Precise fertilization method and device for potatoes

Abstract

The invention provides a precise fertilization method and device for potatoes. The method comprises the following steps: acquiring a remote sensing image of a target growth stage of the potatoes; inputting the remote sensing image into a trained nutrient absorption amount inversion model, and obtaining the actual nutrient absorption amount corresponding to each pixel output by the nutrient absorption amount inversion model; based on the total nutrient demand and the actual nutrient absorption amount corresponding to each pixel, determining the nutrient deficiency amount corresponding to each pixel; and based on the nutrient deficiency amount corresponding to each pixel, by taking the plot corresponding to the pixel as a unit, carrying out precise fertilization on the potatoes. According to the precise fertilization method and device for the potatoes, the QUEFTS model and nutrient deficiency remote sensing diagnosis are combined, the nutrient deficiency amount corresponding to each pixel is determined, precise fertilization is conducted on the potatoes with the land parcels corresponding to the pixels as units, and the utilization rate of the potato fertilizer is increased.

Description

Technical Field

[0001] The present invention relates to the field of agricultural information technology, and particularly to a precise potato fertilization method and device. Background Art

[0002] Reasonable fertilization is the core strategy to improve nutrient use efficiency and reduce nutrient loss.

[0003] Current potato fertilization strategies mainly include: traditional soil testing and formulated fertilization method, nitrate testing method, chlorophyll diagnosis method, and canopy nitrogen nutrition diagnosis method. Among them, the traditional soil testing and formulated fertilization method has a long cycle and cannot adjust the fertilization strategy in time; although the nitrate testing method and chlorophyll diagnosis method can provide nitrogen management suggestions in a short time, they can only adjust the topdressing amount and cannot accurately determine the demand for basal fertilizer. The sensitivity of the canopy nitrogen diagnosis method decreases in the later stage of crop growth, especially its application effect is limited in high-density crop populations.

[0004] In summary, traditional potato fertilization methods rely on technologies such as soil testing for fertilization, nitrate detection, and chlorophyll diagnosis, and have disadvantages such as long nutrient monitoring cycle, poor real-time performance, and inability to dynamically adjust the fertilization amount, resulting in technical problems of low fertilizer utilization rate and serious nutrient loss. Summary of the Invention

[0005] The present invention provides a precise potato fertilization method and device to solve the technical problems of low fertilizer utilization rate and serious nutrient loss in the prior art.

[0006] The present invention provides a precise potato fertilization method, including: Obtaining remote sensing images of the target growth stage of potatoes; Inputting the remote sensing images into a trained nutrient uptake inversion model to obtain the actual nutrient uptake corresponding to each pixel output by the nutrient uptake inversion model; the nutrient uptake inversion model is trained based on remote sensing images of different growth stages of potatoes and the actual nutrient uptake in the plots corresponding to each pixel of the remote sensing images; Based on the total nutrient demand and the actual nutrient uptake corresponding to each pixel, determining the nutrient deficit corresponding to each pixel; the total nutrient demand is obtained based on the QUEFTS model; Based on the nutrient deficit corresponding to each pixel, precisely fertilizing the potatoes in units of the plots corresponding to the pixels.

[0007] In some embodiments, it further includes: Obtaining potato yield data under different fertilization gradient treatments; Based on the potato yield data under the different fertilization gradient treatments, the total nutrient demand is determined using the QUEFTS model.

[0008] In some embodiments, the determining the total nutrient demand using the QUEFTS model based on the potato yield data under the different fertilization gradient treatments includes: Based on the potato yield data under the different fertilization gradient treatments, the QUEFTS model is used to fit the nutrient absorption curve of the potato ground object; Based on the nutrient absorption curve of the potato ground object, the optimal target yield of the potato, as well as the nutrient absorption amount of the above-ground part and the nutrient absorption amount of the underground part of the potato under the optimal target yield are determined; Based on the nutrient absorption amount of the above-ground part of the potato and the nutrient absorption amount of the underground part of the potato, the total nutrient demand is determined.

[0009] In some embodiments, it further includes: Obtaining remote sensing images of different growth stages of the potato, as well as the actual nutrient absorption amount in the plot corresponding to each pixel of the remote sensing image; Based on the remote sensing images of different growth stages of the potato and the actual nutrient absorption amount in the plot corresponding to each pixel of the remote sensing image, an integrated learning algorithm is used for model training to obtain the nutrient absorption amount inversion model.

[0010] In some embodiments, the integrated learning algorithm includes any one of the following algorithms: Random forest algorithm; AdaBoost algorithm; Gradient Boosting algorithm; XGBoost algorithm; GBDT algorithm.

[0011] In some embodiments, the precise fertilization of the potato in units of the plot corresponding to the pixel based on the nutrient deficit amount corresponding to each pixel includes: Based on the nutrient deficit amount corresponding to each pixel, the fertilization amount corresponding to each pixel is determined; In units of the plot corresponding to the pixel, precise fertilization is performed on the plot corresponding to each pixel according to the fertilization amount corresponding to each pixel.

[0012] The present invention also provides a potato precise fertilization device, including: An acquisition module for acquiring remote sensing images of the target growth stage of the potato; An inversion module, configured to input the remote sensing image into a trained nutrient absorption amount inversion model, and obtain the actual nutrient absorption amount corresponding to each pixel output by the nutrient absorption amount inversion model; the nutrient absorption amount inversion model is obtained by training based on remote sensing images of different growth stages of potatoes and the actual nutrient absorption amount in the plot corresponding to each pixel of the remote sensing image; A determination module, configured to determine the nutrient deficit amount corresponding to each pixel based on the total nutrient demand and the actual nutrient absorption amount corresponding to each pixel; the total nutrient demand is obtained based on the QUEFTS model; A fertilization module, configured to perform precise fertilization on potatoes in units of the plots corresponding to the pixels based on the nutrient deficit amount corresponding to each pixel.

[0013] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, where when the processor executes the computer program, the potato precise fertilization method as described in any one of the above is implemented.

[0014] The present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the potato precise fertilization method as described in any one of the above is implemented.

[0015] The present invention also provides a computer program product, including a computer program, and when the computer program is executed by a processor, the potato precise fertilization method as described in any one of the above is implemented.

[0016] The potato precise fertilization method and device provided by the present invention combine the QUEFTS model with remote sensing diagnosis of nutrient deficit, determine the nutrient deficit amount corresponding to each pixel, and perform precise fertilization on potatoes in units of the plots corresponding to the pixels, improving the utilization rate of potato fertilizers. Description of the Drawings

[0017] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention, and for those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0018] Figure 1 It is a schematic flowchart of the potato precise fertilization method provided by the present invention.

[0019] Figure 2 It is a schematic diagram of the potato precise fertilization principle provided by the present invention.

[0020] Figure 3It is one of the schematic diagrams of the nutrient absorption curves of potatoes under different potential yields fitted by the QUEFTS model provided by the present invention.

[0021] Figure 4 It is the satellite remote sensing image of the potato plants during the tuber formation period provided by the present invention.

[0022] Figure 5 It is one of the inversion result diagrams of PNA, PKA and PPA of the Planet image of the potato field during the tuber formation period provided by the present invention.

[0023] Figure 6 It is the monitoring diagram of the nutrient deficiency amount of potatoes provided by the present invention.

[0024] Figure 7 It is one of the fertilization decision diagrams of nitrogen, phosphorus and potassium for potatoes during the tuber formation period provided by the present invention.

[0025] Figure 8 It is the schematic diagram of the potato yield and harvest index provided by the present invention.

