CN115015132A - Winter wheat moisture condition monitoring method based on data fusion - Google Patents
Winter wheat moisture condition monitoring method based on data fusion Download PDFInfo
- Publication number
- CN115015132A CN115015132A CN202210616907.5A CN202210616907A CN115015132A CN 115015132 A CN115015132 A CN 115015132A CN 202210616907 A CN202210616907 A CN 202210616907A CN 115015132 A CN115015132 A CN 115015132A
- Authority
- CN
- China
- Prior art keywords
- data
- canopy
- plant
- water content
- spectral
- 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
- 241000209140 Triticum Species 0.000 title claims abstract description 78
- 238000012544 monitoring process Methods 0.000 title claims abstract description 66
- 235000021307 Triticum Nutrition 0.000 title claims abstract description 54
- 238000000034 method Methods 0.000 title claims abstract description 37
- 230000004927 fusion Effects 0.000 title claims abstract description 26
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims abstract description 128
- 241000196324 Embryophyta Species 0.000 claims abstract description 101
- 230000003595 spectral effect Effects 0.000 claims abstract description 97
- 239000002689 soil Substances 0.000 claims abstract description 79
- 238000012216 screening Methods 0.000 claims abstract description 39
- 238000007637 random forest analysis Methods 0.000 claims abstract description 24
- 238000002310 reflectometry Methods 0.000 claims abstract description 24
- 238000000513 principal component analysis Methods 0.000 claims abstract description 21
- 238000012706 support-vector machine Methods 0.000 claims abstract description 19
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 18
- 238000003066 decision tree Methods 0.000 claims abstract description 15
- 230000000694 effects Effects 0.000 claims abstract description 7
- IJGRMHOSHXDMSA-UHFFFAOYSA-N Atomic nitrogen Chemical compound N#N IJGRMHOSHXDMSA-UHFFFAOYSA-N 0.000 claims description 112
- 229910052757 nitrogen Inorganic materials 0.000 claims description 56
- 238000009825 accumulation Methods 0.000 claims description 27
- 238000005070 sampling Methods 0.000 claims description 23
- 238000012360 testing method Methods 0.000 claims description 23
- 229910052698 phosphorus Inorganic materials 0.000 claims description 22
- 239000000523 sample Substances 0.000 claims description 12
- 238000005303 weighing Methods 0.000 claims description 8
- 210000005069 ears Anatomy 0.000 claims description 7
- 238000005259 measurement Methods 0.000 claims description 7
- 238000001035 drying Methods 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 6
- 230000001186 cumulative effect Effects 0.000 claims description 4
- 238000012549 training Methods 0.000 claims description 4
- 238000011282 treatment Methods 0.000 claims description 4
- 102000004190 Enzymes Human genes 0.000 claims description 3
- 108090000790 Enzymes Proteins 0.000 claims description 3
- 238000007696 Kjeldahl method Methods 0.000 claims description 3
- 238000010276 construction Methods 0.000 claims description 3
- 238000007405 data analysis Methods 0.000 claims description 3
- 230000001066 destructive effect Effects 0.000 claims description 3
- 238000005520 cutting process Methods 0.000 claims description 2
- 101150102320 SWC3 gene Proteins 0.000 claims 3
- 238000000354 decomposition reaction Methods 0.000 claims 1
- 210000000056 organ Anatomy 0.000 description 8
- 238000003973 irrigation Methods 0.000 description 7
- 230000002262 irrigation Effects 0.000 description 7
- XAGFODPZIPBFFR-UHFFFAOYSA-N aluminium Chemical compound [Al] XAGFODPZIPBFFR-UHFFFAOYSA-N 0.000 description 4
- 229910052782 aluminium Inorganic materials 0.000 description 4
- 238000011160 research Methods 0.000 description 4
- 238000010801 machine learning Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 238000007789 sealing Methods 0.000 description 3
- 238000010187 selection method Methods 0.000 description 3
- 238000001228 spectrum Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 235000013339 cereals Nutrition 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 239000003337 fertilizer Substances 0.000 description 2
- 238000009331 sowing Methods 0.000 description 2
- 230000005068 transpiration Effects 0.000 description 2
- XSQUKJJJFZCRTK-UHFFFAOYSA-N Urea Chemical compound NC(N)=O XSQUKJJJFZCRTK-UHFFFAOYSA-N 0.000 description 1
- 238000004847 absorption spectroscopy Methods 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 238000012271 agricultural production Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 229910000389 calcium phosphate Inorganic materials 0.000 description 1
- 239000001506 calcium phosphate Substances 0.000 description 1
- 235000011010 calcium phosphates Nutrition 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 239000004202 carbamide Substances 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000002790 cross-validation Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000006735 deficit Effects 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005553 drilling Methods 0.000 description 1
- 230000008641 drought stress Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000001704 evaporation Methods 0.000 description 1
- 230000008020 evaporation Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000000618 nitrogen fertilizer Substances 0.000 description 1
- 230000005868 ontogenesis Effects 0.000 description 1
- 230000029553 photosynthesis Effects 0.000 description 1
- 238000010672 photosynthesis Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- QORWJWZARLRLPR-UHFFFAOYSA-H tricalcium bis(phosphate) Chemical compound [Ca+2].[Ca+2].[Ca+2].[O-]P([O-])([O-])=O.[O-]P([O-])([O-])=O QORWJWZARLRLPR-UHFFFAOYSA-H 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Images
Classifications
-
- 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
- G01N21/27—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands using photo-electric detection ; circuits for computing concentration
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
- G01B21/00—Measuring arrangements or details thereof, where the measuring technique is not covered by the other groups of this subclass, unspecified or not relevant
- G01B21/28—Measuring arrangements or details thereof, where the measuring technique is not covered by the other groups of this subclass, unspecified or not relevant for measuring areas
-
- 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
- G01N21/55—Specular reflectivity
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
- G01N33/0098—Plants or trees
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N5/00—Analysing materials by weighing, e.g. weighing small particles separated from a gas or liquid
- G01N5/04—Analysing materials by weighing, e.g. weighing small particles separated from a gas or liquid by removing a component, e.g. by evaporation, and weighing the remainder
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N5/00—Analysing materials by weighing, e.g. weighing small particles separated from a gas or liquid
- G01N5/04—Analysing materials by weighing, e.g. weighing small particles separated from a gas or liquid by removing a component, e.g. by evaporation, and weighing the remainder
- G01N5/045—Analysing materials by weighing, e.g. weighing small particles separated from a gas or liquid by removing a component, e.g. by evaporation, and weighing the remainder for determining moisture content
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01W—METEOROLOGY
- G01W1/00—Meteorology
- G01W1/02—Instruments for indicating weather conditions by measuring two or more variables, e.g. humidity, pressure, temperature, cloud cover or wind speed
- G01W1/04—Instruments for indicating weather conditions by measuring two or more variables, e.g. humidity, pressure, temperature, cloud cover or wind speed giving only separate indications of the variables measured
-
- 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
- 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
- G01N21/55—Specular reflectivity
- G01N2021/557—Detecting specular reflective parts on sample
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A20/00—Water conservation; Efficient water supply; Efficient water use
- Y02A20/152—Water filtration
Landscapes
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Chemical & Material Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Pathology (AREA)
- Immunology (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- Theoretical Computer Science (AREA)
- Mathematical Physics (AREA)
- Environmental & Geological Engineering (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Biodiversity & Conservation Biology (AREA)
- Data Mining & Analysis (AREA)
- Atmospheric Sciences (AREA)
- Medical Informatics (AREA)
- Ecology (AREA)
- Environmental Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Botany (AREA)
- Wood Science & Technology (AREA)
- Food Science & Technology (AREA)
- Medicinal Chemistry (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a method for establishing a winter wheat moisture condition monitoring model based on data fusion, which comprises the following steps: acquiring the spectral reflectivity of a wheat plant canopy; measuring the moisture index, and acquiring plant data, soil data and meteorological data; respectively screening spectral indexes with the best correlation to the water content of the canopy, the water content of the plant and the equivalent water thickness of the canopy; performing characteristic selection on soil data, plant data and meteorological data by using a principal component analysis and Pearson correlation coefficient method; fusing different databases of the screened spectral indexes as model input parameters, and establishing a winter wheat moisture condition monitoring model based on data fusion by respectively utilizing decision trees DT, random forest RF and support vector machine SVM algorithm in R language codes; and comparing the prediction effects of different winter wheat moisture condition monitoring models on moisture indexes, and screening out the optimal moisture condition monitoring model. The invention also discloses a winter wheat moisture condition monitoring method based on data fusion.
