Crop rotation-based computer-implemented method for estimating agricultural product consumption in a geographic region

Abstract

The present invention relates to, among other elements, a computer-implemented method for estimating consumption of agricultural products in a geographical region by using historical satellite imaging data for the geographical region and a crop rotation model for classifying crop rotation patterns in the geographical region, and also to the use of historical satellite imaging data for the geographical region in such a method, a system for estimating consumption of agricultural products in a geographical region configured to perform such a method, a computer program element having instructions configured to perform the steps of the method when executed on one or more computing nodes, and a computer-readable medium having the computer program element stored thereon.

Description

[Technical Field]

[0001] The present disclosure relates to a computer-implemented method for estimating consumption of agricultural products in a geographic region, the use in said method of historical satellite imagery data for the geographic region for a previous growing season and at least one additional previous growing season, a system, computer program elements, and computer-readable media for estimating consumption of agricultural products in a geographic region. These and further aspects are reflected in the claims and the following specification. All embodiments, preferences, and aspects described herein for the computer-implemented method are also disclosed in relation to the use of historical satellite imagery data, systems, computer program elements, other computer-implemented methods, and computer-readable media according to the present invention. [Background technology]

[0002] In agriculture, farmers need to apply or spread many agricultural products (e.g., seeds, pesticides, fertilizers, etc.) on fields at specific times or time frames. Therefore, it is very important that the necessary agricultural products can be delivered to farmers at these specified times. However, the production and transportation of many of these agricultural products require a specific lead time before they can be delivered to farmers. For this reason, manufacturers, suppliers, and subcontractors in the production, purchasing, storage, and other agricultural logistics subareas must estimate the required quantities of agricultural products needed by farms in specific periods to ensure that the corresponding quantities of needed agricultural products can be provided. Until now, the required quantities of agricultural products have often been estimated based on manufacturers' years of experience. However, in practice, this also leads to deviations between the estimated quantities and the actual required quantities, which can be a significant disadvantage in cases of overproduction of agricultural products that cannot be easily stored and / or cannot be stored without incurring high costs. In addition, if the estimate is too low, supply bottlenecks may occur. It is further necessary to have accurate estimates as soon as possible in order to initiate appropriate steps to match supply with upcoming demand.

[0003] It has therefore been determined that there is a further need to provide a means for estimating agricultural product consumption in a geographical area as accurately and quickly as possible.

[0004] It is therefore an object of the present invention to provide means, e.g. a method, a system etc., for estimating consumption of agricultural products in a geographical area as early as possible and in an accurate way before demand for agricultural products arises. These and other objects that will become apparent on reading the following description are solved by the subject matter of the independent claims. The dependent claims refer to preferred embodiments of the invention. Summary of the Invention [Means for solving the problem]

[0005] A first aspect of the present disclosure relates to a computer-implemented method for estimating consumption of agricultural products in a geographic region, the method comprising: providing (110) historical satellite imaging data for the geographic region for a previous growing season and at least one additional past growing season; determining (120) historical crop classification data from at least historical satellite imagery data, the historical crop classification data including information regarding crop types grown in fields within the geographic region for at least two past growing seasons; providing (130) a crop rotation model for classifying crop rotation patterns in a geographic region, the model being based at least on historical crop classification data for the geographic region; determining (140) a crop rotation pattern for fields within a geographic region based on historical crop classification data for the fields by using a crop rotation model; determining (150) future crop data including information regarding areas of the geographic region that will be cultivated with particular crops in the current or next growing season based at least on the historical crop classification data for the field by using the crop rotation pattern for the field; providing (160) a product consumption model for agricultural products configured to estimate consumption of the agricultural products based at least on future crop data; using a product consumption model to provide an estimate of agricultural product consumption in a geographic region based at least on future crop data (170); Includes.

[0006] A second aspect of the present disclosure relates to the use of satellite imaging data relating to a geographic region for the previous growing season and at least one further previous growing season in the method defined above.

[0007] A third aspect of the present disclosure relates to a system for estimating consumption of agricultural products in a geographic region, the system comprising: one or more computing nodes; and one or more computer-readable media that, when executed by the one or more computing nodes, provides the system with: providing (110) historical satellite imaging data for the geographic region for a previous growing season and at least one additional past growing season; determining (120) historical crop classification data from at least historical satellite imagery data, the historical crop classification data including information regarding crop types grown in fields within the geographic region for at least two past growing seasons; providing (130) a crop rotation model for classifying crop rotation patterns in a geographic region, the model being based at least on historical crop classification data for the geographic region; determining (140) a crop rotation pattern for fields within a geographic region based on historical crop classification data for the fields by using a crop rotation model; determining (150) future crop data including information regarding areas of the geographic region that will be cultivated with particular crops in the current or next growing season based at least on the historical crop classification data for the field by using the crop rotation pattern for the field; providing (160) a product consumption model for agricultural products configured to estimate consumption of the agricultural products based at least on future crop data; using a product consumption model to provide an estimate of agricultural product consumption in a geographic region based at least on future crop data (170); and one or more computer-readable media having computer-executable instructions configured to cause the

[0008] In a fourth aspect, the present disclosure relates to a computer program element having instructions, which when executed on one or more computing nodes, are configured to perform the steps of the method defined above.

[0009] A fifth aspect of the present disclosure relates to a computer-readable medium having stored thereon the computer program element.

[0010] In a sixth aspect, the present invention relates to a method of producing an agrochemical product in a geographic region, the method comprising: providing (110) historical satellite imaging data for the geographic region for a previous growing season and at least one additional past growing season; determining (120) historical crop classification data from at least historical satellite imagery data, the historical crop classification data including information regarding crop types grown in fields within the geographic region for at least two past growing seasons; providing (130) a crop rotation model for classifying crop rotation patterns in a geographic region, the model being based at least on historical crop classification data for the geographic region; determining (140) a crop rotation pattern for fields within a geographic region based on historical crop classification data for the fields by using a crop rotation model; determining (150) future crop data including information regarding areas of the geographic region that will be cultivated with particular crops in the current or next growing season based at least on the historical crop classification data for the field by using the crop rotation pattern for the field; providing (160) a product consumption model for agricultural products configured to estimate consumption of the agricultural products based at least on future crop data; using a product consumption model to provide an estimate of agricultural product consumption in a geographic region based at least on future crop data (170); providing (180) control data for a manufacturing process relating to the agricultural product based on the estimated consumption of the agricultural product; Producing an agrochemical product by using the control data in a manufacturing process (190); Includes.

[0011] Any disclosures and embodiments described herein relate to the methods, systems, uses, computer program elements, and computer-readable media outlined above, and vice versa. As an advantageous feature, benefits provided by any of the embodiments and examples apply equally to all other embodiments and examples, and vice versa.

