CN114692971A - Crop yield prediction method and device based on yield difference - Google Patents
Crop yield prediction method and device based on yield difference Download PDFInfo
- Publication number
- CN114692971A CN114692971A CN202210330408.XA CN202210330408A CN114692971A CN 114692971 A CN114692971 A CN 114692971A CN 202210330408 A CN202210330408 A CN 202210330408A CN 114692971 A CN114692971 A CN 114692971A
- Authority
- CN
- China
- Prior art keywords
- predicted
- yield
- crop
- crops
- model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 76
- 238000003062 neural network model Methods 0.000 claims abstract description 30
- 230000008569 process Effects 0.000 claims abstract description 29
- 238000000342 Monte Carlo simulation Methods 0.000 claims abstract description 12
- 238000004590 computer program Methods 0.000 claims description 15
- 238000009826 distribution Methods 0.000 claims description 15
- 238000004364 calculation method Methods 0.000 claims description 14
- 230000006870 function Effects 0.000 claims description 13
- 238000011160 research Methods 0.000 claims description 13
- 238000012549 training Methods 0.000 claims description 9
- 238000004422 calculation algorithm Methods 0.000 claims description 7
- 238000003860 storage Methods 0.000 claims description 7
- 238000009499 grossing Methods 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 6
- 230000003595 spectral effect Effects 0.000 claims description 6
- 238000001556 precipitation Methods 0.000 claims description 5
- 230000005855 radiation Effects 0.000 claims description 5
- 238000001514 detection method Methods 0.000 claims description 4
- 238000001914 filtration Methods 0.000 claims description 3
- 230000007246 mechanism Effects 0.000 abstract description 8
- 230000008901 benefit Effects 0.000 abstract description 6
- 238000010586 diagram Methods 0.000 abstract description 6
- 238000005259 measurement Methods 0.000 abstract description 4
- 238000013135 deep learning Methods 0.000 abstract description 3
- 238000004891 communication Methods 0.000 description 5
- 238000013136 deep learning model Methods 0.000 description 5
- 238000013528 artificial neural network Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 230000018109 developmental process Effects 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 239000013598 vector Substances 0.000 description 3
- 208000005156 Dehydration Diseases 0.000 description 2
- 230000015572 biosynthetic process Effects 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 238000010168 coupling process Methods 0.000 description 2
- 238000005859 coupling reaction Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000006353 environmental stress Effects 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 230000035882 stress Effects 0.000 description 2
- 239000002028 Biomass Substances 0.000 description 1
- 241000209140 Triticum Species 0.000 description 1
- 235000021307 Triticum Nutrition 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 235000013339 cereals Nutrition 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000013213 extrapolation Methods 0.000 description 1
- 235000013305 food Nutrition 0.000 description 1
- 230000008014 freezing Effects 0.000 description 1
- 238000007710 freezing Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000000053 physical method Methods 0.000 description 1
- 238000011002 quantification Methods 0.000 description 1
- 238000002310 reflectometry Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000001502 supplementing effect Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000017260 vegetative to reproductive phase transition of meristem Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N7/00—Computing arrangements based on specific mathematical models
- G06N7/01—Probabilistic graphical models, e.g. probabilistic networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/02—Agriculture; Fishing; Forestry; Mining
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Economics (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Strategic Management (AREA)
- Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Computational Linguistics (AREA)
- Tourism & Hospitality (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- Agronomy & Crop Science (AREA)
- Mathematical Optimization (AREA)
- Mining & Mineral Resources (AREA)
- Mathematical Analysis (AREA)
- Computational Mathematics (AREA)
- Algebra (AREA)
- Animal Husbandry (AREA)
- Primary Health Care (AREA)
- Probability & Statistics with Applications (AREA)
- Marine Sciences & Fisheries (AREA)
- Pure & Applied Mathematics (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
Abstract
The invention provides a crop yield prediction method and a device based on yield difference, wherein the method comprises the following steps: calibrating WOFOST crop model parameters by adopting a Markov chain-Monte Carlo method; inputting meteorological data into a calibrated WOFOST model to generate the potential yield of the crop to be predicted; subtracting the potential yield of the crop to be predicted from the field observed yield to obtain the actual yield difference of the crop to be predicted; taking satellite data, meteorological data and potential yield of actual measurement sample crops in a growth period as input samples, taking actual yield difference of the crops to be predicted as a label, and building a neural network model based on a process model; and operating a WOFOST model by crop grids to generate potential yield, inputting the potential yield into a neural network model to generate a predicted yield difference, and subtracting the predicted yield difference from the potential yield to generate a predicted yield space diagram. The method disclosed by the invention integrates the advantages of a process model and deep learning, fully utilizes rich information of a mechanism model and remote sensing data, and improves the crop yield prediction precision.
Description
Technical Field
The invention relates to the technical field of agricultural remote sensing, in particular to a crop yield prediction method and device based on yield difference.