Description
Technical Field
The invention belongs to the technical field of crop moisture condition monitoring, and relates to a winter wheat moisture condition monitoring method based on multi-source data fusion.
Background
In the past decades, climate change and water resource shortage seriously affect agricultural sustainable development, and the grain safety is critical by improving the utilization efficiency of agricultural water. Wheat is one of the main grain crops in the world, water is one of the main factors influencing the physiological growth, photosynthesis and yield of winter wheat in the growth and development process of the winter wheat, and timely monitoring and managing water information is the key for ensuring high quality and high yield of the wheat [1] . Water deficit can lead to winter wheat yield and qualityThe water utilization efficiency is reduced due to over irrigation, and the waste of water resources is caused [2-3] . Therefore, timely acquiring the moisture information of the winter wheat is very critical to agricultural production and field moisture management.
The change of the moisture of crops is a very complicated process and can be influenced by various aspects such as soil environment, meteorological factors, plant physiological action and the like. Currently, most studies diagnose and direct field moisture management based only on monitoring of a certain aspect of the monitoring index. Soil moisture content is one of the most common indicators for monitoring soil moisture status and making moisture management decisions [4] The growth and moisture status of crops are influenced by the water content of the soil and the temperature of the soil [5] . The climate factors such as temperature, evapotranspiration, wind speed and relative humidity are the main factors influencing soil evaporation and crop transpiration [6] . The leaf area and the stomatal conductance can directly influence the strength of plant transpiration [7] 。
The remote sensing technology is widely used for monitoring the moisture condition of crops due to the characteristics of rapidness and no damage [8] . Research shows that the spectral index can be effectively used for estimating moisture indexes such as canopy water content, plant water content, canopy equivalent water thickness and the like [9-10] . The remote sensing technology can be used for timely and quickly acquiring large-range information, and is helpful for guiding irrigation management decision and evaluating drought stress degree. However, the complex canopy structure and the spectral indexes constructed at different growth stages in the crop ontogenesis have different sensitivities to the crop moisture evaluation, so that the accuracy of the monitoring method only based on the spectral indexes in the moisture condition diagnosis is limited to a certain extent [11] 。
Therefore, the influence of various factors on the moisture condition of the winter wheat is comprehensively considered, and the construction and use method of the winter wheat moisture condition monitoring model based on data fusion and the related technical important are researched and obtained, so that the method has very important scientific research value and practical significance.
Reference documents:
[1] qiume, Zhao Long, Mao Peng Jun, Zhang Hai Feng, in Ruifeng, the study on the influence of different water shortage on the physiological indexes of crops [ J ] Chinese agricultural chemistry report 2016,37(4):260 plus 263.
[2]Bicego B,Sapkota A,Torrion JA.Differential nitrogen and water impacts on yield and quality of wheat classes[J].Agronomy Journal,2019,111(6):2792-2803.
[3]Gong X,Zhang H,Ren C,Sun D,Yang J.Optimization allocation of irrigation water resources based on crop water requirement under considering effective precipitation and uncertainty[J].Agricultural Water Management,2020,239,106264.
[4]Luo L,Yu Z,Wang D,Zhang Y,Shi Y.Effects of plant density and soil moisture on photosynthetic characteristics of flag leaf and accumulation and distribution of dry matter in wheat[J].ActaAgronomicaSinica,2011,37(6):1049-1059.
[5]Reth S,Reichstein M,Falge E.The effect of soil water content,soil temperature,soil pH-value and the root mass on soil CO 2 efflux–A modified model[J].Plant and Soil,2005,268:21-33.
[6]Sawan Z M,Hanna L I,McCuistion W L,Foote R J.Egyptian cotton(Gossypiumbarbadense)flower and boll production as affected by climatic factors and soil moisture status[J].Theoretical and Applied Climatology,2010,99:217-227.
[7]Liu E,Mei X,Yan C,Gong D,Zhang Y.Effects of water stress on photosynthetic characteristics,dry matter translocation and WUE in two winter wheat genotypes[J].Agricultural Water Management,2016,167:75-85.
[8]Zhou J,Zhang Y,Han Z,Liu X,Jian Y,Hu C,Dian Y.Evaluating the performance of hyperspectral leaf reflectance to detect water stress and estimation of photosynthetic capacities[J].Remote Sensing,2021,13,2160.
[9]Becker E,Schmidhalter U.Evaluation of yield and drought using active and passive spectral sensing systems at the reproductive stage in wheat[J].Frontiers in Plant Science,2017,8,379.
[10]Winterhalter L,Mistele B,Jampatong S,Schmidhalter U.High throughput phenotyping of canopy water mass and canopy temperature in well-watered and drought stressed tropical maize hybrids in the vegetative stage[J].European Journal of Agronomy,2011,35(1):22-32.
[11] Sun-Dry, Wen Ji, Sun forest, Wang \28156, Zhou Long Fei, Yang Gui Jun, Li Wei Guo, Bao Mei Yan the spectral response of the water content of the canopy of winter wheat under different irrigation conditions [ J ] Chinese agricultural science, 2019,52(14): 2425-.
Disclosure of Invention
Aiming at the problems in the prior art, the invention aims to provide a method for establishing a winter wheat moisture condition monitoring model based on data fusion, which is characterized in that winter wheat soil, plants and meteorological data are linked, a model is established by using three machine learning algorithms of a decision tree, a random forest and a support vector machine, the canopy water content, the plant water content and the canopy equivalent water thickness of winter wheat are estimated, the winter wheat moisture condition monitoring is realized, and guidance is provided for the accurate irrigation management of winter wheat.