[0012] The present disclosure is based, inter alia, on the discovery that a crop rotation model for classifying crop rotation patterns in a geographic region, combined with historical crop classification data containing information about the types of crops grown in fields within the geographic region, can determine which areas of the geographic region will be cultivated with which crops in the current or next growing season. Based on information about the area, i.e., the size of the area, it is possible to estimate the consumption of agricultural products in the geographic region. As a result, manufacturing processes, logistics processes, and warehousing activities can be planned and executed in a safer and more predictable manner, significantly reducing the costs associated with planning and the resulting prediction errors. Furthermore, because agricultural products can also be used until their expiration date, the present disclosure allows for a significant reduction in agricultural products that must be discarded.

[0013] definition The term agricultural product is to be understood broadly in the present disclosure and includes any object or material that is useful / necessary for the treatment of a field. In the context of the present disclosure, the term agricultural product is not limited to: - Chemical products such as fungicides, herbicides, insecticides, acaricides, molluscicides, nematicides, birdsicides, fishcides, rodenticides, insect repellents, bactericides, biocides, safeners, plant growth regulators, urease inhibitors, nitrification inhibitors, denitrification inhibitors or any combination thereof; - biological products such as microorganisms useful as fungicides (bio-fungicides), herbicides (bio-herbicides), insecticides (bio-insecticides), miticides (bio-miticides), molluscicides (bio-molluscicides), nematicides (bio-nematicides), bird killers, fish killers, rodent killers, insect repellents, fungicides, biocides, safeners, plant growth regulators, urease inhibitors, nitrification inhibitors, denitrification inhibitors, or any combination thereof; - Fertilizers and compost, - seeds and seedlings, - water, - Agricultural equipment / devices (e.g. sprayers, harvesters, machinery) and spare parts for such equipment / devices; and / or - any combination thereof, Includes.

[0014] The term "field" relates to cultivated land on which crop plants such as fruits and vegetables, or row crops such as corn or rapeseed, are grown or have been grown and / or grown during at least one previous growing season. The term "field" includes both the entire field and parts thereof, such as half a field or one-third of a field. Fields do not relate to covered facilities such as greenhouses.

[0015] The term "geographical region" should be understood in a broad sense in the present disclosure and can range from an area covering a farmer or several fields of several farmers to a district, a state, an entire country (e.g., Germany), or a continent such as Europe. In one aspect, the term "geographical region" relates to a territory administered by a common legislative body. Preferably, the term geographical region can be understood to mean, for example, a county / district, an area such as southern Bavaria, etc. The geographical region can be selected so that product consumption can be estimated with respect to the producer's delivery area. Particularly preferably, the term geographical region can be understood to mean an area of ​​more than 400 square kilometers. In another aspect, the term "geographical region" relates to an area of ​​the same climate classification, such as a climate zone. Examples of climate classifications are, for example, the Köppen and Geiger classification and the Trewartha climate classification. Climate classifications are strongly correlated with the type of crop cultivated in a geographical region and the length, number, and start / end of the annual period allowing the growth of a particular crop. The term "geographical area" may also relate to the intersection of a zone within a climate class and a territory governed by a common legislative body, such as the Mediterranean countries of the European Union. It may also relate to a subset of territory within a zone of the same climate classification, such as East Germany, Poland, and the Czech Republic.

[0016] The term "crop vegetation index data" should be understood broadly and include any crop growth index data that allows for determining at least the type of crop cultivated in a geographic region, such as by comparison with reference data. In one example, the crop growth index data can be Normalized Difference Vegetation Index (NDVI) data, and the reference data is crop-specific NDVI data. Reference data, such as NDVI data, can be used not only to determine a particular crop by comparing data, but also to determine or compare, for example, whether a particular crop is healthy, damaged, or diseased. The latter can be particularly well compared, identified, and quantified using NDVI data because crop leaves reflect light differently depending on their condition. Historical crop vegetation index data includes data from a previous growing season and at least one additional past growing season. The term "growing season" refers to the portion of the year that allows active plant growth of a particular crop and typically varies between crop types. Thus, the term "last growing season" always refers to the particular crop of interest and describes the last past period in which the crop could be grown in a geographic region under prevailing climatic conditions.

[0017] The term "crop classification data" refers to information about crop types grown in fields within a geographic region. In other words, historical crop classification data classifies fields according to the type of crop grown thereon. Historical crop classification data is determined from at least historical vegetation index data, such as by comparison with plant-specific reference data.

[0018] The term "plant-specific reference data" refers to any data that allows conclusions to be made about a particular crop by comparing provided satellite image data (preferably crop vegetation index data) with the reference data. In one example, the "plant-specific reference data" may be used to train a machine learning algorithm, and the machine learning algorithm may identify the particular crop cultivated in a geographic region.

[0019] The term "machine learning algorithm" should be understood broadly and preferably includes decision trees, naive Bayes classification, nearest neighbors, neural networks, convolutional neural networks, generative adversarial networks, support vector machines, linear regression, logistic regression, random forests, and / or gradient boosting algorithms. Preferably, the machine learning algorithm is configured to process high-dimensional inputs into much lower-dimensional outputs. Such machine learning algorithms are said to be "intelligent" because they can be "trained." The algorithm can be trained using records of training data. A record of training data includes training input data and corresponding training output data. The training output data of a record of training data is the result expected to be produced by the machine learning algorithm when given the training input data of the same record of training data as input. The deviation between this expected result and the actual result produced by the algorithm is observed and evaluated by a "loss function." This loss function is used as feedback to adjust parameters of the machine learning algorithm's internal processing chain. For example, parameters may be tuned with an optimization goal of minimizing the value of a loss function, obtained when all training input data is fed to the machine learning algorithm and the results are compared with the corresponding training output data. As a result of this training, even if a relatively small number of training data records are given as "ground truth," the machine learning algorithm can perform well on a number of input data records that are orders of magnitude larger.

