CN111241912A - Multi-vegetation index rice yield estimation method based on machine learning algorithm - Google Patents
Multi-vegetation index rice yield estimation method based on machine learning algorithm Download PDFInfo
- Publication number
- CN111241912A CN111241912A CN201911311109.6A CN201911311109A CN111241912A CN 111241912 A CN111241912 A CN 111241912A CN 201911311109 A CN201911311109 A CN 201911311109A CN 111241912 A CN111241912 A CN 111241912A
- Authority
- CN
- China
- Prior art keywords
- yield
- rice
- model
- machine learning
- data
- 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.)
- Pending
Links
- 235000007164 Oryza sativa Nutrition 0.000 title claims abstract description 52
- 235000009566 rice Nutrition 0.000 title claims abstract description 52
- 238000004422 calculation algorithm Methods 0.000 title claims abstract description 33
- 238000000034 method Methods 0.000 title claims abstract description 33
- 238000010801 machine learning Methods 0.000 title claims abstract description 22
- 240000007594 Oryza sativa Species 0.000 title 1
- 241000209094 Oryza Species 0.000 claims abstract description 54
- IJGRMHOSHXDMSA-UHFFFAOYSA-N Atomic nitrogen Chemical compound N#N IJGRMHOSHXDMSA-UHFFFAOYSA-N 0.000 claims abstract description 16
- 230000012010 growth Effects 0.000 claims abstract description 13
- 238000012360 testing method Methods 0.000 claims abstract description 12
- 229910052757 nitrogen Inorganic materials 0.000 claims abstract description 8
- 238000005516 engineering process Methods 0.000 claims abstract description 5
- 238000011156 evaluation Methods 0.000 claims abstract description 4
- 238000001228 spectrum Methods 0.000 claims abstract description 4
- 239000003337 fertilizer Substances 0.000 claims description 23
- 238000007637 random forest analysis Methods 0.000 claims description 18
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 15
- 238000012706 support-vector machine Methods 0.000 claims description 14
- 230000005855 radiation Effects 0.000 claims description 13
- 238000005259 measurement Methods 0.000 claims description 11
- 238000012795 verification Methods 0.000 claims description 9
- 235000013339 cereals Nutrition 0.000 claims description 7
- 238000010276 construction Methods 0.000 claims description 7
- 238000012937 correction Methods 0.000 claims description 7
- 230000000694 effects Effects 0.000 claims description 7
- 229910019142 PO4 Inorganic materials 0.000 claims description 6
- KWYUFKZDYYNOTN-UHFFFAOYSA-M Potassium hydroxide Chemical compound [OH-].[K+] KWYUFKZDYYNOTN-UHFFFAOYSA-M 0.000 claims description 6
- 238000005286 illumination Methods 0.000 claims description 6
- NBIIXXVUZAFLBC-UHFFFAOYSA-K phosphate Chemical compound [O-]P([O-])([O-])=O NBIIXXVUZAFLBC-UHFFFAOYSA-K 0.000 claims description 6
- 239000010452 phosphate Substances 0.000 claims description 6
- 229940072033 potash Drugs 0.000 claims description 6
- 235000015320 potassium carbonate Nutrition 0.000 claims description 6
- BWHMMNNQKKPAPP-UHFFFAOYSA-L potassium carbonate Substances [K+].[K+].[O-]C([O-])=O BWHMMNNQKKPAPP-UHFFFAOYSA-L 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000007726 management method Methods 0.000 claims description 4
- 238000011160 research Methods 0.000 claims description 4
- 208000027066 STING-associated vasculopathy with onset in infancy Diseases 0.000 claims description 3
- 238000007605 air drying Methods 0.000 claims description 3
- 238000004458 analytical method Methods 0.000 claims description 3
- 238000002790 cross-validation Methods 0.000 claims description 3
- 238000005520 cutting process Methods 0.000 claims description 3
- 230000001419 dependent effect Effects 0.000 claims description 3
- 238000003306 harvesting Methods 0.000 claims description 3
- 239000012535 impurity Substances 0.000 claims description 3
- 239000000618 nitrogen fertilizer Substances 0.000 claims description 3
- 238000012805 post-processing Methods 0.000 claims description 3
- 230000003595 spectral effect Effects 0.000 claims description 3
- 238000005303 weighing Methods 0.000 claims description 3
- 241001116459 Sequoia Species 0.000 claims description 2
- 230000008451 emotion Effects 0.000 claims description 2
- 238000001556 precipitation Methods 0.000 claims 1
- 239000002699 waste material Substances 0.000 abstract description 3
- 238000012549 training Methods 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 238000007689 inspection Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 238000004519 manufacturing process Methods 0.000 description 2
- 210000000056 organ Anatomy 0.000 description 2
- 238000002310 reflectometry Methods 0.000 description 2
- 239000002028 Biomass Substances 0.000 description 1
- 241000196324 Embryophyta Species 0.000 description 1
- 229920002472 Starch Polymers 0.000 description 1
- 238000010521 absorption reaction Methods 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 238000012271 agricultural production Methods 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 235000013336 milk Nutrition 0.000 description 1
- 239000008267 milk Substances 0.000 description 1
- 210000004080 milk Anatomy 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 235000016709 nutrition Nutrition 0.000 description 1
- 230000035764 nutrition Effects 0.000 description 1
- 230000000050 nutritive effect Effects 0.000 description 1
- 230000029553 photosynthesis Effects 0.000 description 1
- 238000010672 photosynthesis Methods 0.000 description 1
- 102000004169 proteins and genes Human genes 0.000 description 1
- 108090000623 proteins and genes Proteins 0.000 description 1
- 235000019698 starch Nutrition 0.000 description 1
- 239000008107 starch Substances 0.000 description 1
- 238000013179 statistical model Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 230000009105 vegetative growth Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/188—Vegetation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/20—Ensemble learning