Background
With the latest progress of data science, deep learning and development of big data technology, the remote sensing big data provides an informatization and intelligentized development approach for agricultural application, and the progress of agricultural remote sensing estimation technology is promoted. Timely, accurate and large-scale crop yield estimation has very important practical significance for guiding scientific production, guaranteeing the income of farmers and ensuring the grain safety in regions.
Currently, the common methods for crop assessment can be roughly divided into two categories, one being of the theoretical drive type and the other being of the data drive type. The method is characterized in that a mechanical model is used for carrying out quantitative simulation on the growth and development and yield formation processes of crops, the obtaining and quantification methods of model parameters have problems, the data processing process is complex, and the prediction precision and efficiency are yet to be improved. The data-driven production estimation method completely depends on available marked data to establish mapping between driving variables and yield, and has the problems of insufficient mechanicalness, weak space-time universality, insufficient generalization capability of models and the like.
Process models and deep learning models are generally viewed as two distinct domains, with very different scientific paradigms, the former being theory-driven and the latter data-driven. Where physical methods are in principle directly interpretable and offer extrapolation potential beyond observation conditions, while data-driven methods are highly flexible in adapting to data. Thus, coupling the two methods for crop yield prediction can take full advantage of the potential of physical knowledge and data analysis.
The existing remote sensing data is applied to a crop yield estimation method, the formation of biomass and yield of crops is mostly simulated by taking a vegetation index of spectral data or spectral combination as an index, and generally the inversion capability of the method is insufficient due to the complex growth process of the crops; the remote sensing data has better representation of stress factors (such as drought, freezing damage and the like) suffered by crops in the growth process, the yield loss caused by the environmental stress of the crops is inverted through the remote sensing data, and the research of estimating the yield by combining a mechanism model is not reported yet.
Disclosure of Invention
The invention provides a crop yield prediction method and device based on yield difference, which are used for overcoming the defects of poor crop yield prediction precision and efficiency and weak universality in the prior art, improving the crop yield prediction precision and efficiency and enhancing the universality.
The invention provides a crop yield prediction method and a device based on yield difference, comprising the following steps:
step S1, collecting field observation data of sample plots in the growth period of crops to be predicted in the research area and corresponding meteorological data, and calibrating WOFOST crop model parameters by adopting a Markov chain-Monte Carlo method based on Bayesian theory;
step S2, inputting the meteorological data of the sample plot in the growth period of the crop to be predicted into a calibrated WOFOST model, and outputting the potential yield of the crop to be predicted;
step S3, subtracting the field observed yield of the crop to be predicted from the potential yield of the crop to be predicted in the step S2 to obtain the actual yield difference of the crop to be predicted;
step S4, taking the satellite remote sensing vegetation index data and the meteorological data of the crops to be predicted in the growth period and the potential yield of the crops to be predicted in the step S2 as input sample variables, and taking the actual yield difference of the crops to be predicted in the step S3 as a label to build a neural network model based on a process model;
step S5, generating potential yield of the crops to be predicted in each crop grid of the target area by the WOFOST model, inputting the potential yield of the crops to be predicted in each crop grid of the target area, corresponding meteorological data and satellite remote sensing vegetation indexes in a growth period into the neural network model set up in the step S4, and generating a predicted yield difference of the crops to be predicted in each crop grid;
and step S6, operating the step S5 one by one crop grid, subtracting the predicted yield difference from the potential yield of the crop to be predicted, and outputting the final predicted yield of the crop to be predicted in each crop grid.
According to the crop yield prediction method and device based on the yield difference, provided by the invention, the WOFOST crop model parameters are calibrated by adopting a Markov chain-Monte Carlo method based on a Bayesian theory, and the method comprises the following steps of:
obtaining posterior distribution of parameters of the WOFOST model according to LAI, observed yield and prior distribution of the parameters of the WOFOST model in the growth period of the crop to be predicted in the sample plot based on a Markov chain-Monte Carlo method of Bayesian theory;
the posterior probability density function expression of the posterior distribution is as follows:
wherein θ and y represent parameters and simulated output values of the WOFOST model, respectively; x represents meteorological data input to the wobest model; p (θ/x, y) is a posterior probability density function of the parameter; f (y/theta, x) is an observation value likelihood function; g (θ) is a prior distribution of the parameter;
and taking the median or the mean of the posterior distribution of each parameter as the calibration result of each parameter.
According to the crop yield prediction method based on the yield difference, provided by the invention, the formula of the objective function for training the neural network model based on the process model is as follows:
R(W)＝λ1‖W‖1+λ2‖W‖2；
wherein,
predicted yield difference for the study area
And the actual yield difference Y, n is the number of observed records in the field of the crop to be predicted in the research area, YiThe actual yield difference corresponding to the ith observation record of the crop to be predicted,
the predicted yield difference corresponding to the ith observation record of the crop to be predicted, W is a regularization term, lambda1And λ2Is a preset coefficient.