[0020] The term "crop," when used in, for example, "crop type," may refer to a plant, such as a grain, fruit, or vegetable, that is grown in large quantities. Preferred crops include onion (Allium cepa), pineapple (Ananas comosus), peanut (Arachis hypogaea), asparagus (Asparagus officinalis), oats (Avena sativa), beet (Beta vulgaris spec. altissima) (sugar beet), beet (Beta vulgaris spec. rapa) (turnip) (Beta vulgaris spec. rapa), rapeseed (Brassica napus var. napus), rutabaga (Brassica napus var. napobrassica), and Brassica sylvastris (winter tuna rape) (Brassica rapa var.silvestris, kale (Brassica oleracea), black mustard (Brassica nigra), tea plant (Camellia sinensis), safflower (Carthamus tinctorius), pecan (Carya illinoinensis), lemon (Citrus limon), orange (Citrus sinensis), Arabica coffee (Coffea arabica), (Robusta coffee (Coffea canephora)), Liberica coffee (Coffea liberica), cucumber (Cucumis sativus), horsegrass (Cynodon dactylon), wild carrot (Daucus carota), oil palm (Elaeis guineensis), wild strawberry (Fragaria vesca), soybean (Glycine max), and upland cotton (Gossypium hirsutum), (Gossypium arboreum, Gossypium herbaceum, Gossypium vitifolium, sunflower (Helianthus annuus), rubber tree (Hevea brasiliensis), barley (Hordeum vulgare), hops (Humulus lupulus), sweet potato (Ipomoea batatas), oak walnut (Juglans regia), lentil (Lens culinaris), flax (Linum usitatissimum), tomato (Lycopersicon lycopersicum), apple (Malus spec.), cassava (Manihot esculenta), alfalfa (Medicago sativa), musa (Musa spec.), tobacco (Nicotiana tabacum) (N. rustica), olive (Olea europaea), rice (Oryza sativa), lima bean (Phaseolus lunatus), common bean (Phaseolus vulgaris), Norway spruce (Picea abies), pine (Pinus spec.), pistachio (Pistacia vera), pea (Pisum sativum), sweet cherry (Prunus avium), peach (Prunus persica), pear (Pyrus communis), apricot (Prunus armeniaca), black cherry (Prunus cerasus), almond (Prunus dulcis) and plum (Prunus domestica), Ribes sylvestre, castor bean (Ricinus communis), sugarcane (Saccharum officinarum), rye (Secale cereale), white mustard (Sinapis alba), potato (Solanum tuberosum), sorghum (Sorghum bicolor) (S. vulgare), cocoa (Theobroma cacao) cacao, red clover (Trifolium pratense), bread wheat (Triticum aestivum), triticale, durum wheat (Triticum durum), broad beans (Vicia faba), European grape (Vitis vinifera), and corn (Zea mays). The most preferred crops are peanut (Arachis hypogaea), sugar beet (Beta vulgaris spec. altissima), rapeseed (Brassica napus var.napus, Brassica oleracea, lemon (Citrus limon), orange (Citrus sinensis), Arabica coffee (Coffea arabica), (Coffea canephora, Coffea liberica), Cynodon dactylon, soybean (Glycine max), Gossypium hirsutum, (Gossypium arboreum, Gossypium herbaceum, Gossypium vitifolium), sunflower (Helianthus annuus), barley (Hordeum vulgare), Chinese walnut (Juglans regia), lentil (Lens culinaris), flax (Linum usitatissimum), tomato (Lycopersicon lycopersicum), apple (Malus spec.), alfalfa (Medicago sativa), tobacco (Nicotiana tabacum) (N. rustica), olive (Olea europaea), rice (Oryza sativa), lima bean (Phaseolus lunatus), common bean (Phaseolus vulgaris), pistachio (Pistacia vera), pea (Pisum sativum), almond (Prunus dulcis), sugarcane (Saccharum officinarum), rye (Secale cereale), potato (Solanum tuberosum), sorghum (Sorghum These include S. bicolor (S. vulgare), triticale (Triticale), bread wheat (Triticum aestivum), durum wheat (Triticum durum), broad beans (Vicia faba), European grape (Vitis vinifera), and corn (Zea mays).

[0021] The term "crop rotation model" should be understood broadly in this disclosure and refers to any computer-operable model, method, or mathematical algorithm that can be used to classify crop rotation patterns in a geographic region, where the model is based on past crop classification data for the geographic region. Thus, the crop rotation model summarizes past crop classification data for the geographic region and further classifies fields according to a crop rotation pattern. The term "crop rotation pattern" refers to the sequence of crops over at least two subsequent growing periods, typically at least the last three growing periods, particularly at least four growing periods, e.g., at least five growing periods. For example, past crop classification data for a geographic region may indicate that for a particular area, a particular crop A is typically followed by crop B (pattern 1) in the subsequent growing season, while in other areas of the geographic region, it is followed by crop C (pattern 2). In such a case, the crop rotation model may include information of two different crop rotation patterns. The crop rotation model may also take into account more than past crop classification data. In one embodiment, the crop rotation model may also be based on traditional crop rotation patterns, such as three-field crop rotation systems, four-field crop rotation systems, and considerations of soil organic matter, pest management, nutrients, soil erosion, hybridization, and impacts on surrounding fields. For example, the crop rotation model may include information on a four-field crop rotation system of the classic crops wheat, turnip, barley, and clover, or rapeseed, wheat, barley, and legumes. In another example, the crop rotation model may include information on a two-field crop rotation system, such as rapeseed and wheat. In another embodiment, the crop rotation model may further consider regulations and recommendations regarding crop rotation patterns from local governments and environmental agencies, such as government agencies and agricultural organizations. Such regulations and recommendations may vary by region, including their goals, such as increasing yield, soil conservation, biodiversity, or reducing the use of crop protection agents. Therefore, region-specific information regarding regulations on crop rotation practices is typically taken into account in the crop rotation model.

[0022] The "area of ​​a geographical region cultivated with a specific crop" refers to the area of ​​the geographical region where the specific crop is planted or will be planted. In other words, the "area of ​​a geographical region cultivated with a specific crop" is the area obtained by adding up all the fields in the geographical region where the specific crop is grown or will be grown. For example, if wheat is grown in 100 individual fields, each 2 hectares in size, and the individual fields can be distributed arbitrarily in the geographical region, the "area of ​​a geographical region cultivated with a specific crop" is 200 hectares. Based on this area, the consumption of agricultural products required for the treatment of the determined area can be estimated. In one example, the "area" can be expressed in hectares, square meters, square kilometers, etc. In the term "determining future crop data including information about areas of a geographic region cultivated with a particular crop (150)," the term "area of ​​a geographic region cultivated with a particular crop" used in step (150) includes determining several areas cultivated with different crops, such as determining an area of ​​a geographic region cultivated with winter wheat and an area cultivated with corn. Thus, the method of the present invention can not only achieve an estimation of consumption of agricultural products based on the cultivated area of ​​one particular crop, but can also take into account consumption based on the cultivated area of ​​various crop types, such as when an agricultural product can be used for more than one crop.

[0023] The term "crop growth model" should be understood broadly in this disclosure and refers to any computer-operable model, method, or mathematical algorithm that can be used to calculate the growth stage of a particular crop at time t2 based on the growth stage of the particular crop at time t1. An example of such a crop growth model is described in the paper "Use time series NDVI and EVI to develop dynamic crop growth metrics for yield modeling" (Sadia Alam Shammi et al., Ecological Indicators, Volume 121, February 2021). The growth stage at time t1 can be derived from satellite imagery data, preferably crop vegetation indices, at time t1, such as by regression analysis similar to that described for crop classification data, in particular by using machine learning-based models trained with annotated reference data containing information about the crop growth stage.