-
- 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
- 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—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/02—Agriculture; Fishing; Mining
Abstract
The invention discloses a rice yield estimation method of multiple vegetation indexes based on a machine learning algorithm, which takes a typical large-scale farmland block of a Sanjiang seven-star farm established in Heilongjiang province, which is a main planting area of rice in a cold area as an object, acquires spectrum data of rice jointing Stage (SE) and heading stage (HD) in 2017 and 2018 of the cold area by setting field scale tests with different nitrogen levels and combining an unmanned aerial vehicle remote sensing technology, constructs a regression model based on the multiple vegetation indexes and the actually measured yield of the rice in the cold area in the two periods, and performs model evaluation by combining the actually measured yield of the crop, can quickly and effectively diagnose and estimate the growth vigor and the yield of the crop, provides a quick and efficient management tool for large-scale planting and management, effectively solves the problem that a single vegetation index estimation method cannot comprehensively consider the influence of other vegetation indexes, and simultaneously solves the phenomenon of remote sensing data waste, the practicability is strong.
Description
Technical Field
The invention relates to the technical field of agricultural production, in particular to a multi-vegetation index rice yield estimation method based on a machine learning algorithm.
Background
At present, a satellite remote sensing method is mostly used for large-scale crop estimation, but the influence of the satellite remote sensing on weather, spatial resolution and climate is large. For farmland management, satellite remote sensing is difficult to realize high-precision yield estimation, and particularly in a rainy and poor-lighting rice planting area, a satellite image with high resolution is acquired in a proper period, so that high-precision yield estimation is difficult to realize. In recent years, the unmanned aerial vehicle platform with the advantages of low cost, strong mobility, simple operation, large observation range and the like is developed rapidly, remote sensing can provide a remote sensing image with high space-time resolution based on an unmanned aerial vehicle remote sensing system, a suitable image with higher precision can be provided, and a new way is provided for farmland information acquisition and field scale yield estimation.
In addition, many scholars at home and abroad develop the research of remote sensing technology for estimating crop yield, which mainly estimates the yield by establishing a statistical model between vegetation indexes and the yield and using remote sensing data. However, the single vegetation index estimation method cannot comprehensively consider the influence of other vegetation indexes, and simultaneously causes the waste of remote sensing data of other vegetation indexes to be solved.
Disclosure of Invention
Technical problem to be solved
Aiming at the defects of the prior art, the invention provides a multi-vegetation index rice yield estimation method based on a machine learning algorithm, and solves the problems that the influence of other vegetation indexes cannot be comprehensively considered in the conventional single vegetation index estimation method, and the waste of remote sensing data of other vegetation is caused.
(II) technical scheme
In order to achieve the purpose, the invention is realized by the following technical scheme: a rice yield estimation method based on multiple vegetation indexes of a machine learning algorithm is characterized in that a typical large-scale farmland block of a Sanjiang seven-star farm built in Heilongjiang province in cold regions is taken as an object, spectrum data of rice jointing Stages (SE) and heading stages (HD) in the cold regions in 2017 and 2018 are obtained by combining a field block scale test with different nitrogen levels and an unmanned aerial vehicle remote sensing technology, a regression model based on the multiple vegetation indexes and the actually measured yield of the cold regions is constructed, and model evaluation is carried out by combining the actually measured yield of crops, and the method comprises the following steps:
① Experimental arrangement of research area, establishing seven-star farm (47.01 degree N-47.29 degree N, 132.31 degree E-134.14 degree E) of the administration of three kingdom branches in the general Times of agricultural reclamation in the northeast of China, wherein the main climate type belongs to the continental monsoon climate of the cold and warm zone, the annual average sunshine duration is 2300 minus one-year-old 2600 hours, the frost-free period is 110 minus one-year-old 135 days, the annual rainfall is 500-600 mm, 72% is concentrated in 6-9 months, and different nitrogen levels, varieties (Longjing 21, Longjing 31) and densities (27 holes/m) are developed in 2017 and 2018233 points/m2) And 3 repetitions are set, each treatment cell field area being 7 x 9m2When the rice is raised in the greenhouse in the middle of 4 months, the transplanting time is 5 middle of the month, the harvesting time is 9 late of the month, and nitrogen fertilizer is applied in the transplanting period, the tillering period and the jointing period respectively, wherein the base fertilizer: and (3) fertilizing the tillers: the panicle fertilizer is 4: 3: 3, the application amount of all the phosphate fertilizers and the potash fertilizers is the same, and 50kg of P is completely applied to the phosphate fertilizers before the rice is transplanted2O5ha-1. The application amount of the potash fertilizer is 105kg K2O ha-1Respectively applying 50% of the fertilizer before transplanting and in the jointing stage of the rice.