[0026] Figure 9 It is the schematic diagram of the relationship between potato yield and nutrient absorption of the above-ground potato plants under different parameter a and d values provided by the present invention.

[0027] Figure 10 It is the second schematic diagram of the nutrient absorption curves of potatoes under different potential yields fitted by the QUEFTS model provided by the present invention.

[0028] Figure 11 It is the result diagram of the correlation analysis between the yield and different parameter indicators provided by the present invention.

[0029] Figure 12 It is the construction and verification accuracy diagram of the inversion models of PNA, PPA and PKA of the potato Planet image based on the RF algorithm provided by the present invention.

[0030] Figure 13 It is the inversion result diagram of PNA, PPA and PKA of the potato plot Planet image based on the RF algorithm provided by the present invention.

[0031] Figure 14 It is the second inversion result diagram of PNA, PKA and PPA of the potato field Planet image during the tuber formation period provided by the present invention.

[0032] Figure 15 It is the second fertilization decision diagram of nitrogen, phosphorus and potassium for potatoes during the tuber formation period provided by the present invention.

[0033] Figure 16 It is the structural schematic diagram of the potato precision fertilization device provided by the present invention.

[0034] Figure 17 It is a schematic structural diagram of the electronic device provided by the present invention. Detailed implementation manners

[0035] To make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below with reference to the accompanying drawings in the present invention. Apparently, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present invention without making creative efforts shall fall within the protection scope of the present invention.

[0036] Figure 1 It is a schematic flow diagram of the potato precise fertilization method provided by the present invention. As Figure 1 shown, the method includes the following: Step 101: Obtain the remote sensing image of the target growth stage of the potato.

[0037] Step 102: Input the remote sensing image into the trained nutrient absorption amount inversion model to obtain the actual nutrient absorption amount corresponding to each pixel output by the nutrient absorption amount inversion model; the nutrient absorption amount inversion model is trained based on the remote sensing images of different growth stages of the potato and the actual nutrient absorption amount in the plot corresponding to each pixel of the remote sensing image.

[0038] Step 103: Determine the nutrient deficit amount corresponding to each pixel based on the total nutrient demand and the actual nutrient absorption amount corresponding to each pixel; the total nutrient demand is obtained based on the QUEFTS model.

[0039] Step 104: Based on the nutrient deficit amount corresponding to each pixel, perform precise fertilization on the potato with the plot corresponding to the pixel as a unit.

[0040] In some embodiments, it further includes: Obtain the potato yield data under different fertilization gradient treatments; Based on the potato yield data under different fertilization gradient treatments, use the QUEFTS model to determine the total nutrient demand.

[0041] Specifically, Figure 2 It is a schematic diagram of the potato precise fertilization principle provided by the present invention. As Figure 2As shown in the figure, the present invention evaluates soil fertility and crop yield based on the QUEFTS model, determines the characteristics of the crop nutrient uptake curve, the optimal target yield, and the best nutrient ratios of nitrogen, phosphorus, and potassium in the study area, and is used for calculating the precise fertilization amount. By correlation analysis, indicators significantly affecting yield (PNA, PPA, PKA) are screened, the nitrogen, phosphorus, and potassium contents of potatoes at different growth stages are retrieved using remote sensing images, and a potato fertilization model driven by remote sensing monitoring is constructed in combination with the QUEFTS model. This model monitors the nutrient status of the field area through satellite remote sensing data, generates precise fertilization prescriptions, realizes precise fertilization during the critical growth stages of potatoes, improves fertilizer utilization efficiency, and reduces environmental pollution.

[0042] Specifically, basic agronomic data is obtained first.

[0043] For the study area, a crop nitrogen (N), phosphorus (P), and potassium (K) nutrient gradient experiment is carried out to obtain crop planting-related data, including soil nutrient information, fertilization amount information, crop nutrient information at different growth stages of the crop, biomass information, crop yield, harvest index, etc.; relevant data can also be collected and sorted according to historical materials.

[0044] Through the crop nitrogen, phosphorus, and potassium nutrient gradient experiment in the study area, crop planting-related data is systematically obtained, including: (1) soil nutrient information (such as available nitrogen, organic matter, available potassium, available phosphorus, pH value, etc.); (2) fertilization amount information (total amount of nitrogen, phosphorus, and potassium fertilizers, base fertilizer and top dressing amounts); (3) crop nutrient information (above-ground nitrogen, phosphorus, and potassium contents at critical growth stages and nitrogen, phosphorus, and potassium contents in tubers at maturity); (4) crop biomass information (fresh and dry weights of each organ); (5) yield information (economic yield (yield of the harvestable part of the crop) and biological yield (biomass information of the above-ground part of the crop at maturity)). These data can be collected through on-site experiments or sorted from historical materials, providing data support for studying the relationship between crop nutrient uptake and yield.

[0045] Then, based on the experimental data of the study area obtained or the historical literature data sorted out, the nitrogen, phosphorus, and potassium nutrient uptake characteristics of the crop are analyzed, the crop yield and the nitrogen, phosphorus, and potassium uptake efficiency in the study area are evaluated; the QUEFTS model is used to simulate the upper and lower limit correspondence between the nitrogen, phosphorus, and potassium nutrient uptake of the crop and the achievable yield, establish the corresponding yield range under the limitation of nitrogen, phosphorus, and potassium nutrient elements, obtain the crop nutrient uptake curve under different potential yields of the crop, and determine the optimal target yield of potatoes in the study area and the best nitrogen, phosphorus, and potassium requirements of the crop based on the analysis of the curve characteristics.

[0046] In some embodiments, the total nutrient demand is determined using the QUEFTS model based on the potato yield data under different fertilization gradient treatments, including: Based on the potato yield data under the different fertilization gradient treatments, use the QUEFTS model to fit the nutrient absorption curve of potato ground objects; Based on the nutrient absorption curve of the potato ground objects, determine the optimal target yield of the potato, as well as the nutrient absorption amount of the above-ground part of the potato and the nutrient absorption amount of the underground part of the potato under the optimal target yield; Based on the nutrient absorption amount of the above-ground part of the potato and the nutrient absorption amount of the underground part of the potato, determine the total nutrient demand.

[0047] Specifically, first analyze the internal efficiency of nutrients of the potato.

[0048] Based on the field test data, this invention analyzes the tuber yield and nutrient absorption characteristics of the potato. By collecting the yield data under different fertilization gradient treatments and combining with the nutrient absorption amount at the crop growth stage, evaluate the yield of the potato and the absorption efficiency of nitrogen, phosphorus, and potassium.

[0049] Internal efficiency of nutrients (IE, kg / kg): It refers to the yield produced by the crop absorbing a unit of nutrients (N, P, K), reflecting the efficiency of the nutrients absorbed by the above-ground part of the crop in forming the yield. According to the data of the nitrogen, phosphorus, and potassium nutrient accumulations ( , , )(unit: kg / ha) of the above-ground part of the crop and the data of the tuber yield M (unit: kg / ha) of the crop, use formula (1-3) to calculate the internal efficiency parameters IE_N, IE_P, and IE_K of nitrogen, phosphorus, and potassium nutrients of the above-ground part of the potato respectively. This parameter is used to characterize the tuber yield produced by the above-ground part of the plant absorbing 1 kg of nitrogen, phosphorus, and potassium nutrients.

[0050] IE_N= (1) IE_P= (2) IE_K= (3) M is the tuber yield of the crop (kg / ha); is the nitrogen, phosphorus, and potassium nutrient accumulations of the above-ground part of the potato (kg / ha).

