Method for determining a productivity variability index of an agricultural area

Abstract

Disclosed is a computer-implemented method of determining a productivity variability index of an agricultural area. The method may comprise a step of providing a time-series of remotely sensed images of an agricultural area. The method may comprise a step of selecting from the time series of remotely sensed images of the agricultural area at least one subset of remotely sensed images based on a critical period for crop growth on the agricultural area. The method may comprise a step of transforming each subset of the remotely sensed images into a productivity image indicating a productivity distribution within the agricultural area to obtain at least one set of productivity images. The method may comprise a step of determining, based on the at least one set of productivity images, the productivity variability index of the agricultural area, wherein the productivity variability index indicates a degree of heterogeneity for the productivity distribution.

Description

[0001] METHOD FOR DETERMINING A PRODUCTIVITY VARIABILITY INDEX OF AN AGRICULTURAL AREA

[0002] TECHNICAL FIELD

[0003] The present disclosure generally relates to the field of precision farming, and more particularly to techniques for determining a productivity variability index of an agricultural area.

[0004] BACKGROUND

[0005] In the technical field of digital agriculture, it is desired to make farming more sustainable. Precision farming or precision agriculture is seen as one of the ways to achieve good sustainability and reducing environmental impact. One of the main challenges is to provide field treatment, soil treatment, seeding, plant treatment well targeted in the desired amount of treatment action and / or of agricultural product (application product I treatment product).

[0006] This is because productivity levels (crop biomass or crop yield) within an agricultural field can strongly vary in different parts of a field. Therefore, Variable Rate Applications (VRA) of crop inputs like seeds, fertilizers or crop protection products are a key element of precision farming and aim at applying different rates of each crop input product in different parts of an agricultural field. This makes sure that e.g., the dosage of a fertilizer corresponds to the nutrient demand or that the dosage of a fungicide corresponds to the disease infestation level in a certain part of the field.

[0007] In practice, however, VRA is implemented in a generic way which means that fields are subdivided into fixed numbers of treatment zones or some treatment zones independent of the actual inherent variability of the field. This has led to outcomes below expectations, as well as limited benefits and limited increase of harvest or yield.

[0008] It is therefore an objective of the present disclosure to provide a way of efficiently determine whether VRA is applicable to an agricultural area, thereby overcoming the above-mentioned disadvantages of the prior art at least in part and thus to provide a more robust, reliable, and efficient agricultural treatment.

[0009] SUMMARY OF THE DISCLOSURE

[0010] The objective is solved by the subject-matter defined in the independent claims. Advantageous modifications of embodiments of the present disclosure are defined in the dependent claims as well as in the description and the figures.

[0011] In general, the aspects of the present disclosure provide a way of quantifying the productivity variability within an agricultural area in order to determine whether an agricultural area is suitable for VRA or whether an agricultural area can equally be treated with a flat-rate / fixed application. For VRA suitable agricultural areas, the present disclosure further provides a way of determining the appropriate number of treatment zones and zone-specific operation configurations (e.g., treatment rates) for each treatment zone.

[0012] One aspect of the present disclosure relates to a computer-implemented method of determining a productivity variability index of an agricultural area. The method may comprise a step of providing a time-series of remotely sensed images of an agricultural area. The method may comprise a step of selecting from the time series of remotely sensed images of the agricultural area at least one subset of remotely sensed images based on a critical period for crop growth on the agricultural area. The method may comprise a step of transforming each subset of the remotely sensed images into a productivity image indicating a productivity distribution within the agricultural area to obtain at least one set of productivity images. The method may comprise a step of determining, based on the at least one set of productivity images, the productivity variability index of the agricultural area, wherein the productivity variability index indicates a degree of heterogeneity for the productivity distribution.

[0013] Based on productivity variability measurements (e.g., remotely sensed images) the productivity variability index indicating the heterogeneity for the productivity distribution within the agricultural area can be determined. The productivity variability index provides an efficient way of determining whether the agricultural area is suitable for VRA (e.g., for a high productivity variability index) or not (e.g., for a low productivity variability index). Based on the productivity variability index, a more precise way of operating the agricultural area is provided which allows an operation configuration to be adjusted accordingly.

[0014] Throughout the present disclosure, a productivity variability index can be understood as a metric which based on measurements (e.g., remotely sensed images) indicates a variability of crop growth (e.g., a crop growth distribution). A high variability index may show that a more individual or flexible operation on the agricultural area is required whereas a low variability index may show that a substantially fixed or homogenous operation can be used on the agricultural area.

[0015] In other words, detecting a homogeneous and / or unified distribution of a quality parameter on a specific field may be represented as a low variability index and / or a low production variability index. Such a low variability index indicates that a homogeneous treatment may be applied to the field. In this way a specific sophisticated treatment of single sub-areas may be prevented. Thus, such a low variability index may indicate that a simple treatment of the field may be possible, and a farmer may not have to provide specific machines that are of high cost and / or difficult to set up. A flat rate treatment may be seen as such a simple treatment. A treatment may be seeding, spraying of a liquid or a mechanical treatment to the field and / or the soil of the field.

[0016] Having a high variability index may indicate a heterogenous, irregular and / or patchy distribution of a quality parameter. In an example, a field may also be a combination of a plurality of fields located at different geographical locations and virtually considered as a large field. The quality parameter may be a metric indicating growth condition, growth quality and / or a growth performance factor for a crop. It may be considered that different fields are clustered according to their variability index and fields with a high variability index may be treated at a time where for example a sophisticated variable rate machine may be rented. In this way the variability index may support scheduling of the work.

[0017] In an example of a field, a combined field and / or a field cluster and having a high variability index it may be useful to treat sub-areas with a high growth performance differently than sub areas with a low growth performance. It may be desirable to treat high performance area with a higher treatment rate and / or treatment frequency than a low performance area. This may be the case when seeding is used as a treatment method and a higher seeding rate and / or a higher seeding density is applied at high performance areas compared to low performance areas.

[0018] In another example, like treating a field with a fertilizer, a low performance area may need a higher dosage, rate and / or application frequency than a high-performance area.

[0019] In one example the treatment may be targeted to align and / or harmonize the inhomogeneity in the field. In this case the treatment rate and / or treatment frequency may be reciprocal and / or reverse to the distribution of the quality parameter.

[0020] In another example, treatment may be targeted to use the inhomogeneity in the field and consequently the treatment rate and / or treatment frequency may be aligned to the distribution of the quality parameter.

[0021] In both cases of an inhomogeneous field the different sub-areas of the field may be treated as individual as possible. However, the physical dimensions of a treatment device may limit the smallest treatable area. In an example the size of nozzles and / or seeding units of a treatment device may determine how small the sub-areas may be for an individual treatment. The smaller the physical dimensions, the more individual sub-areas may be treated and the smaller the sub-areas may be dimensioned. It may be a desire to adapt the sub-areas as close as possible to the distribution of the quality parameter however, due to the physical dimensions of the treatment device only a zone wise treatment may be possible. In other words, the zone size may be substantially predetermined by the dimensions of the treatment device, but the zone size may be larger than the variation of the quality parameter within the zone. Consequently, the zone may be treated equally even a more delicate and / or finer treatment may be necessary. The dimensions of the treatment device may determine the resolution reachable by the treatment device and / or the zone size. Therefore, a good balance between zone size and treatment rate may be to be determined.

[0022] A large treatment device with for example large spraying patterns may have a coarser resolution and / or sampling pattern than a smaller treatment device with smaller physical dimensions. The smaller treatment device may better approach the shape of small performance areas than a larger dimensioned treatment device. In an example small treatment roboter may allow for well treating small areas on a field and may allow to work with a higher number of zones than larger dimensioned treatment devices.

[0023] According to another aspect of the present disclosure, each remotely sensed image may comprise a plurality of pixels and transforming a remotely sensed image into a productivity image may comprises a step of determining, for each pixel of the remotely sensed image, a crop growth proxy value for a corresponding pixel of the productivity image. The productivity distribution within the agricultural area may be represented by the distribution of the proxy values of the pixels of the productivity image.