[0024] The term "product consumption model" should be understood broadly in this disclosure and refers to any computer-operable model, method, or mathematical algorithm that can be used to calculate / estimate the consumption of an agricultural product at a specific time or over a period of time. The consumption of such a product may be based at least on future crop data, including information about the area of ​​a geographic region cultivated with a particular crop. However, it is not excluded that additional parameters may be used in the product consumption model. In particular, the product consumption model may use the results of a crop growth model to determine the exact timing when an agrochemical product is needed. Thus, in addition to future crop data, the product consumption model may use the estimated crop growth stage at time t2 determined by the crop growth model as an input parameter.

[0025] Additionally, agricultural product shelf life data, farmers' expected planting decisions, pest pressure data, agricultural product regulatory data, etc., may also be taken into account here to improve the accuracy of the product consumption estimates. The relationship between the determined area during the season and the demand for agricultural products may be derived from previously observed demand patterns and practical experience (e.g., as described in the "Pflanzenschutzberater-Kloster Muehle," which describes the normal use of various agricultural products in various plant conditions). However, numerous other sources / recommendations are known that may be used in this regard. Additionally, specialized or trained consumption models for this purpose may be used for each agricultural product. For example, a herbicide consumption model, a fertilizer consumption model, a pesticide consumption model, etc. may be used. However, these consumption models may also be combined into a single consumption model. In one example, a consumption model may be provided with a standard recommended application rate, for example, for Cantus Gold (a fungicide for treating maturity diseases of rapeseed), with a standard application rate of 0.5 l / ha recommended.

[0026] As used herein, "determining" also includes "initiating a determination or causing a determination," "generating" also includes "initiating a generation or causing a generation," and "providing" also includes "initiating a determination, generation, selection, transmission, or reception, or causing a determination, generation, selection, transmission, or reception." "Initiating or causing the performance of an action" includes any processing signal that triggers a computing device or processing unit to perform the respective action. "Determining [...] by a / that processing unit" relates to an automatic decision made by a processing unit without human interaction.

[0027] Detailed Description of the Invention and Preferred Embodiments The method (100) of the present invention is directed to estimating consumption of agricultural products. The agricultural products may be fungicides, herbicides, insecticides, acaricides, molluscicides, nematicides, birds, fish, rodenticides, repellents, bactericides, biocides, safeners, plant growth regulators, urease inhibitors, nitrification inhibitors, denitrification inhibitors, fertilizers, nutrients, seeds / seedlings, and / or combinations thereof. In one embodiment, the agricultural product is a fungicide. In another embodiment, the agricultural product is plant propagation material such as seeds.

[0028] The first step (110) of the method (100) of the present invention involves providing historical satellite imaging data for a geographic region. In one embodiment, the historical satellite imaging data is based on data obtained using synthetic aperture radar (SAR) or satellite-based light detection and ranging (LIDAR).

[0029] In one embodiment, the historical satellite imaging data includes time-resolved imaging data. The term time-resolved imaging data refers to a series of imaging data captured over the previous growing season and at least one additional previous growing season, preferably a series of imaging data for both the previous growing season and at least one additional previous growing season. Accuracy for purposes of the invention described herein increases with the number of measurements per growing season. Typically, one set of image data for one field per week is captured over the growing season.

[0030] The historical satellite imaging data includes data for at least the previous growing season and one additional previous growing season. Typically, the historical satellite imaging data includes data for at least two previous growing seasons. In one embodiment, the historical satellite imaging data is for at least three previous growing seasons, preferably at least four previous growing seasons, more preferably at least five previous growing seasons, and especially at least the last six previous growing seasons. It should be understood that the method will produce more reliable results the more previous growing seasons for which historical crop vegetation index data is available.

[0031] In a second step (120), the method (100) of the present invention determines historical crop classification data from at least historical satellite imaging data.

[0032] In one embodiment, step (120) includes determining (125) historical time-resolved vegetation index data from the satellite imagery data, followed by determining from the vegetation index data information regarding crop types grown in fields within the geographic region for at least two previous growing seasons.

[0033] Vegetation indices typically relate the intensities of specific light waves to provide specific information about the growth stage, health, nutrition, water supply, and other important characteristics of crop plants, and their calculation from imaging data is known to those skilled in the art and is described, for example, at https: / / en.wikipedia.org / wiki / Vegetation_Index or in https: / / earthobservatory.nasa.gov / features / MeasuringVegetation / measuring_vegetation_2.php.

[0034] In one embodiment, the historical satellite imaging data is time-resolved imaging data and the historical vegetation index data is time-resolved vegetation index data selected from Normalized Difference Vegetation Index (NDVI) data and / or Leaf Area Index (LAI) data, Normalized Difference Water Index (NDWI), and Enhanced Vegetation Index (EVI) data.

[0035] The classification of plants grown in a field can be performed with high accuracy from satellite imagery data, preferably vegetation index data. Thus, a processing unit can determine which type of plant is growing in the field. The determination (120, 125) can be achieved by using a classification model. The classification model can be obtained by various methods, such as machine learning, particularly supervised learning using reference data. Useful techniques for supervised learning to obtain a classification model are logistic regression, perceptron algorithms, Bayesian classification, naive Bayesian classification, k-nearest neighbor algorithms, artificial neural networks, and decision tree-based modeling, such as random forest algorithms. Reference data for generating a suitable model is typically obtained by annotation of satellite imagery data using ground truth data collected by agronomic advisors, users, or field-based machines, such as a BASF Smart Sprayer.

[0036] The highest accuracy of predictions is typically achieved by using time-resolved satellite imaging data, preferably vegetation index data, to determine the types of crop plants growing thereon. If a vegetation index is used, a single index such as the NDVI or LAI index may be sufficient, however, it goes without saying that a combination of different vegetation indices will greatly improve the accuracy of the determination.

[0037] Other methods useful in steps (120, 125) are extensive statistical modeling techniques. The accuracy achievable with such methods is typically at least 90%, usually at least 95%. High accuracy in determining steps (120, 125) is particularly important when a large number of fields are evaluated, since even small errors can lead to considerable uncertainty.

[0038] Thus, the determination in steps 120, 125 first results in a certain probability that a crop plant was grown in the field. This information can be provided directly in steps 130 and 150. Alternatively, the field can be classified as a field in which a crop plant was or was not grown during at least one previous growing season based on predefined benchmarks, and then this classified information can be provided in steps 130 and 150.

[0039] The method of the present invention may typically include a step (115) of determining field boundaries in historical satellite imagery data, which may typically be performed after step (110) and before step (120). However, it is also possible to perform step (115) after step (120), such as before step (125), or after step (125) and before step (130).