[0051] According to the nitrogen, phosphorus, and potassium nutrient accumulations ( , , )(unit: kg / kg) data of the underground tubers of the crop and the data of the tuber yield M (unit: kg / ha) of the crop, use formula (4-6) to calculate the internal efficiency of nitrogen, phosphorus, and potassium nutrients of the underground tubers of the potato respectively.Tuber IE_N ,Tuber IE_P , Tuber IE_K A parameter, which is used to characterize the tuber yield produced by the crop per 1 kg of nitrogen, phosphorus, and potassium nutrients absorbed.

[0052] Tuber IE_N = (4) Tuber IE_P = (5) Tuber IE_K = (6) M is the tuber yield of the crop (kg / ha); is the cumulative amount of nitrogen, phosphorus, and potassium nutrients in potato tubers (kg / ha).

[0053] Nutrient absorption for producing 1000 kg of grain (Reciprocal Internal Efficiency, RIE): It refers to the total amount of nitrogen, phosphorus, and potassium nutrients required to be absorbed by the above-ground plants of the tuber crop to produce 1000 kg of tubers, and is an important indicator for measuring the nutrient requirements of the crop.

[0054] According to formulas (7 - 9), the nitrogen, phosphorus, and potassium nutrient absorption amounts RIE_N, RIE_P, and RIE_K parameters of the above-ground plants of the crop per 1000 kg of grain are calculated respectively. This parameter is used to characterize the nutrient absorption amount of the above-ground plants required to produce 1000 kg of tubers.

[0055] RIE-N= (7) RIE-P= (8) RIE-K= (9) According to formulas (10 - 12), the nitrogen, phosphorus, and potassium nutrient absorption amounts of the underground tubers of the crop per 1000 kg of grain are calculated respectively Tuber RIE_N ,Tuber RIE_P ,Tuber RIE_K The parameter is used to characterize the nutrient absorption amount of the tubers required to produce 1000 kg of tubers.

[0056] Tuber RIE_N = (10) Tuber RIE_P = (11) Tuber RIE_K = (12) Taking the local potato harvest index (HI) in the study area as the screening index of the QUEFTS model, which is used to measure the proportion of the economic yield of the crop in the total biomass. During the analysis process, first calculate the harvest index HI of the crops under different fertilization treatments. The calculation of the harvest index HI can be seen in formula (13), and select the data with HI > 0.4 to participate in the subsequent analysis.

[0057] Harvest index (HI): It is the ratio of the economic yield of the crop (i.e., the harvestable part) to the total biomass (i.e., the above-ground biomass), reflecting the ability of the crop to convert photosynthetic products into economic yield, and is an important indicator to measure the production efficiency of the crop.

[0058] HI = (13) M is the tuber yield of the crop (kg / ha); DW is the above-ground dry matter accumulation of the crop (kg / ha).

[0059] Harvest index of nitrogen, phosphorus, and potassium nutrients ( , ): The harvest index of nitrogen, phosphorus, and potassium nutrients of the crop refers to the ratio of the accumulation of nitrogen, phosphorus, and potassium in potato tubers to the total accumulation of nitrogen, phosphorus, and potassium in the whole plant, which is used to measure the efficiency of nutrient conversion into economic yield in the crop. See formulas (14 - 16) for details.

[0060] = (14) = (15) = (16) Among them, , are the harvest indexes of nitrogen, phosphorus, and potassium nutrients; is the accumulation of nitrogen (or phosphorus, potassium) in the crop tubers (kg / ha); is the accumulation of nitrogen (or phosphorus, potassium) in the crop plants.

[0061] Then, the nutrient uptake parameters of potatoes are determined. Based on the potato field trial data, this invention uses three sets of parameter sets of the 2.5th (Set Ⅰ), 5th (Set Ⅱ), and 7.5th (Set Ⅲ) of the intrinsic nutrient efficiency of potatoes as the parameter intervals for the maximum accumulation boundary (a) and the maximum dilution boundary (d) of nitrogen, phosphorus, and potassium, and uses the QUEFTS model to fit the nutrient uptake amounts under different potential yields. By adjusting the parameters, the range of the intrinsic nutrient efficiency is gradually narrowed to ensure the accuracy of the nutrient demand estimation. Finally, by comparing the nutrient uptake curves under different parameter conditions, the parameter group that has less impact on the linear part of the model is selected as the key boundary parameters (a and d) of the QUEFTS model.

[0062] Then, the optimal target yield of potatoes is determined. Based on the optimal crop nitrogen, phosphorus, and potassium nutrient maximum accumulation parameters (a) and maximum dilution parameters (d) of the selected QUEFTS model, Figure 3 is one of the schematic diagrams of the potato nutrient uptake curves under different potential yields fitted by the QUEFTS model provided by this invention. As Figure 3 shown, this invention sets different potential yields of the crop, uses the QUEFTS model to fit the nutrient uptake amount of the crop, and obtains the nitrogen, phosphorus, and potassium nutrient uptake curves of the crop's ground plants. The crop nutrient uptake curve shows a parabola-platform curve. When the target yield is before reaching 60%-70% of the potential yield, the curve shows a linear upward trend. At this time, the nutrient uptake amount and the crop yield have a linear increasing relationship. When the crop target yield exceeds 60%-70% of the yield potential, the fitted curve changes from the linear part to the platform area, and the absorption of nitrogen, phosphorus, and potassium by the crop changes, and the effect of fertilization on increasing the yield decreases.

[0063] This invention determines the best potential yield of the research area based on the best nutrient uptake curve YU fitted by the QUEFTS model. The linear part of the YU curve corresponds to the stage with the highest nutrient uptake efficiency of the crop, and the position where the curve changes from linear to platform corresponds to the optimal target yield (Ya) of the crop in the research area. YA represents the maximum accumulation boundary line, indicating the state with the lowest yield and the highest unit nutrient content under the same nutrient uptake; YD is the maximum dilution boundary line, indicating the state with the highest yield and the lowest unit nutrient content under the same nutrient uptake.

[0064] Then, the nitrogen, phosphorus, and potassium nutrient demands are determined. The best above-ground nutrient uptake amount of the crop can be calculated from the optimal target yield Ya and the nitrogen, phosphorus, and potassium nutrient uptake parameters per ton of grain RIE corresponding to the linear part of the best nutrient uptake curve (YU) of the crop.

[0065] Among them, for the potato tubers with the production target yield Ya, the best above-ground nitrogen nutrient uptake amount of the potato Plant N uptake and the best above-ground phosphorus nutrient uptake amount of the potatoPlant P uptake and the optimal potassium nutrient uptake of the above-ground part of the potato Plant K uptake is calculated as follows: Plant N uptake (kg / ha)= Ya / 1000×RIE_N (17) Plant P uptake (kg / ha)= Ya / 1000×RIE_P (18) Plant K uptake (kg / ha)= Ya / 1000×RIE_K (19) where Ya is the target production yield of potato tubers, and the optimal nitrogen nutrient uptake, Tuber N uptake optimal phosphorus nutrient uptake, Tuber P uptake and optimal potassium nutrient uptake Tuber K uptake of the underground part of the potato are calculated as follows: Tuber N uptake (kg / ha)= Ya / 1000 × Tuber RIE_N (20) Tuber P uptake (kg / ha)= Ya / 1000 × Tuber RIE_P (21) Tuber K uptake (kg / ha)= Ya / 1000 × Tuber RIE_K (22) Based on the QUEFTS model, the present invention simulates the optimal nutrient demand of crops in the study area. According to the yield characteristics and fertilization decision indexes of potatoes, taking each production target yield of potatoes as the benchmark, the nitrogen, phosphorus, and potassium nutrient demands of the above-ground plants of the crops and the nutrient demands of the potato tubers corresponding to the target yield are calculated, so as to calculate the total nutrient demand value of the final yield: total nutrient N demand N Demand ; total nutrient P demand P Demand ; total nutrient K demand K Demand , and the specific calculation formula is shown in (23 - 25).

