CN115392120A - LSTM-based soil organic matter content high-spectrum modeling method - Google Patents
LSTM-based soil organic matter content high-spectrum modeling method Download PDFInfo
- Publication number
- CN115392120A CN115392120A CN202211011513.3A CN202211011513A CN115392120A CN 115392120 A CN115392120 A CN 115392120A CN 202211011513 A CN202211011513 A CN 202211011513A CN 115392120 A CN115392120 A CN 115392120A
- Authority
- CN
- China
- Prior art keywords
- soil
- model
- organic matter
- data
- samples
- 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
- 239000004016 soil organic matter Substances 0.000 title claims abstract description 38
- 238000000034 method Methods 0.000 title claims abstract description 29
- 238000001228 spectrum Methods 0.000 title claims abstract description 20
- 239000002689 soil Substances 0.000 claims abstract description 58
- 230000003595 spectral effect Effects 0.000 claims abstract description 45
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 21
- 238000010801 machine learning Methods 0.000 claims abstract description 18
- 238000002310 reflectometry Methods 0.000 claims abstract description 18
- 238000013135 deep learning Methods 0.000 claims abstract description 10
- 238000012545 processing Methods 0.000 claims abstract description 9
- 239000005416 organic matter Substances 0.000 claims abstract description 6
- 238000007781 pre-processing Methods 0.000 claims abstract description 6
- 238000009499 grossing Methods 0.000 claims description 11
- KMUONIBRACKNSN-UHFFFAOYSA-N potassium dichromate Chemical compound [K+].[K+].[O-][Cr](=O)(=O)O[Cr]([O-])(=O)=O KMUONIBRACKNSN-UHFFFAOYSA-N 0.000 claims description 6
- 238000012937 correction Methods 0.000 claims description 5
- 238000001035 drying Methods 0.000 claims description 4
- 238000000227 grinding Methods 0.000 claims description 4
- 230000006403 short-term memory Effects 0.000 claims description 4
- 238000012706 support-vector machine Methods 0.000 claims description 4
- 238000012360 testing method Methods 0.000 claims description 4
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 238000013499 data model Methods 0.000 claims description 3
- 238000013501 data transformation Methods 0.000 claims description 3
- 238000010438 heat treatment Methods 0.000 claims description 3
- 230000007787 long-term memory Effects 0.000 claims description 3
- 238000007873 sieving Methods 0.000 claims description 3
- 238000012549 training Methods 0.000 claims description 3
- 238000012795 verification Methods 0.000 claims description 3
- 230000001590 oxidative effect Effects 0.000 claims description 2
- 230000001131 transforming effect Effects 0.000 abstract description 2
- 239000000523 sample Substances 0.000 description 20
- 238000005070 sampling Methods 0.000 description 6
- 238000010238 partial least squares regression Methods 0.000 description 5
- 238000011160 research Methods 0.000 description 5
- 238000005259 measurement Methods 0.000 description 4
- 235000015097 nutrients Nutrition 0.000 description 4
- 238000011161 development Methods 0.000 description 3
- 238000002203 pretreatment Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 239000010410 layer Substances 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 238000012628 principal component regression Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 1
- 241000196324 Embryophyta Species 0.000 description 1
- 244000166124 Eucalyptus globulus Species 0.000 description 1
- 235000008331 Pinus X rigitaeda Nutrition 0.000 description 1
- 235000011613 Pinus brutia Nutrition 0.000 description 1
- 241000018646 Pinus brutia Species 0.000 description 1
- 229910052799 carbon Inorganic materials 0.000 description 1
- 238000004177 carbon cycle Methods 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 239000003337 fertilizer Substances 0.000 description 1
- 230000002068 genetic effect Effects 0.000 description 1
- 239000012535 impurity Substances 0.000 description 1
- 238000002329 infrared spectrum Methods 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 230000014759 maintenance of location Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 230000003647 oxidation Effects 0.000 description 1
- 238000007254 oxidation reaction Methods 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000004856 soil analysis Methods 0.000 description 1
- 239000002344 surface layer Substances 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
- G01N21/31—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry
- G01N21/35—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light
- G01N21/3563—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light for analysing solids; Preparation of samples therefor
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
- G01N21/31—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry
- G01N21/35—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light
- G01N21/359—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light using near infrared light
-
- 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