[0024] This way, the remotely sensed images are transformed into a format (i.e., productivity image) which allows to draw conclusions about the productivity distribution based on the pixel values representing crop growth proxy values. According to another aspect of the present disclosure determining the productivity variability index of the agricultural area may comprise a step of generating for each set of productivity images an overall productivity image representative of all productivity images of the set of productivity images. Determining the productivity variability index of the agricultural area may be based on the at least one overall productivity image.

[0025] This way, the information underlying each productivity image may be bundled into a single overall productivity image which may be used as a representative for determining the productivity variability index.

[0026] Throughout the present disclosure, an "overall productivity image” may be understood as an image representative of all productivity images within a set of productivity images. The overall productivity image may be generated by averaging the proxy values of the pixels at the same location of a predefined number of remotely sensed image, in particular of each image of the subset of the remotely sensed images.

[0027] According to another aspect of the present disclosure the overall productivity image may comprises a plurality of pixels and generating the overall productivity image may comprises a step of determining, for each pixel of the plurality of pixels, a value depending on corresponding pixel values of each productivity image of the set of productivity images.

[0028] This way, the pixel values and thus the overall productivity image can be used as a representative for all productivity images within a set of productivity images. Performing the intermediary step of generating the overall productivity image and to determine the productivity variability index based on a single image (i.e., the overall productivity image) is more efficient with respect to computation effort than determining the productivity variability on all productivity images of a set of productivity images. The overall productivity image may eliminate the time component of the set of productivity images. According to another aspect of the present disclosure determining the productivity variability index of the agricultural area may further comprise a step of determining for the at least one overall productivity image a statistical metric. Determining the productivity variability index may be based on the at least one statistical metric.

[0029] According to another aspect of the present disclosure at least a first statistical metric for a first overall productivity image of a first set of productivity images and a second statistical metric for a second overall productivity image of a second set of productivity images may be determined. Determining the productivity variability index of the agricultural area may be based on an aggregation of the first and second statistical metric.

[0030] Aggregating the statistical metrics to determine the productivity variability index allows to merge information of different subsets of productivity images and thus different overall productivity images (e.g.., a first overall productivity image of a first subset of productivity images and a second overall productivity image of a second subset of productivity images).

[0031] According to another aspect of the present disclosure selecting the at least one subset of remotely sensed images based on the critical period for crop growth on the agricultural area may comprise a step of determining a field history comprising a plurality of time periods from seeding to harvesting on the agricultural area. Selecting the at least one subset of remotely sensed images based on the critical period for crop growth on the agricultural area may further comprise a step of determining, from the plurality of time periods, a time period indicative of maximum environmental stress exposure on crop of the agricultural area and / or a time period indicative of a final yield of crop on the agricultural area and / or a time period indicate of a maximum susceptibility of the crop on the agricultural area as the critical period for crop growth on the agricultural area.

[0032] This way, it is ensured that only these remotely sensed images are selected for determining the productivity variability index which are actually relevant for the productivity within the agricultural area and / or from the planned location and / or for the planned crop type. The selection criteria for the at least one subset of remotely sensed images may be at least one of the agricultural area, the location of agricultural area and / or the planned crop type.

[0033] The critical period may depend on the crop type and / or region. For corn in Brazil the critical period may be 8-10 weeks after sowing. In Germany for winter wheat it may be 28-30 weeks after sowing. Soya beans, barley, oil seed rape, sunflowers, and other crops all have their specific critical period.

[0034] According to another aspect of the present disclosure the method may further comprise a step of classifying the agricultural area into a variability category of a plurality of variability categories according to the productivity variability index of the agricultural area. Each variability category of the plurality of variability categories may be associated with a classification threshold. An amount of variability categories of the plurality of variability categories and / or the associated classification thresholds may depend on a set of operating characteristics of the agricultural area. According to another aspect of the present disclosure the set of operating characteristics of the agricultural area may comprise a location of the agricultural area, a crop type cultivated in the agricultural area, a crop variety cultivated in the agricultural area, soil properties of the agricultural area, environmental factors such as weather and / or climate, and / or task details, such as seeding, fertilization or crop protection, and / or product master data for a product to be applied.

[0035] This way, a flexible way (i.e., depending on the set of operating characteristics) of classifying the agricultural area into a variability category is provided. As operating characteristics of agricultural areas may vary heavily (e.g., depending on the location) using classification thresholds depending on the operating characteristics of the agricultural area ensures a meaningful classification result.

[0036] The classification threshold in an example may depend on the number of zones selected for a specific field, a combination of fields and / or a cluster of fields. In this way the threshold may depend on the physical dimensions of a treatment device.

[0037] According to another aspect of the present disclosure the method may further comprise a step of generating, based on the variability index of the agricultural area, a treatment map indicating an operation configuration for the agricultural area for an agricultural apparatus. The agricultural area may be highlighted within the treatment map according to the productivity variability index of the agricultural area. The operation configuration may comprise either a variable rate application or a fixed rate application for operating the agricultural apparatus on the agricultural area. In other words, the productivity variability index may allow to indicate for an agricultural area whether a variable rate application or a fixed rate application may be necessary. In an example a high productivity variability index may be an indication for a variable rate application (VRA) and a low productivity variability index may be an indication for fixed rate application and / or flat rate application, i.e. treatment with a constant rate.

[0038] The treatment map generated this way allows to adjust the operation configuration of an agricultural apparatus in an efficient manner. The treatment map may be seen as instructions for a treatment device, e.g. an agricultural apparatus.

[0039] Throughout the present disclosure, an "operation configuration” with respect to an agricultural apparatus may be understood as any configuration affecting the operation of an agricultural apparatus. The operation configuration may comprise a hardware equipment to be used for operating an agricultural area (e.g., a certain type of sprayer) and / or a certain way of configuring / using a hardware equipment (e.g., settings such as spot or flat spraying, seeding distance of a seeder, amount of fertilizer used etc.).

[0040] According to another aspect of the present disclosure in which the operation configuration comprises the variable rate application, the method may further comprise a step of dividing the agricultural area into a plurality of treatment zones according to the variability category of the agricultural area and a step of determining, for each treatment zone of the plurality of treatment zones, a zone-specific operation configuration for the agricultural apparatus. The operation configuration may further comprise the zone-specific operation configurations.

[0041] In case the variable application is applicable, the present method does further provide a way of determining the appropriate number of treatment zones and / or of the location, in particular of the borders of the zones. The agricultural area may then be divided into the determined number of treatment zones with each treatment zone having its zone-specific configuration, e.g. dosage, application rate, application threshold, and / or seeding rate. According to another aspect of the present disclosure generating the treatment map may further comprise a step of displaying the plurality of treatment zones within the agricultural area. Each treatment zone may be highlighted according to the zone-specific operation configuration.

[0042] In an example a layer of a map may be generated. One layer may indicate the productivity variability index for the complete agricultural area, another layer may show the zone distribution for the agricultural area.

[0043] This way, a farmer or user may be served with an improved representation of the productivity variability within an agricultural area based on which the farmer / user can either provide additional input (e.g., define different relevant crop seasons, define different start and end points of the critical periods, change the amount of variability categories, use different statistical metrics etc.) to re-execute the determining of the productivity variability index or can use the treatment map for guidance and operation of an agricultural apparatus. In another example the additional input may be derived from a database.

[0044] Another aspect of the present disclosure relates to a data-processing device or system, the device or system comprising a processor and a memory and being configured to perform the method of any one of the preceding aspects.

[0045] Another aspect of the present disclosure relates to a computer program or a computer-readable medium comprising a computer program, wherein the computer program comprises instructions which when executed by a computer cause the computer to perform the method of any one of the preceding aspects.

[0046] Another aspect of the present disclosure relates to a treatment map for an agricultural apparatus on an agricultural area, wherein the treatment map is generated according to aspects of the present disclosure and wherein the treatment map is configured to adjust an operation configuration of the agricultural apparatus.

[0047] Another aspect of the present disclosure relates to an agricultural apparatus comprising means for processing a treatment map according to the aspects of the present disclosure, wherein as a result of the processing of the treatment map an operation configuration of the agricultural apparatus is adjusted. In this way the method and / or the device and / or the system may be seen as a measurement method and measurement device and measurement system, respectively.

[0048] BRIEF DESCRIPTION OF THE DRAWINGS

[0049] The disclosure may be better understood by reference to the following drawings:

[0050] Fig. 1 : A flowchart of a method for determining a productivity variability index in accordance with embodiments of the present disclosure.

