prior to scheduled fertilizer application and corresponding timestamps indicating when the images were taken; and generating at least one vegetation index (SX) indicating the crop nutrient status based on the image data of the multiple image data on the multiple dates. i,j t(m-n) ..., SX i,j t(m) The process involves: assessing the rate of change of at least one vegetation index generated between each of multiple dates for at least one vegetation index; selecting at least one of multiple image data and its corresponding vegetation index based on the corresponding rate of change; and determining crop nutrient status based on at least one selected vegetation index.
[0011] This method avoids suboptimal image data and allows older images to be used to compensate for suboptimal data, even when there are time differences.
[0012] According to another embodiment, assessing the rate of change of at least one vegetation index further includes averaging the vegetation indices for each given image data with respect to at least a portion of agricultural land and determining the corresponding difference between the averaged at least one vegetation index.
[0013] This method allows for the advantageous selection of the most suitable image dataset.
[0014] According to another embodiment, multiple received image data indicating the crop nutrient status of at least one agricultural field on multiple dates prior to scheduled fertilizer application, and a single image indicating when the image data was taken, are selected based on an assessed rate of change, and the determination of the crop nutrient status is based on the selected images, and the crop nutrient status is adjusted based on the time difference between the timestamp of the selected images and the scheduled fertilizer application.
[0015] According to this method, temporal differences for the selected image data are taken into account. According to another embodiment, remote data comprising multiple image data further includes selecting image data from the received image data where the number of valid pixels of a corresponding image data is higher than a predetermined threshold, wherein the validity of the pixels is received within the image data.
[0016] This method discards unsuitable image data before any evaluation, reducing computational workload.
[0017] According to another embodiment, adjusting crop nutrient status includes determining the crop dry matter value at the time of the received image data used to generate at least one vegetation index; simulating the evolution of crop dry matter of at least one crop in at least one agricultural field between the time of the image data and a scheduled fertilizer application, wherein simulating the evolution of crop dry matter further includes receiving weather data and iteratively updating the crop dry matter based on weather data between the timestamp of the received image data and the scheduled application date; and adjusting the crop nutrient status based on the simulation of crop dry matter.
[0018] This method allows for weather-compensated adjustments to the nutrient status of crops.
[0019] According to another embodiment, receiving weather data for at least one agricultural field further includes receiving daily temperature and solar irradiance values, and wherein iteratively updating crop dry matter values further includes updating the dry matter values by means of a generated value calibrated by a temperature efficiency factor in proportion to the daily absorbed photosynthetic radiation.
[0020] This method adjusts crop dry matter values based on temperature and solar irradiance data, enabling more precise control over crop nutrient status.
[0021] According to another embodiment, receiving weather data further includes receiving historical precipitation data, and wherein iteratively updating crop dry matter values further includes adjusting the iteratively updated dry matter values based on soil water content estimated from water balance calculations based on historical precipitation data.
[0022] This method can solve the water supply shortage.
[0023] According to another embodiment, determining variable fertilizer recommendations based on adjusted crop nutrient status further includes using additional agronomic parameters, such as mineralization potential, yield forecasts, or weather forecasts.
[0024] Using this method, additional field parameters can be considered in fertilizer recommendations.
[0025] According to another embodiment, variable fertilizers are recommended for generating machine-readable prescription maps for controlling fertilizer application systems.
[0026] Using this method, agricultural machinery can automatically recommend fertilizers.
[0027] According to another embodiment, scheduling fertilizer application further includes determining the application date and time based on at least one of the following: user input, received or generated weather forecasts, field crop data, predictions of phenological stages, and / or predictions of crop development based on crop growth.
[0028] This method can be used to determine the appropriate application date.
[0029] According to another aspect, systems, data processing devices, computer-readable storage media, and computer program products configured to perform the above methods are envisioned within this disclosure. Attached Figure Description
[0030] The accompanying drawings, which are included to provide a further understanding of this disclosure and are incorporated in and constitute a part of this specification, illustrate embodiments of the disclosure and, together with the description, serve to explain the principles of the disclosure.
[0031] Figure 1 Agricultural fields are shown in accordance with the application area of this disclosure.
[0032] Figure 2 A schematic representation of a system according to an embodiment of the present disclosure is shown.
[0033] Figure 3 and 4 Flowcharts of methods according to different embodiments of the present disclosure are shown.
[0034] Figure 5 For the purposes of this application, examples of possible evolution of vegetation indices over time periods representing crop seasons are shown.
[0035] Figure 6 A detailed view of a preceding figure is shown, illustrating one of the embodiments according to this disclosure.
[0036] Figure 7 This shows the evolution of vegetation indices for different crop types over different growing seasons according to different standards.
[0037] Figure 8 The rate of change of vegetation indices for different crop types over the growing season was generated.