② remote sensing data acquisition, using unmanned aerial vehicle to synchronously carry a multispectral camera, wherein the camera comprises four wave bands of green light (central wavelength G550nm, wave band width 40nm), red light (R660 nm, 40nm), red edge (RE 735nm, 10nm) and near infrared (NIR 790nm, 40nm), and is also provided with an RGB sensor, the focal length of the camera multispectral sensor lens is 3.98nm, the image pixel size is 1280 x 960, in addition, the camera is also provided with an illumination sensor and a calibration white board simultaneously, the white board image is acquired for radiation calibration before takeoff, the radiation illumination information is acquired during each shooting, the radiation correction of the data during the post-processing is convenient, the navigation time is selected to be about 10: 00-14: 00 at noon, no wind and little cloud are required during the measurement, the flight control software is adopted to plan the flight route and set flight parameters during the data acquisition, after the flight is finished, the image data with high-precision position information is derived and copied into the splicing software for automatic splicing, the whole experimental area is acquired, and the reflectivity image file after the radiation correction, the GPS coordinate information is displayed correctly.
③ data acquisition of yield comprises selecting 3 samples of 1 square meter with uniform growth at each position during the mature period of rice, cutting off all rice plants in the samples, air drying and threshing, removing impurities and empty and shriveled grains by winnowing, weighing to obtain yield per square meter, measuring water content of grain sample by rapid water content meter, and converting the yield to standard yield of 14% water content.
Yield calculation formula: the yield was expressed in t/ha units, which is the actual yield per meter (g) per meter (1-actual water content)/(1-standard water content 0.14) per meter (0.01).
④ construction and verification of multiple vegetation index combination estimation model based on machine learning algorithm, wherein ten vegetation indexes with better estimation yield, namely NDVI, RVI, DVI, SAVI, GOSAVI, GNDVI, GRVI, OSAVI, NREI, MNDI, NDRE, WDRVI, are selected through multiple tests.
And randomly dividing test data of 2017 and 2018, wherein 70% of data sets are used for model construction, 30% of data sets are used for model verification, regression models between different vegetation indexes and rice yield in different periods are constructed, and a machine learning method is adopted for modeling.
Two very popular machine learning algorithms of Random Forest (RF) and Support Vector Machine (SVM) are selected for modeling analysis, the two algorithms can well solve the problems of classification and prediction of small sample data, ten times of cross validation is adopted, and optimal parameters are searched for the RF and the SVM through network search.
Respectively taking the actual measurement yield of the rice in the jointing stage and the heading stage as dependent variables, taking all vegetation indexes as independent variables, and constructing a remote sensing inversion model of the yield by using a random forest algorithm regression algorithm; thirdly, constructing a remote sensing inversion model with yield by using a support vector machine, and evaluating and verifying the effect of the constructed model by selecting three indexes of a decision coefficient (R2), a Root Mean Square Error (RMSE) and a Relative Error (RE), wherein the closer to 1 the R2 is, the closer to 0 the RMSE and the RE are, the better the model effect is, and the more 10% the RE is, the better the model performance is; the model performance is general when the ratio is 20-30%; greater than 30% indicates poor performance. The specific formula is shown as formula (1), formula (2) and formula (3).
Wherein, YiActual measurement of crop yield for the ith sample point (t/hm)2)；EiFor the model calculated crop yield estimate for the ith sample point (t/hm)2)；
Is the average yield (t/hm) of actual measurement2)；
Average yield (t/hm) for model estimation2)。
The unmanned aerial vehicle is an ebee SQ unmanned aerial vehicle, and the multispectral camera carried by the unmanned aerial vehicle is a Sequoia multispectral camera.
The flight control software is eMotion Ag 3.5.0 flight control software, the flight height is set to be 106.1 meters, the side-by-side overlapping rate is set to be 75 percent, and the resolution of image pixels is 0.1 meter.
The splicing software is Pix4Dag 3.2.23 splicing software, and a spectral reflectance image file which can cover the whole experimental area and is subjected to radiation correction is obtained by accurate positioning and splicing of a GPS.
(III) advantageous effects
The invention provides a machine learning algorithm-based rice yield estimation method for multiple vegetation indexes. The method has the following beneficial effects:
the invention can obviously improve the precision of the estimated model by using a machine learning method, the random forest algorithm has higher precision than a support vector machine algorithm model, and the training set R in the heading period2Can reach 0.94, and the verification set RMSE is 0.59t/hm2The rice yield estimation method based on the multiple vegetation indexes of the machine learning algorithm can quickly and effectively diagnose and estimate the growth vigor and the yield of crops, provides a quick and efficient management tool for large-scale planting and management, effectively solves the problem that the single vegetation index estimation method cannot comprehensively consider the influence of other vegetation indexes, fully utilizes the data acquired by remote sensing by utilizing the estimation method of the multiple vegetation indexes, solves the problem that the remote sensing data of other vegetation only utilizing single vegetation is wasted, and has stronger practicability and higher accuracy.
