Procedure for the production of high accuracy precision prescription maps for arable land

Abstract

The subject of the patent application is a method for the production of a high-accuracy precision prescription map for arable land, in which a given area of land is divided into 10 x 10 m cells, a base map and a multipolygon zone map are produced for the cells, and soil sampling and soil analysis measurements are carried out. The base map is created as an accurate mapping of the 10 x 10 m resolution vector layer by vector layer of the 10 x 10 m cells of the contoured field, resulting in a vector base map which contains NDVI images, topographic parameters and fertility data, attributes in digital form, which are used to create the multipolygon zone map. The base map and the multipolygon zone map are uploaded to a database. Based on the multipolygon zone map, soil sampling is carried out for each multipolygon zone, soil analysis measurements are performed on the soil samples and the results of the soil analysis measurements are uploaded to the database. The obtained soil analysis results are evaluated and then used to produce a fertilisation prescription map, a sowing prescription map, a crop protection prescription map, an irrigation prescription map and a tillage prescription map for the 10 x 10 m cells of the field.

Description

[0001] Procedure for the production of high accuracy precision prescription maps for arable land

[0002] The subject of the patent application is a procedure for the production of a high- accuracy precision prescription map of arable land.

[0003] The procedure produces precision prescription maps at 10x10 m resolution based on the land topography and multipolygon zones based on historical fertility data and the yield target resulting from the yield estimation. The instructions of the prescription maps can be carried out by means of field power and machinery equipped with positioning systems in 10x10 m cells of the fields. The prescription map can be a fertilisation, sowing, crop protection, irrigation and tillage prescription map at 10x10 m resolution for the examined field.

[0004] The use of agricultural inputs has long been a key issue in agricultural research. Soil and agrochemical research over the last century has made it possible to apply fertilisers site-specifically in the same cultivated area, i.e. at the "field level".

[0005] Improving fertiliser conversion ratio and reducing losses can only be achieved through nutrient management planning. However, the nutrient supply of a crop grown under specific climatic conditions, adapted to its intended use, requires knowledge of the growing site and the measurement of important physico-chemical parameters that determine the nutrient supply capacity of the soil and the effectiveness of the nutrients applied.

[0006] Nutrient management planning has undergone significant changes in recent years. Technological progress has made it possible to relate the soil analysis results from soil samples from the land under cultivation to the field sections represented by the composite sample and to make interventions based on this information.

[0007] Since the introduction of yield measurement and yield mapping in the late 1980s, precision farming has developed in a complex and diverse way. Today, we have the possibility to apply practically all agricultural inputs in a precision, differentiated way within the field, from fertiliser to sowing seed to irrigation water, while the need to know the soil on which we grow is still present and even more so as environmental and food safety requirements become more stringent.

[0008] In the last decades, precision farming's soil sampling strategy has evolved from "grid" sampling to "management zone" sampling based on topographic parameters, satellite imagery and other sensor measurements. As in the last century, the development of soil mapping and nutrient management has gone hand in hand. High-resolution digital soil mapping can use digital elevation models (DEMs), derived terrain parameters (slope, curvature, exposure), complex parameters such as potential drainage density (PDD), complex parameters such as surface roughness or the probability of water accumulation, satellite imagery and combinations of these. By using the same parameters, as well as other sensory measurements (such as soil conductivity - EC) in different combinations, management zones within the field can be established, which are related to yield data of the land. They can also be used as a basis for differentiating irrigation, crop protection or tillage in addition to fertilisation.

[0009] There are known solutions to implement precision farming, such as the subject of US patent application US20230010675 A1 , which is a method and system for technologies for increasing agricultural production fertility, a method and system for analysing and managing agricultural production comprising a measurement algorithm, an evaluation, a decision and a response to select optimal measures to increase agricultural fertility. The information used in the process allows the selection of plant density, the planning of fertilisation or the development of strategies for the future harvesting schedule. The process consists of several stages: loading historical data (yield data from previous years), assessing current data, forecasting meteorological and hydrological events, creating an agricultural management protocol and transmitting the agricultural management protocol to the users.