According to the crop yield prediction method based on the yield difference, the potential yield of the crop to be predicted in each crop grid of the target area is generated by the WOFOST model, the potential yield of the crop to be predicted in each crop grid of the target area is input into the neural network model set up in the step S4 together with the corresponding meteorological data and the satellite remote sensing vegetation index in the growth period, and the predicted yield difference of the crop to be predicted in each crop grid is generated, and the method comprises the following steps:
carrying out cloud removing processing on the satellite remote sensing vegetation index data based on a cloud shadow detection algorithm;
arranging spectral bands corresponding to the satellite remote sensing vegetation index data after cloud removing according to the growing season of the crops to be predicted;
filtering and smoothing the arrayed satellite remote sensing vegetation index data based on a moving window least square polynomial smoothing algorithm;
inputting the filtered and smoothed satellite remote sensing vegetation index data, meteorological data and potential yield into a neural network model based on a process model, and outputting the predicted yield difference of the crops to be predicted.
According to the crop yield prediction method based on the yield difference, the satellite remote sensing vegetation index data comprise EVI, LWSI and REP.
According to the crop yield prediction method based on the yield difference, the meteorological data comprise average temperature, highest temperature, lowest temperature, precipitation, radiation and wind speed.
The present invention also provides a crop yield prediction apparatus comprising:
the calibration module is used for collecting field observation data of sample plots of crops to be predicted in a research area in a growth period and corresponding meteorological data and calibrating WOFOST crop model parameters by adopting a Markov chain-Monte Carlo method based on a Bayesian theory;
the first prediction module is used for inputting the meteorological data of the sample plot in the growth period of the crop to be predicted into a calibrated WOFOST model and outputting the potential yield of the crop to be predicted;
the first calculation module is used for subtracting the field observed yield of the crop to be predicted from the potential yield of the crop to be predicted to obtain the actual yield difference of the crop to be predicted;
the building module is used for building a neural network model based on a process model by taking satellite remote sensing vegetation index data and meteorological data of the crops to be predicted in the growth period of the crops to be predicted and the potential yield of the crops to be predicted in the first prediction module as input sample variables and taking the actual yield difference of the crops to be predicted in the first calculation module as a label;
the second prediction module is used for generating the potential yield of the crops to be predicted in each crop grid of the target area through the WOFOST model, inputting the potential yield of the crops to be predicted in each crop grid of the target area, corresponding meteorological data in a growth period and a satellite remote sensing vegetation index into the neural network model constructed in the building module, and generating the predicted yield difference of the crops to be predicted in each crop grid;
and the second calculation module is used for generating the predicted yield difference of the crops to be predicted one by one in the crop grids, subtracting the predicted yield difference from the potential yield of the crops to be predicted and outputting the final predicted yield of the crops to be predicted in each crop grid.
The invention also provides an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor executes the program to realize the steps of the method for predicting the crop yield based on the yield difference.
The present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of the method for crop yield prediction based on yield difference as described in any one of the above.
The present invention also provides a computer program product comprising a computer program which, when executed by a processor, performs the steps of the method for predicting crop yield based on yield difference as described in any one of the above.
According to the crop yield prediction method and device based on the process model of the yield difference and the deep learning coupling, the potential yield simulated by the crop mechanism model WOFOST, the satellite remote sensing vegetation index data in the crop growth period and the meteorological data are input into the neural network based on the process model together to obtain the crop predicted yield difference, and the crop predicted yield is obtained by subtracting the potential yield difference. The difference between the simulated yield and the yield observation value of the crop mechanism model based on the physical process can be considered as an error caused by imperfect knowledge of the crop model, and the deep learning model can help to identify and understand the content which is not shown and expressed in the crop model, so that the advantages of the models in two different fields are integrated, the crop yield difference is accurately identified, and the accuracy of crop yield prediction is improved.
Drawings
In order to more clearly illustrate the technical solutions of the present invention or the prior art, the drawings needed for the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and those skilled in the art can also obtain other drawings according to the drawings without creative efforts.
FIG. 1 is a schematic flow chart of a method for predicting crop yield based on yield difference according to the present invention;
FIG. 2 is a schematic diagram of the structure of the device for predicting crop yield based on yield difference provided by the present invention;
fig. 3 is a schematic structural diagram of an electronic device provided in the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is obvious that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be obtained by a person skilled in the art without inventive step based on the embodiments of the present invention, are within the scope of protection of the present invention.