[0051] Fig. 2: An example of determining a productivity variability index in accordance with embodiments of the present disclosure.

[0052] Fig. 3: An exemplary visualization of productivity images in accordance with embodiments of the present disclosure.

[0053] Fig. 4: An exemplary visualization of a treatment region in accordance with embodiments of the present disclosure.

[0054] Fig. 5: An exemplary visualization of a treatment map for a heterogeneous agronomical area of medium productivity variability index in accordance with embodiments of the present disclosure.

[0055] Fig. 6: An exemplary visualization of a treatment map for a homogenous productivity variability index below a threshold in accordance with embodiments of the present disclosure.

[0056] DETAILED DESCRIPTION

[0057] In the following, representative embodiments illustrated in the accompanying drawings will be explained. It should be understood that the illustrated embodiments and the following descriptions refer to examples which are not intended to limit the embodiments to one preferred embodiment.

[0058] Fig. 1 illustrates a flowchart of a method 100 for determining a productivity variability index in accordance with embodiments of the present disclosure. The method 100 provides an approach for determining a productivity variability index based on data from previous crop seasons which allows to determine a more efficient way of operating an agricultural apparatus with respect to a certain use case (e.g., seeding or fertilization tasks). The method 100 may be applied to a plurality of agricultural areas of an agricultural field with the same target crop (e.g., corn) which is intended to grow. In case, an agricultural field comprises only one agricultural area, the agricultural field may be used as a synonym for the agricultural area. In an example a farmer may be responsible for a plurality of agricultural areas and / or fields which may be located in the same and / or different regions. A season may be understood as the time period from seeding to harvesting. A season may be a period smaller or equal to a year. Accordingly, when applying the method 100 to a plurality of agricultural areas of an agricultural field, a productivity variability index may be determined for each agricultural area based on which the operating efficiency within the agricultural field can be improved.

[0059] In step 102, a time-series of remotely sensed images of an agricultural area may be provided. Additionally, other maps such as a time-series of yield maps or high resolution soil maps of the agricultural area may be provided and fused with the remotely sensed images to generate the variability index. The time series may cover one or more years, preferably at least two years. The remotely sensed images may be obtained from a satellite or other suitable recording means (e.g., a drone). The remotely sensed images recorded this way may be stored within a suitable storage medium (e.g., a database) and provided once the present method is executed. For example, upon executing the present method, a request may be issued and the recorded images maybe loaded from the storage medium. Before step 104 is executed, the method may comprise a step of evaluating each image of the time series of remotely sensed images using an image processing technique (e.g., a heuristic or a machine learning model such as a CNN) to confirm that each image is usable and not impaired by clouds, cloud shadows, image errors or other issues or artifacts. In cases where a LAI may be used as growth proxy value an area covered by plant may be determined. If however this LAI value may be too high, e.g. over 45%, it may be an indication that artefacts in the image, a cloud, and / or a cloud shadow cover the area and the image cannot be used s for classification for the productivity variability index. For classification, a value between 0 - 10% may be assessed as homogenous, a value from 10%-20% may be assessed as medium, and a value 20% to 30% may be assessed as heterogenous and a value above 45% may be assessed as impaired.

[0060] A Leaf Area Index (LAI) may measure the total leaf area per unit ground area, expressed as one-sided leaf area (m2) per unit ground area (m2). It can be measured directly through field sampling techniques or indirectly via remote sensing techniques using multi-spectral data, which involve complex algorithms and models. In an example, a Machine Learning approach is used for determining the LAI. Such Machine Learning algorithm may be the XGBOOST regression model, which takes substantially all spectral information and satellite geometrical information like view-zenith angle and azimuth angle as input. The output includes canopy structure like average leaf angle, LAI, leaf area index, leaf optical information (chlorophyll, leaf water content, leaf dry matter), and other parameters. LAI shows the biomass and is helpful for understanding processes such as photosynthesis, transpiration, and energy balance in ecosystems. It is also important in modeling plant growth, ecosystem productivity, and understanding carbon and water cycles. In another example the Normalized Difference Vegetation Index (NDVI) is used. The NDVI may be derived from remote sensing data, such as satellite imagery or aerial photography, which measures reflectance in the red and near-infrared (NIR) bands. The NDVI is a dimensionless index calculated using the formula: NDVI = (NIR - Red) I (NIR + Red). NDVI provides a quick and indirect assessment of vegetation health, indicating crop density and chlorophyll content. High NDVI values suggest healthy, photosynthetically active vegetation, while low or negative values indicate sparse or stressed vegetation.

[0061] Even if LAI and NDVI are options to be used as basis for the variability index each parameter may be used for different embodiments and / or scenarios. LAI may allow for precise calculations of a variability index. The LAI increases when the crop is growing. NDVI may provide limited information such as growth density but may be easier to determine and may allow for a quicker calculation.

[0062] The selection of a useful parameter in an example may be part of the determination of the variability index method. In step 104, at least one subset of remotely sensed images is selected from the time series of remotely sensed images of the agricultural area based on a critical period for crop growth on the agricultural area.

[0063] Selecting the at least one subset of remotely sensed images based on the critical period for crop growth on the agricultural area may comprise a step of determining a field history comprising a plurality of time periods from seeding to harvesting on the agricultural area (i.e., relevant crop seasons). The field history may indicate which type of crop has been planted / grown on the agricultural area in a particular season or period in the past. Based on the field history, relevant crop seasons in one or more previous years may be determined for the agricultural area and the corresponding target crop. The critical period(s) within each relevant crop season for crop growth may then be selected. As explained below, said selection may be based on one or more factors such as environmental stress exposure (e.g., water stress, differences in soil conditions), crop yield and / or crop susceptibility. For example, if the target crop is corn, the critical period for corn growth are usually during flowering occurring around 8-10 weeks after seeding the corn.

[0064] The critical period may be gathered from historical data by statistical evaluation. In an example the critical period may be aligned with a calendar year. In another example the critical period may be aligned with a growth season. In a further example there may be a plurality of critical periods within a calendar year and / or within a period of 12 months. For example, the Brazilian region Goias may have within a period of 12 months a critical period in a growth season for corn from March until May and a second period from December to March with the longer growth season. In contrast, in Germany the growth season for corn may be only a single growth season within a period of 12 months from April until October, with the critical period lying in May to July. Consequently, the critical period may be dependent on the geographical location. For example, the Brazilian region Goias has a critical period for corn from March until May. However, the more southern located Brazilian region Parana has a critical period from March until June.

[0065] The statistical parameter for a statistical evaluation for an evaluation may be derived from data recorded in a data base, e.g. in a field managing tool, or official sources such as agricultural chambers.

[0066] Since the critical periods are crop and country specific and / or crop and location specific many combinations of specific crop types in specific countries, regions and / or location may be examined in order to provide variability index for each such location-crop-type combination. Even within a single country, the critical periods may be region specific. Thus, trials may be made for the location-crop-type combinations like corn, soy, cotton, sugarcane in Brazil, like corn, soy, sunflower, winter wheat in Argentina, like corn, soy, cotton in USA, like winter wheat, winter barley, oil seed rape, sugar beet, potato, corn in Germany and France and canola, winter wheat in Canada, like rice, soybean, sugarbeet in Japan. For each location-crop-type combination a productivity variability index may be provided. In this way a productivity variability index for cotton in Brazil, a productivity variability index for corn in Germany, a productivity variability index for soybean in Brazil, a productivity variability index for cotton in Brazil, a productivity variability index for corn in Argentina etc may be provided. The location may be provided with its specific latitude and longitude. In this way a location-crop-type productivity variability index may be generated. In other words, the productivity variability index for a substantially similar field may be different dependent on where the field may be located and / or what crop is grown on the specific field.

[0067] Selecting the at least one subset of remotely sensed images based on the critical period for crop growth on the agricultural area may further comprise a step of determining, from the plurality of time periods, a time period indicative of maximum environmental stress exposure on crop of the agricultural area and / or a time period indicative of a final yield of crop on the agricultural area and / or a time period indicate of a maximum susceptibility of the crop on the agricultural area as the critical period for crop growth on the agricultural area.