[0010] The system according to the invention links and controls: 1 ) the techniques and design of sample collection, 2) the determination of sample quality, 3) the monitoring and optimisation of information quality with the necessary local and remote sensors, 4) the development and implementation of a statistical correlator, 5) the development and implementation of a time correlator, and 6) the development and implementation of an economic correlator.

[0011] In order to increase fertility, the main factors affecting the fertility of the area to be interfered with are analysed on the basis of information from sensors in the area or from several databases.

[0012] Using satellite imagery and maps produced by combining information from the green indices, they identify growth trends, terrain problems, internal deficiencies of the area, and farmers’ management problems. The images are used to collect information on the quality, quantity and availability of surface water, soil topography and geology, soil nutrient content and properties such as compaction rate, electrical conductivity, organic matter content, pH. Information from multispectral sensors capable of analysing spectral images of elements present on the soil surface or leaves is used to estimate soil nutrient composition. This information is analysed by the appropriate module of the platform.

[0013] The system and method described in US patent document US20020022929 A1 generates field attribute maps for site-specific management, field application of agricultural products. Data collected from a field is used to create an attribute map, which is converted into a format used to create application maps. The process involves cleaning and validating the data collected from a given agricultural field in the mapping system. The cleaning process corrects any data errors and converts the data into a standard format, and the validation process checks the latitude and longitude of the data. The data is then converted into a two-dimensional grid format, where each cell contains the agricultural data associated with the corresponding latitude and longitude. The two-dimensional grid format allows the mapping system to produce application maps. The main components of the site-specific farming system are the mapping software, the field data collection system, the harvest data collection system and the application control system. The agricultural information (e.g. soil analysis results, soil surveys, field boundaries and survey information, yield data) collected by the field data collection system and the harvest data collection system is processed by the mapping software and used to generate an application map, which is used by the application control system to apply agricultural products (e.g. seeds, fertilisers, pesticides and any other soil conditioners or additives) to the field.

[0014] US Patent Application US20060074560 A1 is a method and system for evaluating the performance of crop production by characterising the environmental impact of a geographic region or areas within a region on crop production. Environmental measurements obtained by the method, which may include soil data, weather data, associated with a particular geographic region, are used to determine an estimated performance characteristic for a particular crop planted in that geographic region. Within a geographical region, boundaries are defined for one or more uniform performance areas, usually with uniform performance characteristics, using decision tree analysis.

[0015] The method and process disclosed in international patent application WO / 2018 / 009257 A1 relate to the preparation of a site-specific prescription map based on a multi-year yield analysis. The method is used to examine variables in multi-year yield data to monitor agricultural activities and to determine management zones for sowing / agricultural events in a given area.

[0016] The variables are coverage data, as the percentage of the delimited field covered by the crop; uniformity data, as the coefficient of variation of the crop yield in the delimited field; age data, as the number of years since the crop event information was collected; and weather data containing temperature values and precipitation values for a given year.

[0017] The variables as input data form a yield information layer for a single crop in a limited area. The information layer consists of several variables, which include coverage data, uniformity data, age data and weather data.

[0018] The method calculates the weight of each variable in the set of variables, which includes: the weight of the coverage data, which represents a decrease in coverage data at a rate starting from a coverage factor of 90% and approaching zero between a coverage factor of 35% and 65%; the weight of the uniformity data, which represents a linear decrease of the uniformity data from a uniformity factor of 80% to a uniformity factor of 10%, approximating zero; the weight of the age data based on the collected age of the yield event, if the total number of yield events exceeds the specified number of years; and a weather data weight based on increasing decline as the precipitation value deviates further from a normal precipitation value and as the temperature value deviates further from a normal temperature value, for the period representing the layer; calculating combined weights for each of the many variables into a single weighted value representing the yield information; and mapping the yield to one or more sitespecific use maps in the delimited field. The disadvantage of the above solution is that the input data does not include parameters for the topography of the delimited areas and calculates a weighted average of all input data for management zones, which includes irrelevant outliers.

[0019] The aim of this procedure is to identify the heterogeneity of the arable land in terms of soil and fertility and, on the basis of this information, to produce precision maps at a resolution of 10x10 m, which can be executed by means of field machinery equipped with positioning systems, in 10x10 m cells.