[0066] N Demand (kg / ha) = (Plant N uptake +Tuber N uptake ) (23) P Demand (kg / ha)= (Plant P uptake +Tuber P uptake ) (24) K Demand (kg / ha)=(Plant K uptake +Tuber K uptake ) (25) Among them, N Demand ，P Demand and K Demand are the total N demand, total P demand, and total K demand of the target-yield potatoes respectively; Ya is the optimal target yield of potatoes in the study area; Plant N uptake ,Plant P uptake ,Plant K uptake are the N uptake, P uptake, and K uptake by the plants respectively; Tuber N uptake ,Tuber P uptake and Tuber K uptake are the N uptake, P uptake, and K uptake by the tubers respectively.

[0067] Finally, the demands for nitrogen, phosphorus, and potassium fertilizers are determined. According to the utilization rates of nitrogen, phosphorus, and potassium fertilizers and the proportions of the effective components in nitrogen, phosphorus, and potassium fertilizers (N, P, and K are 46%, 46%, and 50% respectively), the present invention converts the total demands for nitrogen, phosphorus, and potassium nutrients required for the production of the target yield into N, P2O5, and K2O. The specific formulas are shown in Formulas (26 - 28).

[0068] N 总肥量 = (26) P 2 O 5总肥量 = (27) K 2 O 总肥量 = (28) Among them, is the local nitrogen fertilizer utilization rate in the study area; is the nitrogen nutrient content in common nitrogen fertilizers; is the local phosphorus fertilizer utilization rate in the study area; is the phosphorus nutrient content in common nitrogen fertilizers; is the local potassium fertilizer utilization rate in the study area; is the potassium nutrient content in common nitrogen fertilizers.

[0069] In some embodiments, it further includes: Obtaining remote sensing images of different growth stages of potatoes, and the actual nutrient absorption amount in the plot corresponding to each pixel of the remote sensing image; Based on the remote sensing images of different growth stages of the potatoes and the actual nutrient absorption amount in the plot corresponding to each pixel of the remote sensing image, using an ensemble learning algorithm for model training to obtain the nutrient absorption amount inversion model.

[0070] The ensemble learning algorithm includes any one of the following algorithms: Random Forest algorithm; AdaBoost algorithm; Gradient Boosting algorithm; XGBoost algorithm; GBDT algorithm.

[0071] Specifically, obtaining remote sensing data (drone images, satellite images) of the critical growth stages of crops; based on the remote sensing data and crop nutrient data; screening the spectral parameters sensitive to crop nitrogen and yield; based on the selected parameters, constructing the actual nutrient absorption amount of crops in the critical growth stages, including the remote sensing monitoring models of the actual nitrogen nutrient absorption amount of plants (PNA), the actual phosphorus nutrient absorption amount of plants (PPA), and the actual potassium nutrient absorption amount of plants (PKA), and obtaining the real-time spatial distribution map of the nutrients of the above-ground plants in the target farmland plot. The calculation formulas for the nitrogen, phosphorus, and potassium nutrient absorption amounts of plants are shown in Formulas (29 - 31).

[0072] The actual nitrogen nutrient absorption amount of plants (Plant Nitrogen Accumulation, PNA): refers to the total cumulative amount of nitrogen elements contained in the above-ground plants of crops (including parts such as stems and leaves).

[0073] PNA (kg / ha) = + (29) is the nitrogen content in crop leaves (%), is the nitrogen content in above-ground stems (%), is the dry matter weight of crop leaves (g / m 2 ) and is the dry matter weight of crop stems (g / m2 ).

[0074] Actual phosphorus uptake of plants (Plant Phosphorus Accumulation, PPA): It refers to the total cumulative amount of nitrogen element contained in the above-ground parts of crops (including stems, leaves, etc.).

[0075] PPA (kg / ha) = + (30) is the phosphorus content in crop leaves (%), is the phosphorus content in the above-ground stem (%), is the dry matter weight of crop leaves (g / m 2 ), and is the dry matter weight of crop stems (g / m 2 ).

[0076] Actual potassium uptake of plants (Plant potassium Accumulation, PKA): It refers to the total cumulative amount of nitrogen element contained in the above-ground parts of crops (including stems, leaves, etc.).

[0077] PKA (kg / ha) = + (31) is the potassium content in crop leaves (%), is the potassium content in the above-ground stem (%), is the dry matter weight of crop leaves (g / m 2 ), and is the dry matter weight of crop stems (g / m 2 ).

[0078] In some embodiments, it further includes processing satellite remote sensing images.

[0079] Figure 4 is the satellite remote sensing image map of potato tuber formation stage plants (Planet) provided by the present invention, such as Figure 4 , obtaining satellite remote sensing images of key growth stages of crops. The images are mainly high-spatial-resolution images, including but not limited to data such as Sentinel-2, Planet, and high-resolution satellites. Image preprocessing includes radiometric correction, atmospheric correction, and geometric correction. According to the spectral band settings of different satellites, vegetation indices based on information of different bands are constructed for crop nutrient monitoring.

[0080] Taking Planet data as an example, the spectral parameters that can be constructed are shown in Table 1.

[0081] Table 1 Construction of potato vegetation indices based on Planet images

[0082] Among them, B represents the blue band, G represents the green band, R represents the red band, RE represents the red edge band, and NIR represents the near-infrared band.

[0083] Taking the integrated learning algorithm as the random forest algorithm as an example. Based on satellite remote sensing images of the critical growth stages of crops and measured data of the actual nutrient uptake of nitrogen, phosphorus, and potassium in crops (PNA, PPA, and PKA), the random forest (RF) algorithm is used to construct (train) inversion models for the nutrient uptake PNA, PPA, and PKA of potato plants at the critical growth stages using the extracted spectral parameters.

[0084] Random forest is an integrated learning technique that contains multiple decision trees and is used for classification, regression, and other tasks. Each tree selects a random subset of the dataset during training and a random subset of the best features at each branch point of the decision tree. This randomness helps reduce the variance of the model and improve the generalization ability.

[0085] Random forest achieves the final prediction by integrating the prediction results of multiple decision trees: calculating the mean of the predictions of all trees in the regression task and using majority voting in the classification task. The algorithm constructs a training set by sampling about 63.2% of the samples from the initial training set through the bootstrap sampling method to increase the heterogeneity of the dataset and improve the generalization ability. When selecting features, a partial feature set is randomly selected at each decision tree node and the optimal feature is selected for partitioning, thereby enhancing the difference of the sub-models and improving the model stability. Finally, the results of all decision trees are calculated in parallel and weighted averaged to reduce noise interference and obtain accurate prediction results. The calculation formula for its weighted prediction value is shown in Equation (32).

[0086] (32) In the formula,[[]]END]] Pt is the output value predicted for the test sample,[[]]END]] Yi 、 Wi are respectively the output value and the weight of the i th regression decision tree,[[]]END]] T is the number of decision trees. At the same time, random forest can evaluate the importance of each feature variable during the regression process and provide a reference for users in feature input selection.