[0068] In step 106, each subset of the remotely sensed images is transformed into a productivity image indicating a productivity distribution within the agricultural area to obtain at least one set of productivity images.

[0069] The productivity distribution may be a crop growth proxy distribution such as a biomass distribution, a leaf area index distribution and / or a crop yield distribution. The productivity distribution may be represented as a pixel distribution or raster value distribution or zoned map.

[0070] Each remotely sensed image may comprise a plurality of pixels and transforming a remotely sensed image into a productivity image may comprise a step of determining, for each pixel of the remotely sensed image, a crop growth proxy value for a corresponding pixel of the productivity image. The productivity distribution within the agricultural area may be represented by the distribution of the proxy values of the pixels of the productivity image. For example, the crop growth proxy value may be a crop physiological parameter such as a Leaf Area Index (LAI) value, soil type (humus, silt, clay, sand), pH-value, biomass, nutrient content. The crop proxy values or pixels may have different values, e.g. dependent on the different productivity of the soil in a field. The higher the crop proxy values within an image of an agricultural area differ, the higher the productivity variable index may be (i.e., the productivity variability index may depend on the number of different crop growth proxy values). The variability productivity index may be represented as a percentage value indicating a number of different crop growth proxy values of the agricultural area compared to a maximum number of different growth proxy values. The maximum possible growth proxy values may depend on a region (e.g., geographical area) in which the agricultural area is located. It is also possible to represent the variability productivity index as an absolute number (e.g., number of proxy values) or in another suitable way (e.g., by a statistical metric).

[0071] Each pixel of the plurality of pixels may be associated with a geo coordinate within the agricultural area. Accordingly, determining a crop growth proxy value may be done for each geo coordinate within the agricultural area.

[0072] Before step 108 is executed, the method may comprise a step of evaluating each productivity image using an image processing technique (e.g., a heuristic or a machine learning model such as a CNN) to confirm that each image is usable and not impaired by clouds, cloud shadows, image errors or other issues or artifacts.

[0073] In step 108, the productivity variability index of the agricultural area is determined based on the at least one set of productivity images wherein the productivity variability index indicates a degree of heterogeneity for the productivity distribution. Additionally the productivity variability index may incorporate the time-series of yield data of the agricultural area. As explained above, the productivity variability index may be understood as a metric. The metric may be a coefficient of variation of the different crop growth proxy values appearing in the at least one set of productivity images.

[0074] Determining the productivity variability index of the agricultural area may comprise a step of generating for each set of productivity images an overall productivity image representative of all productivity images of the set of productivity images. Determining the productivity variability index of the agricultural area may be based on the at least one overall productivity image.

[0075] The overall productivity image may comprise a plurality of pixels and generating the overall productivity image may comprise a step of determining, for each pixel of the plurality of pixels, a value depending on corresponding pixel values of each productivity image of the set of productivity images. For example, the value depending on corresponding pixel values of each productivity image of the set of productivity images may be determined using an average of weighted average of the pixel values of each productivity image of the set of productivity images. Weighting may be based on a time at which the corresponding productivity images were taken (e.g., at the start, the middle or the end of critical periods).

[0076] Determining the productivity variability index of the agricultural area may further comprise a step of determining for the at least one overall productivity image a statistical metric. Determining the productivity variability index may be based on the at least one statistical metric. For example, the statistical metric may be a coefficient of variation or a quantile range of pixel values of the overall productivity image.

[0077] At least a first statistical metric for a first overall productivity image of a first set of productivity images and a second statistical metric for a second overall productivity image of a second set of productivity images may be determined. Determining the productivity variability index of the agricultural area may be based on an aggregation of the first and second statistical metric. For example, the aggregation may be an average or weighted average.

[0078] The method 100 may further comprise a step of classifying the agricultural area into a variability category of a plurality of variability categories according to the determined productivity variability index of the agricultural area. Each variability category of the plurality of variability categories may be associated with a classification threshold. An amount of variability categories of the plurality of variability categories and / or the associated classification thresholds may depend on a set of operating characteristics of the agricultural area. The amount of variability categories and / or the associated classification thresholds may be pre-defined based on the set of operating characteristics.

[0079] For example, defining the classification thresholds and / or variability categories may be done using a machine learning model trained to output a number of variability categories and / or (corresponding) classification thresholds based on a plurality of sets of operating characteristics of a plurality of agricultural areas on which the same type of crop is / was cultivated. Alternatively, or additionally, defining the classification thresholds and / or variability categories may be based on agronomic expert knowledge. The threshold may further consider the crop type, location and / or use case.

[0080] In another example, defining the classification thresholds and / or variability categories may be done using an unsupervised-learning model (e.g., a clustering method) which receives as input a plurality of sets of operating characteristics of a plurality of agricultural areas on which the same type of crop is / was cultivated, processes the plurality of sets to identify one or more clusters (e.g., the number of clusters may be predefined or dynamically determined) and outputs the classification thresholds and / or variability categories based on the identified one or more clusters (e.g., based on properties of each cluster such as the centroid or the like).

[0081] The set of operating characteristics of the agricultural area may comprise a location of the agricultural area, a crop type cultivated in the agricultural area, a crop variety cultivated in the agricultural area, soil properties of the agricultural area, environmental factors such as weather and / or climate, and / or task details, such as seeding, fertilization or crop protection, and / or product master data for a product to be applied.

[0082] For example, the variability categories may be defined as very low (i.e., very homogenous), low (i.e., homogenous), medium, high (i.e., heterogeneous) and / or very high (i.e., highly heterogeneous). For example, if an agricultural area is located in a region with highly varying environmental factors, the agricultural area may be classified into one of the rather higher variability categories such as high or very high, whereas an agricultural area which is located in a region with stable environmental factors may be classified into one of the rather lower variability categories such as very low or low. In another example, the threshold values associated with each variability category may vary for different agricultural areas based on the associated set of operating characteristics of the agricultural area.

[0083] The method 100 may further comprise a step of generating, based on the variability index of the agricultural area, a treatment map indicating an operation configuration for the agricultural area for an agricultural apparatus. Alternatively, in embodiments in which the agricultural area was classified into a variability category, the treatment map may be generated on the variability category.

[0084] The agricultural area may be highlighted within the treatment map according to the productivity variability index and / or the variability category of the agricultural area. A color code may be determined and the agricultural area or parts of it (i.e., treatment zones) maybe colored according to the determined color code.

[0085] The complete field map may be colored in the same color code related to the productivity variability index indicating one single parameter for the field. This parameter or color code provides the information whether the field may substantially be treated with flat rate and / or with variable rate and whether it may be worth taking the effort for a variable rate application.

[0086] The operation configuration may comprise either a variable application (e.g., variable rate application) or a fixed application (e.g., fixed rate application) for operating the agricultural apparatus on the agricultural area. Whether the operation configuration comprises the variable application or the fixed application may be based on the productivity variability index and / or the variability category of the agricultural area. For example, a high production variability index may be an indication for using a certain operation configuration (e.g., a variable rate seeding apparatus, a spot spraying apparatus) whereas a low production variability index may indicate to use another operation configuration (e.g., a flat-rate seeding apparatus, flat spraying apparatus). The indication whether a variable or fixed application should be used may be represented by a binary variable (e.g., 1 for variable and 0 for fixed) which value is set based on the productivity variability index.

[0087] In this way the treatment of the field may be planned by an operator and the appropriate treatment device may be selected. If the productivity variability index and / or the variability category indicates that the operation configuration comprises a variable application (i.e. , variable application is suitable for the corresponding agricultural area), the method may further comprise a step of dividing the agricultural area into a plurality of treatment zones according to the variability category of the agricultural area and a step of determining, for each treatment zone of the plurality of treatment zones, a zone-specific operation configuration for the agricultural apparatus. The operation configuration may further comprise the zone-specific operation configurations. The method may further comprise, prior to dividing, a step of determining a number of treatment zones for the agricultural area based on the productivity variability index and / or the variability category.

[0088] For example, an initial map comprising the agricultural area may be generated, a number of treatment zones may be determined based on the productivity variability index and / or the variability category and the agricultural area within the initial map may be divided accordingly to generate the treatment map with each treatment zone being associated with a corresponding zone-specific operation configuration.