[0020] The aim is to accurately identify the variability in soil properties resulting from the topography of the land.

[0021] The aim is to reduce pressure on the environment (preventing leaching of easily mobilisable forms of nutrients, reducing unnecessary chemical inputs, preventing under- and over-irrigation, etc.), thus reducing the ecological footprint of agriculture, making farming more sustainable while generating additional income in the longer and shorter term.

[0022] The aim is to help users / producers plan field operations and implement precision farming by creating an interactive interface that manages the data generated during farming in a complex system and supports users / producers in making decisions.

[0023] The purpose is to produce a series of maps of the land in the examined area, providing accurate and detailed information for users / producers.

[0024] The objectives are achieved by processing spatial information to produce digital maps with a resolution of 10x10 m, the base map and the multipolygon zone map, containing the characteristics and attributes of the examined land, i.e. the field. Based on the combined assessment of the topographic parameters provided by the base map, selected satellite images, soil fertility data and soil sampling data from the multipolygon zone map, we will produce prescription maps, applicable to the 10x10 m cells of the field, which will be used to apply just enough nutrients, seeds, pesticides and water to the field to ensure that they are still used without causing any environmental pollution.

[0025] In the main steps of the procedure according to the invention, a special selection and processing of satellite images of a given land area, a contoured field, over many years, and the calculation of relevant terrain parameters (DEM, Slope degree, PDD) are used to create a vector layer of polygons with a resolution of 10x10 m, the base map, which contains the terrain parameters and data, attributes indicating the fertility of the examined land (e.g. Based on this base map, a multivariate cluster analysis is used to produce a multi-polygon zone map of the land, on the basis of which soil sampling and soil analysis measurements are carried out.

[0026] The base map and the multipolygon zone map, as well as the soil analysis results for the soil samples taken from the multipolygon zone map, are uploaded to a database and evaluated.

[0027] Using the resulting data, we produce a fertilisation prescription map, a sowing prescription map, a crop protection prescription map, an irrigation prescription map and a tillage prescription map.

[0028] To produce a fertilisation prescription map, a nutrient balance calculation is carried out for each 10x10 m cell based on soil analysis results and fertility. The calculated active ingredient requirement is also stored in the database and can be used to calculate the fertiliser requirement for the selected fertilisers. The calculated fertiliser requirement, the fertiliser prescription map, is also stored in the database and is used as a basis for the fertilisation operations that can be carried out by the field machines for the 10x10 m cells, based on the 10x10 m resolution prescription map.

[0029] The yield target for a given 10x10 m cell for a given crop species is the basis for the production of the sowing prescription map. The plant density is determined according to the plant density response of the variety / hybrid. The determined plant density recommendation is stored in the database and can be used to produce a 10x10 m resolution crop prescription map.

[0030] The results of the soil analysis following the multipolygon zoning are used as a basis for differentiated plant protection interventions. The phytotoxicity of preemergent herbicides is influenced by the colloid content of the soil. Soil colloids, such as clay minerals and humus, adsorb part of the active ingredient of the herbicide, while the other part is present in the soil solution and can be taken up by the plants. The ratio between the amount of herbicide adsorbed and the amount of herbicide in solution is therefore influenced by the colloid content, mainly humus, and the herbicide dose can be used to produce a crop protection prescription map accordingly. According to the invention, the basis for precision irrigation management is a grid with a resolution of 10x10 m, each cell of which can be characterised by fertility and topography parameters. Based on the spatial variability of these parameters, we delimit the multipolygon zones, which are well characterised by soil heterogeneity, with similar soil physical properties and water management properties within the zones, but with significant differences between the zones. Soil moisture measurements in the stand, which are essential for irrigation management, are therefore carried out continuously throughout the irrigation season in each multipolygon zone. Soil moisture is measured in the upper 60 cm soil layer every 10 cm at 5 minute intervals. The measurement principle is capacitive soil moisture measurement. The signal transmitted by the sensor is proportional to the volumetric percentage soil moisture content, which is thus converted to volumetric percentage soil moisture content by means of the appropriate calibration curve. The irrigation data are entered into the database and used to determine the irrigation norm, thus preventing waterlogging, acidification or secondary salinisation due to over-irrigation.