Drawings
FIG. 1 is a multispectral image acquired by an unmanned aerial vehicle in the heading stage of 2018;
fig. 2 is a graph comparing estimated production with actual production in different growth periods under different modeling algorithms based on unmanned aerial vehicle remote sensing.
Detailed Description
The technical solutions in the embodiments of the present invention will be described clearly and completely with reference to the accompanying drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The embodiment of the invention provides a method for estimating rice yield based on multiple vegetation indexes of a machine learning algorithm, which takes a typical large-scale farmland block of a Sanjiang seven-star farm established in Heilongjiang province of cold region rice as an object, acquires spectrum data of rice jointing Stage (SE) and heading stage (HD) in 2017 and 2018 of the cold region by setting field scale tests with different nitrogen levels and combining an unmanned remote sensing technology, constructs a regression model based on the multiple vegetation indexes and the actual measured yield of the cold region rice in the two periods, and performs model evaluation by combining the actual measured yield of crops, and comprises the following steps:
① Experimental arrangement of research area, establishing seven-star farm (47.01 degree N-47.29 degree N, 132.31 degree E-134.14 degree E) of the administration of three kingdom branches in the general Times of agricultural reclamation in the northeast of China, wherein the main climate type belongs to the continental monsoon climate of the cold and warm zone, the annual average sunshine duration is 2300 minus one-year-old 2600 hours, the frost-free period is 110 minus one-year-old 135 days, the annual rainfall is 500-600 mm, 72% is concentrated in 6-9 months, and different nitrogen levels, varieties (Longjing 21, Longjing 31) and densities (27 holes/m) are developed in 2017 and 2018233 points/m2) And 3 repetitions are set, each treatment cell field area being 7 x 9m2When the rice is raised in the greenhouse in the middle of 4 months, the transplanting time is 5 middle of the month, the harvesting time is 9 late of the month, and nitrogen fertilizer is applied in the transplanting period, the tillering period and the jointing period respectively, wherein the base fertilizer: and (3) fertilizing the tillers: the panicle fertilizer is 4: 3: 3, the application amount of all the phosphate fertilizers and the potash fertilizers is the same, and 50kg of P is completely applied to the phosphate fertilizers before the rice is transplanted2O5ha-1. The application amount of the potash fertilizer is 105kg K2O ha-1Respectively applying 50% of the fertilizer before transplanting and in the jointing stage of the rice.
TABLE 12017 and 2018 shows the amount of fertilizer applied in each treatment of the test field
Table 1 shows the amount of fertilizer applied in each of the test plots in 2017 and 2018.
② remote sensing data acquisition, using unmanned aerial vehicle to synchronously carry a multispectral camera, wherein the camera comprises four wave bands of green light (central wavelength G550nm, wave band width 40nm), red light (R660 nm, 40nm), red edge (RE 735nm, 10nm) and near infrared (NIR 790nm, 40nm), and is also provided with an RGB sensor, the focal length of the camera multispectral sensor lens is 3.98nm, the image pixel size is 1280 x 960, in addition, the camera is also provided with an illumination sensor and a calibration white board simultaneously, the white board image is acquired for radiation calibration before takeoff, the radiation illumination information is acquired during each shooting, the radiation correction of the data during the post-processing is convenient, the navigation time is selected to be about 10: 00-14: 00 at noon, no wind and little cloud are required during the measurement, the flight control software is adopted to plan the flight route and set flight parameters during the data acquisition, after the flight is finished, the image data with high-precision position information is derived and copied into the splicing software for automatic splicing, the whole experimental area is acquired, and the reflectivity image file after the radiation correction, the GPS coordinate information is displayed correctly.
③ the data acquisition of yield is carried out by selecting 1 square meter of uniform growth at each position in the mature period of rice, cutting all rice plants in the sample, air drying and threshing, removing impurities and empty and shriveled grains by winnowing, weighing to obtain yield per square meter, measuring water content of grain sample by a rapid water content tester, converting the yield into standard yield of 14% in water content state, dividing the mature period of rice into milk stage, wax stage and complete stage according to time intervals, and selecting the sample by paying attention to the sample at the complete stage to avoid misleading information caused by error of data acquisition.
Yield calculation formula: the yield was expressed in t/ha units, which is the actual yield per meter (g) per meter (1-actual water content)/(1-standard water content 0.14) per meter (0.01).
④ construction and verification of multiple vegetation index combination estimation model based on machine learning algorithm, wherein ten vegetation indexes with better estimation yield, namely NDVI, RVI, DVI, SAVI, GOSAVI, GNDVI, GRVI, OSAVI, NREI, MNDI, NDRE, WDRVI, are selected through multiple tests.
TABLE 2 vegetation index selected
Table 2 shows the vegetation index names and calculation methods, where RE is the red band, G is the green band, and NIR is the near infrared band.