[0089] Generating the treatment map may further comprise a step of displaying (or causing a display of) the plurality of treatment zones within the agricultural area. Each treatment zone may be highlighted according to the zone-specific operation configuration.

[0090] The productivity variability index and / or the variability category and the zone-specific operation configuration and / or zones may be realized as different layers of a map wherein the different layers may be combined to an overly map to a map of the agricultural area wherein the different layers may be switched on and off as desired.

[0091] For example, if the productivity variability index indicates that the agricultural area is homogenous (e.g., classified into the variability category "very low”), a fixed rate application may be used because productivity within the agricultural area is homogenous. In this example, a corresponding treatment map may comprise only one treatment zone for which the treatment map indicates an operation configuration comprising a fixed-rate application. As the treatment map only comprises one treatment zone (i.e., the agricultural area), the treatment map -when being displayed- may be highlighted with only one color code.

[0092] For example, if the productivity variability index indicates that the agricultural area is highly heterogenous (e.g., classified into the variability category "very high”), the variable rate application may be used because productivity within the agricultural area is heterogeneous. In this example, if the operation configuration comprises a variable rate application, the operation configuration may comprise for each treatment zone of the agricultural area a zone-specific operation configuration.

[0093] The number of treatment zones may depend on the variability category of the agricultural area. For example, an agricultural area classified into the variability category "very high” may be divide into 5 zones or more, an agricultural area classified into the variability category "medium” may be divided into 2 to 3 zones and an agricultural area classified into the variability category "low” may be divided into 2 zones. Accordingly, a corresponding treatment map of the agricultural area divided into the treatment zones may indicate an operation configuration which comprises a zone-specific operation configuration for each treatment zone. The treatment map -when being displayed- may be highlighted with different color codes (e.g., a different color code for each treatment zone and / or a different color code for variability category). The number of treatment zones and / or the corresponding geographical information (e.g., position etc.) for each zone may be determined based at least in part on the productivity variability index as explained with respect to Fig. 3.

[0094] In one example, an agricultural apparatus comprising means for processing a treatment map according to the aspects of the present disclosure, wherein as a result of the processing of the treatment map an operation configuration of the agricultural apparatus is adjusted is provided. The agricultural apparatus may be used to operate (e.g., physically apply an agricultural product) on an agricultural area. The agricultural apparatus may be a semiautomated or fully automated agricultural apparatus. The agricultural apparatus may comprise a treatment device or application device and a processing device for controlling the treatment device. The agricultural apparatus may be configured to traverse the agricultural field. The treatment device may be adapted for different purposes of treatment of a field. A treatment device may represent any device being configured to provide and / or spread seeds, plants, fertilizers, and / or herbicides onto the soil of an agricultural field. In general, an agricultural apparatus comprising a treatment device may be realized as a ground or an air vehicle, e.g. a rail vehicle, a tractor, as a robot, and / or as an UAV (Unmanned Aerial Vehicle) or drone. The treatment may be a chemical, biological and / or mechanical treatment of the field, in particular a treatment may be a treatment of the soil and / or of the plants growing on the field. The agricultural apparatus can be an autonomous or a non-autonomous apparatus.

[0095] In case, the treatment device is used to apply a first agricultural product and a second agricultural product at substantially the same time, the treatment device may comprise at least two product tanks. However, in an example, two treatment devices may be used at substantially the same time, each comprising at least one product tank for the first or second agricultural product. In case, the agricultural products are applied at different times, the treatment device(s) may comprise only one single product tank. A treatment device in an example may be used for spraying an application product to the field. For example, a spraying device, e.g. a smart sprayer may be configured to treat the weed, the insects and / or the pathogens. It may also be used to spray fertilizer to the field.

[0096] In another example the treatment device may be designed as a seeding device for distributing seeds on the field. In yet another example the treatment device is implemented as a harvesting device in order to collect the yield of a field.

[0097] The treatment in the agricultural area may be further based on ecological and economical rules. Applying ecological and economical rules may substantially be related to reducing the quantity of the applied product and results in the technical effect of being more sustainable. In the cloud platform the rules are executed in the ADE (Agronomical Decision Engine).

[0098] The agricultural apparatus and / or the treatment device is / are controlled by control data generated by that ADE. The way of providing control data at an output interface may be in form of a treatment map generated as described herein. The treatment map may comprise a dosage, an applied quantity of an application product and / or an instruction for a treatment related to a geographical coordinate.

[0099] It may be a goal of an ADE to provide an appropriate treatment schema to a farmer. The ADE may be adapted to prevent recommending a variable rate application if it is not necessary. Thus, the additional effort for variable rate application may be reduced.

[0100] A general approach to determine the number of zones may be based on a huge database of field data, e.g. stored in a field management platform and / or in a field management device. Those data may be used to determine a threshold to categorize the fields according to their heterogeneity and / or variability in relation to the huge amount of data stored in the field management database. The higher the heterogeneity and / or variability, the higher the number of zones. For example, a homogeneous field may be treated flat-rate, i.e. with the same rate for the complete field, thus it is one single zone field. Since agricultural machine-capabilities may depend on physical parameter like working width, latency, i.e. the delay before a process begins or "reaction time”, how quickly the agricultural treatment device, e.g. a seeder, spreader, sprayer .react to a change in the signal controlling the agricultural treatment device.

[0101] In other words, according to its physical dimensions, resolution and / or physical properties a treatment device may generate zones as the smallest unit to be treated. Within this substantially smallest working size or within this zone the rate of the treatment device may not be adapted. Zones are the result of the fact that actions like changing the seed rate, the amount of fertilizer, the dose rate may take some time that may be transformed by the moving velocity of the treatment device into a physical extension. And further by the breadth of the treatment device, e.g. the breadth of a nozzle spray pattern. Those physical dimensions in at least a two dimensional space may prevent a small application to the ground. The reaction to a change to a property of a field zone, footprint and / or grids may may limit the arial zone sizes. In this way it may not be possible to substantially reproduce the shape of the inhomogeneity of a productivity parameter and / or of a crop proxy value. Structures and / or variations of the crop growth proxies that are smaller than the zone size may not be treated on their fine shape. Sampling of such structures with the treatment device may hide their fine structures and may only let the zones appear. In one example 1 to 5 zones may be mapped to five categories, in particular to field variability categories.

[0102] An appropriate zone size may consider the dimensions of the treatment device. The zone size may also have impact to the number of zones that may be mapped to a specific variability index. As a general rule a large number of fields is stored in a database such as a field management database. Each of the fields comprise filed data. The field data may be a set of pixels with allocated crop growth proxy values. Statistical parameter like the coefficient of variation may be generated for all the fields in a specific level or location area, eg. fields in a region, fields in a country, or fields in a state.. Substantially any value for the crop growth proxy values may be available for the totality of pixels for a specific crop, location and / or critical period. In order to map a certain value range of crop growth proxy values to a specific zone threshold values are defined in order to quantize a certain value range to a specific zone. In an example a diagram with all growth proxy values for all fields in a specific region may be provided and a cluster algorithm may be used in order to find the threshold values for the zones so that substantially the same number of values are allocated to a zone. Verifying the generated threshold values with the physical dimensions of a treatment device may help to generate useful threshold values.

[0103] In an example the clustering algorithm may suggest using 5 different field variability categories and allocate threshold for them- For example a field variability index and / or productivity variability index of smaller then 0,16 (16%) may map to the field variability category of very low and / or very homogeneous and only 1 zone may be suggested, i.e. flat rate. A field variability index and / or productivity variability index of smaller then 0,22 (22%), i.e. between 16% and 22%, may map to the field variability category of low and / or homogeneous and 2 zones are suggested. A field variability index and / or productivity variability index of smaller then 0,30 (30%) may map to the field variability category of medium and / or moderate and 3 zones are suggested. A field variability index and / or productivity variability index of smaller then 0,40 (40%) may map to the field variability category of high and / or heterogenous and 4 zones are suggested. A field variability index and / or productivity variability index of smaller then 0,55 (55%) may map to the field variability category of very high and / or very heterogenous and may suggest to use 5 zones for that field. The number of zones for the totality of fields in the field management device may be lower or higher than the 5 zones in other regions.