[0038] The accompanying drawings are provided to aid in the easy understanding of the technical concepts of this disclosure, and it should be understood that the concepts of this disclosure are not limited to the drawings. The concepts of this disclosure should be interpreted as extending to any modifications, equivalents, and substitutions other than those shown in the drawings. Several embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. Detailed Implementation
[0039] As used herein, the singular forms “a,” “an,” and “the” include both the singular and plural unless the context clearly indicates otherwise. The term “comprising” as used below is synonymous with “including” or “contains” and is inclusive or open-ended, and does not exclude additional unmentioned parts, elements, or method steps. When this description refers to a product or process that “comprising” a particular feature, part, or step, it means that other features, parts, or steps may also be present, but it may also refer to an embodiment containing only the listed features, parts, or steps.
[0040] The numerical values listed with the aid of the accompanying drawings include all values and fractions within those ranges, as well as the referenced endpoint values. The term "approximately" as used when referring to measurable values (such as parameters, quantities, time periods, etc.) is intended to include a specified value and variations from that specified value of + / - 10% or less, preferably + / - 5% or less, more preferably + / - 1% or less, and even more preferably + / - 0.1% or less, provided that the variation applies to the disclosure herein. It should be understood that the values referred to by the term "approximately" itself are also disclosed.
[0041] Unless otherwise defined, all terms appearing in this disclosure, including technical and scientific terms, have the meanings that are commonly given to them by those skilled in the art. For further guidance, definitions are included to further explain the terms used in the description of this disclosure.
[0042] Figure 1 Agricultural fields 1, including crops, are depicted, along with other systems and devices that system 100 can interoperate with. System 100 of this disclosure is configured to determine fertilizer recommendations to be implemented in farms and agricultural fields. Agricultural fields 1 may include crop and forestry areas and may further include: a network of dedicated agricultural sensors 270, including soil sensors, moisture sensors, and other state-of-the-art sensors; and a weather station 260, including rain sensors, wind sensors, temperature sensors, solar irradiance sensors, humidity sensors, etc. Figure 1Common remote sensing devices mentioned in this disclosure are further described. Reference is made to remote sensing devices (such as satellite imaging system 250) and other aircraft 240 (such as aircraft or unmanned aerial vehicles (UAVs)).
[0043] Figure 2 A schematic representation of a system according to one embodiment of this disclosure is shown. According to this disclosure, system 100 includes several components, such as a memory unit 110, a processor 120, a wired / wireless communication unit 130, and an input / output unit 140. System 100 can also be operatively connected to another personal or mobile device 200 via the communication unit 130.
[0044] System 100 includes a remote agricultural recommendation engine 220 (if it is remote in nature) that the system can remotely connect to via communication unit 130. In this case, the remote agricultural recommendation engine 220 can be represented by a computer, a remotely accessible server, other client-server architecture, or any other electronic device generally covered under the term data processing equipment. System 100 does not need to be located near the agricultural fields where recommendations should be made.
[0045] System 100 may also be represented by a laptop computer or handheld device operated directly by the farmer or user at or outside the agricultural field location, having an integrated agricultural recommendation engine 150 capable of operating entirely at the farm location, and may include a GPS unit 180 or any other suitable positioning tool, as well as a fully remote computer or server configured to communicate with the user from another person or mobile device 200 on their operating system 100. It should be understood that the presence of the remote recommendation engine and the integrated recommendation engine is not mutually exclusive. The integrated agricultural recommendation engine 150 may be a local copy of the remote agricultural recommendation engine 250 or a lightweight version thereof to support periods of low network connectivity and offline operation. Furthermore, the mobile or personal device 200 is considered to be any state-of-the-art mobile computing device that allows user input and output of data and includes common features.
[0046] System 100 and the remote or integrated agricultural recommendation engine may include field and farm data as well as external data and / or be configured to receive said data, wherein the external data includes weather data, remote data including remote imagery data, and additional data provided by weather forecast providers or other third parties. Field data may include, in particular, current and past data on at least one of the following: field and geographic identifiers relating to the geometry of the boundaries of agricultural fields, including the presence of unmanaged areas within the agricultural fields; topographic data; crop identifiers for current and past crops (crop variety and type, growth status, planting data and date, plant nutrition and health status); harvest data (yield, value, product quality, estimated or recorded historical values); soil data (type, pH, soil organic matter (SOM) and / or cation exchange capacity (CEC)). Farm data may include additional data on planned and past tasks (such as field maintenance practices and agricultural practices), fertilizer application data, pesticide application data, irrigation data, and other field reports, as well as historical series of data, thereby allowing comparison of data with past data and processing of additional administrative data, such as shifts, logs, and other organizational data. Planned and past tasks may include further activities such as monitoring plants and pests, applying pesticides, fungicides or crop nutrients, measuring at least one farm or field parameter, maintaining and repairing ground hardware, and other similar activities.
[0047] System 100 can be further configured to receive and / or retrieve soil data from available online soil databases (such as SoilGrid from World Soil Information, SSURGO (from the USDA Soil Survey Geographic Database), or any similar soil data repository) and via user input.