The following describes a method for predicting crop yield based on yield difference according to the present invention with reference to fig. 1, which comprises: step S1, collecting field observation data and corresponding meteorological data of sample plot in growth period of crops to be predicted in a research area, and calibrating WOFOST (World Food Studies) crop model parameters by adopting a Markov Chain-Monte Carlo (MCMC) method based on Bayesian theory;
wherein the research area is an area needing crop yield prediction. The crop to be predicted can be winter wheat and the like, and the embodiment is not particularly limited;
wherein, the data sources of the field observation sample plot include but are not limited to agricultural meteorological observation stations, test stations of all levels and field observation experiment records;
determining parameters of the WOFOST model, and calibrating and obtaining according to a Leaf Area Index (LAI) and an actually measured yield of a crop to be predicted in a growth period;
the WOFOST model key parameters include TSUM1 (emergence to flowering temperature), SPAN (leaf life cycle at 35 ℃) and TDWI (initial dry weight);
calibrating WOFOST model parameters according to an MCMC method, wherein posterior distribution of the parameters of the WOFOST model is obtained according to LAI, observed data of the actually measured yield and prior distribution of key parameters of the WOFOST model in the growth period of the actually measured sample crops; the posterior probability density function calculation formula of the posterior distribution is as follows:
wherein θ and y represent parameters and simulated output values, respectively, of the WOFOST model; x represents meteorological data input to the wobest model; p (θ/x, y) is a posterior probability density function of the parameter; f (y/theta, x) is an observation value likelihood function; g (θ) is a prior distribution of the parameter; and taking the median or the mean of the posterior distribution of each parameter as the calibration result of each parameter.
Step S2, inputting the meteorological data of the sample plot in the growth period of the crop to be predicted into a calibrated WOFOST model, and outputting the potential yield of the crop to be predicted;
step S3, subtracting the field observed yield from the crop model simulated potential yield to be predicted in the step S2 to obtain the actual yield difference of the crop to be predicted, wherein the calculation formula is as follows:
wherein, yiIs the actual yield difference, Y, corresponding to the ith measured sample cropREAiIs the measured yield, Y, corresponding to the ith measured sample cropPHYiAnd (4) simulating the model potential yield corresponding to the ith measured sample crop.
Step S4, training by taking the satellite remote sensing vegetation index data and the meteorological data of the crops to be predicted in the growth period of the crops to be predicted and the potential yield of the crops to be predicted in the step S2 as input sample variables and taking the actual yield difference of the crops to be predicted in the step S3 as a label, and building a neural network model based on a process model;
optionally, the satellite remote sensing vegetation index data is determined according to data obtained by fusing Landsat8 and Sentinel-2, and a time sequence data set with the spatial resolution of 30 meters is formed by fusion.
The process model-based Neural network model is selected from a DNN (Deep Neural Networks) model. The neural network model includes an input layer, an output layer, and a plurality of fully-connected layers. The input layer is used for inputting weather data of a time sequence, satellite remote sensing vegetation index data and potential yield of crops to be predicted, the input characteristic vectors reach the output layer through full-connection layer transformation, and predicted yield difference estimation results are obtained on the output layer. Wherein the first fully connected layer is represented as:
the remaining fully connected layers and the final output layer are represented as:
wherein W represents a weight matrix, b represents a bias term, x is an input vector, f is an activation function, aiIs the output vector, L is the number of network layers,
the final output of the model.
Optionally, the sample data of the research area is divided into training data and test data according to a ratio of 7: 3.
During training, satellite remote sensing vegetation index data, meteorological data and potential yield of actual measurement sample crops in a research area in a growth period are used as sample characteristics of a training set, and a predicted yield difference of the actual measurement sample crops is used as a target variable and is input into a neural network model for training; and evaluating the training result of the model by using the test data.
In this embodiment, the formula of the objective function for training the neural network model based on the process model is as follows:
wherein,
predicted yield difference for the study region
And the actual yield difference Y, n is the number of observed records in the field of the crop to be predicted in the research area, YiThe actual yield difference corresponding to the ith observation record of the crop to be predicted,
the predicted yield difference corresponding to the ith observation record of the crop to be predicted, W is a regularization term, and lambda1And λ2Is a preset coefficient.
Step S5, generating potential yield of the crops to be predicted in each crop grid of the target area by the WOFOST model, inputting the potential yield of the crops to be predicted in each crop grid of the target area, corresponding meteorological data and satellite remote sensing vegetation indexes in a growth period into the neural network model set up in the step S4, and generating a predicted yield difference of the crops to be predicted in each crop grid;
and step S6, operating the step S5 one by one crop grid, subtracting the predicted yield difference from the potential yield of the crop to be predicted, and outputting the final predicted yield of the crop to be predicted in each crop grid.
On the basis of the above embodiment, the inputting the index data of the satellite remote sensing vegetation, the meteorological data and the potential yield into the neural network model based on the process model in the growth period of the field observation sample crop in the embodiment includes: carrying out cloud removing processing on the satellite remote sensing vegetation index data based on a cloud shadow automatic detection algorithm;
and preprocessing the satellite remote sensing vegetation index data before using the satellite remote sensing vegetation index data to predict the yield difference, wherein the preprocessing comprises cloud removing processing and snow removing processing. The present embodiment does not limit the kind of the cloud shadow automatic detection algorithm.