[0087] Figure 5 is one of the inversion result maps of PNA, PKA, and PPA of the Planet images of the potato tuber formation period in the field provided by the present invention, as shown in Figure 5As shown in the figure, the present invention applies the RF inversion model constructed based on ground sample data to the potato field area to realize the inversion of the actual nitrogen nutrient absorption amount PNA, the actual phosphorus nutrient absorption amount PPA, and the actual potassium nutrient absorption amount PKA of potato plants during the key growth period (tuber formation period) of the target plot.

[0088] In some embodiments, the precise fertilization of potatoes is carried out in units of the plots corresponding to the pixels based on the nutrient deficiency amount corresponding to each pixel, including: Determining the fertilization amount corresponding to each pixel based on the nutrient deficiency amount corresponding to each pixel; Precisely fertilizing the plots corresponding to each pixel according to the fertilization amount corresponding to each pixel in units of the plots corresponding to the pixels.

[0089] Specifically, based on the nutrient requirement characteristics of the crop, the base fertilizer and topdressing amount are set in a certain proportion. The base fertilizer is used to provide the main nutrients during the early growth stage of the potato. Generally, when sowing the crop, 60% of the total nutrients are first applied as the base fertilizer; during the key period of crop topdressing, the remaining 40% of the nutrients are supplemented according to the nutrient requirements of the crop.

[0090] During the key period of topdressing, with the pixel as the fertilization target, based on the remote sensing monitoring results of the total nutrient demand of the crop and the actual nutrient absorption amount of the crop plants, the nutrient deficiency diagnosis of field crops is carried out, and the necessary nutrients are supplemented according to the actual nutrient gap of the crop. Nutrient N deficiency amount N Deficit , nutrient P deficiency amount P Deficit , and nutrient K deficiency amount K Deficit The calculation formulas are shown in Formulas (33 - 35).

[0091] N Deficit (kg / ha) = N Demand - PNA (33) P Deficit (kg / ha) = P Demand - PPA (34) K Deficit (kg / ha) = K Demand - PKA (35) Among them, N Demand is the total nutrient demand for nutrient N; P Demand is the total nutrient demand for nutrient P; KDemand is the total nutrient requirement for K; PNA, PPA, and PKA are the remotely sensed monitoring values of the actual nutrient uptake during the key growth stages of fertilization.

[0092] Figure 6 is the monitoring map of potato nutrient deficiency provided by the present invention. As Figure 6 shown, during the potato tuber formation period in the study area, there are significant differences in the nutrient deficiency of potatoes within the plot. For the pixels with a tuber nutrient deficiency amount < 0, it indicates that the overall growth of the crop is vigorous, and the nutrient uptake of the above-ground plants reaches the optimal nutrient uptake amount during the potato tuber formation period. At this time, excessive application of nitrogen fertilizer will cause the plants to grow excessively and the maturity period to be delayed, seriously affecting the yield. Therefore, for the pixels with a small amount or no fertilizer treatment, for the pixels with a nutrient deficiency amount > 0, quantitative and precise fertilizer supplementation needs to be carried out according to the nutrient deficiency situation.

[0093] Taking the plot as the research object, for the pixels in the plot, calculate its variable fertilization weight . The calculation formula is as follows: (36) Among them, represents the pixels in the plot with a nutrient deficiency amount greater than 0 (respectively representing N Deficit、 P Deficit、 K Deficit values), represents the total sum of the nutrient deficiency amounts of all pixels greater than 0 in the plot.

[0094] Then, the calculation formulas for the topdressing amounts of nitrogen, phosphorus, and potassium for the pixels with a nutrient deficiency index ( N Deficit、 P Deficit、 K Deficit ) greater than 0 are as follows: (37) Among them, is the fertilization amount of the i-th pixel, is the total fertilization amount minus the base fertilizer amount, where: (38) (39) (40) Figure 7 is one of the nitrogen, phosphorus, and potassium fertilization decision maps for potato tuber formation provided by the present invention. As Figure 7As shown, taking the plot corresponding to the pixel as the unit, precise fertilization is carried out on the plot corresponding to each pixel according to the fertilization amount corresponding to each pixel.

[0095] Finally, the present invention compares the topdressing amount and fertilizer utilization rate to specifically analyze the influence of the traditional uniform fertilization strategy and the topdressing strategy adopted by the present invention on the potato yield. In the traditional uniform fertilization strategy, the calculation formula for the topdressing amount during the critical growth period of the crop is as follows: (41) (42) (43) where m is the total number of pixels in the plot, and S Pixel is the area of each pixel.

[0096] The evaluation and analysis of fertilizer utilization rate are of great significance for optimizing fertilizer use, increasing crop yield, and reducing environmental pollution in agricultural production. By evaluating the fertilizer utilization rate, it can help agricultural producers find ways to improve fertilizer efficiency, avoid over-fertilization and resource waste. The formula for fertilizer utilization rate is shown below.

[0097] Fertilizer utilization rate = × 100% (44) Through this formula, the contribution of different fertilization strategies to crop yield can be evaluated, thus helping agricultural producers find ways to improve fertilizer efficiency. The present invention uses the comparative analysis of fertilizer utilization rate to demonstrate the advantages of the precise fertilization strategy in enhancing potato yield and nutrient absorption efficiency.

[0098] The present invention combines the QUEFTS model with remote sensing technology. Based on the law of soil nutrient supply and crop demand, it simulates the nitrogen, phosphorus, and potassium nutrient absorption curves of different target yields of potatoes, and determines the optimal target yield and nutrient demand. Using ground test data and Planet satellite images, a remote sensing inversion model based on PNA, PPA, and PKA is constructed to dynamically monitor the crop nutrient status, and determine the best fertilization timing and fertilization amount. Through the dynamic adjustment of base fertilizer and topdressing, the fertilization plan is optimized. Compared with the traditional strategy, it significantly improves the utilization rates of nitrogen, phosphorus, and potassium fertilizers, realizes large-scale precise fertilization management, enhances fertilizer utilization efficiency, and reduces the environmental burden.

[0099] The following further illustrates the present invention with a specific example.

[0100] The present invention selects the Science and Technology Park of Keshan Farm, Qiqihar Branch of Heilongjiang Land Reclamation Bureau as the research area, obtains data through field tests, and constructs a recommended fertilization model for potatoes. The soil characteristics of the potato production area are shown in Table 2.

[0101] Table 2 Soil Characteristics of Potato Production Area

[0102] The present invention conducted nitrogen fertilizer and potassium fertilizer gradient treatment experiments on potatoes in 2022 and 2023, respectively.

[0103] 2022 trial: The sowing date was May 13, and the experimental varieties were Kenshu 1 (P1) and Nei-9 (P2). The experiment set up 5 nitrogen fertilizer gradients (0, 60, 150, 200, 250 kg N / ha, N0, N1, N2, N3, N4 respectively) and 5 potassium fertilizer gradients (0, 60, 150, 200, 250 kg K2O / ha, K0, K1, K2, K3, K4 respectively), with 3 replicates for each treatment, for a total of 30 plots. For nitrogen fertilizer and potassium fertilizer, each plot was applied at a ratio of 7:3 of basal fertilizer and topdressing during the potato sowing and budding period. The basal fertilizer of nitrogen fertilizer and potassium fertilizer in each plot was 70% of the total nitrogen fertilizer, and 30% was applied during the budding period, and nitrogen, phosphorus, and potassium fertilizers were applied as basal fertilizers at one time.