[0048] System 100 may be further configured to receive any of the aforementioned data manually input by a user / farmer via input / output unit 140 or mobile / handheld device 200, or received by communication unit 130 from dedicated sensors or data processing equipment. Furthermore, system 100 and the agricultural recommendation engine may be configured to receive weather data from nearby weather stations 260 and / or external crop / farm sensors 270. The nearby weather stations 260 and / or external crop / farm sensors 270 are configured to communicate via one or more networks. In another embodiment, the weather data is provided by an external weather forecasting company. The weather data may further include current and past series of data or historical weather data of at least one of the following: temperature, cumulative precipitation, relative humidity, wind speed, solar irradiance, cumulative sunshine hours, and daily absorbed photosynthetic radiation, as well as forecasts of these parameters.
[0049] System 100 may further be operatively connected to agricultural equipment 300. Examples of agricultural equipment 300 include tractors, combine harvesters, harvesters, planters, trucks, fertilizer equipment, and any other physical machinery or hardware (typically mobile machinery that can be used in agriculture-related tasks). In one embodiment, system 100 may be configured to communicate with agricultural equipment 300 via a wireless network to perform variable, site-specific fertilizer application on a determined crop. System 100 may be further configured to generate machine-readable script files for the agricultural equipment to perform fertilizer application.
[0050] Figure 3 A method according to a main embodiment of this disclosure is described. The method of this disclosure provides 1500 fertilizer recommendations for crops at scheduled time points, for variable site-specific applications. The method of this disclosure is configured to identify 1000 at least one agricultural field comprising at least one crop and schedule 1100 fertilizer application for at least one agricultural field.
[0051] This method, in a non-limiting example, can provide a user with a predetermined field for determining at least one agricultural field. However, the method of this disclosure can be configured to automatically retrieve and recommend targeted fields based on farm and / or field data. Agricultural fields can also be determined based on the user's location, depending on where the user can implement the method of this disclosure; this location can be provided by the mobile device 200, the GPS unit 180 of system 100, or the agricultural equipment 300. In another embodiment, system 100 is configured to determine the boundaries of the determined agricultural fields, even if the determined agricultural fields do not have data regarding boundary locations in existing farm data.
[0052] The method of this disclosure is further configured to determine the crop nutrient status of at least 1200 agricultural fields comprising at least one crop. The method of this disclosure is configured to determine the crop nutrient status via remote imagery, as will be explained below.
[0053] As a non-limiting example, this method can be configured to schedule fertilizer application based on user input, determining whether fertilizer application should be scheduled immediately or at another given time. As another example, system 100 can be configured to schedule fertilizer application during appropriate time periods for execution, as described in further embodiments below.
[0054] This application utilizes suitable remote data to remotely determine crop nutrient content. Remote data may refer to data provided by an imaging satellite 250 or a suitable manned or unmanned imaging vehicle 240. These satellite or vehicle systems are configured to communicate via dedicated networks and common methods not disclosed herein. Among the various remote data available, satellite data is now widely available from numerous public (NASA's LANDSAT, ESA's SENTINEL) and / or private providers. However, this method is not limited to satellite data platforms, as the spectral bands that can be used in this method are provided across a wide range of standard satellite data available in both public and private contexts. Due to the differences across different satellite and optical sensor platforms, it is not intended to limit the support of this disclosure to precise and specific wavelengths, nor to provide a given wavelength for orientation. While different factors and corrections can be introduced to address these variability, it should be understood that wavelengths close to those mentioned below are used, as the specifications of the platforms vary accordingly.
[0055] In this embodiment, the remote data is obtained from the Sentinel-2 satellite. The Sentinel-2 mission includes an MSI (Multispectral Instrument) capable of acquiring high spatial resolution data to monitor the Earth's surface. The MSI operates passively by collecting sunlight reflected from the Earth, and is therefore a more efficient and energy-saving detection method. Sentinel-2 consists of 13 bands with different spatial resolutions (10m, 20m, or 60m) located in the visible, near-infrared, and short-wave infrared portions of the spectrum. In this embodiment, the method uses image data associated with spectral bands having at least a plurality of wavelengths approximately between 700 and 850 nm. In another embodiment, the method uses data associated with spectral bands having wavelengths approximately 740 and 780 nm and 900 and 970 nm. Using Sentinel-2 spectral bands from the MSI produces measurements with high resolution (approximately 20m), and is therefore preferred for the implementation of this disclosure. However, Sentinel-2 data has a known drawback: it is not available on cloudy days.
[0056] In embodiments, remote data may include data associated with different spectral bands determined for improving the method. Depending on the nature and source of the remote data, additional compensation and calibration algorithms are considered in this application.