Arranging spectral bands corresponding to the satellite remote sensing vegetation index data after cloud removal according to the growing season of the crops to be predicted;
and arranging the spectral bands corresponding to each data point in the satellite remote sensing vegetation index data after the cloud removal treatment according to the time sequence of the growing seasons of the crops to be predicted.
Filtering and smoothing the arrayed satellite remote sensing vegetation index data based on a moving window least square polynomial smoothing algorithm;
inputting the filtered and smoothed satellite remote sensing vegetation index data, meteorological data and potential yield into a neural network model based on a process model, and outputting the predicted yield difference of the crops to be predicted.
On the basis of the foregoing embodiments, in this embodiment, the satellite remote sensing Vegetation Index data includes an Enhanced Vegetation Index (EVI), a Leaf Water Stress Index (LWSI), and a Red-Edge Position (REP). The present embodiments are not limited to these satellite remote sensing vegetation index data.
The satellite remote sensing vegetation index data is selected from EVI, LWSI and REP time sequences of crops to be predicted in a growth period, and is used for reflecting crop growth and environmental stress effects and supplementing and perfecting the situation that a crop model cannot be completely considered. Wherein EVI reflects crop growth, LWSI reflects crop water stress, and REP reflects crop frost stress.
The calculation formulas of EVI, LWSI and REP are as follows:
where ρ isNIR、ρRED、ρBLUEAnd ρSWIRRespectively, the reflectivities of near infrared, red, blue and short wave infrared bands, and B4 to B7 are the number of bands of Sentinel-2.
On the basis of the above embodiments, the meteorological data in this embodiment includes average temperature, highest temperature, lowest temperature, precipitation, radiation, and wind speed.
The meteorological data includes the average temperature, the maximum temperature and the minimum temperature of the crop to be predicted throughout the growth period. Optionally, the precipitation, radiation and wind speed in the meteorological data are the precipitation monthly mean, radiation monthly mean and wind speed monthly mean of the crops to be predicted in the whole growth period.
In the embodiment, the potential yield of crops output by the crop mechanism model WOFOST, satellite remote sensing vegetation index data in the crop growth period and meteorological data are input into the deep learning model together, the difference value between the output result and the observed value of the crop mechanism model based on the physical process can be considered as an error caused by imperfect knowledge of the crop model reflecting the actual growth status of the crops, the deep learning model can help to identify and understand the content which is not displayed and expressed in the crop model, the advantages of the models in two different fields are integrated, the accurate identification of the crop yield difference is realized, and the accuracy of crop yield prediction is improved.
The following describes a crop yield prediction apparatus provided by the present invention, and the crop yield prediction apparatus described below and the crop yield prediction method described above may be referred to in correspondence with each other.
As shown in fig. 2, the apparatus includes a calibration module 201, a first prediction module 202, a first calculation module 203, a construction module 204, a second prediction module 205, and a second calculation module 206, wherein:
the calibration module 201 is used for collecting field observation data of sample plots in a growth period of crops to be predicted in a research area and corresponding meteorological data, and calibrating WOFOST crop model parameters by adopting a Markov chain-Monte Carlo method based on a Bayesian theory;
the first prediction module 202 is used for inputting the meteorological data of the sample plot in the growth period of the crop to be predicted into a calibrated WOFOST model and outputting the potential yield of the crop to be predicted;
the first calculation module 203 is configured to subtract the field observed yield of the crop to be predicted from the potential yield of the crop to be predicted to obtain an actual yield difference of the crop to be predicted;
the building module 204 is used for building a neural network model based on a process model by taking the satellite remote sensing vegetation index data and the meteorological data of the crops to be predicted in the growth period of the crops to be predicted and the potential yield of the crops to be predicted in the first prediction module as input sample variables and taking the actual yield difference of the crops to be predicted in the first calculation module as a label;
the second prediction module 205 is configured to generate a potential yield of the crop to be predicted in each crop grid of the target area by using the wobest model, input the potential yield of the crop to be predicted in each crop grid of the target area into the neural network model constructed in the construction module together with the corresponding meteorological data and the satellite remote sensing vegetation index in the growth period, and generate a predicted yield difference of the crop to be predicted in each crop grid;
the second calculation module 206 is configured to generate predicted yield differences of the crops to be predicted one by one in a crop grid, subtract the predicted yield differences from the potential yields of the crops to be predicted, and output a final predicted yield of the crops to be predicted in each crop grid.
In the embodiment, the potential yield simulated by the crop mechanism model WOFOST, the satellite remote sensing vegetation index data in the crop growth period and the meteorological data are input into the neural network based on the process model together to obtain the crop predicted yield difference, and the crop predicted yield is finally obtained by subtracting the potential yield. The difference between the simulated yield and the yield observation value of the crop mechanism model based on the physical process can be considered as an error caused by imperfect knowledge of the crop model, and the deep learning model can help to identify and understand the content which is not shown and expressed in the crop model, so that the advantages of the models in two different fields are integrated, the crop yield difference is accurately identified, and the accuracy of crop yield prediction is improved.