[0031] Variable depth tillage and variable intensity tillage provide the possibility to take into account the different erosion exposures within the field and to adapt the management of stem residues to the topography. Shallower and less intensive tillage reduces erosion directly due to less disturbance and indirectly due to more stem residues remaining on the field. Multipolygon zoning is used to delimit zones that are considered homogeneous in terms of topographic parameters, and thus identical in terms of slope steepness and fragmentation (PDD), so that the characteristic value of these parameters within a zone can be used to determine the erosion exposure of the zone and the depth / intensity of cultivation applied there. The resulting prescription map can be implemented by properly equipped tillage machines and will reduce soil degradation.

[0032] The main steps of the invention procedure are:

[0033] First, the base map is prepared in digitised form, then, using the attributes of the base map, the multipolygon zone map is prepared, on the basis of which soil sampling and soil analysis measurements are carried out and, after data processing, the prescription maps are prepared. The base map is a vector map containing the geometric elements and attributes needed to produce precision prescription maps.

[0034] When making a base map, we first define the exact area of land, or field, for which we are making a prescription map. The contour of the field can be determined from data received from the farmer or on the soil sampling car's tablet before soil sampling. The information used to create the base map, i.e. satellite images and terrain parameters, is selected for the field delimited by a contour.

[0035] A vector layer with a resolution of 10x10 m is created for the contoured field, which is an accurate representation of the 10x10 m cells of the land. The attribute table of this map, i.e., the base map, contains the information needed to delimit the multipolygon zones and the attributes characterising the fertility: the NDVI values of the NDVI images of the selected, cloud-free, developed vegetation, the digital elevation model (DEM) of the field and the terrain parameters that can be derived from it.

[0036] The selection and filtering of NDVI images to be included in the attribute table of the base map consists of the following steps:

[0037] First, the contour of the field is transformed to the UTM (Universal Transverse Mercator) projection of the overlying Sentinel tile, and then the 45 m and 10 m buffers are created separately for the contour, to filter out the NDVI images that do not match. The 45-meter buffer is used to filter out cloud-free images, and the 10-meter buffer is used to filter out images representative of the developed stand.

[0038] Cuts are made from the raster layers of the B10 band of the Sentinel satellite images using the 45 m buffer, within which we check for the presence of a value of 1 (cloud pixel, cloud mask layer) or image without data (NODATA). If either is present, the image is not used in subsequent processes.

[0039] In the next step, the 10 m buffer is used to extract NDVI raster layers generated from the B4 and B8 bands of the Sentinel satellite images. If the NDVI value of the 10 m buffered clips is lower than 0.4, the image is discarded from the subsequent analysis. Within the original contour (without buffer), NDVI values are extracted from the grids of all Sentinel images considered as matching the previous two filters and are then put into a vector layer of 10x10 m polygons with the same geometry as the Sentinel image grid In the attribute table of the finished vector layer, the NDVI values from the NDVI images taken at the given time points are examined and a field average is calculated from these values for the given time points, and if this value is below 0.7, it is deleted from the attribute table.

[0040] The next step in is to take annual averages for each cell of the NDVI values that have passed the previous filters, and then take the annual averages to form the multi-year NDVI average (NDVI_mean).

[0041] After determining the multi-year NDVI average, the topographic parameters are determined.

[0042] Based on the point vector layer containing the elevation data of the field, a digital elevation model (DEM) of the field is created using interpolation. From the DEM, a slope degree map is created.

[0043] For each cell, we determine the direction of maximum elevation drop (flow direction), and then construct the flow accumulation map by summing for each cell the number of cells in which the direction of maximum elevation drop points to that cell. If this number is high, it indicates the formation of a water flow.

[0044] From the flow map (flow accumulation), we select the cells with a flow accumulation greater than 20 (for 10x10 m cells), i.e. the number of cells with the highest elevation drop pointing to the given cell greater than 20, and construct a flow network map (Flow 20).