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/02—Reliability analysis or reliability optimisation; Failure analysis, e.g. worst case scenario performance, failure mode and effects analysis [FMEA]
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Theoretical Computer Science (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Biochemistry (AREA)
- Analytical Chemistry (AREA)
- Development Economics (AREA)
- Educational Administration (AREA)
- Chemical & Material Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Entrepreneurship & Innovation (AREA)
- Evolutionary Computation (AREA)
- Strategic Management (AREA)
- General Health & Medical Sciences (AREA)
- Marketing (AREA)
- Game Theory and Decision Science (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Operations Research (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Quality & Reliability (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The invention discloses an LSTM-based soil organic matter content high spectrum modeling method which is characterized by comprising the following steps of: 1) Collecting and processing a soil sample; 2) Measuring the organic matter content of the soil sample and the spectral reflectance data of the soil sample; 3) Preprocessing soil spectral reflectivity data; 4) Transforming the spectral reflectivity data; 5) Establishing a soil organic matter prediction model by adopting a machine learning algorithm and a deep learning algorithm; 6) Establishing a hyperspectral prediction model; 7) And (6) evaluating the accuracy of the model. The method can predict in real time, quickly and accurately, and has good practical application value.
Description
Technical Field
The invention relates to the technical field of soil spectrum acquisition and analysis, in particular to a soil organic matter content high spectrum modeling method based on a Long Short-Term Memory network (LSTM).
Background
Measurement of Soil Organic Matter (SOM) content is a key procedure in carbon cycle research and forest management. SOM not only provides nutrients required by crops and improves the physical structure of soil, but also is beneficial to water and fertilizer retention, so that the SOM content can be quickly and accurately estimated, and the SOM has important significance for improving the grain yield and helping the sustainable development of agriculture. However, the conventional SOM content estimation method is costly, time-consuming, and labor-intensive, and thus cannot meet the current demand of production management.
Fortunately, the development of hyperspectral remote sensing technology has led to several new methods of soil analysis. The prediction of the content of the soil nutrients has wide research and application bases. With the development of machine learning, many new spectral model regression prediction algorithms are continuously proposed and applied. Due to the complexity of the visible near-infrared spectrum, various methods are applied for the pre-treatment of the soil spectrum, such as the Savitzky-Golay smoothing, normalization and normalization methods. And processing the soil spectrum data by adopting a method of combining SG smoothing and scattering correction so as to reduce irrelevant and useless information in the spectrum to the maximum extent and increase the correlation between the spectrum and the measured value. By selecting the optimal pretreatment method combination to process the soil vis-NIR data, not only can interference factors be eliminated to the maximum extent, but also the complementary relation among the pretreatment methods can be utilized to improve the prediction precision of the network model. In the existing literature, researchers mostly focus on preprocessing of spectral data, and the recommendation and improvement of related spectral regression models are less. The high-performance spectral data modeling method can simplify the preprocessing requirement of spectral data and is also the key for ensuring the accuracy of spectral prediction. As regression prediction develops, more and more linear regression methods are applied to soil nutrient prediction, such as Principal Component Regression (PCR) and Partial Least Squares Regression (PLSR) methods. Then, a random forest, a genetic algorithm, a least squares support vector machine (LS-SVM), and a stereotactic method in machine learning are also used to improve the model prediction capability. However, compared with the traditional mathematical modeling and machine learning methods, the neural network model has higher computational efficiency and stronger modeling capability, and can independently extract effective characteristic structures from complex spectral data for learning.
Disclosure of Invention
The invention aims to provide a soil organic matter content high-spectrum modeling method based on LSTM, aiming at the defects of the prior art. The method can predict in real time, quickly and accurately, and has good practical application value.