[0104] In other words, the generated field variability index and / or productivity variability index of a single field may be compared to totality of fields up to a certain level or in the specific country, or state, or region. In this way the field variability index and / or productivity variability index may be seen as a relative parameter. If an individual field, a plurality of fields of a single farmer and / or a cluster of fields may have a value smaller than 40% for that specific field and this value may be compared with the field variability index and / or productivity variability index of the totality of fields in this level, region, state or country a number of 4 zones may be derived for that individual field. The size of the 4 zones for the individual field may differ from the thresholds used for assessing the totality of fields in the field management platform.

[0105] In an example dependent on the level different numbers of zones are allocated for the same field. A field that may have 4 zones on country level may have 5 zones on region level. The field variability index and / or productivity variability index may indicate how large the differences of performance in a field are and whether a flat and / or a variable rate application may be useful to treat such a field with corresponding differences in field performance.

[0106] To distinguish the variability of fields, different levels of granularity may be used. In an example, the data to determine the categories may have a specific level of detail. The data to determine the categories may be level 1 and / or country-wide, i.e. for determining the categories for the variability index, the field variability index or productivity variability index only fields are used located in a specific country. This level 1 may represent the coarsest level.

[0107] In a further example the data to determine the categories may be level 2 and / or state-wide i.e. for determining the categories for the variability index, the field variability index or productivity variability index only fields are used located in a specific state. This level 2 may represent a medium level.

[0108] In another example the data to determine the categories may may be level 3 and / or region-wide i.e. for determining the categories for the variability index, the field variability index or productivity variability index only fields are used located in a specific region. Level 3 may be the finest level and would rate the field of interest within the given regional geography. In other words, the field of interest by applying the different levels of details may be compared to a larger number of fields and / or to more distant fields the lower the number of level. In this way the local character of the field may be considered. In yet other words, the perimeter around the field may be considered to a more far and / or nearer degree. For the different levels may a different number of fields stored in a database such as a field management database, which may be relevant to be considered.

[0109] If the same farmer may have another field where 3 zones may be useful the firstly described field may still have 4 zones. With each of these maximum number of zones for a specific field the application and / or treatment map may be generated such that the 3 zone field may distinguishes three color codes and the 4 zone field may distinguishes four color codes.

[0110] In a further example the field variability index may also be used to rate the general productivity of a field. If in a specific example, two fields in a region are available and each of the fields may have a determined a high field variability index and consequently five zones will be provided for each of the fields. Even both fields fall under the category "high field variability”, and each field may have a high field variability index allocated, and thus five zones allocated, the first of the two fields may have high biomasses, i.e. a high LAI values, in the different zones and the second field may have low biomasses, i.e., low LAI values, in the different zones. This may be the result of the fact that the thresholds are a relative factor. The thresholds used for identifying the classification of the field variability index and allocating the number of zones may be based on the relative variability and / or on a statistical distribution of the variability in the huge number of fields however may be substantially independent from the identified absolute LAI value. The field variability index and / or productivity variability index may indicate whether VRA (Variable Rate Application) or flat application may be useful. The variability index and / or productivity variability index may be used as an input parameter for other methods which use the provided number of zones and / or variability index and determines the inputs such as seed rates, or the fertilizer amounts, or the spraying dosages, or the mechanical treatments, that may be necessary for each of the zones and / or geographic locations within a field. Input may also be the quantity of spaying material such as herbicide, fungicide, pesticide and / or the quantity of fertilizer and / or the seeding material.

[0111] The other methods which may be used for determining the inputs for the specific zone may be adapted to provide the input to be applied to a zone as an application map. Dependent on the scenario the inputs for the zones may be a rate and / or frequency, e.g. a seeding rate, a liquid spraying rate and / or a mechanical treatment rate as well as corresponding material, liquid or product. The algorithm used for determining the application rate for each zone, determining the input for each zone and / or determining the application map uses the provided number of zones and link it with the relevant input for each zone, e.g. an application rate. Thus, in the example of the two fields each having allocated 5 zones, each having a high variability and / or each having a high variability index the first field may have higher inputs into each zone than the second field.

[0112] The input in an example may be a seeding rate and thus a high biomass zone has a high seeding rate and / or corresponding material. In another example the input may be a fertilizer application rate and thus a zone with a high biomass may have a low input. In other words the input may be proportional to the quantity of biomass or opposite proportional to the quantity of biomass dependent on the respective application.

[0113] Fig. 2 illustrates an example 200 of determining a productivity variability index in accordance with embodiments of the present disclosure.

[0114] A time-series of remotely sensed images 202 of an agricultural area is provided from which at least one subset of remotely sensed images 204 is selected based on a critical period for crop growth on the agricultural area and / or based on the planned target crop type, e.g. corn, barley, wheat, oil seed, rape. The selection may be made by a filter filtering the images based on their meta data, e.g. Exchangeable Image File Format (EXIF). As one can see, the time-series of remotely sensed images 202 of this example is taken over 5 subsequent years (e.g., 2018-2022). While the remotely sensed images of this example were taken by a satellite, it is to be understood that other means are also suitable for taking the remotely sensed images as long as they provide a birds-eye view image of the agricultural area. For example, drones or aircrafts may be used to take the remotely sensed images. It may also be possible that images taken by a combination of means are used for providing the remotely sensed images (e.g., a combination of aircraft(s), satellite(s) and drone(s)). As explained with respect to Fig.1 , selecting the at least one subset of remotely sensed images 204 may comprise determining a field history comprising information about a crop history and a plurality of time periods from seeding to harvesting on the agricultural area. As one can see, the critical period may relate to a certain period during a relevant crop season. The critical period may correspond to a critical phenological period.

[0115] Once the at least one subset of remotely sensed images 204 is selected, said subset 204 may be transformed into a corresponding set of productivity images 206. For this purpose, each image of the at least one subset of remotely sensed images 204 may be transformed into a corresponding productivity image. Each productivity image indicates a productivity distribution within the agricultural area at the corresponding point in time.

[0116] The at least one set of productivity images may then further be aggregated to generate an overall productivity image representative of all productivity images of the corresponding set. For example, the set of productivity images 206 comprises four productivity images which may be aggregated to generate one overall productivity image representative of all four productivity images. Generating the overall productivity images may be done as explained with respect to Fig 1. Determining the productivity variability index may then be done by calculating a corresponding productivity variability score. Said score may for example be a statistical metric such as a coefficient of variation or a quantile range of pixel values of the overall productivity image.

[0117] As one can see, in case another set of productivity images 208 was generated for which again an overall productivity image (the overall productivity image is not shown in Fig. 2) was generated, the productivity variability index may be determined based on an aggregation of both overall productivity images (i.e., a first overall productivity image representative of all productivity images of the set 206 and a second overall productivity image representative of all productivity images of the set 208).

[0118] For example, a first statistical metric (i.e., a first productivity variability score) for the first overall productivity image and a second statistical metric (i.e., a second productivity variability score) for the second overall productivity image may be determined. Both metrics may then be aggregated (e.g., average or weighted average) to determine the productivity variability index. Weighting may be done based on a time at which the images were taken (e.g., previously taken images may be weighted stronger compared to images which were taken further into the past). Fig. 3 illustrates an exemplary visualization 300 of productivity images 302-304 in accordance with embodiments of the present disclosure.

[0119] The productivity images 302-304 are represented as raster grids wherein each grid cell represent a pixel of the remotely sensed original image. The greyscale applied to the raster grids represent the productivity achieved at the corresponding position associated with the corresponding pixel within the agricultural field. A dark grid cell indicates a high productivity whereas a light grey grid cell indicates a low productivity. As one can see, the productivity image 302 shows a heterogenous agricultural area (e.g., a field) with a high productivity variability with a plurality of different values 302', 302”, 302'” within comparison area 301. The productivity image 304 shows a substantially homogenous agricultural area (e.g., a field) with a low productivity variability indicated by less different values 304', 304” within comparison area 301. The area 301 may be an agricultural area and / or agricultural field.