[0045] For each cell, we sum up the number of image elements within a circle of radius 5 cells of the flow network map (Flow20) affected by water flows, to obtain the result map, the complex parameter characterising the probability of water confluences, the potential drainage density (PDD) parameter.

[0046] The DEM, Slope and PDD values are added as additional attributes to the base map vector layer.

[0047] In the next step, we determine the percentage deviation of the NDVI mean of the cells from the field mean (we take the field mean as 100), i.e. the relative fertility.

[0048] The relative fertility for the nthcell is calculated using the following equation: NDVI_meann

[0049] Relative fertilityn= — — - * 100

[0050] / i NDVI_meann\

[0051] \ N

[0052] Where NDVI_mearin is the NDVI_average of the nthcell.

[0053] N: total number of cells

[0054] As a next step, we determine the fertility classes based on the multi-year average NDVI value, and enter the yield target per class for the planned yield in the attribute table of the base map.

[0055] For the cells, 15 fertility classes are defined, based on the multi-year average NDVI values (range). The lowest fertility class 1 is for cells with multi-year average NDVI values less than 0.6333; the highest fertility class 15 is for cells with multi-year average NDVI values greater than 0.9547.

[0056] For each fertility class, we assign typical yield targets for the main arable crops, such that the yield difference between successive classes for the main arable crops is: 1 t / ha maize, 0.7 t / ha winter wheat, 0.35 t / ha sunflower, 0.4 t / ha winter coleseed, 0.3 t / ha soybean and 0.6 t / ha winter barley.

[0057] After preparing the base map, the next main step is to create the multipolygon zonal map, which is the basis for the zonal sampling map, combining information from the multi-year NDVI average and topographic parameters to characterise fertility. We run multivariate cluster analysis resulting in spatially non-contiguous zones and the optimal number of zones is determined by taking into account the pseudo-F statistics (CALINSKI, T. & HARABASZ, J., 1974. A dendrite method for cluster analysis. Communications in Statistics. 3. 1 -27.).

[0058] The variables included in the cluster analysis are: multi-year NDVI mean (NDVI_mean), DEM, Slope degree, and PDD for each cell.

[0059] We define the zones that can be considered homogeneous with respect to the four variables included and denote each zone by a different number .

[0060] Zones of unsuitable size for sampling are considered to be those with an area of less than 0.5 ha. Such small zones are merged into adjacent zones where the area of the zone into which they are merged does not exceed 5 ha and their combined area after merging is greater than 0.5 ha but less than 5 ha. Zones of more than 5 ha are divided into zones of more than 0.5 ha but less than 5 ha.

[0061] The base map and the multipolygon zone map are loaded into the database.

[0062] In the next step, we will carry out a soil analysis of the field based on the multipolygon zone map of the field, make a composite sample of 20-25 subsamples per zone and then carry out laboratory tests.

[0063] The soil sampling vehicles are equipped with an automatic soil drill, a tablet and an automated label printer. Data for sampling, such as the multipolygon zone map, is fed into the soil sampling application on the tablet from the database and sampling documentation, such as access routes and sampling points, is saved directly to the database. The printing of a label to identify the soil samples is automatic.

[0064] A sample record containing the data required for the laboratory test is automatically generated on the car's tablet when the sampling is completed.

[0065] Soil samples are identified by means of a sample list and a label on the samples.

[0066] The following parameters will be determined during the laboratory analysis: PHKCI, Arany’s Plasticity Index (KA ), total water soluble salts, humus, lime content, AL-P2O5, AL-K2O, AL-Na, KCI-soluble NO3+NO2-N, KCI-soluble Mg, KCI-soluble SO4-S, EDTA- Mn, EDTA-Zn, EDTA-Cu.

[0067] The laboratory test results and other data needed for the calculation are loaded into the database.

[0068] Other data required for the calculation, such as the crop grown, the planned yield, information on the previous crop, and data on organic fertilisation (date, dose, type) are also uploaded to the database and the soil analysis results are stored in the multipolygon zones.

[0069] Data processing, displaying and data visualisation are done in PrecZone applications.