The invention adopts the following technical scheme:
a method for establishing a winter wheat moisture condition monitoring model based on data fusion comprises the following steps: the spectral index with the best correlation to the moisture index is screened by constructing the spectral index, soil data, plant data and meteorological data are fused, a winter wheat moisture condition monitoring model is established by utilizing different feature selection methods and a machine learning algorithm, and the model with the best prediction effect is screened out. The method comprises the following steps:
step (1), data acquisition: acquiring the spectral reflectivity of a wheat plant canopy; sampling the plants, and determining the moisture index: canopy Water Content (CWC), Plant Water Content (PWC), Canopy Equivalent Water Thickness (CEWT), obtain plant data (P): leaf Area Index (LAI), dry leaf weight (LDM), dry aerial part weight (ADM), Leaf Nitrogen Concentration (LNC), Plant Nitrogen Concentration (PNC), leaf nitrogen accumulation amount (LNA), and plant nitrogen accumulation amount (PNA); acquiring soil data (S): carrying out layered sampling on 0-60cm of soil to respectively obtain the soil water content of 0-20cm (SWC1), the soil water content of 20-40cm (SWC2) and the soil water content of 40-60cm (SWC 3); obtaining winter wheat in the growth periodMeteorological data (C): maximum temperature (T) max ) Lowest temperature (T) min ) Average temperature (T) avg ) Wind Speed (WS), Evapotranspiration (ET) and Relative Humidity (RH);
step (2), establishing a spectral index of any two-waveband combination based on the spectral reflectivity of the wheat plant canopy: normalizing the spectral index (NDSI), the Ratio Spectral Index (RSI) and the Difference Spectral Index (DSI), respectively establishing a linear model with the Canopy Water Content (CWC), the Plant Water Content (PWC) and the Canopy Equivalent Water Thickness (CEWT), and determining the coefficient (R) of the linear model 2 ) And Root Mean Square Error (RMSE), respectively screening the spectral indices having the best correlation to Canopy Water Cut (CWC), Plant Water Cut (PWC) and Canopy Equivalent Water Thickness (CEWT);
step (3), performing feature selection on the variables of the obtained soil data, plant data and meteorological data by using two methods, namely Principal Component Analysis (PCA) and Pearson Correlation Coefficient (PCC), and establishing 7 databases (S, P, C, S + P, S + C, P + C, S + P + C) consisting of feature variables on the basis of the feature variables in the screened soil data, plant data and meteorological data;
step (4), different databases selected in the step (2) and established in the spectral index fusion step (3) are used as model input parameters, and a winter wheat moisture condition monitoring model established based on data fusion is established by respectively utilizing Decision Trees (DT), Random Forests (RF) and Support Vector Machine (SVM) algorithms in R language codes;
and (5) comparing the prediction effects of the different winter wheat moisture condition monitoring models constructed in the step (4) on moisture indexes, screening out that the optimal monitoring model for the water content CWC of the canopy is RF-PCC- (S + P + C), namely the random forest-Pearson correlation coefficient- (soil data + plant data + meteorological data), the optimal monitoring model for the water content PWC of the plant is RF-PCC- (S + P + C), and the optimal monitoring model for the equivalent water thickness CEWT of the canopy is SVM-PCC- (S + P + C), namely the support vector machine-Pearson correlation coefficient- (soil data + plant data + meteorological data).
In the step (1), obtaining the spectral reflectance of the wheat plant canopy: the method comprises the steps of using an ASD (automatic absorption spectroscopy) hyperspectral radiometer to carry out measurement in clear, cloudless and windless conditions, measuring the time of 10:00-14:00, correcting by using a white board before measurement, measuring the wave band range of 350-2500 nm, measuring the height of a sensor probe vertically downward to a wheat canopy, measuring the field angle of the spectroradiometer to be 25 degrees and the diameter of a ground field range to be 0.44m, measuring 3 sampling points in each sampling cell, measuring each sampling point for 5 times (totaling 15 spectrums), and taking the average value as the spectral reflectivity of the wheat plant canopy in a processing cell.
The ASD hyperspectral radiometer can specifically adopt a FieldSpec FR Pro hyperspectral radiometer manufactured by Analytical Spectral Device (ASD) company in the United states.
The test subjects should be dark colored garments.
Destructive sampling of wheat plant sample is carried out while testing the spectral reflectance of the wheat plant canopy, wheat with uniform growth is randomly selected in a treatment cell, and 0.15m is framed 2 Cutting off roots, separating the sample of the overground part into four leaves, the remaining leaves, the stem sheaths and the ear parts at the top of the canopy, weighing to obtain the fresh weights of the four leaves, the remaining leaves, the stem sheaths and the ears at the top of the canopy, and calculating to obtain the fresh weight of the plant; measuring the area of the leaf by using an LI-3000C portable leaf area meter, and calculating the leaf area index by using a specific leaf weight method; deactivating enzyme for 30 minutes at 105 ℃, drying to constant weight at 80 ℃, weighing to obtain the dry weight of four leaves at the top of the canopy, the dry weight of the rest leaves, the dry weight of the stem sheath and the dry weight of the spike, and calculating to obtain the dry weight of the plant; respectively calculating the water content of the canopy, the water content of the plant and the equivalent water thickness of the canopy according to the fresh weight, the dry weight and the leaf area index of the sample; and (3) determining the total nitrogen content of the dried and weighed plant sample by a Kjeldahl method to obtain the nitrogen concentration of the leaves, the stem sheaths and the ears, and calculating the leaf nitrogen accumulation amount and the plant nitrogen accumulation amount by multiplying the dry weight by the nitrogen content.
Canopy water content (%) - (fresh weight of top four leaves-dry weight of top four leaves)/fresh weight of top four leaves × 100
Plant water content (%) - (fresh weight of plant-dry weight of plant)/fresh weight of plant × 100
Leaf equivalent water thickness (cm) ═ leaf fresh weight-leaf dry weight)/(leaf area × water density)
Canopy equivalent water thickness (cm) is equal to blade equivalent water thickness multiplied by blade area index
Accumulation of Nitrogen in organs (kg/hm) 2 ) Organ nitrogen content x dry organ weight
Plant nitrogen accumulation (kg/hm) 2 ) The sum of the accumulated nitrogen in each organ
Plant nitrogen concentration (%). plant nitrogen accumulation/plant dry matter weight x 100
And respectively collecting 0-20cm, 20-40cm and 40-60cm soil samples between any side of the middle point of the wheat plant sampling row by using a 1m soil drill, drying the samples in an oven at 105 ℃ until the weight is constant, and calculating to obtain the water content of the soil with different soil layers of 0-20cm, 20-40cm and 40-60 cm.
Soil water content (%) - (fresh soil weight-dried soil weight)/dried soil weight × 100
Downloading data of daily maximum temperature, minimum temperature, average temperature, wind speed, evapotranspiration and relative humidity in a winter wheat growth period at a local meteorological site, taking data within 10 days before each plant sampling as a test data analysis period, taking the maximum temperature within 10 days as the maximum value of the maximum temperature, taking the minimum temperature within 10 days as the minimum value of the minimum temperature, taking the average temperature, wind speed and relative humidity as the average values within 10 days, and taking the cumulative evapotranspiration within 10 days as the evapotranspiration.
In the step (2), the spectral reflectivity of the canopy of the wheat plant is processed by adopting ViewSpecPro software, and the spectral reflectivity of the canopy under each wave band within the wave band range of 350-2500 nm is obtained at intervals of 1 nm; respectively constructing the spectral indexes of any two-waveband combination based on spectral reflectivity data by using Matlab software: normalizing the spectral index (NDSI), the Ratio Spectral Index (RSI), the Difference Spectral Index (DSI), and establishing a linear model of the spectral index and the moisture indicator (CWC, PWC, and CEWT); coefficient of determination (R) from linear model 2 ) And Root Mean Square Error (RMSE), screening out the spectral index with the best correlation with different moisture indexes, and screening out the spectral index with the best correlation with the moisture content of the canopy (CWC) as DSI (2015,2375) The spectral index which has the best correlation to Plant Water Content (PWC) is NDSI (2175,2245) The spectral index with the best correlation to Canopy Equivalent Water Thickness (CEWT) is RSI (720,1200) 。
NDSI＝(R λ1 -R λ2 )/(R λ1 +R λ2 )
RSI＝R λ1 /R λ2
DSI＝R λ1 -R λ2
Wherein R is λ1 、R λ2 Respectively is the spectral reflectivity of any wave band within the wave band range of 350-2500 nm.
In the step (3), SPSS software is used for carrying out principal component analysis on all characteristic variables (soil data, plant data and meteorological data), the number of the selected variables is gradually reduced according to the reduction of the proportion of different principal components, and the variable with larger load in each principal component is selected as a screening result. The feature variables screened based on Principal Component Analysis (PCA) were: the screening results of the soil data (S) are 0-20cm soil water content (SWC1) and 20-40cm soil water content (SWC2), the screening results of the plant data (P) are Leaf Nitrogen Concentration (LNC), Plant Nitrogen Accumulation (PNA) and overground part dry matter weight (ADM), and the screening results of the meteorological data (C) are maximum temperature (T) max ) Average temperature (T) avg ) And an amount of Evapotranspiration (ET).