[0040] Such determination can be achieved by a field boundary detection model, which is typically obtained using supervised machine learning. Suitable machine learning methods can include deep learning techniques, particularly the use of convolutional neural networks for segmentation, such as UNet or ResUNet-A (Diakogiannis, F.I., Waldner, F., Caccetta, P., Wu, C., 2019, Resunette-a: a deep learning framework for semantic segmentation of remotely sensed data. arXiv preprint arXiv:1904.00592, https: / / www.tensorflow.org / ).

[0041] Training images for these supervised machine learning techniques should be as uniform as possible by selecting observations with the least cloud coverage from a specified interval (e.g., within three months). Gaps in cloud-selected observations can be filled with other non-cloud observations. This technique can generate images that are both complete and minimize artificial disturbances in the image, which can result from freely combining images from different observation dates. Additionally, an additional layer can be created that encodes the observation time of each pixel to indicate artificial disturbances from exchanges to the model. Thus, the model can learn to identify the disturbances. Furthermore, Sobel filters applied to individual satellite bands are typically generated to enhance the visibility of optical edges in the image.

[0042] The training data, selected in vector form, is typically rasterized to match the satellite data. Three different targets can be derived to train the model, as described in Waldner et al. (Deep learning on edge: extracting field boundaries from satellite images with a convolutional neural network; 2020; arXiv preprint arXiv:1910.12023v2): a binary mask of the field boundaries, a binary mask of the extent of the fields, and a field-wise normalized distance (distance to closest boundary).

[0043] To train the model, the satellite data and training data can be sliced ​​into smaller portions (e.g., 128x128 pixel images) to fit into GPU memory. Models based on the TensorFlow library (www.tensorflow.org) are typically optimized by minimizing the Tanimoto loss (see Waldner, F., Diakogiannis, F.I., 2020: Deep learning on edge: extracting field boundaries from satellite images with a convolutional neural network. arXiv preprint arXiv:1910.12023v2).

[0044] The trained segmentation model can then be used to derive field boundary and field extent predictions for new targeted areas, typically utilizing the mechanisms described above for the training process. Satellite imagery is selected, preprocessed, and sliced ​​down to generate inputs for inference. Inference is performed independently for multiple time points over one or more seasons to mitigate changes in field appearance due to vegetation processes throughout the year. These predictions for the slices are then recombined into larger units (tiles).

[0045] The field boundary predictions may finally be combined to generate a vector-format field boundary. Individual predictions for areas at different times may be merged (e.g., via an average or max operation), artificially expanded to a higher resolution (e.g., 10 m to 2 m), and smoothed. This expansion allows the system to compensate for the effects of coarse pixels in the satellite data and result in smoother field boundaries. Probabilistic model predictions, with continuous values ​​between 0 and 1, are thresholded to obtain binary values. These binary masks may then be combined into a final field boundary mask by subtracting the binary field boundary prediction from the binary field extent prediction. Finally, the raster-formatted segments are vectorized, and minor adjustments such as smoothing the boundaries and filling smaller gaps within the field may be performed. The resulting field boundaries are stored in file storage and referenced in a database for easy access and use.

[0046] In a subsequent step (130), a crop rotation model for classifying crop rotation patterns in a geographic region is provided, the model being based at least on historical crop classification data for the geographic region.

[0047] In a subsequent step (140), the crop rotation pattern for fields within the geographic region, i.e., for each individual field, is determined. This is accomplished by using a crop rotation model and historical crop classification data. This is typically done by comparing the field's historical crop classification data with different rotation patterns in the crop rotation model and classifying the field according to the crop rotation pattern that best explains the historical crop classification data. The accuracy of the method of the present invention is improved if the historical crop classification data is not just for the last growing season and one additional past growing season, but for at least the last two growing seasons, preferably at least the last three growing seasons, more preferably at least the last four growing seasons, e.g., at least the last five growing seasons.

[0048] In a next step (150), the method involves determining future crop data, including information regarding areas of the geographic region that will be cultivated with a particular crop in the current or next growing season, based at least on the field's past crop classification data by using the field's crop rotation pattern. In other words, the future crop data includes information regarding the type of crop already grown or that will be grown in a particular area within the geographic region. For example, the future crop data may include information regarding which areas within the geographic region are currently growing winter wheat or will be growing winter wheat in the next growing season. For the avoidance of doubt, it is emphasized that the term "current growing season" refers to a situation in which the growing season has already begun but the crop has not yet been sown, or the seedlings have already been sown but the seedlings have not yet emerged above the soil, or the seedlings are still too small to allow for crop classification based on the crop vegetation index data. In other words, the term "current growing season" refers to a situation in which the growing season has already begun but crop classification data for the field using the crop vegetation index data is not yet available. Preferably, the future crop data includes information regarding areas of a geographic region that will be cultivated with a particular crop in the next growing season.

[0049] The future crop data is obtained from the historical crop classification data by using the field's crop rotation pattern. In other words, the historical crop classification data for one, several, or even all fields in a geographical region is compared with the field's crop rotation pattern to determine at which stage of the crop rotation cycle the field is currently cultivated, and based on the field's crop rotation pattern, which crop type is currently cultivated or will be cultivated in the next growing season is predicted. For example, if the field's crop rotation pattern follows the sequence -(ABC)- and the historical crop classification data further determines that the field was last cultivated with crop B and cultivated with crop A the previous year, the future crop data indicates that crop C will be grown in the next growing season. Because the future crop data is a prediction based on a crop rotation model, it is usually associated with a certain accuracy that can vary depending on the amount of underlying data. Therefore, the data accuracy can also be determined and further used in subsequent steps of the inventive model.

[0050] The future crop data may be determined for one particular crop, several particular crops, or for all crops grown in fields within a geographic region. In one embodiment, the future crop data is determined for areas in a geographic region cultivated with one particular crop.

[0051] In one embodiment of a computer-implemented method for estimating consumption of agricultural products, future crop data is determined for a group of specific pre-selected crops, hi one example, the group of specific pre-selected crops is a combination of winter rapeseed, winter rye, sugar beet, and winter wheat determined in parallel.

[0052] In a next step (160), the method involves providing a product consumption model for the agricultural product configured to estimate consumption of the agricultural product based at least on the future crop data. In a subsequent step (170), the method involves using the product consumption model to provide an estimate of consumption of the agricultural product in the geographic region based at least on the future crop data.

[0053] In one embodiment, the product consumption model for an area cultivated with a particular crop is based on the results of a machine learning algorithm configured to estimate consumption of agricultural products based at least on the area of ​​a geographic region cultivated with the particular crop. However, statistically based product consumption models can also be used. Furthermore, such product consumption models are not limited to using only area; i.e., additional data can be used in this regard. Agricultural product shelf life data, farmers' expected planting decisions, pest pressure data, agricultural product regulatory data, etc. can also be considered here to improve the accuracy of the product consumption estimates.