The technical scheme for realizing the purpose of the invention is as follows:
an LSTM-based soil organic matter content high spectrum modeling method comprises the following steps:
1) Collecting and processing soil samples: collecting a plurality of samples, naturally drying and grinding the samples, and uniformly dividing the samples into two parts;
2) Measuring the organic matter content of the soil sample and the spectral reflectance data of the soil sample in the step 1): sieving a sample through a 0.2nm soil sieve, oxidizing and heating by using potassium dichromate, determining the SOM content, passing the sample through the 0.149nm soil sieve, and acquiring hyperspectral data by using an ASD field Spec 14 Hi-Res ground object spectrometer, wherein the spectrum range comprises visible light and a near infrared region, namely a region with the wavelength of 350-2500 nm;
3) Preprocessing the soil spectral reflectivity data in the step 2): removing the edge wave band with larger noise in the spectral reflectivity data and smoothing the spectral reflectivity data, wherein the edge wave band with larger noise is 350-399nm and 2401-2500nm, and the smoothing is Savitzky-Golay smoothing on the spectral reflectivity data;
4) Spectral reflectance data transformation: carrying out conversion processing of first order differential, second order differential, standard normal variable and multivariate scattering correction on the spectral reflectivity;
5) The method comprises the steps of establishing a soil organic matter prediction model by adopting a machine learning algorithm and a deep learning algorithm, wherein the machine learning algorithm is partial least square regression and a support vector machine, the deep learning algorithm is a long-term and short-term memory network, 4/5 of the total number of samples is used as a training set, 1/5 of the total number of samples is used as a verification set, an LSTM model is established by adopting a keras library in Pychar software and adopting Python3.8 language, a PLSR model and an SVM model are realized by calling corresponding machine learning modules in a skearn interface, establishing an inversion model between R, 1DR, 2DR, SVN and MSC five kinds of spectral reflectance data and soil organic matter content, and performing data model fitting by adopting Origin 2021 in an initial model test;
6) Establishing a hyperspectral prediction model: selecting an optimal model in the step 5);
7) And (3) evaluating the model precision: evaluating the soil organic matter prediction model established in the step 5) by adopting the decision coefficient and the root mean square error, and determining the precision, the stability and the prediction performance of the soil organic matter prediction model.
Compared with the prior art, the technical scheme has the advantages and beneficial effects that:
compared with the machine learning method for modeling, the technical scheme can simplify the modeling process, solve the problems of multiple collinearity among all wave bands and the like, has the advantages of small workload, low cost, high precision, high reliability and the like, has high generalization of the model, and is suitable for different soils in different regions;
the technical scheme fills the blank of research of a deep learning algorithm on a soil organic matter spectrum inversion model, and provides the machine learning method which is simple in model, high in operation efficiency, analysis capability and accuracy and better in prediction performance.
By collecting and processing soil samples, measuring soil organic matters and spectral reflectance data, transforming spectral reflectance data, establishing a machine learning algorithm and a deep learning algorithm model, establishing a hyperspectral prediction model and evaluating indexes, compared with a machine learning method, the technical scheme can predict in real time, quickly and accurately and has good practical application value.
The method can be widely applied to engineering practice, and provides a basis for the treatment and utilization of subsequent land resources.
Drawings
FIG. 1 is a schematic flow chart of an exemplary method;
FIG. 2 is a schematic diagram of a sample point of a study area in an embodiment;
FIG. 3 is a graph illustrating an average spectrum of soil reflectivity for different organic content in examples;
fig. 4 is a comparison graph of the real value and the predicted value of the soil organic matter on the test set in the example.
Detailed Description
The invention will be further described with reference to the following drawings and examples, but the invention is not limited thereto.