When the Pearson correlation coefficient is used for feature selection, the correlation between each feature variable and the moisture index in soil data, plant data and meteorological data is analyzed respectively, the variable with the best moisture index determination coefficient in each type of data is selected, the Pearson correlation coefficient of the residual variable and the variable with the best moisture index determination coefficient is analyzed, and the variables with the selection coefficient smaller than 0.5 jointly form the variable screening result of the data (see Table 4).
In the step (4), the different databases established in the spectral index fusion step (3) screened in the step (2) are used as model input parameters, and the specific steps are as follows:
for CWC, the spectral indexes of the constructed models are all DSI (2015,2375) On the basis, screening variables of the database S, P, C, S + P, S + C, P + C, S + P + C are added respectively, wherein the variable results selected based on Principal Component Analysis (PCA) are respectively S: SWC1, SWC2, P: LNC, PNA, ADM, C: tavg, T max 、ET，S+P：SWC1、SWC2、LNC、PNA、ADM，S+C：SWC1、SWC2、T avg 、T max 、ET，P+C：LNC、PNA、ADM、T avg 、T max 、ET，S+P+C：SWC1、SWC2、LNC、PNA、ADM、T avg 、T max ET, wherein variable results selected based on the Pearson correlation coefficient PCC are S: SWC3, P: LNC, PNA, ADM, C: t is avg 、WS、ET，S+P：SWC3、LNC、PNA、ADM，S+C：SWC3、T avg 、WS、ET，P+C：LNC、PNA、ADM、T avg 、WS、ET，S+P+C：SWC3、LNC、PNA、ADM、T avg WS, ET, all variables and spectral indices DSI for each combination scheme (2015,2375) As input parameters of the model, 42 prediction models of the CWC, namely 2 feature selection methods, 7 databases and 3 machine learning algorithms, are constructed by utilizing three algorithms of a Decision Tree (DT), a Random Forest (RF) and a Support Vector Machine (SVM).
For PWC, NDSI is used for spectral indexes of all constructed models (2175,2245) On the basis, screening variables of the database S, P, C, S + P, S + C, P + C, S + P + C are added respectively, wherein the variable results selected based on Principal Component Analysis (PCA) are respectively S: SWC1, SWC2, P: LNC, PNA, ADM, C: t is avg 、T max 、ET，S+P：SWC1、SWC2、LNC、PNA、ADM，S+C：SWC1、SWC2、T avg 、T max 、ET，P+C：LNC、PNA、ADM、T avg 、T max 、ET，S+P+C：SWC1、SWC2、LNC、PNA、ADM、T avg 、T max ET, wherein variable results selected based on the Pearson correlation coefficient PCC are S: SWC3, P: LNC, PNA, ADM, C: t is avg 、WS、ET，S+P：SWC3、LNC、PNA、ADM，S+C：SWC3、T avg 、WS、ET，P+C：LNC、PNA、ADM、T avg 、WS、ET，S+P+C：SWC3、LNC、PNA、ADM、T avg WS, ET, all variables and spectral indices NDSI of each combination scheme (2175,2245) And as input parameters of the model, 42 prediction models of the PWC are constructed by using three algorithms of a decision tree DT, a random forest RF and a support vector machine SVM.
For CEWT, spectral indices of the constructed model are all RSI (720,1200) On the basis, adding the database S,P, C, S + P, S + C, P + C, S + P + C, wherein the variable results selected based on principal component analysis PCA are S: SWC1, SWC2, P: LNC, PNA, ADM, C: t is avg 、T max 、ET，S+P：SWC1、SWC2、LNC、PNA、ADM，S+C：SWC1、SWC2、T avg 、T max 、ET，P+C：LNC、PNA、ADM、T avg 、T max 、ET，S+P+C：SWC1、SWC2、LNC、PNA、ADM、T avg 、T max ET, wherein variable results selected based on the Pearson correlation coefficient PCC are S: SWC3, P: LNA, ADM, C: t is min 、T avg 、WS、ET、RH，S+P：SWC3、LNA、ADM，S+C：SWC3、T min 、T avg 、WS、ET、RH，P+C：LNA、ADM、T min 、T avg 、WS、ET、RH，S+P+C：SWC3、LNA、ADM、T min 、T avg WS, ET, RH, all variables and spectral indices RSI of each combination scheme (720,1200) As input parameters of the model, 42 prediction models of the CEWT are constructed by using three algorithms of a decision tree DT, a random forest RF and a support vector machine SVM.
Decision tree DT, random forest RF and SVM algorithms for constructing models in the R language respectively use rpart, randomForest and e1071 toolkits in RStudio software.
In step (5), 70% of the data is used as training set and 30% of the data is used as test set, according to the decision coefficient (R) 2 ) Comparing and evaluating the winter wheat moisture condition monitoring model by Root Mean Square Error (RMSE) and Mean Absolute Error (MAE), and respectively screening out the optimal monitoring models for canopy water content, plant water content and canopy equivalent water thickness.
The optimal monitoring model for canopy water content comprises 40-60cm soil water content (SWC3), Leaf Nitrogen Concentration (LNC), Plant Nitrogen Accumulation (PNA), overground part dry matter weight (ADM), and average temperature (T) avg ) A database of Wind Speed (WS) and Evapotranspiration (ET).
The plant water content optimal monitoring model comprises 40-60cm of soil water content (SWC3), Leaf Nitrogen Concentration (LNC), plant nitrogen accumulation amount (PNA), overground part dry matter weight (ADM), and average temperature (T) avg ) A database of Wind Speed (WS) and Evapotranspiration (ET).
The optimal monitoring model of the equivalent water thickness of the canopy uses the soil water content (SWC3) of 40-60cm, the accumulated Leaf Nitrogen (LNA), the dry matter weight (ADM) of the overground part and the lowest temperature (T) min ) Average temperature (T) avg ) A database of Wind Speed (WS), Evapotranspiration (ET) and Relative Humidity (RH).
The invention also aims to provide a winter wheat moisture condition monitoring method based on data fusion, which comprises the following steps:
step (1), data acquisition: acquiring the spectral reflectivity of a wheat plant canopy; collecting soil data: soil moisture content of 40-60cm (SWC3), plant data: leaf Nitrogen Concentration (LNC), Leaf Nitrogen Accumulation (LNA), Plant Nitrogen Accumulation (PNA), overground part dry matter weight (ADM), meteorological data: lowest temperature (T) min ) Average temperature (T) avg ) Wind Speed (WS), Evapotranspiration (ET) and Relative Humidity (RH);
step (2), data processing: respectively calculating spectral index DSI by using canopy spectral reflectivity data (2015,2375) 、NDSI (2175,2245) And RSI (720,1200) (ii) a For monitoring of Canopy Water Content (CWC) and Plant Water Content (PWC), 40-60cm soil water content (SWC3), Leaf Nitrogen Concentration (LNC), Plant Nitrogen Accumulation (PNA), overground dry matter weight (ADM), mean temperature (T.sub.T.sub. avg ) The Wind Speed (WS) and the Evapotranspiration (ET) form a database 1; for monitoring Canopy Equivalent Water Thickness (CEWT), soil moisture content (SWC3), Leaf Nitrogen Accumulation (LNA), overground dry matter weight (ADM), and minimum temperature (T) of 40-60cm were used min ) Average temperature (T) avg ) A database 2 consisting of Wind Speed (WS), Evapotranspiration (ET) and Relative Humidity (RH);
step (3), obtaining a moisture index: optimal monitoring model RF-PCC- (S + P + C) based on the Canopy Water Content (CWC) of claim 1, using the spectral index DSI (2015,2375) And the database 1 is used as an input variable, and the Canopy Water Content (CWC) is obtained through a random forest algorithm RF;
use of the spectral index NDSI based on the Plant Water Cut (PWC) optimal monitoring model of claim 1 RF-PCC- (S + P + C) (2175,2245) And the database 1 is used as an input variable, and the water content PWC of the plants is obtained through a random forest algorithm RF;
based on the optimal monitoring model SVM-PCC-S + P + C of claim 1 Canopy Equivalent Water Thickness (CEWT), using the spectral index RSI (720,1200) And database 2 as input variables, and the Canopy Equivalent Water Thickness (CEWT) is obtained by a support vector machine algorithm SVM.