[0057] Receiving remote data includes receiving image data of at least one agricultural field, wherein each image data includes a timestamp indicating when the image was taken. Once the image data is received, the method is configured to generate at least one coefficient (or vegetation index) derived from the image data indicating the nutrient status of the crop. Different coefficients or indices are used in the literature to obtain different agricultural, soil, and vegetation information, such as the differential vegetation index and the normalized differential vegetation index (NDVI). However, the NDVI is sensitive to the effects of soil brightness, natural color, atmosphere, clouds, cloud shadows, and canopy shadows, and requires remote sensing calibration. In this sense, other envisioned coefficients may include: the atmospheric impedance vegetation index (ARVI), to reduce dependence on atmospheric influences; and the soil-adjusted vegetation index (SAVI) or the soil type atmospheric impedance vegetation index (TSARVI), which takes into account the difference between vegetation and different types of soil background. In addition to the standard vegetation indices included in the prior art, this application may also utilize two additional indices, which will be described in more detail below.
[0058] In one embodiment, in order to provide a reliable vegetation index or coefficient indicating aboveground nitrogen uptake present in vegetation, one of the other indices of this application may include different wavelengths between 670 and 800 nm at the so-called red edge of the vegetation.
[0059] In another embodiment, in order to provide a reliable vegetation index or coefficient indicating the water content present in the vegetation, the remote data may include additional remote data, which includes additional wavelengths at or near the water absorption zone, such as approximately 950 nm, 1100, 1450, or 1950 nm, to determine the canopy water content.
[0060] In another embodiment, the formula for the vegetation index under consideration can be expressed as: SC = f(R760, R730), where R760 and R730 represent the reflectance related to the corresponding wavelength received in the long-range data, or the values closest to the reflectance among the corresponding satellite platforms as described above. Therefore, this index is sensitive to the chlorophyll content present in the vegetation and can be directly correlated with the total aboveground nitrogen uptake within the canopy. In this case, f can be a mathematical relationship following established common vegetation indices.
[0061] In another embodiment, in order to provide a reliable vegetation index or coefficient indicating aboveground fresh or dry matter in vegetation, the remote data may include additional remote data, which includes additional wavelengths at or near the water absorption zone, such as approximately 970 nm, 1100 nm, 1450 nm, or 1950 nm.
[0062] In another embodiment, the fresh biomass can be calculated, for example, according to SW = g(R900, R970), where, as previously described, R900 and R970 represent the reflectivity at the corresponding wavelengths received with remote data, or values close to the reflectivity in the corresponding satellite platform. The function g can be a function similar to the function f described above.
[0063] The series of coefficients (or indices) mentioned above are not intended to be limiting. The different coefficients used in this method to calculate other properties of agronomic fields (such as soil moisture) can vary considerably. While these other properties do not directly indicate crop nutrient content, they can help determine fertilizer recommendations by further understanding other field parameters.
[0064] Once at least one coefficient indicating the crop status within an agricultural field is generated, the crop nutrient status can be determined based on it. Therefore, site-specific, location-related nutrient recommendations can be generated by determining various factors such as chlorophyll, nitrogen uptake, and biomass. However, these methods based on remote data cannot account for cloud formation and the limitations imposed by this event.
[0065] Due to the presence of clouds and the limited availability of remote imagery data caused by the revisit time of remote imagery providers (e.g., Sentinel platform has a revisit time of 2-3 days in Europe), or the use of only drone or aircraft imagery, remote imagery data may at best be data from a few days ago, and during cloudy seasons it may be data from several weeks ago. Since scheduled applications may also be carried out with some time lag, this time lag between remote imagery data and fertilizer application dates needs to be compensated for.
[0066] Therefore, this method is configured to: receive remote data, which includes image data of at least one agricultural field and a timestamp indicating when the image was taken; and adjust crop nutrient status based on the time difference between the image timestamp and the scheduled fertilizer application, wherein a variable site-specific fertilizer recommendation is generated based on the crop nutrient status.
[0067] Therefore, this method is configured to adjust 1300 determined crop nutrient states based on the time difference between the timestamp of the image and the scheduled fertilizer application. In another embodiment, the adjustment of the determined crop nutrient states can be performed as a function proportional to the time difference between the date of the received image data and the scheduled fertilizer application. In another embodiment, the method can be configured to adjust the determined crop nutrients based on the expected crop development using a crop growth model and / or weather data between the date of the received image data and the scheduled fertilizer application. As mentioned above, due to the limited availability of remote image data on cloudless days, crop nutrient states and corresponding fertilizer recommendations can be based on "outdated" images. Improved fertilizer recommendations and application are achieved by considering the time difference between the two events.
[0068] This method is further configured to determine 1,500 variable site-specific fertilizer recommendations for at least one agricultural field based on adjusted crop nutrient status.