Fig. 3 illustrates a physical structure diagram of an electronic device, which may include, as shown in fig. 3: a processor (processor)310, a communication Interface (communication Interface)320, a memory (memory)330 and a communication bus 340, wherein the processor 310, the communication Interface 320 and the memory 330 communicate with each other via the communication bus 340. The processor 310 may invoke logic instructions in the memory 330 to perform a crop yield prediction method comprising: calibrating WOFOST crop model parameters by adopting a Markov chain-Monte Carlo method; inputting meteorological data into a calibrated WOFOST model to generate the potential yield of the crop to be predicted; subtracting the potential yield of the crop to be predicted from the field observed yield to obtain the actual yield difference of the crop to be predicted; taking satellite data, meteorological data and potential yield of an actually measured sample in a crop growth period as input samples, taking actual yield difference of crops to be predicted as a label, and building a neural network model based on a process model; and operating a WOFOST model by crop grids to generate potential yield, inputting the potential yield into a neural network model to generate a predicted yield difference, and subtracting the predicted yield difference from the potential yield to generate a predicted yield space diagram.
In addition, the logic instructions in the memory 330 may be implemented in the form of software functional units and stored in a computer readable storage medium when the software functional units are sold or used as independent products. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk, and various media capable of storing program codes.
In another aspect, the present invention also provides a computer program product, the computer program product comprising a computer program, the computer program being stored on a non-transitory computer readable storage medium, wherein when the computer program is executed by a processor, the computer is capable of executing the crop yield prediction method provided by the above methods, the method comprising: calibrating WOFOST crop model parameters by adopting a Markov chain-Monte Carlo method; inputting meteorological data into a calibrated WOFOST model to generate the potential yield of the crop to be predicted; subtracting the potential yield of the crop to be predicted from the field observed yield to obtain the actual yield difference of the crop to be predicted; taking satellite data, meteorological data and potential yield of actual measurement sample crops in a growth period as input samples, taking actual yield difference of the crops to be predicted as a label, and building a neural network model based on a process model; and operating a WOFOST model by crop grids to generate potential yield, inputting the potential yield into a neural network model to generate a predicted yield difference, and subtracting the predicted yield difference from the potential yield to generate a predicted yield space diagram.
The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment may be implemented by software plus a necessary general hardware platform, and may also be implemented by hardware. With this understanding in mind, the above-described technical solutions may be embodied in the form of a software product, which can be stored in a computer-readable storage medium such as ROM/RAM, magnetic disk, optical disk, etc., and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the methods described in the embodiments or some parts of the embodiments.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (10)
1. A method for predicting crop yield based on yield difference, comprising:
step S1, collecting field observation data of sample plots in the growth period of crops to be predicted in the research area and corresponding meteorological data, and calibrating WOFOST crop model parameters by adopting a Markov chain-Monte Carlo method based on Bayesian theory;
step S2, inputting the meteorological data of the sample plot in the growth period of the crop to be predicted into a calibrated WOFOST model, and outputting the potential yield of the crop to be predicted;
step S3, subtracting the field observed yield of the crop to be predicted from the potential yield of the crop to be predicted in the step S2 to obtain the actual yield difference of the crop to be predicted;
step S4, taking the satellite remote sensing vegetation index data and the meteorological data of the crops to be predicted in the growth period and the potential yield of the crops to be predicted in the step S2 as input sample variables, and taking the actual yield difference of the crops to be predicted in the step S3 as a label to build a neural network model based on a process model;
step S5, generating potential yield of the crops to be predicted in each crop grid of the target area by the WOFOST model, inputting the potential yield, the meteorological data and the satellite remote sensing vegetation index in the corresponding growth period into the neural network model set up in the step S4 together, and generating a predicted yield difference of the crops to be predicted in each crop grid;
and step S6, operating the step S5 one by one crop grid, subtracting the predicted yield difference from the potential yield of the crop to be predicted, and outputting the final predicted yield of the crop to be predicted in each crop grid.
2. The method of predicting crop yield based on yield difference according to claim 1, wherein the calibrating wocost crop model parameters using a markov chain-monte carlo method based on bayes theory comprises:
obtaining posterior distribution of parameters of the WOFOST model according to LAI, observed yield and prior distribution of the parameters of the WOFOST model in the growth period of the crop to be predicted in the sample plot based on a Markov chain-Monte Carlo method of Bayesian theory;
the posterior probability density function expression of the posterior distribution is as follows:
wherein θ and y represent parameters and simulated output values of the WOFOST model, respectively; x represents the meteorological data input into the wocost model; p (θ/x, y) is a posterior probability density function of the parameter; f (y/theta, x) is an observation value likelihood function; g (θ) is a prior distribution of the parameter;
and taking the median or the mean of the posterior distribution of each parameter as the calibration result of each parameter.