The invention has the beneficial effects that:
the invention provides an innovative winter wheat moisture condition monitoring mode based on soil, plant and meteorological data fusion, the content comprises data acquisition of a monitoring system, establishment of a monitoring method and a winter wheat moisture index monitoring model which can be used for practical application, influence factors on moisture conditions can be comprehensively considered, the capability of monitoring the moisture conditions of crops is improved, crop moisture information is accurately acquired, implementation of farmland management decision is guided, and technical support is provided for accurate agricultural management.
Drawings
Fig. 1 is a contour diagram of spectral index determination coefficients based on a moisture indicator and any two-band combination.
FIG. 2 shows Pearson's correlation coefficient between variables and correlation (R) between each variable and moisture indicator 2 )。
FIG. 3 is a comparison of the performance of different models of the training set; wherein, S: variables for soil data screening; p: variables for plant data screening; c: a variable for climate data screening; s + P: variables for soil and plant data screening; s + C: variables for soil and meteorological data screening; p + C: plant and meteorological data screening variables; s + P + C: soil, plants and meteorological data screening variables.
FIG. 4 is a comparison of the performance of different models of a test set; wherein, graphs a-c represent the determining coefficients (R) of CWC, PWC and CEWT, respectively 2 ) Graphs d-f represent Root Mean Square Error (RMSE) for CWC, PWC, and CEWT, respectively, and graphs g-i represent Mean Absolute Error (MAE) for CWC, PWC, and CEWT, respectively. A: decision tree-pearson correlation coefficient (DT-PCC), B: random forest-pearson correlation coefficient (RF-PCC), C: support vector machine-pearson correlation coefficient (SVM-PCC), D: decision tree-masterCompositional analysis (DT-PCA), E: random forest-principal component analysis (RF-PCA), F: support vector machine-principal component analysis (SVM-PCA).
FIG. 5 is a correlation between measured values and predicted values of CWC, PWC and CEWT; wherein, the graphs a-c are three models with the best correlation between the CWC measured value and the predicted value; fig. d-f are three models with best correlation between the PWC measured value and the predicted value; the graphs g-i are the three models with the best correlation between the CEWT observed values and the predicted values.
Detailed Description
The technical solutions of the present invention are further illustrated by the following examples, which should be understood as merely illustrative and not restrictive, and various equivalent modifications of the present invention by those skilled in the art after reading the present invention fall within the scope of the appended claims.
In order to better research the winter wheat water condition monitoring based on multi-source data fusion, a water condition monitoring model is established, a wheat season test is carried out in 2019-2021 in 35.3 degrees N and 113.9 degrees E in a rain-proof shed measuring pit (35.3 degrees N and 113.9 degrees E) of a seven-Ringying base of a farm irrigation research institute of China agricultural science institute in New county City in Henan province, and the specific test conditions are shown in Table 1.
The test is carried out for two years, the planting variety is Zhoumai No. 27, manual drilling is adopted, the basic seedling is 15 ten thousand per mu, the row spacing is 25cm, and the area of each test cell is 6.67m 2 (2 m.times.3.33 m), three repeated random block tests were set for all treatments, nitrogen fertilizer application was carried out before sowing and in the jointing stage using granular urea at a topdressing ratio of 5:5, 120kg of P was applied before sowing, respectively 2 O 5 ·ha -1 And applying 120kgK to the calcium phosphate fertilizer 2 O·ha -1 K of 2 SO 4 And (3) a fertilizer. The water treatment was carried out 3 times in total, and irrigation was carried out in the jointing stage, heading stage and grouting stage, respectively, with the amount of irrigation per time shown in table 1.
TABLE 1 field test arrangement and setup
1. Acquisition of multi-source data
Collecting the spectral reflectivity of the wheat plant canopy: the tester should wear the dark clothing, use the FieldSpec FR Pro hyperspectral radiometer produced by American Analytical Spectral Device (ASD), test when clear and cloudless, windless or very small wind speed, the test time is 10:00-14:00, use the white board to correct before the test, the sensor probe is downward vertically during the measurement, the height is 1.0m from the wheat canopy, the field angle of the spectrometer is 25 degrees, the diameter of the ground field range is 0.44m, each test cell determines 3 sampling points, each sampling point repeats the measurement 5 times (15 spectra in total), take the average value as the sampling spectrum of each processing cell, namely the wheat plant canopy Spectral reflectance. And carrying out white board correction once every two cell data are collected.
Acquisition of plant data (P): destructive sampling is carried out on the wheat plants while testing the spectral reflectivity of the canopy of the wheat plants, the wheat with uniform growth vigor is randomly selected in the cell to be tested, and the frame is 0.15m 2 After the roots are cut off, separating the sample of the overground part into four leaves, the remaining leaves, the stem sheaths and the ear parts at the top of the canopy in time, weighing to obtain the fresh weight of the four leaves, the fresh weight of the remaining leaves, the fresh weight of the stem sheaths and the fresh weight of the ears at the top of the canopy, calculating to obtain the fresh weight of plants, simultaneously measuring the leaf area by using an LI-3000C portable leaf area meter, and calculating the Leaf Area Index (LAI) by using a specific leaf weight method; putting all samples into an oven, deactivating enzyme at 105 ℃ for 30 minutes, then adjusting to 80 ℃, drying for 24 hours to constant weight, weighing, calculating to obtain the dry weight of four leaves at the top of the canopy, the dry weight of the rest leaves, the dry weight of stem sheaths and the dry weight of ears, and calculating to obtain the dry weight of plants; and (3) measuring the nitrogen concentration of the dried and weighed plant sample by a Kjeldahl method to respectively obtain the nitrogen concentrations of the leaf, the stem sheath and the ear, and calculating the leaf nitrogen accumulation amount (LNA) and the plant nitrogen accumulation amount (PNA) by multiplying the dry matter weight by the nitrogen concentration.
Accumulation of Nitrogen in organs (kg/hm) 2 ) Organ nitrogen content x dry organ weight
Plant nitrogen accumulation (kg/hm) 2 ) The sum of the accumulated nitrogen in each organ
Plant nitrogen concentration (%). plant nitrogen accumulation/plant dry matter weight x 100
Acquiring a moisture index: in the plant sampling process, the fresh weight and the dry weight of four leaves, the residual leaves, the stem sheaths and the ears at the top of the canopy are obtained, and the Canopy Water Content (CWC), the Plant Water Content (PWC) and the Canopy Equivalent Water Thickness (CEWT) are respectively obtained through calculation.