[0104] 2023 trial: The sowing date is May 20, and the experimental varieties are Vilas (P3) and Nei-9 (P2). The trial set up 4 nitrogen fertilizer gradients (0, 60, 150, 250 kg N / ha, respectively N0, N1, N2, N3) and 4 potassium fertilizer gradients (0, 60, 150, 250 kg K2O / ha, respectively K0, K1, K2, K3), each treatment has 3 replications, a total of 24 plots. The fertilization ratio of nitrogen fertilizer and potassium fertilizer is also 70% basal fertilizer and 30% topdressing, and topdressing is applied during the budding stage.

[0105] Analysis of intrinsic efficiency of crop nutrients.

[0106] Based on the analysis of potato field trial data in 2022 and 2023 (e.g. Figure 8 As shown in Table 3, the average yield of potato tubers in the study area was 20808.594 kg / ha (dry weight, including no fertilization treatment). The test results show that the sample data with harvest index HI ≥ 0.40 are stable, with an average value of 0.572 kg / kg. According to the research of Setiyono et al., data with HI < 0.40 have little effect on the fitting results of the QUEFTS model. Therefore, the present invention retains some sample data with HI < 0.40 for further analysis and optimization of the fertilization decision model.

[0107] Table 3 Nutrient absorption characteristics of potatoes

[0108] Among them, 25%Q: first quartile; 50%Q: second quartile; 75%Q: third quartile.

[0109] Based on formula (1-6), the present invention calculates two parameters of the intrinsic efficiency of crop nutrients and the nutrient uptake per ton of grain. The results are shown in Table 4. The average values of IE for nitrogen, potassium, and phosphorus in potatoes are 550.483 kg / kg N, 2828.968 kg / kg P, and 380.353 kg / kg K, respectively. The average values of RIE for nitrogen, potassium, and phosphorus are 1.835 kg N / t, 0.358 kg P / t, and 2.651 kg K / t, respectively. The demand for phosphorus is relatively low.

[0110] Table 4 Characteristics of the intrinsic efficiency IE and RIE of nitrogen, phosphorus, and potassium nutrients per ton of grain in the whole potato plant

[0111] Among them, 25%Q: the first quartile; 50%Q: the second quartile; 75%Q: the third quartile.

[0112] Determination of crop nutrient uptake parameters.

[0113] First, taking the potential yield of 60000 kg / ha as an example, the present invention uses three groups of parameter values of the 2.5th (Set Ⅰ), 5th (Set Ⅱ), and 7.5th (Set Ⅲ) of the potato nutrient intrinsic efficiency data (Table 4) as the crop nutrient uptake parameters, that is, the slopes of the maximum accumulation (a) and maximum dilution (d) boundary lines of nitrogen, phosphorus, and potassium (Table 5), and conducts a fitting analysis on the nutrient uptake of potatoes. The results are as Figure 9 shown.

[0114] Table 5 Maximum accumulation (a) and maximum dilution boundary (d) of nitrogen, potassium, and phosphorus nutrient uptake in the above-ground part of potatoes

[0115] The results show that the a and d parameters of parameter groups Ⅰ, Ⅱ, and Ⅲ successively narrow the range of nutrient intrinsic efficiency. Taking the nitrogen nutrient intrinsic efficiency as an example, the data distributed in the ranges of Set Ⅰ, Set Ⅱ, and Set Ⅲ account for 94.5%, 89.9%, and 84.9% of the nutrient intrinsic efficiency data, respectively. The parameter group Ⅰ (2.5th) representing the largest range of nutrient intrinsic efficiency has the least impact on the nutrient uptake in the linear part of the QUEFTS model output. Therefore, the present invention uses it as the slope parameters of the maximum accumulation (a) and maximum dilution (d) boundary lines of nitrogen, phosphorus, and potassium in the QUEFTS model (as Figure 9 shown).

[0116] Fitting of crop nutrient uptake and estimation of the optimal nutrient demand.

[0117] Based on the QUEFTS model, the present invention simulates the optimal nutrient uptake curves of potatoes at different potential yields (15,000 - 60,000 kg / ha, fresh weight), and the results are as follows Figure 10 shown. The results show that when the target yield is 60% - 70% of the potential yield, the nutrient uptake efficiency is the best, showing a linear upward trend. Applying additional nutrients significantly increases the unit yield; beyond this range, the curve turns into a parabola - plateau shape, and more nutrient input is required to increase the unit yield, resulting in a decrease in fertilization efficiency. Therefore, when the target yield is 60% - 70% of the potential yield, it is the optimal target yield, corresponding to the optimal demand for nitrogen, phosphorus, and potassium nutrients by the crop.

[0118] Table 6 shows the nutrient uptake of potatoes at different yield levels fitted by the QUEFTS model. For the linear part of the fitted curve, for every 1000 kg of tubers produced, the optimal nutrient requirements for total nitrogen, phosphorus, and potassium in potato plants are 18.635 kg N / t, 3.551 kg P / t, and 28.106 kg K / t respectively. At this time, the internal efficiencies of nitrogen, phosphorus, and potassium nutrients are 643.950 kg / kg N, 3379.330 kg / kg P, and 426.955 kg / kg K respectively. At the same time, the QUEFTS model has good simulation results for nitrogen and phosphorus nutrient uptake. There is a large variation in potato potassium nutrient, mainly because the measured values of potassium fertilizer in Heilongjiang Province are significantly higher than the simulated values, and there may be an over - application situation, resulting in excessive potassium uptake by potatoes.

[0119] Table 6 Nutrient uptake of potatoes at different yield levels fitted by the QUEFTS model

[0120] Table 7 shows the above - ground nutrient uptake and tuber nutrient uptake of potatoes simulated using the QUEFTS model. It can be seen from the table that when the target yield reaches 70% of the potential yield, for the linear part fitted by the QUEFTS model, the contents of nitrogen, phosphorus, and potassium in tubers account for 76.7%, 71.3%, and 78.5% of the total plant uptake respectively.

[0121] Table 7 Nutrient uptake of potato tubers fitted by the QUEFTS model

[0122] Remote sensing image acquisition and pre - processing.

[0123] Obtain cloudless Planet images during the potato tuber formation period in 2023. The image data of the Planet satellite has high spatio-temporal resolution (spatial resolution of 3 meters and imaging frequency of once a day), which can accurately capture the growth status of potatoes at different times. Image preprocessing includes radiometric correction, atmospheric correction, and geometric correction, and the relevant data processing process is completed in ENVI software. Since the Planet images cover a large area, the images are cropped based on the vector boundary file of the potato experimental area. Secondly, extract the spectral reflectance data of each plot to construct vegetation indices. A total of 34 vegetation indices are constructed in the present invention, as shown in Table 1 specifically; finally, construct the RF model with the constructed vegetation indices, PNA, PPA, and PKA.

[0124] Correlation analysis of crop nutrient and growth parameter indicators with yield.

[0125] Based on two-year experimental data, the present invention conducts correlation analysis on potato yield and parameters such as the nitrogen, phosphorus, and potassium nutrient status and biomass at different growth stages of potatoes. The results are as Figure 11 shown. It can be seen from the figure that the plant nitrogen accumulation amount (PNA), plant phosphorus accumulation amount (PPA), and plant potassium accumulation amount (PKA) at three different growth stages are correlated with the final potato yield. Therefore, these three parameter indicators are introduced and specific fertilization decision-making suggestions are given in combination with the QUEFTS model.