And randomly dividing test data of 2017 and 2018, wherein 70% of data sets are used for model construction, 30% of data sets are used for model verification, regression models between different vegetation indexes and rice yield in different periods are constructed, and a machine learning method is adopted for modeling.
TABLE 3 sample Rice yield statistics
Table 3 shows the statistics of the yield of the rice samples.
Two very popular machine learning algorithms of Random Forest (RF) and Support Vector Machine (SVM) are selected for modeling analysis, the two algorithms can well solve the problems of classification and prediction of small sample data, ten times of cross validation is adopted, and optimal parameters are searched for the RF and the SVM through network search.
Respectively taking the actual measurement yield of the rice in the jointing stage and the heading stage as dependent variables, taking all vegetation indexes as independent variables, and constructing a remote sensing inversion model of the yield by using a random forest algorithm regression algorithm; thirdly, constructing a remote sensing inversion model with yield by using a support vector machine, and evaluating and verifying the effect of the constructed model by selecting three indexes of a decision coefficient (R2), a Root Mean Square Error (RMSE) and a Relative Error (RE), wherein the closer to 1 the R2 is, the closer to 0 the RMSE and the RE are, the better the model effect is, and the more 10% the RE is, the better the model performance is; the model performance is general when the ratio is 20-30%; greater than 30% indicates poor performance. The specific formula is shown as formula (1), formula (2) and formula (3).
Wherein, YiActual measurement of crop yield for the ith sample point (t/hm)2)；EiFor the crop yield estimate (t/hm) for the ith sample point calculated from the model2)；
Is the average yield (t/hm) of actual measurement2)；
Average yield (t/hm) for model estimation2)。
According to a model construction and inspection method, an estimation model of rice yield in the heading stage and the heading stage is constructed, learning ability of a machine learning algorithm of 2 machines is evaluated by utilizing R2, RMSE and RE, and results show that similar to empirical model results, the results are ideal in performance of R2, RMSE and RE in the heading stage, wherein the performance of a random forest regression algorithm is superior to that of a support vector machine algorithm, the modeling precision R2 of a training set in the heading stage reaches 0.94, the RMSE in inspection of a verification set is 0.59t/hm2, and the performance is ideal; the modeling precision R2 of the training set of the SVR algorithm in the heading stage reaches 0.85, the RMSE of the verification set is 0.65t/hm2, and the method is also superior to the empirical model of a single vegetation index.
TABLE 4 correlation between rice yield and vegetation index at different growth periods under different estimation models
RF in Table 4: random forests; SVR: and (4) supporting vector machine regression.
The influence of different growth periods of rice on vegetation index estimation-ground space heterogeneity changes with different growth periods of crops, so that the precision of an estimation model based on the vegetation index of different growth periods of crops is obviously different, vegetative growth is mainly carried out in the jointing stage of the rice, the vegetation index in the period cannot reflect the dry matter accumulation process of yield forming organs, the estimation precision is not high, the heading and filling stage is a stage in which the rice transfers organic matters such as starch, protein and the like generated by photosynthesis from nutritive organs into grains, and the heading stage is the most key stage of yield formation because rice seedlings grow into the rice seedlings and do not grow any longer and are the most complete period of nutrition absorption of the rice.
The estimation accuracy of different vegetation indexes, namely the normalized vegetation index NDVI, is used as a reference quantity of economic and effective practical ground vegetation coverage and growth, is the vegetation index which is most widely applied at present, but can better indicate the growth and biomass of plants when the vegetation coverage is lower, but shows saturation characteristics in a high vegetation coverage area and the indicating capability is reduced, the estimation accuracy of the estimation model based on the NNIR of the cold-region rice heading stage is highest, and R2(n is 80) reaches 0.8.
The estimation precision of different modeling methods in the invention, because the multiple linear regression integrates some vegetation indexes with poor performance, different vegetation index combinations should be optimized in principle to improve the precision of the estimation model. The random forest and vector machine regression are used for modeling the multiple vegetation indexes and the yield in multiple growth periods, a full subset method is utilized, the training sample R2 obtained based on the heading period can reach 0.94, the R2 of the inspection sample also has 0.83, and therefore the method for estimating the rice yield by using the random forest in the heading period has great potential.
It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation. The use of the phrase "comprising one.. said element does not exclude the presence of other, same elements in a process, method, article, or apparatus that comprises the element.
Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (4)
1. A rice yield estimation method based on multiple vegetation indexes of a machine learning algorithm is characterized in that a typical large-scale farmland block of a Sanjiang seven-star farm built in Heilongjiang province in cold regions is taken as an object, spectrum data of rice jointing Stages (SE) and heading stages (HD) in the cold regions in 2017 and 2018 are obtained by setting field scale tests with different nitrogen levels and combining an unmanned aerial vehicle remote sensing technology, a regression model based on the multiple vegetation indexes and the actually measured yield of the cold regions is constructed, and model evaluation is carried out by combining the actually measured yield of crops, and the method comprises the following steps:
① Experimental setup of research area, establishing seven-star farm (47.01 degree N-47.29 degree N, 132.31 degree E-134.14 degree E) of three-river branch administration management bureau in general Jones of agricultural reclamation of Heilongjiang province in northeast China, wherein the main climate type belongs to continental monsoon climate in cold and warm zone, the annual average sunshine duration is 2300 minus one year 2600 hours, the frost-free period is 110 minus one year for 135 days, and the annual precipitation is 500-600mm, 72% of nitrogen concentration in 6-9 months, and different nitrogen levels, varieties (Longjing 21 and Longjing 31) and densities (27 points/m) in 2017 and 2018233 points/m2) And 3 repetitions are set, each treatment cell field area being 7 x 9m2Local rice is usually raised in the greenhouse in about 4 middle of the month, the transplanting time is 5 middle of the month, the harvesting time is 9 late of the month, nitrogen fertilizer is applied in the transplanting period, the tillering period and the jointing period respectively, wherein the basic fertilizer: and (3) fertilizing the tillers: the ear fertilizer is 4: 3: 3, the application amount of all the phosphate fertilizers and the potash fertilizers is the same, and 50kg of P is completely applied to the phosphate fertilizers before the rice is transplanted2O5ha-1. The application amount of the potash fertilizer is 105kg K2O ha-1Respectively applying 50% of the fertilizer before transplanting and in the jointing stage of the rice.
② remote sensing data acquisition, using unmanned aerial vehicle to carry multispectral camera synchronously, the camera includes green light (central wavelength G550nm, wave band width 40nm), red light (R660 nm, 40nm), red side (RE 735nm, 10nm), near infrared (NIR 790nm, 40nm) four wave bands, and is also provided with RGB sensor, the focal length of camera multispectral sensor lens is 3.98nm, image pixel size is 1280 x 960, the camera is also provided with illumination sensor and calibration white board simultaneously, collecting white board image for radiation calibration before taking off, acquiring radiation illumination information during each shooting, facilitating radiation correction of data during post processing, selecting aerial survey time at about 10: 00-14: 00 noon, requiring clear, windless and cloudy during data acquisition, planning air route and setting flight parameters by flight control software, after flight, deriving image data with high precision position information, copying the image data into splicing software for automatic splicing, acquiring spectral reflectance image file capable of covering whole experimental area, and using high precision GPS corrected point coordinate data to display correct position information.
③ data acquisition of yield comprises selecting 3 samples of 1 square meter with uniform growth at each position during the mature period of rice, cutting off all rice plants in the samples, air drying and threshing, removing impurities and empty and shriveled grains by winnowing, weighing to obtain yield per square meter, measuring water content of grain sample by rapid water content meter, and converting the yield into standard yield of 14% water content.
Yield calculation formula: the yield was expressed in t/ha units, which is the actual yield per meter (g) per meter (1-actual water content)/(1-standard water content 0.14) per meter (0.01).
④ construction and verification of multiple vegetation index combination estimation model based on machine learning algorithm, wherein ten vegetation indexes with better estimation yield, namely NDVI, RVI, DVI, SAVI, GOSAVI, GNDVI, GRVI, OSAVI, NREI, MNDI, NDRE, WDRVI, are selected through multiple tests.
And randomly dividing test data in 2017 and 2018, wherein 70% of data sets are used for model construction, 30% of data sets are used for model verification, regression models between different vegetation indexes and rice yield in different periods are constructed, and a machine learning method is adopted for modeling.
Two very popular machine learning algorithms of Random Forest (RF) and Support Vector Machine (SVM) are selected for modeling analysis, the two algorithms can well solve the problems of classification and prediction of small sample data, ten times of cross validation is adopted, and optimal parameters are searched for the RF and the SVM through network search.
Respectively taking the actual measurement yield of the rice in the jointing stage and the heading stage as dependent variables, taking all vegetation indexes as independent variables, and constructing a remote sensing inversion model of the yield by using a random forest algorithm regression algorithm; then, a remote sensing inversion model with yield is built by using a support vector machine, and the effect of the built model is evaluated and verified by selecting three indexes of a decision coefficient (R2), a Root Mean Square Error (RMSE) and a Relative Error (RE), wherein the closer to 1 the R2 is, the closer to 0 the RMSE and RE are, the better the model effect is, and the more 10% the RE is, the better the model performance is; the model performance is general when the ratio is 20-30%; greater than 30% indicates poor performance. The specific formula is shown as formula (1), formula (2) and formula (3).
Wherein, YiActual measurement of crop yield for the ith sample point (t/hm)2)；EiCrop yield estimate (t/hm) for the ith sample point calculated from the model2)；
Is the average yield (t/hm) of actual measurement2)；
Average yield (t/hm) for model estimation2)。
2. The method according to claim 1, wherein the UAV is an ebee SQ UAV, and the multispectral camera carried by the UAV is a Sequoia multispectral camera.
3. The method for estimating rice yield based on multiple vegetation indexes as claimed in claim 1, wherein the flight control software is eMotion Ag 3.5.0 flight control software, the flight height is set to 106.1 m, the sidewise overlap rate is set to 75%, and the image pixel resolution is 0.1 m.