Canopy water content (%) - (fresh weight of top four leaves-dry weight of top four leaves)/fresh weight of top four leaves × 100
Plant water content (%) - (fresh weight of plant-dry weight of plant)/fresh weight of plant × 100
Leaf equivalent water thickness (cm) ═ leaf fresh weight-leaf dry weight)/(leaf area × water density)
Canopy equivalent water thickness (cm) is equal to blade equivalent water thickness multiplied by blade area index
Acquisition of soil data (S): respectively collecting soil samples of 0-20cm, 20-40cm and 40-60cm near a plant sampling point by using a 1m soil drill, immediately filling the soil samples into self-sealing bags, taking the self-sealing bags back to a room, putting the self-sealing bags into a weighed aluminum box, weighing the aluminum box on an electronic balance, putting the aluminum box filled with the soil samples into an oven, drying the aluminum box at 105 ℃ for 8 hours until the weight is constant, weighing and calculating to obtain the soil water content of different soil layers: 0-20cm soil moisture (SWC1), 20-40cm soil moisture (SWC2) and 40-60cm soil moisture (SWC 3).
Soil water content (%) - (fresh soil weight-dried soil weight)/dried soil weight × 100
Acquisition of meteorological data (C): downloading data of the highest temperature, the lowest temperature, the average temperature, the wind speed, the evapotranspiration amount and the relative humidity of each day in the winter wheat growth period at a local meteorological site, wherein the data in 10 days before each plant sampling is a test data analysis period, and the highest temperature (T) max ) Taking the maximum value of the maximum temperature and the lowest temperature (T) within 10 days min ) Taking the minimum value of the lowest temperature within 10 days, and the average temperature (T) avg ) The Wind Speed (WS) and the Relative Humidity (RH) are averaged over 10 days, and the Evapotranspiration (ET) is the cumulative evapotranspiration over 10 days.
2. Construction of spectral index
Processing the wheat plant canopy spectral reflectivity by using ViewSpecpro software, obtaining the spectral reflectivity under each wave band within the wave band range of 350-2500 nm at intervals of 1nm, and constructing a normalized spectral index (NDSI), a Ratio Spectral Index (RSI) and a Difference Spectral Index (DSI) of any two wave band combinations:
NDSI＝(R λ1 -R λ2 )/(R λ1 +R λ2 )
RSI＝R λ1 /R λ2
DSI＝R λ1 -R λ2
wherein R is λ1 、R λ2 Respectively is the spectral reflectivity of any wave band within the wave band range of 350-2500 nm.
By utilizing Matlab software, linear models between spectral indexes (NDSI, RSI and DSI) and moisture indexes (CWC, PWC and CEWT) are established, wherein 70% of data is used for modeling, 30% of data is used for testing and inspecting the models, linear models between water content of a canopy layer and NDSI, RSI and DSI, linear models between water content of a plant and NDSI, RSI and DSI, linear models between equivalent water thickness of the canopy layer and NDSI, RSI and DSI are respectively established, and an optimal core waveband combination under different types of spectral indexes is selected (figure 1). Coefficient of determination (R) from linear model 2 ) And Root Mean Square Error (RMSE), and the spectral index with the best correlation with different moisture indexes is screened out, and the analysis result is shown in table 2. The results show that the spectral index which performs optimally for CWC is DSI (2015,2375) The optimal spectral index for PWC performance is NDSI (2175,2245) Spectral index optimized for CEWT is RSI (720,1200) 。
TABLE 2 screening of optimal spectral indices for different moisture indicators
3. Feature selection
And (3) carrying out principal component analysis on all characteristic variables by using SPSS software, wherein the accumulated contribution rate of the PC1 is 38.93%, variables with factor loads larger than 0.8 are selected as screening results, and 3 variables have higher loads. As the cumulative contribution of PC2, PC3, and PC4 decreases, eachThe number of the principal component selection variables is gradually reduced, the number of the principal component selection variables is respectively 2, 2 and 1, and the variable with large load in each principal component is selected as a screening result. The screening results are shown in Table 3. The soil data screening results are SWC1 and SWC2, the plant data screening results are LNC, PNA and ADM, and the meteorological data screening result is T max 、T avg And ET.
The characteristics are selected by using the Pearson correlation coefficient, the Pearson correlation coefficient among variables in different types of data of soil data (S), plant data (P) and meteorological data (C) and the correlation between the variables and the moisture index are respectively analyzed (figure 2), the variable with the best determination coefficient of the moisture index in each type of data is selected, the Pearson correlation coefficient of the rest variables and the variable with the best determination coefficient of the moisture index is analyzed, the variables with the selection coefficient smaller than 0.5 jointly form the variable screening result of the data, and the result is shown in a table 4.
And according to the screened characteristic variables, different types of data combinations are combined, and a database (S, P, C, S + P, S + C, P + C, S + P + C) of seven selected variable combinations is established as a characteristic variable database input by the model.
TABLE 3 characteristic variable screening results based on principal component analysis
TABLE 4 characteristic variable screening results based on Pearson's correlation coefficient
4. Model build results and validation
Three algorithm models of a decision tree, a random forest and a support vector machine are established by using rpart, randomForest and e1071 packages in R-Studio software, and 42 models are constructed by using two feature selection methods (PCC and PCA) and a database (S, P, C, S + P, S + C, P + C, S + P + C) of seven selection variable combinations. 70% of the data was taken as the training set and ten were performedFold cross validation (FIG. 3), 30% data as test set (FIG. 4), based on the coefficient of determination (R) 2 ) Comparing and evaluating the models by Root Mean Square Error (RMSE) and Mean Absolute Error (MAE), and respectively screening out the optimal models for canopy water content, plant water content and canopy equivalent water thickness. The established model is verified, and the verification result is shown in fig. 5. The result shows that the model with the best prediction effect on the CWC is RF-PCC- (S + P + C) (R) 2 0.86, RMSE 4.10%, MAE 2.21%); the model with the best prediction effect on PWC is RF-PCC- (S + P + C), (R) 2 0.94, 2.24% RMSE, 1.60% MAE); the model with the best prediction effect on CEWT is SVM-PCC- (S + P + C) (R) 2 ＝0.97，RMSE＝48.61μm，MAE＝39.47μm)。
Besides the monitoring model provided in this embodiment, researchers and users can organize tests by themselves, and establish other monitoring models of water conditions for use by using a similar method.
Claims (9)
1. A method for establishing a winter wheat moisture condition monitoring model based on data fusion is characterized by comprising the following steps: the method comprises the following steps:
step (1), data acquisition: acquiring the spectral reflectivity of a wheat plant canopy; sampling the plants, and determining the moisture index: canopy water content CWC, plant water content PWC, canopy equivalent water thickness CEWT acquire plant data P: leaf area index, dry weight of leaves, dry weight of overground parts, nitrogen concentration of leaves, nitrogen concentration of plants, nitrogen accumulation amount of leaves and nitrogen accumulation amount of plants; acquiring soil data S: taking layered sampling on 0-60cm of soil to respectively obtain the water content of 0-20cm of soil, the water content of 20-40cm of soil and the water content of 40-60cm of soil; acquiring meteorological data C in the winter wheat growth period: maximum temperature, minimum temperature, average temperature, wind speed, evapotranspiration and relative humidity;
step (2), establishing a spectral index of any two-waveband combination based on the spectral reflectivity of the wheat plant canopy: normalizing the spectral index NDSI, the ratio spectral index RSI and the difference spectral index DSI, respectively establishing a linear model with the canopy water content CWC, the plant water content PWC and the canopy equivalent water thickness CEWT, and according to the spectral indexesCoefficient of determination R of linear model 2 And root mean square error RMSE, respectively screening the spectral indexes with the best correlation to the canopy water content CWC, the plant water content PWC and the canopy equivalent water thickness CEWT;
and (3) performing feature selection on the obtained soil data, plant data and meteorological data by using a principal component analysis PCA (principal component analysis) and Pearson correlation coefficient PCC (policy and charging control) method, and respectively establishing 7 databases consisting of feature variables on the basis of feature variables in the screened soil data, plant data and meteorological data: s, P, C, S + P, S + C, P + C, S + P + C;
step (4), different databases built in the spectral index fusion step (3) screened in the step (2) are used as model input parameters, and a winter wheat moisture condition monitoring model built based on data fusion is built by respectively utilizing decision trees DT, random forest RF and Support Vector Machine (SVM) algorithms in R language codes;
and (5) comparing the prediction effects of different winter wheat moisture condition monitoring models on moisture indexes, screening out that the optimal monitoring model for the water content CWC of the canopy is RF-PCC- (S + P + C), the optimal monitoring model for the water content PWC of the plant is RF-PCC- (S + P + C), and the optimal monitoring model for the equivalent water thickness CEWT of the canopy is SVM-PCC- (S + P + C).