[0069] Generally, users / farmers can schedule fertilizer application immediately on the same day, or prepare it a few days in advance when suitable conditions (weather conditions, crop growth stage, management, or personal factors) arise. This time delay further impacts the optimal fertilizer recommendation that can be considered for the current application. This may require receiving images taken within a week of the future fertilizer application. However, due to adverse weather conditions, this method may include receiving images taken a month or longer before the future fertilizer application. For example, in some locations, the availability of remote image data less than two days prior for agricultural purposes is less than 30%. While this percentage increases, another 30% of the image data is spaced more than a week apart due to cloud cover. Due to the method disclosed herein, accurate compensation for relative time differences can be considered, while eliminating the need for cloud compensation algorithms that make more remote image data available for each day closest to the fertilizer application date, in order to obtain an accurate determination of the current crop nutrient status.
[0070] When receiving remote image data, common image data providers offer a classification of given pixel categories according to the table below.
[0071]
[0072] Table 1
[0073] Because of the classification provided, excluding images of insufficient quality is not a troublesome procedure. As can be seen below, when the number of pixels in the image data other than those classified as vegetated, non-vegetated, dark, or unclassified (labels 2, 4, 5, 7) exceeds a certain limit, there is no guarantee that the image data will provide meaningful results, and the pixels are classified as invalid. Therefore, in another embodiment, this method can be configured to identify and exclude cloudy image data before processing to generate at least one vegetation index. To avoid processing cloudy images from received remote data, the identification of images to be excluded is performed according to the following rules based on the classification shown in Table 1, where SCL refers to the scene classification following the column in Table 1.
[0074]
[0075] It should be understood that the location of different pixels refers to pixels included within at least one defined agricultural field according to standards, based on the acceptability of the received image data. Therefore, the method is further adapted to determine from each received image data which pixels located within the agricultural field satisfy the conditions and overall relationships listed above. Depending on the number and availability of images, C 云 This can be adjusted by setting a predetermined threshold. In a preferred embodiment, C 云 It is a fixed constant with a value of 0.05. However, C cloud The range can be adjusted to include levels between 0.01 and 0.1. This classification excludes cloudy or defective images, which would otherwise compromise the reliability of the vegetation index. Furthermore, it reduces the number of images processed, thus reducing computational workload.
[0076] In another embodiment, the method of this disclosure is configured to receive remote data, wherein the remote data includes multiple cloud-free image data from multiple dates prior to a scheduled fertilizer application. The multiple received image data include corresponding timestamps indicating when the images were taken. While it is desirable to view the most recently available and cloud-free remote image data to accurately represent the crop state at the time of application, the most recently available and cloud-free remote image data may still be locally affected by clouds, fog, or other atmospheric events of local or other unknown cause (which may blur the image data or affect its quality). This can be seen when the progression of at least one vegetation index does not follow the expected evolution. For example, as in... Figure 5 As can be seen, such attenuation at certain points in time may be caused by the availability of remote image datasets. For example... Figure 5As shown, around late April and early May, there is very little available long-range image data (indicated by the crosses along the lines in the figure). Furthermore, as can be seen in the figure, some available long-range image data does not depict a true representation of the crop's nutrient status due to an unexpected change caused by a decrease in at least one vegetation index described in the figure. Therefore, for example, a user / farmer intending to schedule fertilizer application at the end of the first week of May and begin in the second week of May might find it necessary to use coefficients generated from the most recent image data. While the main embodiment of this application has already provided compensation for the time elapsed between the last available long-range image data and the scheduled fertilizer application date, further advantages of this embodiment will become clear below.
[0077] The method of this application can be further configured to generate at least one vegetation index indicating crop nutrient status based on selected image data from multiple long-range image datasets acquired over multiple captures. For example, SX can be named a specific vegetation index (which can take different forms depending on the corresponding vegetation index), the vegetation index SX t(m) This refers to an index generated based on an image taken at time point m, where m can be the most recently available remote image data. Therefore, if n previous remote image data are available for this method to generate at least one corresponding vegetation index indicating crop nutrient status, the generated index can be called SX. t(m) SX t(m-1) ..., SX t(m-n) .
[0078] Furthermore, at least one generated vegetation index can be calculated at the pixel level rather than the field level, in which case the generated index for the corresponding remote image data at a given pixel (i,j) can be named SX. (I,j) t(m-p) Furthermore, the corresponding average vegetation index for at least one agricultural field SX is calculated as the average of the pixel-level vegetation indices of the corresponding at least one agricultural field.
[0079] In this embodiment, the method can be further configured to evaluate the pixel-level rate of change of at least one vegetation index or the corresponding average of the vegetation index of at least one agricultural field across multiple dates on which remote image data is available. Furthermore, the method can be configured to select at least one of multiple image data sets from which at least one vegetation index is generated, in order to determine crop nutrient status based on the calculated rate of change. Therefore, for a given index SX... m+1 The rate of change (RoC) at time point m+1 is calculated as follows:
[0080] RoC(SX m+1 )=(SXt(m+1) –SX t(m) () / (Number of days between image m and m+1)
[0081] Given the evaluated rate of change for at least one determined index, the method is further configured to select at least one of a plurality of received image data based on the rate of change.