[0070] We use the data in the database to produce prescription maps for 10x10 m cells of the examined area, e.g. fertilisation, sowing, crop protection, irrigation and tillage prescription maps,

[0071] Examples:

[0072] In examples 1 -7, the same field of 26.5 ha with a well-defined contour was considered, with a planned maize yield of 11 .3 t / ha. There were no legumes as previous crop on the field, no stem residue was left on the field and no organic fertiliser was used.

[0073] Example 1 : Presentation of the base map

[0074] Figure 1 shows a map of the multi-year average NDVI of the examined field, which is well correlated with yield potential.

[0075] The figure shows that there are significant differences in the multi-year average NDVI values within the field, which is related to the maize yield. The higher the multi-year NDVI average, the higher the expected yield.

[0076] The digital elevation model is presented first in the representation of maps with topographic parameters.

[0077] Figure 3 shows the digital elevation model (DEM) of this field.

[0078] The elevation data for the digital elevation model were collected using a combine equipped with an antenna capable of receiving RTK correction. With RTK, an accuracy of 2.5 cm can be achieved horizontally. The measurement of the "z" coordinate provides an accurate elevation difference within a field, which makes it suitable for the preparation of a topographic model, but does not provide accurate elevation data in absolute value. The elevation data are shown in Figure 2.

[0079] As shown in Figure 3, there is a difference in elevation of almost 2 m within the field, which may affect soil properties and water management characteristics.

[0080] From the elevation points, a digital copy of the field is constructed using spatial interpolation to create a slope degree map, shown in Figure 5, and then the direction of the water flows is determined, the maximum elevation drop and the probability of waterlogging are determined for each cell. To do this, we first determine the flow accumulation map, from which we construct the flow network map (Flow 20), shown in Figure 4, and then the result map shown in Figure 6, the complex parameter characterising the probability of waterlogging, the potential drainage density (PDD). The maps above show that certain areas of the field are more likely to be subject to damaging waterlogging (the darkest colours on the maps indicate these areas), water run-off, which is important to know for both soil formation and fertility.

[0081] Example 2: the multipolygon zone map and soil analysis

[0082] Based on the multi-year NDVI mean (NDVI_mean), DEM, Slope degree and PDD variables for the cells, a multivariate cluster analysis was used to construct a multi-line zonal map of the area .

[0083] The study area can be divided into eight non-contiguous multipolygon zones as shown in Figure 7.

[0084] Based on the multipolygon zone map, the soil of the field is analysed, with 20-22 readings per zone forming the average sample for zone characterisation. The soil of the field was analysed for the parameters PHKCI, Arany’s Plasticity Index (KA), total water soluble salts (%), humus, lime content (%), AL-P2O5, AL-K2O, AL-Na, KCI- soluble NO3+NO2-N, KCI soluble Mg, KCI soluble SO4-S, EDTA-Mn, EDTA-Zn, EDTA- Cu.

[0085] The soil analysis results are shown in Table 1 , broken down by multipolygon zone.

[0086] Table 1 .

[0087] Example 3: Nutrient management planning and fertiliser prescription map This example shows the steps in nutrient management planning, setting the yield target and determining the nutrient requirements.

[0088] Nutrient management planning starts with setting a yield target and is done at the 10 x 10 metre cell level, so that we can properly estimate the nutrient uptake of the crop. We determine the yield target for each cell, based on the multiple-year average NDVI values of the field, by fertility class.

[0089] Figure 8 shows the projected yields per cell for a given field.

[0090] After the yield target is set, the soil analysis results are used to calculate the active ingredient requirement for each cell and to determine the zone and field averages, which are shown in Table 2.

[0091] An algorithm is used to calculate the active ingredient demand.

[0092] For each cell, the specific active ingredient requirement (active ingredient requirement for 1 t of yield and associated secondary yield) is determ ined for the given plant species and the multipolygon zone in the cell area, depending on the soil nutrient supply.

[0093] For each cell and for each nutrient, the specific nutrient requirement is calculated by multiplying the specific nutrient requirement by the yield target.

[0094] The calculated value is corrected for the active substance content of the pre-sowing stalk residue, the active substance content of the organic fertiliser and the amount of N fixed by the legume pre-sowing.