[0120] This productivity difference is also illustrated by the corresponding probability distribution of the pixel values of the corresponding productivity images 302-304. For the productivity image 302 of a heterogenous field, the variance of the probability distribution is larger compared to the variance of the probability distribution of the productivity image 304 of a homogenous field. In addition, the steepness of a curve (e.g., represented by the excess kurtosis) belonging to a homogenous field may have a negative value and / or be steeper compared to the curve belonging to a heterogenous field. In addition, the inter quartile range may be small (homogenous) or large (heterogenous). An inter quartile range may be considered small if the inter quartile range is smaller or equal to 1 .35* the standard deviation of the distribution. A inter quartile range may be considered large if the inter quartile range is larger than 1 .35* the standard deviation of the distribution.

[0121] The number of treatment zones and / or corresponding geographical information for each treatment zone may be determined based on the pixel distribution within the productivity images. Adjacent pixel regions where pixel values indicate the same productivity may be clustered and considered as one treatment zone. As one can see from the pixel values highlighted within productivity image 302, three treatment zones may be used for the agricultural area associated with the productivity image 302.

[0122] In contrast, the pixels of productivity image 304 all have substantially the same value as they are highlighted in dark grey. Accordingly, the agricultural area associated with productivity image 304 may comprise only one treatment zone.

[0123] The geographical information for each treatment zone may then be determined based on the geographical positions associated with each pixel of the images within the agricultural area.

[0124] Fig. 4 illustrates an exemplary visualization of a treatment region 400 comprising six agricultural areas with their productivity variability index in accordance with embodiments of the present disclosure. The agricultural areas may belong to the same farmer and / or to the same farm and may be shown to the user and / or farmer in a field management application. In the present example, the treatment region 400 was generated for a geographic map comprising six agricultural areas 420a-b, 404a-b and 406a-b. For each of the six agricultural areas, the method according to the aspects of the present disclosure was performed to determine a productivity variability index based on which the corresponding treatment map was generated. As can be seen, for the agricultural areas 402a-b a high productivity variability index was determined. Accordingly, the agricultural areas 402-b were highlighted according to a determined color code (e.g., dark grey; other color codes are possible). Based on the high productivity variability index and the corresponding highlighting, it can be followed that the agricultural areas 402 a-b are very heterogenous. Thus, said agricultural areas may each be divided into a plurality of treatment zones (e.g., 5 zones) wherein each treatment zone has its own zone-specific operation configuration (the treatment zones of the treatment map are not shown in Fig. 4).

[0125] As can be seen, for the agricultural areas 404a-b a medium productivity variability index was determined. Accordingly, the agricultural areas 404-b were highlighted according to a determined color code (e.g., grey; other color codes are possible). Based on the medium productivity variability index and the corresponding highlighting, it can be followed that the agricultural areas 404 a-b are medium. Thus, said agricultural areas may each be divided into a plurality of treatment zones (e.g., 3 zones) wherein each treatment zone has its own zone-specific operation configuration (the treatment zones of the treatment map are not shown in Fig. 4).

[0126] As can be seen, for the agricultural areas 406a-b a low productivity variability index was determined. Accordingly, the agricultural areas 406a-b were highlighted according to a determined color code (e.g., light grey; other color codes are possible). Based on the low productivity variability index and the corresponding highlighting, it can be followed that the agricultural areas 406 a-b are homogenous because the low productivity variability index lays below a predefined threshold. Thus, said agricultural areas may not be divided into a plurality of treatment zones but may instead represent only 1 homogenous treatment zone for which a fixed application and thus only a single operation configuration can be used.

[0127] In order to classify agricultural areas into homogenous or heterogenous in particular low heterogenous, medium heterogeneous, and high heterogenous, a threshold may be provided. This decision may depend on the crop type, in particular the target crop type, the location and / or the use case, e.g. spraying, seeding, fertilizing, applying of plant protection products. If an agricultural area is assessed as heterogenous, i.e. above a predefined threshold for the productivity variability index the number of zones may be determined by determining the absolute number of different crop growth proxy value that have been used in order to determine the productivity variability index. Different crop growth proxy values may be pH-value, nutrition content, soil type, biomass value, soil moisture. Then the minimum and maximum value may be determined and divided by the absolute number crop growth proxy value. In another example where no minimum value may exist, an allocation to a specific category may be made.

[0128] In another example the productivity variability index may be a percentage value indicating the current number of different crop growth proxy values compared to the maximum possible different number of growth proxy values. The maximum possible growth proxy values may depend on a region of the corresponding agricultural area. VRA application may involve more effort than flat application. Thus, the productivity variability index may support a user for deciding about the treatment for each of the user's agricultural area just by analyzing the visualization of the treatment region 400. For agricultural areas identified as useful to apply VRA, a detailed planning may be made by deriving from the productivity variability index a treatment map (see Fig. 5).

[0129] The decision whether to use VRA or flat application is made on a field level, i.e. for each agricultural area independently from another agricultural area. In order to derive the number of treatment zones in one example the soil type and / or a geophysical measurement indicating differences in soil may be used. For example, if the agricultural area comprises 3 treatment zone types a moderate or low productivity variability index may be determined. The treatment zones then may be allocated to each of the available soil type and / or the variations in silt, sand, and clay contents and / or variations in the geophysical measurement values.

[0130] Fig. 5 illustrates an exemplary visualization of a treatment map 404'a for a medium productivity variability index in accordance with embodiments of the present disclosure. The treatment map 404'a may relate to the agricultural area 404a of Fig. 4. Fig. 5 shows an example with three treatment zones 502. A first treatment zone is indicated as the blank area, a second treatment zone is indicated by the vertical dashed area and a third treatment zone is indicated by the horizontal dashed area This configuration of 3 zones may be allocated to a medium productivity variability index. A treatment map with five or more zones may be allocated to a high heterogenous productivity variability index (the five zone configuration is not shown in Fig. 5).

[0131] Accordingly, if an agricultural apparatus is operated using said treatment map 404'a, an operation configuration of the agricultural apparatus may be adjusted according to the zone-specific operation configurations for each treatment zone 502 within the agricultural area 404a. For example, the agricultural apparatus may comprise sensing means (e.g., GPS sensors) which allow to determine a current position of the agricultural apparatus. If processing means of the agricultural apparatus determine that the current position of the agricultural apparatus is within or is about to enter a specific treatment zone 502 of the agricultural area 404'a, the operation configuration of the agricultural apparatus may be adjusted according to the zone-specific operation configuration.

[0132] Fig. 6 illustrates an exemplary visualization of a treatment map 406'a for a homogeneous productivity variability index below a threshold in accordance with embodiments of the present disclosure. The treatment map 406'a may relate to the agricultural area 406a of Fig. 4. Fig. 6 shows an example with one single treatment zone 602 for the complete agricultural area 406a (as indicated by the blank area) which may be allocated to a homogeneous productivity variability index and may result in a flat rate treatment, e.g. flat rate seeding, spraying or fertilizing, Accordingly, if an agricultural apparatus is operated using said treatment map 406'a, an operation configuration of the agricultural apparatus may be adjusted only once according to a fixed operation configuration during operation within the agricultural area 406a. As used herein the term "and / or” includes any and all combinations of one or more of the associated listed items and may be abbreviated

[0133] Although some aspects have been described in the context of an apparatus, it is clear that these aspects also represent a description of the corresponding method, where a block or device corresponds to a method step or a feature of a method step. Analogously, aspects described in the context of a method step also represent a description of a corresponding block or item or feature of a corresponding apparatus.

[0134] Embodiments of the present disclosure may be implemented on a computer system. The computer system may be a local computer device (e.g., personal computer, laptop, tablet computer or mobile phone) with one or more processors and one or more storage devices or may be a distributed computer system (e.g., a cloud computing system with one or more processors and one or more storage devices distributed at various locations, for example, at a local client and / or one or more remote server farms and / or data centers). The computer system may comprise any circuit or combination of circuits. In one embodiment, the computer system may include one or more processors which can be of any type. As used herein, processor may mean any type of computational circuit, such as but not limited to a microprocessor, a microcontroller, a complex instruction set computing (CISC) microprocessor, a reduced instruction set computing (RISC) microprocessor, a very long instruction word (VLIW) microprocessor, a graphics processor, a digital signal processor (DSP), multiple core processor, a field programmable gate array (FPGA), or any other type of processor or processing circuit. Other types of circuits that may be included in the computer system may be a custom circuit, an application-specific integrated circuit (ASIC), or the like, such as, for example, one or more circuits (such as a communication circuit) for use in wireless devices like mobile telephones, tablet computers, laptop computers, two-way radios, and similar electronic systems. The computer system may include one or more storage devices, which may include one or more memory elements suitable to the particular application, such as a main memory in the form of random-access memory (RAM), one or more hard drives, and / or one or more drives that handle removable media such as compact disks (CD), flash memory cards, digital video disk (DVD), and the like. The computer system may also include a display device, one or more speakers, and a keyboard and / or controller, which can include a mouse, trackball, touch screen, voice-recognition device, or any other device that permits a system user to input information into and receive information from the computer system.