Example (b):
in the example, the soil samples are from Guangxi China Huangguan forest field (109 degrees 46 ' E, 24 degrees 37' N) and Guangxi China Yazhi forest field (106 degrees 16'30 ' E, 24 degrees 49' N) which belong to subtropical climate, the annual average rainfall is 1750 and 1057mm, the annual average temperature is 19 ℃ and 16.8 ℃, the soil type of the local area is red soil, the main plants are eucalyptus, fir and pine, due to lack of effective soil nutrient detection and artificial unreasonable planting, the soil degradation of the forest land is increased, the trees in the research area grow uniformly, samples of 0-20cm soil layer are collected by an S-shaped sampling method, 206 samples are collected altogether, and sampling points in the research area are shown in FIG. 2;
referring to fig. 1, the LSTM-based soil organic matter content high spectrum modeling method comprises the following steps:
1) Collecting and processing soil samples: collecting a plurality of samples, naturally drying and grinding the samples, uniformly dividing the samples into two parts, in the embodiment, selecting a forest land with uniform growth and less artificial movement during field sampling, randomly sampling by using W-shaped multiple points, ensuring more than 5 sampling points at each sampling point, collecting soil with the depth of 0-20cm of soil layer away from the surface layer of the ground, uniformly mixing the collected soil samples, taking 1kg of soil sample by a quartering method, naturally drying the soil sample in a laboratory, removing impurities, grinding the soil sample, and screening the ground soil sample by a 0.149mm hole sieve to obtain a soil sample to be detected;
2) Measuring the organic matter content of the soil sample and the spectral reflectance data of the soil sample in the step 1): sieving a sample through a 0.2nm soil sieve, performing oxidation heating by using potassium dichromate, determining the SOM content, passing the sample through the 0.149nm soil sieve, and acquiring hyperspectral data by using an ASD field Spec1 Hi-Res ground object spectrometer, wherein the spectral range comprises a visible light region and a near infrared region, namely a region with the wavelength of 350-2500nm, in the example, the organic carbon content in the soil sample to be determined is determined by using a burning method, the burning temperature is 550 ℃, the spectral reflectance data is determined by using an ASD F4 ground object reflectance spectrometer to determine the spectral reflectance of the soil sample, the spectral resolution is 1nm, and the spectral determination range is 350-2500nm, the field angle of a probe is 12-15 degrees, the incident angle of a light source is 30-45 degrees, the burned soil sample is placed in a tray with the depth of 1.5cm and the diameter of 6cm, the soil surface is flattened, the distance between the probe and the soil surface is 5cm, in order to eliminate the influence of inconsistent external light, a white reference plate is used for calibrating an instrument before measurement, the air background value is deducted, in order to increase the signal-to-noise ratio of the instrument, the repeated scanning times of each measurement are set to be 30 times, the average value of the spectral reflectance data of each measurement is used as the spectral reflectance data of each soil sample, and the average spectral curve of the soil reflectance with different organic matter contents is shown in figure 3;
3) Preprocessing the soil spectral reflectivity data in the step 2): removing edge wave bands with larger noise in the spectral reflectivity data and smoothing the spectral reflectivity data, wherein the edge wave bands with larger noise are 350-399nm wave bands and 2401-2500nm wave bands, and the smoothing is Savitzky-Golay smoothing on the spectral reflectivity data;
4) Spectral reflectance data transformation: carrying out conversion processing of first order differential, second order differential, standard normal variable and multivariate scattering correction on the spectral reflectivity;
5) A soil organic matter prediction model is established by adopting a machine learning algorithm and a deep learning algorithm, wherein the machine learning algorithm is partial least square regression and a support vector machine, the deep learning algorithm is a long-term and short-term memory network, 4/5 of the total number of samples is used as a training set, 1/5 of the total number of samples is used as a verification set, the LSTM model is established by adopting a keras library in PyCharm software and adopting Python3.8 language, the PLSR and SVM models are realized by calling corresponding machine learning modules in a skearn interface, an inversion model between R, 1DR, 2DR, SVN and MSC five kinds of spectral reflectance data and soil organic matter content is established, the initial model test is data model fitting by adopting Origin 2021, and in the embodiment, the quasi-class of the initial model is a quasi-least square regression model and support vector machineCoefficient of determination R for the sum effect 2 The stability of the initial model is checked by the root mean square error RMSE, R 2 The closer to 1, the better the model fitting effect, the smaller the RMSE, the better the stability of the model, and the statistical characteristics of the soil organic matter content are shown in table 1:
table 1: statistical characterization of soil organic matter content
6) The model is established by adopting three models of LSTM, PLSR and SVM, namely R and two transformation forms of 1DR, 2DR, SNV and MSC full wave bands, and is selected from R 2 And from the three evaluation indexes of RMSE, the inversion model established by the SATCN is superior to the TCN model, and the full-wave band R-LSTM, R-PLSR and R-SVM model have the precision R 2 Respectively 0.771, 0.461 and 0.633 which are obviously lower than models established by other spectrum indexes, and show that first-order differential, second-order differential, standard normal variable and multivariate scattering correction transformation can all obviously eliminate redundant information in soil, the sensitivity of hyperspectrum to soil organic matters is enhanced, and REME is respectively 5.477 g.kg -1 、8.393 g·kg -1 And 6.922 g.kg -1 The method is obviously higher than models established by other spectral indexes, which shows that the original spectrum modeling stability is poor, and the accuracy R of the 1DR-LSTM, 1DR-PLSR and 1DR-SVM models 2 0.909, 0.736 and 0.869 respectively, the REME is 3.862 g.kg -1 、6.601 g·kg -1 And 4.646 g.kg -1 The accuracy R of the soil organic matter inversion model, 1DR-LSTM, 1DR-PLSR and 1DR-SVM models, which are all superior to the soil organic matter inversion model established by 2DR, SNV and MSC 2 Are all above 0.7, which shows that the model can well predict the organic matter content of the soil, and the 1DR-LSTM model R 2 0.909,RMSE of 3.862 g.kg -1 The model is an optimal model; the accuracy of the soil organic matter estimation model of different models is shown in table 2:
table 2: model accuracy estimation method for soil organic matters of different models
7) Establishing a hyperspectral prediction model: selecting an optimal model in the step 5), in the embodiment, establishing a model according to a machine learning algorithm and a deep learning algorithm, selecting the optimal model, and comparing the real value and the predicted value of the soil organic matter by adopting origin 2020: selecting a 1DR-SATCN model with highest precision from the soil organic matter content spectrum inversion models established in all bands, verifying the model, and predicting the result based on the FDR-OPLS all band model as shown in figure 4, wherein R is 2 The accuracy is high, the stability is good, and the practical application value is good;
8) And (3) evaluating the model precision: evaluating the soil organic matter prediction model established in the step 5) by adopting the decision coefficient and the root mean square error, and determining the precision, the stability and the prediction performance of the soil organic matter prediction model.
Claims (1)
1. An LSTM-based soil organic matter content high spectrum modeling method is characterized by comprising the following steps:
1) Collecting and processing soil samples: collecting a plurality of samples, naturally drying and grinding the samples, and uniformly dividing the samples into two parts;
2) Measuring the organic matter content of the soil sample and the spectral reflectance data of the soil sample in the step 1): sieving a sample through a 0.2nm soil sieve, oxidizing and heating by using potassium dichromate, determining the SOM content, passing the sample through the 0.149nm soil sieve, and acquiring hyperspectral data by using an ASD field Spec 14 Hi-Res ground object spectrometer, wherein the spectrum range comprises visible light and a near infrared region, namely a region with the wavelength of 350-2500 nm;
3) Preprocessing the soil spectral reflectivity data in the step 2): removing the edge wave band with larger noise in the spectral reflectivity data and smoothing the spectral reflectivity data, wherein the edge wave band with larger noise is 350-399nm and 2401-2500nm, and the smoothing is Savitzky-Golay smoothing on the spectral reflectivity data;
4) Spectral reflectance data transformation: carrying out conversion processing of first order differential, second order differential, standard normal variable and multivariate scattering correction on the spectral reflectivity;
5) The method comprises the steps of establishing a soil organic matter prediction model by adopting a machine learning algorithm and a deep learning algorithm, wherein the machine learning algorithm is partial least square regression and a support vector machine, the deep learning algorithm is a long-term and short-term memory network, 4/5 of the total number of samples is used as a training set, 1/5 of the total number of samples is used as a verification set, an LSTM model is established by adopting a keras library in Pychar software and adopting Python3.8 language, a PLSR model and an SVM model are realized by calling corresponding machine learning modules in a skearn interface, establishing an inversion model between R, 1DR, 2DR, SVN and MSC five kinds of spectral reflectance data and soil organic matter content, and performing data model fitting by adopting Origin 2021 in an initial model test;