[0126] Among them, the numbers 2, 3, and 4 represent the tuber formation period, tuber growth period, and starch accumulation period respectively; Last Yield represents the final yield. SDW, LDW, and ADW represent stem dry weight, leaf dry weight, and total aboveground dry weight respectively; SNA, SKA, and SPA represent stem N accumulation amount, stem K accumulation amount, and stem P accumulation amount respectively; LNA, LKA, and LPA represent leaf N accumulation amount, leaf K accumulation amount, and leaf P accumulation amount respectively; PNA, PKA, and PPA represent plant N accumulation amount, plant K accumulation amount, and plant P accumulation amount.

[0127] Construction of the RF-based inversion model for PNA, PPA, and PKA in potato experimental plots.

[0128] When the present invention uses the RF method to invert PNA, PPA, and PKA, the input remote sensing parameters are vegetation indices constructed based on Planet images, and the total ground measured data set of potatoes in 2023 is divided into 7:3 for model construction and model verification respectively. The results are as Figure 12 . Figure 12 Show the RF model inversion accuracy R of PNA, PPA, and PKA at different growth stages 2 and RMSE.

[0129] Finally, the constructed PNA, PPA, and PKA inversion models were applied to Planet satellite images during the potato tuber formation period, and the results are as Figure 13 shown.

[0130] The present invention applied the PNA, PPA, and PKA inversion models verified by the 2023 plot experiments to the field area during the potato tuber formation period. Figure 14 The predicted results of PNA, PPA, and PKA during the tuber formation period are shown.

[0131] Calculation of crop target yield and total nutrient demand.

[0132] Based on the production of 1000 kg of potatoes per unit, the present invention calculated the nutrient uptake of nitrogen, phosphorus, and potassium in the corresponding plants (N: 1.864 kg, P: 0.355 kg, K: 2.811 kg) and the nitrogen, phosphorus, and potassium uptake in tubers (N: 1.431 kg, P: 0.254 kg, K: 2.206 kg). From this, the total nutrient demand for the optimal target yield was calculated (N: 138.390 kg / ha, P: 25.578 kg / ha, K: 210.714 kg / ha). Table 8 shows the analysis results of fertilization rates.

[0133] Table 8 Analysis results of potato nitrogen fertilizer application rates (based on the production of 1000 kg of potatoes per unit)

[0134] Formulation of fertilization strategy.

[0135] The present invention carried out fertilization according to the ratio of 6:4 for base fertilizer and top dressing. The specific amounts of base fertilizer and top dressing are shown in Table 9. During the potato budding period, using the results of remote sensing monitoring of crop growth and nutrients, variable top dressing was carried out based on the nutrient gap to ensure that the nutrient supply during the growth stage of the crop could meet the growth requirements of the crop.

[0136] Table 9 Potato base fertilizer and top dressing application rates (kg / ha)

[0137] Based on fertilization during the potato budding period, the present invention set two variable fertilization strategies: (1) variable top dressing based on the remote sensing monitoring values and nutrient gaps of PNA, PPA, and PKA; (2) uniform fertilization strategy.

[0138] Fertilization strategy based on PNA, PPA, and PKA values and nutrient gaps.

[0139] The present invention directly generated a fertilization decision-making diagram based on the PNA, PPA, and PKA values and nutrient gaps, and the results are as Figure 15As shown. Under this strategy, fertilization is carried out in the positive value area, and a small amount of fertilization is carried out in the negative value area according to the nutrient gap to precisely meet the crop requirements.

[0140] Comparison of fertilization amounts and analysis of fertilizer utilization rates.

[0141] At the same time, the present invention calculates and compares the total supplementary fertilization amounts of the variable fertilization strategy and the uniform top-dressing strategy. Specifically, see Table 10. The large field area in the image is 91.948 mu.

[0142] Table 10 Comparative analysis of two fertilization strategies for potatoes (based on the large field area in the image)

[0143] It can be seen from this that according to the nitrogen and phosphorus variable top-dressing strategy proposed by the present invention, compared with the uniform top-dressing strategy, the fertilizer savings reach 13%. Since potatoes have a relatively high demand for potassium, the amount of potassium fertilizer used has not changed significantly. Further, the total nutrient requirements for the optimal target yield are calculated (the total requirements for nitrogen, phosphorus, and potassium in plants and tubers are 138.390 kg / ha, 25.578 kg / ha, and 210.714 kg / ha respectively). Converting these nutrient requirements into fertilizer amounts, it is obtained that the required nitrogen fertilizer is 613.975 kg / ha, phosphorus fertilizer is 303.849 kg / ha, and potassium fertilizer is 673.208 kg / ha (expressed in terms of N, P2O5, and K2O).

[0144] Generally speaking, variable fertilization can reasonably increase the fertilization amount in areas with high demand, while reducing or not fertilizing in areas with low demand or no need for fertilization. Thus, with less fertilizer, higher efficiency is achieved while meeting the crop requirements. This also shows that variable fertilization effectively avoids the phenomenon of "over-fertilization" that may occur in uniform fertilization by precisely regulating the nitrogen demand in different areas, saves fertilizer resources, reduces the environmental burden caused by fertilizer surplus, and is more in line with the requirements of sustainable agricultural development.

[0145] The analysis of the fertilizer utilization rates of the traditional uniform fertilization strategy and the top-dressing strategy using the QUEFTS model adopted in the present invention with a plot experiment as an example is shown in Table 11.

[0146] Table 11 Calculation of fertilizer utilization rate

[0147] Among them, 25%Q: the first quartile; 50%Q: the second quartile; 75%Q: the third quartile.

[0148] Generally, the utilization rate of nitrogen fertilizers is usually between 30% - 50%, that of phosphate fertilizers is generally between 10% - 25%, and that of potassium fertilizers is generally 40% - 60%. The data in Table 11 show that traditional fertilization exhibits significant volatility and instability in the utilization rates of nitrogen and potassium fertilizers, with some indicators deviating from the reasonable range; while variable fertilization can significantly improve the stability of fertilizer utilization rates. Among them, the nitrogen fertilizer utilization rate remains within the reasonable range, the potassium fertilizer utilization rate is close to the upper limit of the reasonable range and has less volatility, and the phosphate fertilizer utilization rate is significantly increased and is at a relatively high position in the middle of the reasonable range. Variable fertilization performs better than traditional fertilization in terms of fertilizer use efficiency and waste reduction, and is a more reasonable and efficient fertilization strategy.

[0149] A precise fertilization decision-making technology method and device combining a QUEFTS model with remote sensing diagnosis of potato nutrient deficiency proposed by the present invention constructs a potato fertilization decision-making model by combining remote sensing technology and the QUEFTS model, realizing real-time monitoring of crop nutrient requirements and dynamic adjustment of fertilization strategies. This model effectively solves problems such as fertilizer waste, nutrient loss, and environmental pollution in traditional fertilization methods, significantly improves fertilizer utilization rates, and optimizes crop production efficiency. At the same time, the present invention achieves the goal of precision agriculture management, provides a scientific fertilization management plan for large-scale agricultural production, and promotes the development of sustainable agriculture.

[0150] The potato precise fertilization device provided by the present invention is described below. The potato precise fertilization device described below can be mutually referred to in correspondence with the potato precise fertilization method described above.