[0095] Active ingredient needn= fnx Yield targetn± Corrections

[0096] Where:

[0097] Active ingredient needn: active ingredient need calculated for the nth cell fn: corrected specific active ingredient need calculated for the nth cell Yield targetn: yield planned for the nth cell

[0098] The calculated active ingredient requirement for each cell per nutrient is entered into the database.

[0099] Table 2.

[0100] The data in Table 2 show the amount of each active ingredient needed per zone.

[0101] The calculated active ingredient demand is known not only by field or zone, but also with very high accuracy at the 10 x 10 meter cell level. This allows us to produce a fertilisation prescription map with the same precision in the next step, where we determine which fertilisers we want to cover the active ingredient demand and in which proportions. The fertiliser demand for each zone, for each cell, is saved in the database.

[0102] In Table 1 , we can see that the phosphorus supply of zone 4 and zone 1 is "very good", zone 3 is "excessive", zone 2 is medium and zones 5-8 are poor. The P requirement is provided by MAP fertiliser. Table 2 shows the recommended amount of MAP fertiliser based on the phosphorus supply in each zone, which is used to calculate the MAP fertiliser requirement per cell in Figure 9.

[0103] Example 4: Sowing prescription map Plant density is planned on the basis of a yield target that characterises fertility. According to the yield target, three classes have been established, which are adapted to the specific plant density response of the maize hybrid: cells with an expected yield below 10 t / ha are sown at 75000 seeds / ha, cells with a yield between 10-11 t / ha are sown at 80000 seeds / ha, and cells with a yield above 11 t / ha are sown at 85000 seeds / ha. The sowing prescription map is shown in Figure 10.

[0104] Example 5: Plant protection prescription map

[0105] Total herbicide weed control is only carried out on weedy areas of the field. Determination of which parts of the field have vegetation is based on the current NDVI record, which is shown in Figure 11 .

[0106] The NDVI for bare soil is 0.3, i.e. cells with NDVI values higher than this are treated with herbicides, while the other cells are not. The herbicide prescription map is shown in Figure 12.

[0107] Example 6: Irrigation prescription map

[0108] The average NDVI values and topography characteristics of the multipolygon zones are shown in Table 3. It can be observed that the irrigation water demand of zones 3, 6 and 7, which are located in the lower parts of the field and are prone to waterlogging, is different from that of the higher zones, from which surface and groundwater movement can be initiated towards the former, thus causing waterlogging over the whole area of the field as a result of a uniform irrigation water norm.

[0109] Table 3. To avoid this phenomenon, a reduced irrigation water norm (75%) is applied in zones 3, 6 and 7. Irrigation launching is determined by measuring the soil moisture per zone, while the irrigation norm is determined by taking into account the above-mentioned characteristics of the multipolygon zones. The prescription map for the irrigated areas of the field is shown in Figure 13.

[0110] Example 7: Tillage prescription map

[0111] Based on the slope conditions of the examined field (Figure 3), no significant erosion is expected, but the higher parts of the field (zones 1 and 2) have shallower topsoil, i.e. deep tillage may cause the mixing of subsoil and topsoil. To avoid this, these fields are not cultivated deeper (30 cm), but only with shallow primary tillage (20 cm). The tillage prescription map is shown in Figure 14.