[0054] In one embodiment, the computer-implemented method comprises: providing inventory recommendation data regarding minimum stock levels of agricultural products at a particular time and / or for a period of time based on the estimated consumption of agricultural products; and / or providing inventory recommendation data regarding minimum stock levels of basic materials required for the production of agricultural products at a particular time and / or for a period of time based on the estimated consumption of agricultural products; and / or providing production recommendation data for producing agricultural products based on the estimate of consumption of the agricultural products; and / or providing order recommendation data for ordering quantities of agricultural products and / or quantities of basic materials required for the production of agricultural products based on the estimated consumption of agricultural products; and / or providing summary data on agricultural products required and / or recommended for specific crops; and / or providing control data for manufacturing processes, logistics processes, and / or warehousing processes with respect to the agricultural products based on the estimate of consumption of the agricultural products; The method further includes at least one of:

[0055] The present disclosure will now be described in further detail with reference to the accompanying drawings. [Brief explanation of the drawings]

[0056] [Figure 1] FIG. 1 is a flow diagram of an exemplary method for estimating consumption of agricultural products in a geographic region. [Figure 2] FIG. 1 is a schematic diagram of time-resolved patterns of vegetation indices useful for identifying crop plants growing in a field. [Figure 3] 1 is a schematic diagram of a collection of fields for which agricultural product consumption is estimated. [Figure 4] FIG. 1 illustrates the probability of a particular crop rotation pattern in a geographic region. [Figure 5] A grayscale satellite image of an agricultural area with fields shown in light gray and future crop data indicating that rapeseed will be grown in the next growing season. DETAILED DESCRIPTION OF THE INVENTION

[0057] 1 is a flow diagram of an exemplary method 100 for estimating consumption of agricultural products in a geographic region. For example, the agricultural products may be fungicides, herbicides, insecticides, acaricides, molluscicides, nematicides, birdsicides, piscicides, rodenticides, repellents, bactericides, biocides, safeners, plant growth regulators, urease inhibitors, nitrification inhibitors, denitrification inhibitors, fertilizers, nutrients, seeds / seedlings, and / or combinations thereof.

[0058] In a first step 110, historical satellite imaging data for a geographic region (e.g., Bavaria) is provided. The satellite imaging data is publicly available. The historical satellite imaging data may be provided for the last growing season, e.g., last year, and for further previous growing seasons, e.g., the year before last. Typically, the historical satellite imaging data is provided for at least the last three years, e.g., the last three years.

[0059] Historical satellite imagery data is typically time-resolved, i.e., includes crop vegetation index data for a series of points in time. For example, historical satellite imagery data may include data for each target growing season over a two-week period beginning 15 days after sowing. By considering a period of time rather than just a point in time, the accuracy of determining which areas are planted with which particular crops can be greatly improved.

[0060] In step 120, historical crop classification data is determined from historical satellite imagery data. In a first step (125), crop vegetation index data may be derived from the satellite imagery data. Determining crop vegetation index data from the satellite imagery data (125) is described in further detail below.

[0061] The historical crop vegetation index data may be Normalized Difference Vegetation Index (NDVI) data, Leaf Area Index (LAI) data, Normalized Difference Water Index (NDWI) data, Extended Vegetation Index (EVI) data, or a mixture thereof. Preferably, the historical crop vegetation index data includes historical NDVI data.

[0062] As outlined above, classification of fields by crop type from satellite imaging data, preferably crop vegetation index data, can be achieved by various statistical methods, for example, by comparison with reference data. This can be implemented by using machine learning-based models. Such machine learning-based models can typically be obtained by using annotated satellite imaging data as a training dataset, with annotations based on ground truth data collected by farmers or agronomic advisors. Modeling and classification is discussed and illustrated by FIG. 2. Thus, historical crop classification data includes information on which crop types were grown on a particular area of ​​a geographic region. For example, historical crop classification data may include information on which crops were grown on a particular field over past seasons, e.g., the past three or four years. For one particular field, such information may be that barley was grown in the first year, then chickpea in the second year, then clover in the third year, and barley again in the fourth year (last year). The aggregated information for all fields within a geographic region then represents the historical crop classification data.

[0063] In a subsequent step (130), the historical crop classification data is used to set up a crop rotation model for the geographic region. To this end, the historical crop classification data is analyzed with respect to crop rotation patterns. In other words, the historical crop classification data is further classified on a field level in order to assign fields to specific crop rotation patterns. The crop rotation model may further assign frequency or likelihood data to the crop rotation patterns.

[0064] Such a crop rotation pattern may directly reflect the sequence of crop types grown in the field, but may also be based on further aspects as outlined above. For example, the crop rotation model may include a crop rotation pattern established to ensure good soil quality in terms of nutrients, e.g., nitrogen or carbon content, pest management, e.g., crop-specific fungal diseases such as sclerotinia rot, and soil erosion, e.g., due to uncultivated soils. The crop rotation model may also include recommendations by authorities.

[0065] The crop rotation pattern for each field is then determined (140). This step is accomplished by inputting the field's past crop classification into the crop rotation model. In other words, the sequence of crop plants grown in the field over the last year is compared with all known patterns of crop rotation included in the crop rotation model to determine the most likely crop rotation pattern for the particular field. For example, the crop rotation model may include information that two crop rotation patterns, such as pattern-AB- and pattern-ABC-, exist in a geographic region, and pattern-ABC- predominates across all fields with an 80% likelihood. In one example, the past classification data may indicate that crop A was used last year and crop C was grown the year before last. The crop rotation pattern-ABC- is then selected and used further in subsequent steps.

[0066] As another example, historical crop classification data may indicate that crop A was grown the year before last, while crop B was grown last year. Because the historical crop classification data cannot distinguish between the two rotation patterns, the more likely rotation pattern - A, B - is further used, preferably with the annotation that this information has an 80% likelihood.

[0067] Crop rotation patterns are determined for most (preferably all) of the fields in a geographic region.

[0068] In a further step (150), future crop data is determined. For this purpose, the crop rotation pattern of each individual field is compared with the past crop classification data. For example, if a particular field is determined to fit the pattern -ABC- and the past crop classification data indicates that crop B was grown on the field last year, it is determined that crop C will be grown on the field next year. This procedure is performed for most (preferably all) fields in the geographical area, and the data thus obtained is compiled to obtain future crop data for the geographical area. Thus, the future crop data may include information about the crop types that are most likely to be grown on each field in the geographical area. At a minimum, the future crop data may include only information about which fields in the geographical area a particular crop will be grown on. In one embodiment, the future crop data may include information about all crop types to which a particular agricultural product can be applied.