4. The method for estimating rice yield based on multiple vegetation indexes based on machine learning algorithm according to claim 1, wherein the splicing software is Pix4Dag 3.2.23 splicing software, and the spectral reflectance image file which can cover the whole experimental area and is subjected to radiation correction is obtained by accurate positioning and splicing of GPS.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911311109.6A CN111241912A (en) | 2019-12-18 | 2019-12-18 | Multi-vegetation index rice yield estimation method based on machine learning algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911311109.6A CN111241912A (en) | 2019-12-18 | 2019-12-18 | Multi-vegetation index rice yield estimation method based on machine learning algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111241912A true CN111241912A (en) | 2020-06-05 |
Family
ID=70867638
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911311109.6A Pending CN111241912A (en) | 2019-12-18 | 2019-12-18 | Multi-vegetation index rice yield estimation method based on machine learning algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111241912A (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111798326A (en) * | 2020-06-24 | 2020-10-20 | 安徽大学 | Construction method and application of wheat yield calculation model based on visible light image |
| CN112184703A (en) * | 2020-10-27 | 2021-01-05 | 广东技术师范大学 | Corn ear period unmanned aerial vehicle image alignment method and system based on space-time backtracking |
| CN112345458A (en) * | 2020-10-22 | 2021-02-09 | 南京农业大学 | Wheat yield estimation method based on multispectral image of unmanned aerial vehicle |
| CN112749837A (en) * | 2020-12-22 | 2021-05-04 | 中国农业大学 | Method for estimating grain crop yield based on post-flowering NDVI accumulation value |
| CN112903600A (en) * | 2021-01-15 | 2021-06-04 | 南京农业大学 | Rice nitrogen fertilizer recommendation method based on multispectral image of fixed-wing unmanned aerial vehicle |
| CN113223040A (en) * | 2021-05-17 | 2021-08-06 | 中国农业大学 | Remote sensing-based banana yield estimation method and device, electronic equipment and storage medium |
| CN113505635A (en) * | 2021-05-24 | 2021-10-15 | 中国农业大学 | Method and device for identifying winter wheat and garlic mixed planting area based on optics and radar |
| CN113537645A (en) * | 2021-08-23 | 2021-10-22 | 苏州憨云智能科技有限公司 | Soybean yield prediction method based on machine learning fusion satellite and weather data |
| CN113705937A (en) * | 2021-10-27 | 2021-11-26 | 武汉大学 | Crop yield estimation method combining machine vision and crop model |
| CN113743819A (en) * | 2021-09-15 | 2021-12-03 | 二十一世纪空间技术应用股份有限公司 | Method and device for crop yield estimation, electronic equipment and storage medium |
| CN114254802A (en) * | 2021-11-09 | 2022-03-29 | 北京师范大学 | Prediction method of vegetation coverage space-time change under climate change drive |
| CN114593713A (en) * | 2021-03-02 | 2022-06-07 | 华东师范大学 | Tidal flat vegetation area terrain inversion method and system |
| WO2024019632A1 (en) * | 2022-07-22 | 2024-01-25 | Публичное Акционерное Общество "Сбербанк России" | Device and method for determining crop productivity |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2008079549A (en) * | 2006-09-28 | 2008-04-10 | Mitsubishi Paper Mills Ltd | Method for evaluating tree growth |
| CN109459392A (en) * | 2018-11-06 | 2019-03-12 | 南京农业大学 | A kind of rice the upperground part biomass estimating and measuring method based on unmanned plane multispectral image |
-
2019
- 2019-12-18 CN CN201911311109.6A patent/CN111241912A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2008079549A (en) * | 2006-09-28 | 2008-04-10 | Mitsubishi Paper Mills Ltd | Method for evaluating tree growth |
| CN109459392A (en) * | 2018-11-06 | 2019-03-12 | 南京农业大学 | A kind of rice the upperground part biomass estimating and measuring method based on unmanned plane multispectral image |
Non-Patent Citations (3)
| Title |
|---|
| 朱婉雪等: "基于无人机遥感植被指数优选的田块尺度冬小麦估产", 《农业工程学报》 * |
| 李志铭等: "基于无人机可见光图像的作物分类研究", 《西北农林科技大学学报(自然科学版)》 * |
| 聂红梅等: "基于PCA-SVR的冬小麦土壤水分预测", 《土壤》 * |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111798326A (en) * | 2020-06-24 | 2020-10-20 | 安徽大学 | Construction method and application of wheat yield calculation model based on visible light image |
| CN112345458A (en) * | 2020-10-22 | 2021-02-09 | 南京农业大学 | Wheat yield estimation method based on multispectral image of unmanned aerial vehicle |
| CN112184703A (en) * | 2020-10-27 | 2021-01-05 | 广东技术师范大学 | Corn ear period unmanned aerial vehicle image alignment method and system based on space-time backtracking |
| CN112749837B (en) * | 2020-12-22 | 2023-07-18 | 中国农业大学 | Method for estimating cereal crop yield based on post-flower NDVI accumulation value |
| CN112749837A (en) * | 2020-12-22 | 2021-05-04 | 中国农业大学 | Method for estimating grain crop yield based on post-flowering NDVI accumulation value |