[0082] This method provides a more accurate determination of the crop's nutrient status because it considers which of the latest available images is more suitable for a favorable assessment of the crop's nutrient status.
[0083] In another embodiment, the method may be further configured to compare the corresponding rate of change with a predefined value. In another embodiment, the method may be further configured to compare the corresponding rate of change with the rate of change of the previous interval (e.g., comparing RoC(SX)). t(m+1) ) and RoC(SX) t(m) (Comparison). In another embodiment, the rate of change can be compared to a predefined rate of change calibrated for a specific crop type present in the agricultural field, which is adapted to time frames included between corresponding remote image data defining the RoC. Figure 7 and 8 As can be seen, based on a determined crop, the expected performance of at least one vegetation index can be inferred. Therefore, this method can be further configured to compare a determined rate of change with a threshold. In another embodiment, the threshold for the rate of change of at least one vegetation index can be further determined based on the rate of change of at least one vegetation index over a determined time period, such as based on the existing value of the vegetation index of a given crop during that time period.
[0084] While the expressions used for vegetation indices above have been shown for general representation, this method can be further configured to include pixel-level determination of each of at least one vegetation index and a corresponding determination of the rate of change. For the pixel-level case, the equations listed above will be understood as follows:
[0085] RoC(SX) (I,j) m+1 =(SX) (I,j) t(m+1) –SX (I,j) t(m) () / (Number of days between image m and m+1)
[0086] Furthermore, this method can be configured to process at least one vegetation index of a given remote image data for at least a portion of an agricultural field on a given date, wherein processing the vegetation index further includes assessing the rate of change of at least one vegetation index, and further includes averaging at least one vegetation index of each given image data for at least a portion of the agricultural field and determining the corresponding difference between the averaged at least one vegetation index.
[0087] In another embodiment, adjusting crop nutrient status based on the time difference between image timestamps and scheduled fertilizer application may further include determining the crop dry matter value at the time of the selected image data based on at least one vegetation index, and simulating the evolution of the crop dry matter value of at least one crop in at least one agricultural field between the time of the selected image data and the scheduled fertilizer application. Assuming the absence of water stress and other growth limiting factors, the evolution of crop dry matter value can be accurately predicted based on the simulations disclosed below. This embodiment further includes simulating the evolution of crop dry matter value based on received weather data from at least one agricultural field between the timestamp of the image data and the scheduled fertilizer application, and iteratively updating the crop dry matter value based on weather data between the timestamp of the image data and the scheduled fertilizer application date. Figure 6 express Figure 5 A close-up view of the expected date for fertilizer application, as discussed above. The figure shows a series of values decreasing after a rapid increase in the vegetation index, followed by an even faster increase with some overshoot. Due to the nature of crops, these behaviors are not always reasonable and may indicate image data artifacts or suboptimal readings. Therefore, given the embodiments mentioned above, the method can be further configured to: simulate the evolution of crop dry matter, exemplarily indicated by a blank cross contrasting with the bold cross representing the date image data is available; and subsequently adjust the crop nutrient values accordingly.
[0088] As a non-limiting example, determining the crop dry matter value (DM0) can be based on at least one vegetation index from received remote image data, as explained above.
[0089] In another embodiment, it can be done according to Figure 4 The illustrated flow diagram illustrates the iterative updating of the 1400 crop dry matter value. The method of this application can be further configured to iteratively update the crop dry matter value using a generated value calibrated by a temperature efficiency factor, proportional to the daily absorbed photosynthetic radiation. Once the 1410 dry matter value is determined for the selected image data date, it is presented as shown... Figure 4The described iterative scheme for updating dry matter values based on weather data. According to this embodiment, 1430 daily dry matter values (DMi) are generated based on the dry matter value (DMi) from the previous day and the development factor added to the dry matter value. +1 According to a non-limiting example, the 1420 developmental factor was determined based on daily absorbed photosynthetic radiation, daily temperature evolution, and different calibration factors. This process was then repeated over the number of days between the date of image data and the scheduled fertilizer application until the final dry matter value (DM) was calculated. 终止 Based on the calculated dry matter value, the nutrient status of crop 1440 is adjusted for the scheduled fertilizer application time, thus achieving a more accurate fertilizer recommendation reflecting the current nutrient status of the crop. For example, on the date the image data was taken, the nutrient status was given by the vegetation index and calibrated for a given dry matter for that date. Following an iterative method, the dry matter evolves as follows:
[0090] DM i+1 =DM i +k*h(DM i )*E i *f T (T i )
[0091] This iteration is performed for i = 1, ..., n; where n represents the number of days between the date of the image data and the date of fertilizer application, and k represents a constant that can take different values based on crop data. The function h is proportional to the daily fraction of absorbed photosynthetic radiation; E i For daily solar irradiance, the function f T This is a temperature-calibrated function representing the temperature efficiency factor as it varies with daily temperature. In the embodiment, the function f... T It can be defined as a function of daily temperature evolution. In another embodiment, the function f T It can be defined as a function of the daily average temperature, as well as the maximum and minimum values.