Claims

Patent claims1 . Procedure for the production of a high accuracy precision prescription maps for arable land, whereby a given area of land is divided into 10 x 10 m cells, a base map and a multipolygon zone map are produced, soil sampling and soil analysis measurements are carried out, and after data processing, precision maps are produced in a way that the base map is created as a vector layer of 10 x 10 m resolution of the contoured field, as an exact mapping of the 10 x 1 0 m cells of the field, resulting in a vector base map that contains NDVI images, topographic parameters and fertility data, attributes in digital form, which are used to create the multipolygon zone map, the base map and the multipolygon zone map are uploaded to a database, soil sampling is carried out on the basis of the multipolygon zone map, assigned to multipolygon zones, soil analysis measurements are carried out on the soil samples and the results are also uploaded to the database, the obtained soil analysis results are evaluated and used to produce a fertilisation prescription map, a sowing prescription map, a crop protection prescription map, an irrigation prescription map and a tillage prescription map; when creating the base map, the contour of the field is determined as a first step, then, after transforming the Sentinel tiles onto the Universal Transverse Mercator projection of the field contour, the NDVI images of the field are analysed, sorted and filtered and the resulting data are introduced as attributes into the base map vector layer; when determining the topographic parameters, we first create a digital elevation model of the field, and from the digital elevation model a slope steepness map is created, then for each cell, we determine the direction of maximum elevation drop and a runoff map, and then a result map is created with a complex parameter characterising the probability of waterlogging, the complex parameter values for the digital elevation model, slope steepness and the probability of waterlogging are introduced as additional attributes in the base map vector layer; and then, based on the base map, we use multivariate cluster analysis to construct the multipolygon zone map and assign different numbers to the multipolygon zones, characterised bythe selection and filtering of NDVI images to be included in the base map attribute table consists of the following steps: preparing a 45 m and a 10 m buffer for the contour separately, first, we use the 45 m buffer to make cuts from the B10 bands of the Sentinel satellite images, and images with a value of 1 or no data cells are omitted from further analysis, in the next step, cuts are made from the NDVI raster layers generated from the B4 and B8 bands of the Sentinel satellite images using the 10 m buffer, if the NDVI of the 10 m buffer extracted images is lower than 0.4, the image is excluded from further analysis, within the original contour, NDVI values are extracted from the grids of all Sentinel images left over from the previous two filters and are then date-stamped into a finished vector layer of 10 x 10 m polygons with the same geometry as the Sentinel image grid, in the attribute table of the finished vector layer, the NDVI values from the NDVI images taken at the given time points are examined and a field average is calculated from these values for the given time points, and if this value is below 0.7, it is deleted from the attribute table, annual averages are then calculated for each cell from the NDVI values that have passed the previous filters, and then the annual averages are used to calculate a multiyear NDVI average; from the runoff map, we select the cells for which the number of cells with the highest elevation drop pointing towards the given cell is greater than 20, and create a runoff network map from these cells, for each cell, we sum up the number of cells within a circle of radius 5 cells of the water flow network map affected by water flows to obtain the result map representing the complex parameter characterising the probability of waterlogging; in the next step, we determine the percentage deviation of the cells' multi-year average NDVI values from the field average, the relative fertility, and this data is put into base map, we define fertility classes based on the multi-year average NDVI, and determine the yield target for the planned yield per class and enter these in the attribute table; to determine the multipolygon zones, multivariate cluster analysis is performed for each cellbased on the complex parameter variables characterising the multi-year NDVI average, the digital elevation model, the slope steepness probability and the probability of waterlogging; finally, the fertilisation prescription map, sowing prescription map, crop protection prescription map, irrigation prescription map and tillage prescription map are prepared for the 10 x 10 m cells of the field.

2. The procedure according to claim 1 , characterised by that each multipolygon zone is homogeneous with respect to the multi-year NDVI average, the digital elevation model, the slope steepness, and the complex parameter characterising the probability of waterlogging.

3. The procedure according to claim 1 , characterised by that 15 fertility classes are established for the cells.

4. The procedure according to claim 1 , characterised by that the soil analysis comprises determining the parameters PHKCI , Arany’s Plasticity Index, total water- soluble salts, humus, lime content, AL-P2O5 , AL-K2O, AL-Na, KCI-soluble NO3+NO2-N, KCI-soluble Mg, KCI-soluble SO4-S, EDTA-Mn, EDTA-Zn, EDTA-Cu.

5. The procedure according to claim 1 , characterised by that the printing of a label for identifying the soil samples is automatic when the soil samples are taken.

6. The procedure according to claim 1 , characterised by that zones smaller than 0.5 ha are merged into adjacent zones whose combined size after merging is greater thanO.5 ha but less than 5 ha.

7. The procedure according to claim 1 , characterised by that the zones larger than 5 ha are divided into zones larger than 0.5 ha but smaller than 5 ha.

Images