| CN112903600A (en) * | 2021-01-15 | 2021-06-04 | 南京农业大学 | Rice nitrogen fertilizer recommendation method based on multispectral image of fixed-wing unmanned aerial vehicle |
| CN114593713A (en) * | 2021-03-02 | 2022-06-07 | 华东师范大学 | Tidal flat vegetation area terrain inversion method and system |
| CN113223040A (en) * | 2021-05-17 | 2021-08-06 | 中国农业大学 | Remote sensing-based banana yield estimation method and device, electronic equipment and storage medium |
| CN113505635A (en) * | 2021-05-24 | 2021-10-15 | 中国农业大学 | Method and device for identifying winter wheat and garlic mixed planting area based on optics and radar |
| CN113537645A (en) * | 2021-08-23 | 2021-10-22 | 苏州憨云智能科技有限公司 | Soybean yield prediction method based on machine learning fusion satellite and weather data |
| CN113537645B (en) * | 2021-08-23 | 2023-11-24 | 苏州憨云智能科技有限公司 | Soybean yield prediction method based on machine learning fusion satellite and weather data |
| CN113743819A (en) * | 2021-09-15 | 2021-12-03 | 二十一世纪空间技术应用股份有限公司 | Method and device for crop yield estimation, electronic equipment and storage medium |
| CN113743819B (en) * | 2021-09-15 | 2024-03-26 | 二十一世纪空间技术应用股份有限公司 | Crop yield estimation method, device, electronic equipment and storage medium |
| CN113705937A (en) * | 2021-10-27 | 2021-11-26 | 武汉大学 | Crop yield estimation method combining machine vision and crop model |
| CN114254802A (en) * | 2021-11-09 | 2022-03-29 | 北京师范大学 | Prediction method of vegetation coverage space-time change under climate change drive |
| CN114254802B (en) * | 2021-11-09 | 2023-09-08 | 北京师范大学 | Prediction method for vegetation coverage space-time change under climate change drive |
| WO2024019632A1 (en) * | 2022-07-22 | 2024-01-25 | Публичное Акционерное Общество "Сбербанк России" | Device and method for determining crop productivity |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111241912A (en) | Multi-vegetation index rice yield estimation method based on machine learning algorithm | |
| Zheng et al. | Detection of rice phenology through time series analysis of ground-based spectral index data | |
| CN109829234B (en) | A kind of across scale Dynamic High-accuracy crop condition monitoring and yield estimation method based on high-definition remote sensing data and crop modeling | |
| CN106372592B (en) | A kind of winter wheat planting area calculation method based on winter wheat area index | |
| Qiu et al. | Estimation of nitrogen nutrition index in rice from UAV RGB images coupled with machine learning algorithms | |
| CN110222475B (en) | Method for inverting moisture content of winter wheat plants based on multispectral remote sensing of unmanned aerial vehicle | |
| CN110927082A (en) | Winter wheat yield prediction method based on unmanned aerial vehicle imaging hyperspectral remote sensing | |
| CN111368736B (en) | Rice refined estimation method based on SAR and optical remote sensing data | |
| CN113268923B (en) | Summer corn yield estimation method based on simulated multispectral | |
| Shibayama et al. | Estimating paddy rice leaf area index with fixed point continuous observation of near infrared reflectance using a calibrated digital camera | |
| CN114821362B (en) | Multi-source data-based rice planting area extraction method | |
| CN116108318B (en) | Rape nitrogen fertilizer recommended dressing amount calculation method based on unmanned aerial vehicle multispectral image | |
| CN110909933A (en) | Agricultural drought rapid diagnosis and evaluation method coupling crop model and machine learning language | |
| CN112986158B (en) | Beet nitrogen nutrition detection method and system based on unmanned aerial vehicle multispectral data | |
| Yang et al. | A VI-based phenology adaptation approach for rice crop monitoring using UAV multispectral images | |
| CN112816618A (en) | Method for screening nitrogen-efficient wheat varieties | |
| CN115950838A (en) | Summer corn drought unmanned aerial vehicle rapid monitoring and distinguishing method based on chlorophyll content | |
| CN110954650A (en) | Satellite remote sensing monitoring method for cotton canopy nitrogen | |
| CN106203673A (en) | Consider the crops Biomass remote sensing estimation method of water stress | |
| CN116602106A (en) | Unmanned aerial vehicle-based variable fertilization method in paddy field | |
| Bartoszek | Usefulness of MODIS data for assessment of the growth and development of winter oilseed rape. | |
| Hangbin et al. | MODIS data based NDVI Seasonal dynamics in agro-ecosystems of south bank Hangzhouwan bay | |
| CN112504972A (en) | Method for rapidly monitoring nitrogen content of tobacco | |
| CN114202702B (en) | Based on D-fGWinter wheat dynamic harvest index remote sensing estimation method obtained by parameter remote sensing | |
| He et al. | Diagnosis of nitrogen nutrition in sugar beet based on the characteristics of scanned leaf images |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20200605 |
|
| WD01 | Invention patent application deemed withdrawn after publication |