[0069] In step 160, a product consumption model for an agricultural product is provided, the product consumption model being configured to estimate consumption of the agricultural product based at least on future crop data. For example, a consumption model for Cantus Gold (a fungicide for treating maturity diseases in rapeseed) is provided. The product consumption model may be based on the results of a machine learning algorithm that estimates consumption of the agricultural product. However, it is also possible to use statistically based product consumption models. Furthermore, such product consumption models are not limited to using only area; additional data may be used in this regard. Agricultural product shelf life data, farmers' expected planting decisions, pest pressure data, agricultural product regulatory data, etc. may also be considered here to improve the accuracy of the product consumption estimate.

[0070] An estimate of the consumption of agricultural products in areas cultivated with particular crops is performed in step 170. For example, if it is determined that 1000 hectares will be cultivated with rapeseed in a geographical region next season as included in future crop data, a forecast of the consumption of Cantus Gold may be provided.

[0071] Figure 2 shows the trend of the NDVI index over a season. The NDVI vegetation index considers and relates the intensity relationships between different spectral regions in an image, particularly the near-infrared and visible-infrared portions of the spectrum. The NDVI index for plants fluctuates throughout the season, as shown in the left window of Figure 2. For green, healthy leaves (203), the NIR intensity is significantly higher compared to the red light portion of the spectrum. This is different for young leaves (202) or brown leaves (201). The right side of Figure 2 shows a time-resolved data series of NDVI values ​​over a season for different crop plants (204, 205, 206, 207). Depending on the plant's growth cycle, the NDVI has a very specific pattern for each plant. For example, some crop plants germinate and begin growing at different times during the season. As can be seen from curve (204), the beginning of the curve is quite early and abrupt at the beginning of the winter season, while curve (206) shows a slow and steady rise during the winter season. Finally, the NDVI curve (207) only rises during the summer season at the beginning of June. Distinct patterns for different crop plants are only shown for the NDVI index, but are also present within other vegetation indices, such as the LAI index.

[0072] These patterns can be used to identify the type of crop plant growing in the field. While it is not necessary to record the entire time-resolved pattern of the index, the more data points available, the more accurate the information. Typically, an array of vegetation index factors is used as an input factor, rather than just one vegetation index. As mentioned above, an annotated form of historical vegetation index data can be used as a training dataset for generating machine learning models. Annotation can be achieved by recording farmer data, for example, using a customer front-end tool such as the BASF Xarvio suite. It can also be achieved by using observational data obtained by sales representatives. The annotated training data can be used in supervised machine learning techniques. For example, the training data can include historical vegetation index data for a geographic region, with specific fields annotated with the type of crop growing in the field and the time in the season when the image was captured. Preferably, the imaging data is time-resolved and includes imaging data from at least two points in the season. Machine learning tools typically use a loss function to generate a model that describes the training data in the best possible way. Validation data is typically used to avoid overfitting. The resulting model can then be used to analyze newly acquired satellite imaging data.

[0073] If machine learning techniques are not used, it may be prudent to generate a calibration curve for the vegetation index from the annotated training data, such as by determining the average of various curves captured for the same crop plant. The calibration curve can then be used in a regression technique to classify new data according to the predefined calibration curve. Typically, this is achieved by minimizing the deviation of the newly measured curve from the calibration curve and classifying the field according to the fit with the smallest deviation. Thus, past crop vegetation index data can be analyzed to classify crops grown in the field during previous growing seasons.

[0074] 3 shows a collection of fields 301, 304, 305 from which imaging data is captured by satellite 302. The imaging data is then transferred to a server 303, such as a local server or a cloud computing environment. The fields are cultivated in a crop rotation system. In year 1 (left panel), rapeseed is grown in field 301, chickpea is grown in field 304, and lima bean is grown in field 305. The following year, maize is grown in field 301, lima bean is grown in field 304, and rapeseed is grown in field 305. For each of fields 301, 304, 305, a different crop is grown in year 2 compared to year 1. Therefore, it can be assumed that a crop rotation pattern may exist, but it remains unclear how such a pattern may be used to predict the type of crop that will be grown on the field next season. However, historical crop classification data may be combined with further crop rotation information. For example, if a four-field crop rotation system is assumed, then the field 301 is likely to be cultivated with chickpea or leek pea next season, i.e., there is a 50% probability. If further information, for example, regarding nitrogen management, is taken into account, the probability may even be further affected, such as 80% leek pea vs. 20% chickpea. Thus, if a crop rotation model for a geographical region indicates that more than 90% of farmers use a four-field system, and furthermore, historical crop rotation data for a particular field allows the same assumption for that field, it may be possible to generate moderate to accurate assumptions about the type of crop that will be grown on the field next season (even if the data is very limited). As outlined above, the quality of predictions increases with the amount of underlying data, and less a priori knowledge is required to produce acceptable results. For example, historical crop classification data from the past 6–10 growing seasons will result in highly accurate assumptions of the field's rotation pattern.

[0075] 4 illustrates the probabilities (P) of different crop rotation patterns (401, 402, 403, 404, 405) for a geographic region. This information may reflect a crop rotation model for the geographic region that is based entirely on empirical data, i.e., historical crop classification data. The crop rotation model may include the probability of each crop rotation pattern found in the geographic region, or may simply be a compilation of all observed patterns, preferably with further a priori data on crop rotation management to optimize soil health (carbon or nitrogen content), reduce pest damage, reduce soil erosion, or reduce impacts on neighboring fields.

[0076] FIG. 5 is a grayscale satellite image of an agricultural area, with fields shown in light gray, and future crop data indicating that rapeseed will be grown in the next growing season. Thus, FIG. 5 is an example of future crop data for a geographic region. As noted above, the future crop data may include information about fields where one particular crop is currently grown or will be grown in the next growing season, or it may include information about several crop types. Preferably, it may include information about all crop types relevant to estimating consumption of the particular agricultural product in question. For example, a fungicide may be approved for treating fungal diseases in different related species, such as rapeseed and white mustard. To estimate consumption of the fungicide, it is important to know the size of the areas where white mustard and rapeseed will be grown in the geographic region of interest.