[0092] Therefore, on the date of fertilizer application, this method provides updated dry matter values for identified crops present in agricultural fields, which allows for DM-based... n The adjusted crop nutrient status is then determined. A variable fertilizer recommendation for at least one agricultural field is then determined based on the current values of the crop nutrient status, wherein nitrogen uptake is determined at the fertilizer application data, taking into account the dry matter value calculated for the fertilizer application date.
[0093] In another embodiment, the method may be further configured to receive historical precipitation data. In this embodiment, iteratively updating the crop dry matter value further includes adjusting the iteratively updated dry matter value based on historical precipitation data. For example, in a non-limiting example, soil moisture content can be estimated based on water balance calculations based on historical precipitation data throughout the season, allowing the establishment of a limiting threshold for soil moisture content and the corresponding updating of the iteratively updated dry matter value.
[0094] In another embodiment, the method configured to schedule fertilizer application may further include determining the application date and time based on at least one of: user input, received or generated weather forecasts, field crop data, predictions of phenological stages, and / or predictions of crop development based on crop growth. As mentioned above, users or farmers may wish to schedule fertilizer application at their convenience and determine fixed dates suitable for their schedules. However, this method may be further configured to automatically determine appropriate fertilizer application dates and times. Taking weather forecasts into account, this method may be further configured to determine dates and times based on specific weather events that may impair or reduce the efficiency of fertilizer application. Furthermore, the method may be further configured to determine the scheduled fertilizer application date based on model predictions of crop growth or phenological stages. Therefore, the method of this application can conveniently determine the scheduled fertilizer application date that is most suitable for the crop.
[0095] In another embodiment, the method can be configured to determine variable fertilizer recommendations based on adjusted crop nutrient status, and may further include using at least one of the following additional agronomic parameters: mineralization potential, yield expectation, or weather forecast.
[0096] In another embodiment, the method can be configured to generate a machine-readable application prescription map based on variable fertilizer recommendations to control the fertilizer application system. The method can be further configured to establish direct communication with agricultural machinery 300 via a standard network to automate fertilizer application. In another embodiment, the method can be configured to generate a downloadable script or file that can be used to upload such an application prescription map to the fertilizer application system via physical tools.
[0097] In another embodiment, variable fertilizer recommendation is performed via a fertilizer application system (such as a spreader or specially adapted agricultural machinery).
[0098] Figure 3 and 4Workflows of various embodiments of the methods included in this disclosure are illustrated. While process steps, method steps, algorithms, etc., may be described in a sequential order, such processes, methods, and algorithms may be configured to operate in an alternating order. In other words, any order or sequence of steps that can be described does not necessarily indicate a requirement to perform the steps in that order. The steps of the processes described herein can be performed in any actual order. Furthermore, some steps may be performed simultaneously, in parallel, or concurrently. The various methods described herein can be practiced by combining one or more machine-readable storage media containing code according to this disclosure with suitable standard computer hardware to execute the code contained therein. Apparatus for practicing the various embodiments of this disclosure may involve one or more computers (or one or more processors within a single computer) and a storage system containing or having network access to one or more computer programs encoded according to the various methods described herein, and the method steps of this disclosure may be performed by modules, routines, subroutines, or sub-parts of a computer program product. While various embodiments of this disclosure have been described for the foregoing, other and additional embodiments of this disclosure may be devised without departing from the scope of this disclosure. The scope of this disclosure is defined by the appended claims. This disclosure is not limited to the described embodiments, versions, or examples, which are included to enable those skilled in the art to make and use this disclosure when combined with information and knowledge available to them.
[0099] While plural nouns are preferred when referring to persons (users, farmers) throughout this disclosure to allow for neutral text drafting, there is no limitation on the number of persons to whom this disclosure should be considered relevant. This is in accordance with the guidelines amendments that came into effect on March 1, 2021, which support gender neutrality and serve as other examples.
[0100] While this disclosure has been illustrated by way of description of various embodiments, and these embodiments have been described in considerable detail, the applicant does not intend to limit or in any way restrict the scope of the appended claims to such details. Additional advantages and modifications will readily emerge to those skilled in the art. Therefore, this disclosure is not, in its broader aspects, limited to the specific details, representative devices and methods, and the illustrative examples shown and described.
[0101] Therefore, its detailed description should not be construed as limiting in all respects, but rather as illustrative. The scope of this disclosure should be determined by a reasonable interpretation of the appended claims, and all modifications falling within the scope of equivalents are included within the scope of this disclosure.