2. The method for building a winter wheat moisture status monitoring model based on data fusion as claimed in claim 1, wherein: in the step (1), obtaining the spectral reflectance of the wheat plant canopy: the method comprises the steps of using an ASD (automatic sequence decomposition) hyperspectral radiometer to carry out measurement in clear, cloudless and windless conditions, measuring the time by 10:00-14:00, correcting by using a white board before measurement, measuring the wave band range by 350-2500 nm, measuring the height of a sensor probe vertically downwards to be 1.0m away from a wheat canopy during measurement, measuring the field angle of the spectroradiometer by 25 degrees and the diameter of the ground field range by 0.44m, measuring 3 sampling points in each sampling cell, repeatedly measuring each sampling point for 5 times, and taking the average value as the spectral reflectivity of the wheat plant canopy in a processing cell.
3. The method for building a winter wheat moisture status monitoring model based on data fusion of claim 1, whereinThe method comprises the following steps: in the step (1), destructive sampling of wheat plant samples is carried out while testing the spectral reflectivity of the canopy of the wheat plant, wheat with uniform growth vigor is randomly selected in a treatment cell, and 0.15m is framed 2 Cutting off roots, separating the sample of the overground part into four leaves, the remaining leaves, the stem sheaths and the ear parts at the top of the canopy, weighing to obtain the fresh weights of the four leaves, the remaining leaves, the stem sheaths and the ears at the top of the canopy, and calculating to obtain the fresh weight of the plant; measuring the area of the blade by using an LI-3000C portable blade area meter, and calculating a blade area index; deactivating enzyme for 30 minutes at 105 ℃, drying to constant weight at 80 ℃, weighing to obtain the dry weight of four leaves at the top of the canopy, the dry weight of the rest leaves, the dry weight of the stem sheath and the dry weight of the spike, and calculating to obtain the dry weight of the plant; respectively calculating the water content of the canopy, the water content of the plant and the equivalent water thickness of the canopy according to the fresh weight, the dry weight and the leaf area index of the sample; determining the total nitrogen content of the dried and weighed plant sample by a Kjeldahl method to obtain the nitrogen concentration of leaves, stem sheaths and ears, and calculating the leaf nitrogen accumulation amount and the plant nitrogen accumulation amount by multiplying the dry weight by the nitrogen content;
respectively collecting 0-20cm, 20-40cm and 40-60cm soil samples between any side of the middle point of the wheat plant sampling row by using a 1m soil drill, drying the samples in an oven at 105 ℃ until the samples are constant in weight, and calculating to obtain the water content of the soil with different soil layers of 0-20cm, 20-40cm and 40-60 cm;
downloading data of daily maximum temperature, minimum temperature, average temperature, wind speed, evapotranspiration and relative humidity in a winter wheat growth period at a local meteorological site, taking data within 10 days before each plant sampling as a test data analysis period, taking the maximum temperature within 10 days as the maximum value of the maximum temperature, taking the minimum temperature within 10 days as the minimum value of the minimum temperature, taking the average temperature, wind speed and relative humidity as the average values within 10 days, and taking the cumulative evapotranspiration within 10 days as the evapotranspiration.
4. The method for building a winter wheat moisture status monitoring model based on data fusion of claim 1, wherein: in the step (2), the spectral reflectivity of the canopy of the wheat plant is processed by adopting ViewSpecPro software, and the spectral reflectivity is obtained under each wave band within the wave band range of 350-2500 nm at intervals of 1nm(ii) a canopy spectral reflectance; respectively constructing the spectral indexes of any two-waveband combination based on spectral reflectivity data by using Matlab software: normalizing the spectral index NDSI, the ratio spectral index RSI and the difference spectral index DSI, and establishing a linear model of the spectral indexes and the moisture indexes; coefficient of determination R from a linear model 2 And root mean square error RMSE, and screening out the spectral index with the best correlation with different moisture indexes.
5. The method for building a winter wheat moisture status monitoring model based on data fusion of claim 4, wherein: the spectral index with the best correlation to the CWC of the water content of the canopy is DSI (2015,2375) The best spectral index for the PWC correlation of plant water content is NDSI (2175,2245) The spectral index with the best correlation to the canopy equivalent water thickness CEWT is RSI (720,1200) 。
6. The method for building a winter wheat moisture status monitoring model based on data fusion as claimed in claim 1, wherein: in the step (3), the feature variables screened out based on principal component analysis are: the soil data screening results are 0-20cm of soil water content and 20-40cm of soil water content, the plant data screening results are leaf nitrogen concentration, plant nitrogen concentration and overground part dry matter weight, and the meteorological data screening results are maximum temperature, average temperature and evapotranspiration;
when the Pearson correlation coefficient is used for feature selection, the correlation between each feature variable and the moisture index in soil data, plant data and meteorological data is analyzed respectively, the variable with the best moisture index determination coefficient in each type of data is selected, the Pearson correlation coefficient of the residual variable and the variable with the best moisture index determination coefficient is analyzed, and the variables with the selection coefficient smaller than 0.5 jointly form the variable screening result of the data.
7. The method for building a winter wheat moisture status monitoring model based on data fusion of claim 1, wherein: in the step (4), decision tree DT, random forest RF and SVM algorithms used for model construction in the R language respectively use rpart, randomForest and e1071 toolkits in RStudio software.
8. The method for building a winter wheat moisture status monitoring model based on data fusion of claim 1, wherein: in the step (5), 70% of data is used as a training set, 30% of data is used as a test set, and a coefficient R is determined according to 2 The root mean square error RMSE and the average absolute error MAE are used for comparing and evaluating the winter wheat moisture condition monitoring models, and the optimal monitoring models for the canopy water content, the plant water content and the canopy equivalent water thickness are respectively screened out.
9. A winter wheat moisture condition monitoring method based on data fusion is characterized by comprising the following steps: the method comprises the following steps:
step (1), data acquisition: acquiring the spectral reflectivity of a wheat plant canopy; collecting soil data: 40-60cm soil water content SWC3, plant data: leaf nitrogen concentration LNC, leaf nitrogen accumulation LNA, plant nitrogen accumulation PNA, overground part dry matter weight ADM, and meteorological data: lowest temperature T min Average temperature T avg Wind speed WS, evapotranspiration ET and relative humidity RH;
step (2), data processing: respectively calculating spectral index DSI by using canopy spectral reflectivity data (2015,2375) 、NDSI (2175,2245) And RSI (720,1200) (ii) a For monitoring the water content CWC of the canopy and the water content PWC of the plant, 40-60cm of soil water content SWC3, leaf nitrogen concentration LNC, plant nitrogen accumulation PNA, dry matter weight ADM of the overground part, and average temperature T are used avg The wind speed WS and the evapotranspiration ET form a database 1; for monitoring the equivalent water thickness CEWT of the canopy, the soil water content SWC3, the leaf nitrogen accumulation LNA, the above-ground dry matter weight ADM and the lowest temperature T of 40-60cm are used min Average temperature T avg A database 2 is formed by the wind speed WS, the evapotranspiration ET and the relative humidity RH;
step (3), obtaining a moisture index: based on the optimal monitoring model RF-PCC- (S + P + C) of the canopy water content CWC of the claim 1, the spectral index DSI is utilized (2015,2375) And the database 1 is used as an input variable, and the canopy water content CWC is obtained through a random forest algorithm RF;
the method for optimal monitoring of plant water content PWC model RF-PCC- (S + P + C) according to claim 1 using the spectral index NDSI (2175,2245) And the database 1 is used as an input variable, and the water content PWC of the plants is obtained through a random forest algorithm RF;
based on the claim 1 canopy equivalent water thickness CEWT optimal monitoring model SVM-PCC- (S + P + C), using spectral index RSI (720,1200) And a database 2 is used as an input variable, and the canopy equivalent water thickness CEWT is obtained through a support vector machine algorithm SVM.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210616907.5A CN115015132A (en) | 2022-06-01 | 2022-06-01 | Winter wheat moisture condition monitoring method based on data fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210616907.5A CN115015132A (en) | 2022-06-01 | 2022-06-01 | Winter wheat moisture condition monitoring method based on data fusion |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115015132A true CN115015132A (en) | 2022-09-06 |
Family
ID=83072102
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210616907.5A Pending CN115015132A (en) | 2022-06-01 | 2022-06-01 | Winter wheat moisture condition monitoring method based on data fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115015132A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116484287A (en) * | 2023-06-21 | 2023-07-25 | 云南省地质矿产勘查开发局中心实验室(自然资源部昆明矿产资源检测中心) | Soil moisture prediction method, system, terminal and medium based on error correction |
| CN117907248A (en) * | 2024-03-19 | 2024-04-19 | 中国水利水电科学研究院 | Remote sensing monitoring method and system for root system soil water content in key growth period of winter wheat |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103472009A (en) * | 2013-09-16 | 2013-12-25 | 南京农业大学 | Method for monitoring wheat plant water content under different plant nitrogen content levels |
| CN103528965A (en) * | 2013-08-28 | 2014-01-22 | 南京农业大学 | Wheat leaf equivalent water thickness hyperspectral monitoring method |
| WO2019109383A1 (en) * | 2017-12-05 | 2019-06-13 | 江苏大学 | Water stress detection method for tomatoes in seedling stage based on micro-ct and polarization-hyperspectral imaging multi-feature fusion |
| CN110082300A (en) * | 2019-05-20 | 2019-08-02 | 中国水利水电科学研究院 | Canopy of winter wheat moisture content monitoring model method for building up based on spectral parameter |
| CN111965117A (en) * | 2020-08-04 | 2020-11-20 | 中国水利水电科学研究院 | Winter wheat moisture monitoring method and system based on PROSPECT model |
-
2022
- 2022-06-01 CN CN202210616907.5A patent/CN115015132A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103528965A (en) * | 2013-08-28 | 2014-01-22 | 南京农业大学 | Wheat leaf equivalent water thickness hyperspectral monitoring method |
| CN103472009A (en) * | 2013-09-16 | 2013-12-25 | 南京农业大学 | Method for monitoring wheat plant water content under different plant nitrogen content levels |
| WO2019109383A1 (en) * | 2017-12-05 | 2019-06-13 | 江苏大学 | Water stress detection method for tomatoes in seedling stage based on micro-ct and polarization-hyperspectral imaging multi-feature fusion |
| CN110082300A (en) * | 2019-05-20 | 2019-08-02 | 中国水利水电科学研究院 | Canopy of winter wheat moisture content monitoring model method for building up based on spectral parameter |
| CN111965117A (en) * | 2020-08-04 | 2020-11-20 | 中国水利水电科学研究院 | Winter wheat moisture monitoring method and system based on PROSPECT model |
Non-Patent Citations (2)
| Title |
|---|
| BO SHI ET AL.: "Improving water status prediction of winter wheat using multi-source data with machine learning", EUROPEAN JOURNAL OF AGRONOMY, vol. 139, 27 May 2022 (2022-05-27), pages 126548 - 1, XP087117831, DOI: 10.1016/j.eja.2022.126548 * |
| 哈布热 等: "基于冠层光谱特征的冬小麦植株含水率诊断研究", 灌溉排水学报, vol. 37, no. 10, 15 October 2018 (2018-10-15), pages 9 - 15 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116484287A (en) * | 2023-06-21 | 2023-07-25 | 云南省地质矿产勘查开发局中心实验室(自然资源部昆明矿产资源检测中心) | Soil moisture prediction method, system, terminal and medium based on error correction |
| CN116484287B (en) * | 2023-06-21 | 2023-09-01 | 云南省地质矿产勘查开发局中心实验室(自然资源部昆明矿产资源检测中心) | Soil moisture prediction method, system, terminal and medium based on error correction |
| CN117907248A (en) * | 2024-03-19 | 2024-04-19 | 中国水利水电科学研究院 | Remote sensing monitoring method and system for root system soil water content in key growth period of winter wheat |
| CN117907248B (en) * | 2024-03-19 | 2024-05-28 | 中国水利水电科学研究院 | Remote sensing monitoring method and system for root system soil water content in key growth period of winter wheat |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110082300B (en) | Method for establishing winter wheat canopy water content monitoring model based on spectral parameters | |
| CN109187441B (en) | Method for constructing summer corn nitrogen content monitoring model based on canopy spectral information | |
| Penuelas et al. | Cell wall elasticity and water index (R970 nm/R900 nm) in wheat under different nitrogen availabilities | |
| Piekkielek et al. | Use of a chlorophyll meter to predict sidedress nitrogen requirements for maize | |
| CN107356569B (en) | The construction method of method and its model based on chlorophyll fluorescence prediction wheat grain yield | |
| CN115015132A (en) | Winter wheat moisture condition monitoring method based on data fusion | |
| CN113268923A (en) | Summer corn yield estimation method based on simulation multispectral | |
| Zhu et al. | Monitoring leaf nitrogen in wheat using canopy reflectance spectra | |
| CN110189793B (en) | Hyperspectrum-based wheat nitrogen fertilizer physiological utilization rate estimation model construction and wheat variety classification with different nitrogen efficiencies | |
| CN106718363B (en) | Irrigation test method and test platform for fine agriculture | |
| CN110245420A (en) | The method of coast protection forest biomass and organic C storage monitoring and metering | |
| CN111829965A (en) | Remote sensing inversion model and method for starch accumulation amount of rice overground part | |
| CN113268703A (en) | Nitrogen fertilizer deficiency rapid detection and precision topdressing method applied to wheat field management | |
| CN101044823A (en) | Method for estimating crop energy utilization rate and predetermining the yield | |
| LIU et al. | Leaf area index based nitrogen diagnosis in irrigated lowland rice | |
| CN112129709A (en) | Apple tree canopy scale nitrogen content diagnosis method | |
| CN109425703A (en) | A method of the non-destructive monitoring chamber crop water regime for Precision Irrigation control | |
| Ai et al. | Soil nutrients influence the photosynthesis and biomass in invasive Panicum virgatum on the Loess Plateau in China | |
| Zhang et al. | Key variable for simulating critical nitrogen dilution curve of wheat: Leaf area ratio-driven approach | |
| CN113935541A (en) | Method for predicting spraying time of mechanical cotton picking defoliating agent | |
| CN112990692B (en) | Winter wheat yield estimation method based on multiparameter coupling in sub-fertility stage | |
| CN113504186B (en) | Method for estimating utilization rate of nitrogen fertilizer in wheat by remote sensing | |
| CN113640230B (en) | Rapid detection method and system for field wheat moisture utilization rate | |
| CN112749837B (en) | Method for estimating cereal crop yield based on post-flower NDVI accumulation value | |
| Cao et al. | Estimating rice nitrogen status with the Crop Circle multispectral active canopy sensor |
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 |