[0077] An aspect of the present disclosure relates to a computer program element configured to execute the steps of the above-described method. Accordingly, the computer program element may be stored on a computing unit of a computing device, which may also be part of an embodiment. The computing unit may be configured to execute or direct the execution of the steps of the above-described method. Furthermore, the computing unit may be configured to operate components of the above-described system. The computing unit may be configured to operate automatically and / or execute user instructions. The computing unit may include a data processor. The computer program may be loaded into the working memory of the data processor. Thus, the data processor may be equipped to implement a method according to one of the above-described embodiments. This exemplary embodiment of the present disclosure encompasses both a computer program that uses the present disclosure from the beginning and a computer program that transforms an existing program into a program that uses the present disclosure by means of an update. Furthermore, the computer program element may be capable of providing all steps necessary to perform the procedures of the exemplary embodiment of the above-described method. According to a further exemplary embodiment of the present disclosure, a computer-readable medium, such as a CD-ROM, a USB stick, a downloadable executable file, or the like, is presented, the computer-readable medium having stored thereon a computer program element, the computer program element being described in the previous section. The computer program may be stored and / or distributed on a suitable medium, such as an optical storage medium or a solid-state medium supplied together with or as part of other hardware, or may be distributed in other forms, such as via the Internet or other wired or wireless telecommunications systems, but the computer program may also be present on a network, such as the World Wide Web, and may be downloaded into the working memory of a data processor from such a network.According to a further exemplary embodiment of the present disclosure, a medium is provided that makes a computer program element available for download, the computer program element being arranged to perform a method according to one of the aforementioned embodiments of the present disclosure.

[0078] The present disclosure has been described in conjunction with preferred embodiments as examples. However, those skilled in the art and those practicing the claimed invention will understand and implement other variations upon studying the drawings, the disclosure, and the claims. In particular, the steps presented may be performed in any order; i.e., the present invention is not limited to a particular order of these steps. Furthermore, it is not required that different steps be performed at a particular location or at one node of a distributed system; i.e., each of the steps may be performed at a different node using different equipment / data processing units.

[0079] In the claims and in this description, the word "comprising" does not exclude other elements or steps, and the indefinite article "a" or "an" does not exclude a plurality. A single element or other unit may fulfill the functions of several entities or items recited in the claims. The mere fact that particular measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage in accordance with certain embodiments.

Claims

1. A computer-implemented method (100) for estimating consumption of agricultural products in a geographic region, comprising: providing (110) historical satellite imaging data for the geographic region for a previous growing season and at least one additional past growing season; determining (120) historical crop classification data from at least the historical satellite imagery data, the historical crop classification data including information about the crop types grown in fields within the geographic region for the at least two past growing seasons; providing (130) a crop rotation model for classifying crop rotation patterns in the geographic region, the model being based at least on the historical crop classification data for the geographic region; determining (140) a crop rotation pattern for fields in the geographic region based on the historical crop classification data for the fields by using the crop rotation model; determining (150) future crop data including information regarding areas of the geographic region that will be cultivated with specific crops in the current or next growing season based at least on the past crop classification data for the field by using the crop rotation pattern for the field; providing (160) a product consumption model for the agricultural product configured to estimate consumption of the agricultural product based at least on the future crop data; using the product consumption model to provide an estimate of the consumption of the agricultural product in the geographic region based at least on the future crop data (170); A computer-implemented method (100) comprising:

2. The computer-implemented method of claim 1 , wherein the historical satellite imaging data comprises time-resolved imaging data for the at least two past growing seasons.

3. 3. The method of claim 1, wherein step (120) comprises determining (125) historical vegetation index data from the satellite imaging data, and subsequently determining from the vegetation index data information regarding the crop types grown in the fields within the geographic region for the at least two past growing seasons.

4. 4. The computer-implemented method of claim 3, wherein the historical crop vegetation index data comprises Normalized Difference Vegetation Index (NDVI) data and / or time-resolved crop vegetation index data selected from Leaf Area Index (LAI) data, Normalized Difference Water Index (NDWI), and Extended Vegetation Index (EVI) data.

5. The computer-implemented method of any one of claims 1 to 4, wherein the historical satellite imaging data is for the at least three previous growing seasons.

6. 6. The computer-implemented method of any one of claims 1 to 5, wherein the agricultural product is a fungicide, herbicide, insecticide, acaricide, molluscicide, nematicide, birdcide, fishcide, rodenticide, repellent, bactericide, biocide, safener, plant growth regulator, urease inhibitor, nitrification inhibitor, denitrification inhibitor, fertilizer, nutrient, seed / seedling, and / or combinations thereof.

7. 7. The computer-implemented method of claim 1, wherein the historical satellite imaging data is based on data obtained by using synthetic aperture radar (SAR) or satellite-based light detection and ranging (LIDAR).

8. 8. The computer-implemented method of claim 1, wherein the product consumption model for the area cultivated with the particular crop is based on results of a machine learning algorithm configured to estimate the consumption of the agricultural product based at least on the area of ​​the geographic region cultivated with the particular crop.

9. providing inventory recommendation data regarding minimum stock levels of said agricultural products at a particular time and / or for a period of time based on said estimates of consumption of said agricultural products; and / or providing inventory recommendation data regarding minimum stock levels of basic materials required for the production of said agricultural products at a particular time and / or for a period of time based on said estimates of consumption of said agricultural products; and / or providing production recommendation data for producing said agricultural products based on said estimates of consumption of said agricultural products; and / or providing order recommendation data for ordering quantities of said agricultural products and / or quantities of basic materials required for the production of said agricultural products based on said estimates of consumption of said agricultural products; and / or providing summary data regarding agricultural products required and / or recommended for said particular crop; and / or providing control data for manufacturing, logistics, and / or warehousing processes with respect to the agricultural products based on the estimates of the consumption of the agricultural products; The computer-implemented method of any one of claims 1 to 8, further comprising at least one of:

10. The computer-implemented method of any one of claims 1 to 9, wherein the future crop data is determined for areas in the geographic region cultivated with one particular crop.

11. The method of any one of claims 1 to 10, using historical satellite imaging data for the geographic region for the previous growing season and at least one further previous growing season.

12. 1. A system for estimating consumption of agricultural products in a geographical area, comprising: one or more computing nodes; and one or more computer-readable media that, when executed by the one or more computing nodes, provide the system with: providing (110) satellite imaging data for the geographic region for a previous growing season and at least one additional past growing season; determining (120) historical crop classification data from at least the historical satellite imagery data, the historical crop classification data including information about the crop types grown in fields within the geographic region for the at least two past growing seasons; providing (130) a crop rotation model for classifying crop rotation patterns in the geographic region, the model being based at least on the historical crop classification data for the geographic region; determining (140) a crop rotation pattern for the fields in the geographic region based on the historical crop classification data for the fields by using the crop rotation model; determining (150) future crop data including information regarding areas of the geographic region that will be cultivated with specific crops in the current or next growing season based at least on the past crop classification data for the field by using the crop rotation pattern for the field; providing (160) a product consumption model for the agricultural product configured to estimate consumption of the agricultural product based at least on the future crop data; using the product consumption model to provide an estimate of the consumption of the agricultural product in the geographic region based at least on the future crop data (170); one or more computer-readable media having computer-executable instructions configured to cause the execution of A system comprising:

13. 13. A computer program element comprising instructions configured, when executed on one or more computing nodes, to perform the steps of the method of any one of claims 1 to 10 or for implementation by the system of claim 12.

14. A computer readable medium having stored thereon a computer program element according to claim 13.

Images