Claims
1. A computer-implemented method for providing site-specific variable fertilizer recommendations for crops at a given point in time, the method comprising the steps of: a. Identify at least one agricultural field that includes at least one crop; b. Arrange fertilizer application for at least one of the agricultural fields; c. Remotely determining the crop nutrient status of the at least one agricultural field, wherein remotely determining the crop nutrient status of the at least one agricultural field includes: i. Receive remote data, said remote data including image data of the at least one agricultural field and a timestamp indicating when the image data was taken; ii. Generate at least one vegetation index indicating crop nutrient status based on the image data; iii. Determine crop nutrient status based on the at least one vegetation index; d. Adjusting the crop nutrient status based on the time difference between the timestamp of the image data and the scheduled fertilizer application, wherein adjusting the crop nutrient status includes: i. Determine the crop dry matter value at the time when the received image data is used to generate the at least one vegetation index; ii. Simulate the evolution of crop dry matter of at least one crop in at least one agricultural field between the time of the image data and the scheduled fertilizer application; iii. The simulation of the evolution of the crop dry matter further includes receiving weather data and iteratively updating the crop dry matter value based on weather data between the timestamp of the received image data and the scheduled application date, wherein Receiving the weather data further includes receiving daily temperature and solar irradiance values, and Iteratively updating the crop dry matter value further includes updating the dry matter value by means of a generated value calibrated by a temperature efficiency factor in proportion to the daily absorbed photosynthetic radiation, and by means of a dry matter value (Dm) from the previous day. i ) and developmental factors to generate daily dry matter value (DM) i+1 ), The dry matter value of the crop is iteratively updated as follows: ， The iterations are performed for i = 1, …, n; where n represents the number of days between the date of the image data and the date of fertilizer application, and where k represents a constant that takes different values based on crop data, where the function h is proportional to the daily fraction of absorbed photosynthetic radiation; Ei is the daily solar irradiance, and the function… A temperature-calibrated function representing the temperature efficiency factor as a function of daily temperature variation; and iv. Adjust the nutrient status of the crop based on simulation of the crop's dry matter; e. Determine a variable fertilizer recommendation for the at least one agricultural field based on the adjusted crop nutrient status.
2. The method of claim 1, wherein determining the crop nutrient status of the at least one agricultural field further comprises: a. Receive remote data, the remote data including multiple image data of the at least one agricultural field on multiple dates prior to the scheduled fertilizer application, and corresponding timestamps indicating when the images were taken; b. Generate at least one vegetation index indicating crop nutrient status based on image data from the plurality of image data over multiple dates; c. Assess the rate of change of the at least one vegetation index generated between each of the plurality of dates for the at least one vegetation index; d. Select at least one of the plurality of image data and the corresponding vegetation index based on the corresponding rate of change; e. Determine the crop nutrient status based on at least one selected vegetation index.
3. The method of claim 2, wherein assessing the rate of change of the at least one vegetation index further comprises averaging the vegetation index for each given image data with respect to at least a portion of the agricultural field and determining a corresponding difference between the averaged at least one vegetation index.
4. The method of claim 2 or 3, wherein multiple received image data indicating the crop nutrient status of the at least one agricultural field on multiple dates prior to the scheduled fertilizer application, and a single image indicating when the image data was taken, are selected based on an assessed rate of change, and the determination of the crop nutrient status is based on the selected images, and the crop nutrient status is adjusted based on the time difference between the timestamp of the selected images and the scheduled fertilizer application.
5. The method of claim 2 or 3, wherein the remote data comprising a plurality of image data further comprises selecting image data in the received image data wherein the number of valid pixels of a corresponding image data is higher than a predetermined threshold, wherein the validity of the pixels is received along with the image data.
6. The method of claim 1, wherein receiving weather data further comprises receiving historical precipitation data, and wherein iteratively updating the crop dry matter value further comprises adjusting the iteratively updated dry matter value based on the historical precipitation data.
7. The method according to any one of claims 1 to 3, wherein determining the variable fertilizer recommendation based on the adjusted crop nutrient status further comprises using at least one of the following additional agronomic parameters: growth stage, crop type, crop variety, mineralization potential, yield expectation, or weather forecast.
8. The method according to any one of claims 1 to 3, wherein the variable fertilizer is recommended for generating a machine-readable prescription map for controlling the fertilizer application system.
9. The method according to any one of claims 1 to 3, wherein arranging fertilizer application further comprises determining the application date and time based on at least one of: user input, received or generated weather forecasts, field crop data, predictions of phenological stages, and / or predictions of crop development based on crop growth.
10. A system including a communication unit for providing fertilizer recommendations, the system being configured to perform the method according to any one of claims 1 to 9.
11. The system (100) for providing fertilizer recommendations according to claim 10, comprising a communication unit, wherein the system (100) is configured to communicate with an agricultural device (300) via the communication unit, and wherein the agricultural device is configured to perform fertilizer application specific to a variable site for a determined crop.
12. A data processing apparatus comprising tools for performing the method according to any one of claims 1 to 9.
13. A computer-readable storage medium comprising instructions that, when executed by a computer system, cause the computer system to perform the method according to any one of claims 1 to 9.
14. A computer program product comprising instructions that, when executed by a computer, cause the computer to perform the method according to any one of claims 1 to 9.