[0151] Figure 16 is a schematic structural diagram of the potato precise fertilization device provided by the present invention. As Figure 16 shown, the potato precise fertilization device provided by the present invention includes: An acquisition module 1601 is used to acquire remote sensing images of the target growth stage of potatoes; An inversion module 1602 is used to input the remote sensing images into a trained nutrient absorption amount inversion model to obtain the actual nutrient absorption amount corresponding to each pixel output by the nutrient absorption amount inversion model; the nutrient absorption amount inversion model is trained based on remote sensing images of different growth stages of potatoes and the actual nutrient absorption amount in the plot corresponding to each pixel of the remote sensing image; A determination module 1603 is used to determine the nutrient deficiency amount corresponding to each pixel based on the total nutrient demand and the actual nutrient absorption amount corresponding to each pixel; the total nutrient demand is obtained based on the QUEFTS model; A fertilization module 1604 is used to perform precise fertilization on potatoes in units of the plots corresponding to the pixels based on the nutrient deficiency amount corresponding to each pixel.

[0152] Specifically, the above-mentioned potato precise fertilization device provided by the embodiments of the present application can implement all the method steps implemented by the above-mentioned embodiments of the potato precise fertilization method, and can achieve the same technical effects. Therefore, the same parts and beneficial effects as those in the method embodiments will not be specifically described herein.

[0153] Figure 17 An example of a schematic physical structure diagram of an electronic device is shown as Figure 17 shown. The electronic device may include: a processor 810, a communication interface 820, a memory 830, and a communication bus 840. Among them, the processor 810, the communication interface 820, and the memory 830 complete communication with each other through the communication bus 840. The processor 810 can call the logical instructions in the memory 830 to execute the potato precise fertilization method, and the method includes: Obtaining a remote sensing image of the target growth stage of the potato; Inputting the remote sensing image into the trained nutrient uptake inversion model to obtain the actual nutrient uptake corresponding to each pixel output by the nutrient uptake inversion model; the nutrient uptake inversion model is trained based on the remote sensing images of different growth stages of the potato and the actual nutrient uptake in the plot corresponding to each pixel of the remote sensing image; Based on the total nutrient demand and the actual nutrient uptake corresponding to each pixel, determining the nutrient deficit corresponding to each pixel; the total nutrient demand is obtained based on the QUEFTS model; Based on the nutrient deficit corresponding to each pixel, precise fertilization is performed on the potato in units of the plots corresponding to the pixels.

[0154] In addition, when the logical instructions in the above-mentioned memory 830 are implemented in the form of software functional units and sold or used as an independent product, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or a part of this technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical discs that can store program codes.

[0155] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the potato precise fertilization method provided by each of the above methods, and the method includes: Obtain a remote sensing image of the target growth stage of the potato; Input the remote sensing image into the trained nutrient uptake inversion model to obtain the actual nutrient uptake corresponding to each pixel output by the nutrient uptake inversion model; the nutrient uptake inversion model is obtained by training based on remote sensing images of different growth stages of the potato and the actual nutrient uptake in the plot corresponding to each pixel of the remote sensing image; Based on the total nutrient demand and the actual nutrient uptake corresponding to each pixel, determine the nutrient deficit corresponding to each pixel; the total nutrient demand is obtained based on the QUEFTS model; Based on the nutrient deficit corresponding to each pixel, perform precise fertilization on the potato in units of the plot corresponding to the pixel.

[0156] On another aspect, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it realizes the potato precise fertilization method provided by each of the above methods, and the method includes: Obtain a remote sensing image of the target growth stage of the potato; Input the remote sensing image into the trained nutrient uptake inversion model to obtain the actual nutrient uptake corresponding to each pixel output by the nutrient uptake inversion model; the nutrient uptake inversion model is obtained by training based on remote sensing images of different growth stages of the potato and the actual nutrient uptake in the plot corresponding to each pixel of the remote sensing image; Based on the total nutrient demand and the actual nutrient uptake corresponding to each pixel, determine the nutrient deficit corresponding to each pixel; the total nutrient demand is obtained based on the QUEFTS model; Based on the nutrient deficit corresponding to each pixel, perform precise fertilization on the potato in units of the plot corresponding to the pixel.

[0157] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative labor.

[0158] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0159] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A potato precision fertilization method, characterized in that: include: Acquire remote sensing images of target potato growth stages; Inputting the remote sensing image into a trained nutrient absorption inversion model to obtain the actual nutrient absorption corresponding to each pixel output by the nutrient absorption inversion model; the nutrient absorption inversion model is obtained by model training based on remote sensing images of potatoes at different growth stages and the actual nutrient absorption in the plot corresponding to each pixel of the remote sensing image; Determine the nutrient deficit corresponding to each pixel based on the total nutrient demand and the actual nutrient absorption corresponding to each pixel; the total nutrient demand is obtained based on the QUEFTS model; Based on the nutrient deficit corresponding to each pixel, potatoes are precisely fertilized with the plot corresponding to the pixel as a unit.

2. The potato precision fertilization method according to claim 1, characterized in that: Also includes: Obtain potato yield data under different fertilization gradient treatments; Based on the potato yield data under the different fertilization gradient treatments, the total nutrient requirement is determined using the QUEFTS model.

3. The potato precision fertilization method according to claim 2, characterized in that: The method of determining the total nutrient requirement based on the potato yield data under the different fertilization gradient treatments using the QUEFTS model includes: Based on the potato yield data under the different fertilization gradient treatments, the potato nutrient absorption curve was fitted using the QUEFTS model; Determining the optimal target yield of potatoes, and the nutrient absorption amount of the aboveground part of potatoes and the nutrient absorption amount of the underground part of potatoes under the optimal target yield based on the potato ground nutrient absorption curve; The total nutrient requirement is determined based on the nutrient absorption of the aboveground part of the potato and the nutrient absorption of the underground part of the potato.

4. The potato precision fertilization method according to claim 1, characterized in that: Also includes: Obtain remote sensing images of potatoes at different growth stages, as well as the actual nutrient absorption in the plot corresponding to each pixel of the remote sensing image; Based on the remote sensing images of the potatoes at different growth stages and the actual nutrient absorption in the plot corresponding to each pixel of the remote sensing images, an integrated learning algorithm is used to perform model training to obtain the nutrient absorption inversion model.

5. The potato precision fertilization method according to claim 4, characterized in that: The ensemble learning algorithm includes any one of the following algorithms: Random Forest Algorithm; AdaBoost algorithm; Gradient Boosting algorithm; XGBoost algorithm; GBDT algorithm.

6. The potato precision fertilization method according to claim 1, characterized in that: Based on the nutrient deficiency corresponding to each pixel, the potatoes are precisely fertilized in units of plots corresponding to the pixels, including: Determining the amount of fertilizer to be applied to each pixel based on the nutrient deficiency amount to be applied to each pixel; Taking the plot corresponding to the pixel as the unit, the plot corresponding to each pixel is precisely fertilized according to the fertilizer amount corresponding to each pixel.

7. A potato precision fertilization device, characterized in that: include: An acquisition module, used to acquire remote sensing images of target potato growth stages; An inversion module is used to input the remote sensing image into a trained nutrient absorption inversion model to obtain the actual nutrient absorption corresponding to each pixel output by the nutrient absorption inversion model; the nutrient absorption inversion model is obtained by model training based on remote sensing images of potatoes at different growth stages and the actual nutrient absorption in the plot corresponding to each pixel of the remote sensing image; A determination module, used to determine the nutrient deficit corresponding to each pixel based on the total nutrient demand and the actual nutrient absorption corresponding to each pixel; the total nutrient demand is obtained based on the QUEFTS model; The fertilization module is used to accurately fertilize potatoes based on the nutrient deficit corresponding to each pixel and in units of the plots corresponding to the pixels.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the potato precision fertilization method according to any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the potato precision fertilization method according to any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the potato precision fertilization method according to any one of claims 1 to 6 is implemented.

Images