[0135] Some or all of the method steps may be executed by (or using) a hardware apparatus, like for example, a processor, a microprocessor, a programmable computer or an electronic circuit. In some embodiments, some one or more of the most important method steps may be executed by such an apparatus.

[0136] Depending on certain implementation requirements, embodiments of the present disclosure can be implemented in hardware or in software. The implementation can be performed using a non-transitory storage medium such as a digital storage medium, for example a floppy disc, a DVD, a Blu-Ray, a CD, a ROM, a PROM, and EPROM, an EEPROM or a FLASH memory, having electronically readable control signals stored thereon, which cooperate (or are capable of cooperating) with a programmable computer system such that the respective method is performed.

[0137] Therefore, the digital storage medium may be computer readable.

[0138] Some embodiments according to the present disclosure comprise a data carrier having electronically readable control signals, which are capable of cooperating with a programmable computer system, such that one of the methods described herein is performed.

[0139] Generally, embodiments of the present disclosure can be implemented as a computer program product with a program code, the program code being operative for performing one of the methods when the computer program product runs on a computer. The program code may, for example, be stored on a machine-readable carrier.

[0140] Other embodiments comprise the computer program for performing one of the methods described herein, stored on a machine-readable carrier.

[0141] In other words, an embodiment of the present disclosure is, therefore, a computer program having a program code for performing one of the methods described herein, when the computer program runs on a computer.

[0142] A further embodiment of the present disclosure is, therefore, a storage medium (or a data carrier, or a computer- readable medium) comprising, stored thereon, the computer program for performing one of the methods described herein when it is performed by a processor. The data carrier, the digital storage medium or the recorded medium are typically tangible and / or non-transitory. A further embodiment of the present disclosure is an apparatus as described herein comprising a processor and the storage medium.

[0143] A further embodiment of the present disclosure is, therefore, a data stream or a sequence of signals representing the computer program for performing one of the methods described herein. The data stream or the sequence of signals may, for example, be configured to be transferred via a data communication connection, for example, via the internet. A further embodiment comprises a processing means, for example, a computer or a programmable logic device, configured to, or adapted to, perform one of the methods described herein.

[0144] A further embodiment comprises a computer having installed thereon the computer program for performing one of the methods described herein.

[0145] A further embodiment according to the present disclosure comprises an apparatus or a system configured to transfer (for example, electronically or optically) a computer program for performing one of the methods described herein to a receiver. The receiver may, for example, be a computer, a mobile device, a memory device or the like. The apparatus or system may, for example, comprise a file server for transferring the computer program to the receiver. In some embodiments, a programmable logic device (for example, a field programmable gate array) may be used to perform some or all of the functionalities of the methods described herein. In some embodiments, a field programmable gate array may cooperate with a microprocessor in order to perform one of the methods described herein. Generally, the methods are preferably performed by any hardware apparatus.

Claims

CLAIMS1. A computer-implemented method (100) of determining a productivity variability index of an agricultural area, the method comprising: providing (102) a time-series of remotely sensed images (202) of an agricultural area; selecting (104) from the time series of remotely sensed images (202) of the agricultural area at least one subset of remotely sensed images (204) based on a critical period for crop growth on the agricultural area; transforming (106) each subset of the remotely sensed images (204) into a productivity image (302, 304) indicating a productivity distribution within the agricultural area to obtain at least one set of productivity images; determining (108), based on the at least one set of productivity images (302, 304), the productivity variability index of the agricultural area, wherein the productivity variability index indicates a degree of heterogeneity for the productivity distribution.

2. The method of any one of the preceding claims, wherein each remotely sensed image of the time-series of remotely sensed images (202) comprises a plurality of pixels and wherein transforming a remotely sensed image into a productivity image comprises: determining, for each pixel of the remotely sensed image, a crop growth proxy value for a corresponding pixel of the productivity image; wherein the productivity distribution within the agricultural area is represented by a distribution of the proxy values of the pixels of the productivity image.

3. The method of any one of the preceding claims, wherein determining the productivity variability index of the agricultural area comprises: generating for each set of productivity images an overall productivity image representative of all productivity images of the set of productivity images; wherein determining the productivity variability index of the agricultural area is based on the at least one overall productivity image.

4. The method of claim 3, wherein the overall productivity image comprises a plurality of pixels and wherein generating the overall productivity image comprises: determining, for each pixel of the plurality of pixels, a value depending on corresponding pixel values of each productivity image of the set of productivity images.

5. The method of any one of claims 3 or 4, wherein determining the productivity variability index of the agricultural area further comprises:determining for the at least one overall productivity image a statistical metric; wherein determining the productivity variability index is based on the at least one statistical metric.

6. The method of claim 5, wherein at least a first statistical metric for a first overall productivity image of a first set of productivity images and a second statistical metric for a second overall productivity image of a second set of productivity images is determined, and wherein determining the productivity variability index of the agricultural area is based on an aggregation of the first and second statistical metric.

7. The method of any one of the preceding claims, wherein selecting the at least one subset of remotely sensed images based on the critical period for crop growth on the agricultural area comprises: determining a field history comprising a plurality of time periods from seeding to harvesting on the agricultural area; and determining, from the plurality of time periods, a time period indicative of maximum environmental stress exposure on crop of the agricultural area and / or a time period indicative of a final yield of crop on the agricultural area and / or a time period indicate of a maximum susceptibility of the crop on the agricultural area as the critical period for crop growth on the agricultural area.

8. The method of any one of the preceding claims further comprising: classifying the agricultural area into a variability category of a plurality of variability categories according to the productivity variability index of the agricultural area; wherein each variability category of the plurality of variability categories is associated with a classification threshold and wherein an amount of variability categories of the plurality of variability categories and / or the associated classification thresholds depend on a set of operating characteristics of the agricultural area.

9. The method of claim 8, wherein the set of operating characteristics of the agricultural area comprises: a location of the agricultural area, a crop type cultivated in the agricultural area, a crop variety cultivated in the agricultural area, soil properties of the agricultural area, environmental factors such as weather and / or climate, and / or task details, such as seeding, fertilization or crop protection, and / or product master data for a product to be applied.

10. The method of any one of the preceding claims, further comprising: generating, based on the variability index of the agricultural area, a treatment map indicating an operation configuration for the agricultural area for an agricultural apparatus ; wherein the agricultural area is highlighted within the treatment map according to the productivity variability index of the agricultural area; andwherein the operation configuration comprises either a variable rate application or a fixed rate application for operating the agricultural apparatus on the agricultural area.11 . The method of any one of the claims 9-10, wherein the operation configuration comprises the variable rate application; and wherein the method further comprises: dividing the agricultural area into a plurality of treatment zones according to the variability category of the agricultural area; and determining, for each treatment zone of the plurality of treatment zones, a zone- specific operation configuration for the agricultural apparatus; wherein the operation configuration further comprises the zone-specific operation configurations.

12. The method of claims 10 and 11, wherein generating the treatment map further comprises: displaying the plurality of treatment zones within the agricultural area; wherein each treatment zone is highlighted according to the zone-specific operation configuration.13 A data-processing device or system, the device or system comprising a processor and a memory and being configured to perform the method of any one of the preceding claims.

14. A computer program or a computer-readable medium having stored there on a computer program, the computer program comprising instructions which when executed by a computer cause the computer to perform the method of any one of the claims 1-12.

15. A treatment map (404' a, 406'a) for an agricultural apparatus on an agricultural area, wherein the treatment map is generated according to any one of the claims 10-12; and wherein the treatment map is configured to adjust an operation configuration of the agricultural apparatus.

Images