3. The method of claim 1, wherein the formula of the objective function for training the neural network model based on the process model is as follows:
R(W)＝λ1‖W‖1+λ2‖W‖2；
wherein,
predicted yield difference for the study region
And the actual yield difference Y, n is the number of observed records in the field of the crop to be predicted in the research area, YiThe actual yield difference corresponding to the ith observation record of the crop to be predicted,
the predicted yield difference corresponding to the ith observation record of the crop to be predicted, W is a regularization term, lambda1And λ2Is a preset coefficient.
4. The method for predicting crop yield based on yield difference according to claim 1, wherein the generating potential yield of the crop to be predicted in each crop grid of the target area by the wobest model, together with the meteorological data and the satellite remote sensing vegetation index in the corresponding growth period, is input into the neural network model set up in the step S4 to generate the predicted yield difference of the crop to be predicted in each crop grid, comprises:
carrying out cloud removing processing on the satellite remote sensing vegetation index data based on a cloud shadow detection algorithm;
arranging spectral bands corresponding to the satellite remote sensing vegetation index data after cloud removal according to the growing season of the crops to be predicted;
filtering and smoothing the arrayed satellite remote sensing vegetation index data based on a moving window least square polynomial smoothing algorithm;
inputting the filtered and smoothed satellite remote sensing vegetation index data, meteorological data and potential yield into a neural network model based on a process model, and outputting the predicted yield difference of the crops to be predicted.
5. The method of crop yield prediction based on yield differences according to any of claims 1-4, wherein the satellite remote sensing vegetation index data comprises EVI, LWSI and REP.
6. The method of any one of claims 1 to 4, wherein the meteorological data comprises average temperature, maximum temperature, minimum temperature, precipitation, radiation and wind speed.
7. A crop yield prediction apparatus based on yield difference, comprising:
the calibration module is used for collecting field observation data of sample plots of crops to be predicted in a research area in a growth period and corresponding meteorological data and calibrating WOFOST crop model parameters by adopting a Markov chain-Monte Carlo method based on a Bayesian theory;
the first prediction module is used for inputting the meteorological data of the sample plot in the growth period of the crop to be predicted into a calibrated WOFOST model and outputting the potential yield of the crop to be predicted;
the first calculation module is used for subtracting the field observed yield of the crop to be predicted from the potential yield of the crop to be predicted to obtain the actual yield difference of the crop to be predicted;
the building module is used for building a neural network model based on a process model by taking satellite remote sensing vegetation index data and meteorological data of the crops to be predicted in the growth period of the crops to be predicted and the potential yield of the crops to be predicted in the first prediction module as input sample variables and taking the actual yield difference of the crops to be predicted in the first calculation module as a label;
the second prediction module is used for generating the potential yield of the crops to be predicted in each crop grid of the target area through the WOFOST model, inputting the potential yield of the crops to be predicted in each crop grid of the target area, corresponding meteorological data in a growth period and a satellite remote sensing vegetation index into the neural network model constructed in the building module, and generating the predicted yield difference of the crops to be predicted in each crop grid;
and the second calculation module is used for generating the predicted yield difference of the crops to be predicted one by one in the crop grids, subtracting the predicted yield difference from the potential yield of the crops to be predicted and outputting the final predicted yield of the crops to be predicted in each crop grid.
8. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor when executing the program performs the steps of the method for predicting crop yield based on yield difference according to any one of claims 1 to 6.
9. A non-transitory computer readable storage medium having a computer program stored thereon, wherein the computer program when executed by a processor implements the steps of the method for predicting crop yield based on yield difference according to any one of claims 1 to 6.