6) Establishing a hyperspectral prediction model: selecting an optimal model in the step 5);
7) And (3) evaluating the model precision: evaluating the soil organic matter prediction model established in the step 5) by adopting the decision coefficient and the root mean square error, and determining the precision, the stability and the prediction performance of the soil organic matter prediction model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211011513.3A CN115392120A (en) | 2022-08-23 | 2022-08-23 | LSTM-based soil organic matter content high-spectrum modeling method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211011513.3A CN115392120A (en) | 2022-08-23 | 2022-08-23 | LSTM-based soil organic matter content high-spectrum modeling method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115392120A true CN115392120A (en) | 2022-11-25 |
Family
ID=84119785
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211011513.3A Pending CN115392120A (en) | 2022-08-23 | 2022-08-23 | LSTM-based soil organic matter content high-spectrum modeling method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115392120A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116818687A (en) * | 2023-06-21 | 2023-09-29 | 浙江大学杭州国际科创中心 | Soil organic carbon spectrum prediction method and device based on spectrum guide integrated learning |
-
2022
- 2022-08-23 CN CN202211011513.3A patent/CN115392120A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116818687A (en) * | 2023-06-21 | 2023-09-29 | 浙江大学杭州国际科创中心 | Soil organic carbon spectrum prediction method and device based on spectrum guide integrated learning |
| CN116818687B (en) * | 2023-06-21 | 2024-02-20 | 浙江大学杭州国际科创中心 | Soil organic carbon spectrum prediction method and device based on spectrum guide integrated learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Banerjee et al. | High-throughput phenotyping using digital and hyperspectral imaging-derived biomarkers for genotypic nitrogen response | |
| CN111855589A (en) | Remote sensing inversion model and method for rice leaf nitrogen accumulation | |
| CN102305772A (en) | Method for screening characteristic wavelength of near infrared spectrum features based on heredity kernel partial least square method | |
| CN112485204A (en) | Hyperspectrum-based rice panicle nitrogen nutrition monitoring and diagnosis method and application | |
| CN111855590A (en) | Remote sensing inversion model and method for rice leaf starch accumulation | |
| CN110687053B (en) | Regional organic matter content estimation method and device based on hyperspectral image | |
| CN111829965A (en) | Remote sensing inversion model and method for starch accumulation amount of rice overground part | |
| CN111855591A (en) | Rice overground part carbon-nitrogen ratio remote sensing inversion model and method | |
| CN114460013B (en) | Coastal wetland vegetation overground biomass GAN model self-learning remote sensing inversion method | |
| CN111855593A (en) | Remote sensing inversion model and method for starch content of rice leaf | |
| CN112270131A (en) | Remote sensing inversion model and method for rice leaf area index based on ARD regression algorithm | |
| CN110596028A (en) | High-spectrum inversion method for content of deposited rare earth La element | |
| CN113514410A (en) | Real-time quantitative monitoring method for vertical distribution of nitrogen utilization rate of summer corn in whole growth period based on canopy hyperspectral technology | |
| CN113436153A (en) | Method for predicting carbon components of undisturbed soil profile based on hyperspectral imaging and support vector machine technology | |
| CN110569605A (en) | Non-glutinous rice leaf nitrogen content inversion model method based on NSGA2-ELM | |
| CN115392120A (en) | LSTM-based soil organic matter content high-spectrum modeling method | |
| CN111855592A (en) | Remote sensing inversion model and method for upper dry matter weight in unit area of rice | |
| CN112129709A (en) | Apple tree canopy scale nitrogen content diagnosis method | |
| CN113971989A (en) | Forest soil organic carbon content high-spectrum modeling method based on OPLS | |
| CN110779875A (en) | Method for detecting moisture content of winter wheat ear based on hyperspectral technology | |
| CN113049499A (en) | Indirect remote sensing inversion method for water total nitrogen concentration, storage medium and terminal equipment | |
| CN115392305A (en) | Soil organic matter content high spectrum modeling method based on improved time convolution network | |
| CN111595806A (en) | Method for monitoring soil carbon component by using mid-infrared diffuse reflection spectrum | |
| CN114878508A (en) | Hyperspectral imaging-based method for detecting surface water content of cultural relic in earthen site | |
| CN111521568A (en) | Soil water content prediction method based on spectrum angle |
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 |