10. A computer program product comprising a computer program, wherein the computer program when executed by a processor implements the steps of the method for crop yield prediction based on yield difference according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210330408.XA CN114692971B (en) | 2022-03-30 | 2022-03-30 | Crop yield prediction method and device based on yield difference |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210330408.XA CN114692971B (en) | 2022-03-30 | 2022-03-30 | Crop yield prediction method and device based on yield difference |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114692971A true CN114692971A (en) | 2022-07-01 |
| CN114692971B CN114692971B (en) | 2024-08-20 |
Family
ID=82140539
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210330408.XA Active CN114692971B (en) | 2022-03-30 | 2022-03-30 | Crop yield prediction method and device based on yield difference |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114692971B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115753625A (en) * | 2022-11-02 | 2023-03-07 | 中国农业大学 | Method and device for estimating yield of regional crops, electronic equipment and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104134095A (en) * | 2014-04-17 | 2014-11-05 | 中国农业大学 | Crop yield estimation method based on scale transformation and data assimilation |
| CN108509836A (en) * | 2018-01-29 | 2018-09-07 | 中国农业大学 | Crop yield estimation method based on double-polarized synthetic aperture radar and crop model data assimilation |
| CN109754125A (en) * | 2019-01-18 | 2019-05-14 | 中国农业大学 | Crop yield forecast method based on crop modeling, history and meteorological forecast data |
| CN110222870A (en) * | 2019-05-05 | 2019-09-10 | 中国农业大学 | Assimilate the Regional Fall Wheat yield estimation method of satellite fluorescence data and crop growth model |
| CN112667955A (en) * | 2020-12-24 | 2021-04-16 | 青岛大学 | Method for estimating regional scale corn potential yield and yield difference based on remote sensing and application |
| CN113537604A (en) * | 2021-07-21 | 2021-10-22 | 中国农业大学 | Crop yield prediction method and device by coupling process model and deep learning |
-
2022
- 2022-03-30 CN CN202210330408.XA patent/CN114692971B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104134095A (en) * | 2014-04-17 | 2014-11-05 | 中国农业大学 | Crop yield estimation method based on scale transformation and data assimilation |
| CN108509836A (en) * | 2018-01-29 | 2018-09-07 | 中国农业大学 | Crop yield estimation method based on double-polarized synthetic aperture radar and crop model data assimilation |
| CN109754125A (en) * | 2019-01-18 | 2019-05-14 | 中国农业大学 | Crop yield forecast method based on crop modeling, history and meteorological forecast data |
| CN110222870A (en) * | 2019-05-05 | 2019-09-10 | 中国农业大学 | Assimilate the Regional Fall Wheat yield estimation method of satellite fluorescence data and crop growth model |
| CN112667955A (en) * | 2020-12-24 | 2021-04-16 | 青岛大学 | Method for estimating regional scale corn potential yield and yield difference based on remote sensing and application |
| CN113537604A (en) * | 2021-07-21 | 2021-10-22 | 中国农业大学 | Crop yield prediction method and device by coupling process model and deep learning |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115753625A (en) * | 2022-11-02 | 2023-03-07 | 中国农业大学 | Method and device for estimating yield of regional crops, electronic equipment and storage medium |
| CN115753625B (en) * | 2022-11-02 | 2024-04-19 | 中国农业大学 | Yield estimation method and device for regional crops, electronic equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114692971B (en) | 2024-08-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Tian et al. | A deep learning framework under attention mechanism for wheat yield estimation using remotely sensed indices in the Guanzhong Plain, PR China | |
| Sharma et al. | Wheat crop yield prediction using deep LSTM model | |
| CN111027752A (en) | Crop yield estimation method based on deep spatiotemporal feature joint learning | |
| Wang et al. | A deep learning framework combining CNN and GRU for improving wheat yield estimates using time series remotely sensed multi-variables | |
| Dong et al. | Comparison and analysis of data assimilation algorithms for predicting the leaf area index of crop canopies | |
| Li et al. | A novel multichannel long short-term memory method with time series for soil temperature modeling | |
| CN115018105A (en) | Winter wheat meteorological yield prediction method and system | |
| CN114529097A (en) | Multi-scale crop phenological period remote sensing dimensionality reduction prediction method | |
| CN113537645A (en) | Soybean yield prediction method based on machine learning fusion satellite and weather data | |
| CN114692971B (en) | Crop yield prediction method and device based on yield difference | |
| CN116343046A (en) | Multi-source remote sensing urban river water quality monitoring method based on self-optimizing machine learning | |
| Wang et al. | Application of hyperspectral technology combined with bat algorithm-AdaBoost model in field soil nutrient prediction | |
| Stathakis et al. | Neuro-fuzzy modeling for crop yield prediction | |
| Yang et al. | Prediction of corn variety yield with attribute-missing data via graph neural network | |
| Chen et al. | Integrating a crop growth model and radiative transfer model to improve estimation of crop traits based on deep learning | |
| CN116629453B (en) | Remote sensing yield estimation method suitable for whole crop growth period | |
| Kumawat et al. | Time-Variant Satellite Vegetation Classification Enabled by Hybrid Metaheuristic-Based Adaptive Time-Weighted Dynamic Time Warping | |
| CN117764238A (en) | Crop pest prediction method based on complex graph characterization, electronic equipment and storage medium | |
| Monica et al. | Soil NPK prediction using enhanced genetic algorithm | |
| CN117131758A (en) | Training method and device of irrigation water estimation model and irrigation water estimation method | |
| CN116151454A (en) | Method and system for predicting yield of short-forest linalool essential oil by multispectral unmanned aerial vehicle | |
| Cheng et al. | Quantifying corn LAI using machine learning and UAV multispectral imaging | |
| Zhou et al. | Extracting tobacco planting areas using LSTM from time series Sentinel-1 SAR data | |
| CN118097435B (en) | Supergraph neural network-based corn lodging classification method and device | |
| Honda et al. | Prediction of Photosynthetic Carbon Assimilation Rate of Individual Rice Leaves under Changes in Light Environment Using BLSTM-Augmented LSTM. |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant |