CN109460789B - Soil moisture fusion method based on Bayes maximum entropy - Google Patents
Soil moisture fusion method based on Bayes maximum entropy Download PDFInfo
- Publication number
- CN109460789B CN109460789B CN201811316588.6A CN201811316588A CN109460789B CN 109460789 B CN109460789 B CN 109460789B CN 201811316588 A CN201811316588 A CN 201811316588A CN 109460789 B CN109460789 B CN 109460789B
- Authority
- CN
- China
- Prior art keywords
- soil moisture
- data
- soil
- vegetation index
- fusion
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/251—Fusion techniques of input or preprocessed data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/217—Validation; Performance evaluation; Active pattern learning techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/02—Agriculture; Fishing; Mining
Abstract
The invention relates to a Bayesian maximum entropy-based soil moisture fusion method, which fully utilizes soil moisture obtained by manual soil borrowing, soil moisture (EC-5) monitored by a sensor in real time and soil moisture information inverted by hyperspectral, applies a data fusion technology to fuse soil moisture data from three different sources so as to obtain the distribution condition of the soil moisture in a regional scale, eliminates possible redundancy and contradiction between information, reduces uncertainty of irrigation decision information and fuzzy degree of decision reasoning, improves irrigation decision precision, provides decision irrigation suggestions and suggestions for government drought-resistant work, and guides water management departments in irrigation areas to scientifically allocate water resources.
Description
Technical Field
The invention relates to a soil moisture fusion method based on Bayes maximum entropy.
Background
At present, water resource shortage is a worldwide problem, and safe, efficient and reasonable utilization of water resources is a focus of global attention.
Accurate irrigation is realized, information is obtained quickly and accurately, and timely and scientific irrigation decision making is the core. In recent years, with the development of fine, intensive, information-based agriculture and digital earth, various high-tech and high-precision sensors are widely applied to the field of agriculture, so that the information sources of irrigation decisions are wider and richer, and the expression forms of irrigation decision information are more diversified. Because each sensor has respective characteristics, usually, one sensor can only reflect the appropriate degree of the moisture of crops from a certain aspect or in a certain range, only local or one-sided information is obtained, and the information quantity is limited; meanwhile, due to the influence of manual operation, external interference, noise and the like, information obtained by a single sensor has certain uncertainty and ambiguity, and if irrigation decision is made by information provided by a certain sensor or experience of an expert alone, the existing information is caused to be bedridden, a large amount of information is wasted, and irrigation precision is influenced.
The development of the information fusion technology provides possibility for multi-index irrigation decision. The multi-source information fusion technology is a data processing technology which utilizes a computer technology to comprehensively analyze and process sensor monitoring information from multiple sources under a certain criterion so as to obtain valuable information which cannot be obtained by a single information source and finally finish the aim. Because various irrigation information collected by the sensors are different in space, time and expression modes, different in reliability and uncertainty, different in emphasis and application, and the risk of irrigation decision can be increased if the irrigation information is not screened. Therefore, in order to eliminate redundancy and contradiction possibly existing among information, reduce uncertainty of irrigation decision information and fuzzy degree of decision reasoning and improve accuracy of irrigation decision, a multi-source information fusion technology is introduced into the field of farmland irrigation, an information fusion method suitable for the irrigation field is researched, some difficulties and key problems existing in the current irrigation decision process are solved, the utilization rate of irrigation information and the capability of correct decision in a changeable environment are improved, and the method has certain practical significance for realizing increment of irrigation information and establishing a new-generation irrigation management mode which is efficient and convenient.
Disclosure of Invention
The invention provides a soil moisture fusion method based on Bayesian maximum entropy aiming at the fact that information obtained by a single sensor has certain uncertainty and ambiguity. The method fully utilizes soil moisture obtained by manual soil sampling, soil moisture (EC-5) monitored by a sensor in real time and soil moisture information inverted by hyperspectral technology, and applies a data fusion technology to fuse soil moisture data of three different sources so as to obtain the distribution condition of the soil moisture in a regional scale, provide irrigation decision suggestions and suggestions for government drought resistance work, and guide water management departments in irrigation areas to carry out water resource scientific allocation.
In order to solve the technical problems, the technical scheme adopted by the invention is as follows:
a soil moisture fusion method based on Bayes maximum entropy comprises the following specific steps:
the method comprises the following steps: collecting soil moisture information once every 7 days in the winter wheat growth period to obtain soil moisture content, wherein the sampling depth is 0-20cm, 20-40 cm, 40-60 cm, 60-80 cm and 80-100 cm;
step two: laying soil moisture sensors by adopting a land statistical analysis method, and monitoring the change condition of soil moisture in real time, wherein the laying depth is 0-20cm, 20-40 cm, 40-60 cm, 60-80 cm and 80-100 cm respectively;
step three: collecting canopy hyperspectral data from the elongation stage to the maturity stage of the winter wheat by using a surface feature spectrometer, and constructing a hyperspectral-based soil moisture inversion model by using a vegetation index as an independent variable and soil moisture as a dependent variable;
step four: fusing the soil moisture data obtained in the first step, the second step and the third step by adopting a Bayesian maximum entropy theory;
step five: and carrying out uncertainty evaluation on the fusion result.
Further, a preferred embodiment of the present invention is: the second step comprises the following steps: a statistical analysis method is adopted to carry out optimal layout of soil moisture content monitoring points, the method is based on a half variance function and an interpolation method, the reasonable spatial distribution positions of the soil moisture content monitoring points are determined, and soil moisture sensors are laid according to the determined positions.
Further, a preferred embodiment of the present invention is: the third step comprises:
1) preprocessing the collected hyperspectral data by using ViewSpec software, and calculating a vegetation index formed by combining original spectrums of any wave band by using formulas (1) to (8), wherein OSAVI is used for adjusting the vegetation index for optimizing soil, EVI2 is an enhanced vegetation index II, NDVI is a normalized vegetation index, mNDVI is a corrected normalized difference vegetation index, MSRI is a corrected simple ratio vegetation index, RVI is a ratio vegetation index, NVI is a novel vegetation index, and WI is a moisture vegetation index;
OSAVI＝1.16×(R800-R670)/(R800+R670+0.16) (1)
EVI2＝2.5×(R800-R660)/(1+R800+2.4×R660) (2)
NDVI＝(R810-R680)/(R810+R680) (3)
mNDVI＝(R750-R705)/(R750+R705-2R445) (4)
MSRI＝(R750-R445)/(R750+R445) (5)
RVI＝R810/R680 (6)
NVI＝(R777-R747)/R673 (7)
WI＝R900/R970 (8)
2) analyzing 8 the correlation between the vegetation index and the soil moisture, screening out a vegetation index with a higher correlation coefficient with the soil moisture, and performing inversion on the soil moisture;
3) and establishing a hyperspectral winter wheat soil moisture estimation model by using the screened vegetation index as an independent variable x and soil moisture of a 1m soil layer as a dependent variable y.
Further, a preferred embodiment of the present invention is: the winter wheat soil moisture estimation model adopts a leave-one-out cross verification method to test the simulation precision of the model.
Further, a preferred embodiment of the present invention is: the fourth step comprises:
1) the Bayesian maximum entropy is based on actually measured hard data (soil moisture data obtained by soil sampling), soft data with certain error and uncertainty and data from other prior information and other various sources are blended to analyze the variation trend of the soil moisture, the soft data comprises three soft data forms,
firstly, soil moisture ET-5 is monitored by a sensor in real time;
secondly, converting soil moisture obtained by hyperspectral inversion into interval soft data, wherein the conversion process is to calculate and obtain a simulated soil moisture value according to an established hyperspectral vegetation index soil moisture inversion model, establish a linear relation between the simulated soil moisture value and the interval soft data by taking actual measured soil moisture data as a dependent variable and inverted soil moisture as an independent variable, and calculate a standard variance of a linear fitting equation residual error
Calculating a soil moisture fitting value through linear regression, converting the inverted soil moisture data into interval data, and respectively calculating an upper limit value SM and a lower limit value SM of the interval data by adopting a formula (9) and a formula (10)uAnd SML，
SMu＝SMestimation-σε (9)
SML＝SMestimation+σε (10)
In formulae (9) and (10), SMuAnd SMLRespectively upper and lower limits of interval data, SMestimationFor estimated soil moisture value, σεIs the square root of the standard deviation of the residual.
Thirdly, converting soil moisture subjected to hyperspectral inversion into Gaussian distribution soft data, namely converting the soil moisture into corresponding mean values and variances;
2) fusing by using a Bayes maximum entropy basic principle, constructing a covariance model, and calculating by using a formula (11) and a formula (12);
in the formulae (11) and (12), C (r) is covariance, N is the number of covariance structures, and c0iThe variance contribution rate of the ith covariance Structure, cri(r) represents a spatial function, the fusion process uses an exponential function, ariIs a spatial range;
3) three different combination forms are adopted for fusion,
firstly, actually measured soil moisture data is used as hard data, and EC-5 sensor monitoring data is used as soft data to be fused and recorded as BME1 (soil sampling + EC-5);
secondly, the actually measured soil moisture data is used as hard data, the EC-5 sensor monitoring data is used as soft data, the soil moisture data inverted by the vegetation index is converted into interval soft data, and the interval soft data is recorded as BME2 (soil sampling + EC-5+ inverted soil moisture is converted into interval soft data);
and thirdly, taking actually measured soil moisture data as hard data, taking EC-5 sensor monitoring data as soft data, taking vegetation index inverted soil moisture data as probability soft data of Gaussian distribution, and recording as BME3 (Gaussian distribution probability soft data converted from soil and EC-5+ inverted soil moisture).
Further, a preferred embodiment of the present invention is: the fifth step comprises the following steps:
and performing uncertainty evaluation on the quality of the fusion result, wherein the evaluation process adopts a mode of modeling by adopting one group of data and verifying the other group of data, and the method specifically comprises the steps of selecting m samples for modeling and verifying the rest (n-m) samples on the assumption that a total of n samples exist. The fusion results were evaluated using the following 5 evaluation indexes (formula 13 to formula 17) in the claims, which are the standard square error (SME), the mean of absolute error (ABSMPE), the mean square error (AVAR), the Root Mean Square Error (RMSE), and the Root Mean Square of Standard Error (RMSSE), respectively.
In the above 5 formulae, xiAnd yiRespectively the measured value and the predicted value of the soil moisture, sigma2Is the variance, and n is the number of samples.
Further, a preferred embodiment of the present invention is: the method for obtaining the water content of the soil is a soil-taking and drying method.
The invention has the beneficial effects that:
the method fully utilizes soil moisture obtained by manual soil sampling, soil moisture (EC-5) monitored by a sensor in real time and soil moisture information inverted by hyperspectral technology, applies data fusion technology,
the method has the advantages that the soil moisture data from three different sources are fused to obtain the distribution condition of soil moisture in the regional scale, the redundancy and contradiction possibly existing among information are eliminated, the uncertainty of irrigation decision information and the fuzzy degree of decision reasoning are reduced, the irrigation decision precision is improved, irrigation decision suggestions and suggestions are provided for the drought resisting work of the government, and water management departments in irrigation areas are guided to carry out water resource scientific allocation.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention.
FIG. 1 is a flow chart of a soil moisture fusion method based on Bayesian maximum entropy of the present invention;
FIG. 2 is a graph of the results of fusion performed in the form of BME 1;
FIG. 3 is a graph of the results of fusion performed in the form of BME 2;
FIG. 4 is a graph showing the results of fusion in the form of BME 3.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Example 1
As shown in fig. 1, a soil moisture fusion method based on bayes maximum entropy includes the following steps:
s101, collecting soil moisture information once every 7 days in the winter wheat growth period by using a soil drill, and obtaining the soil moisture content by adopting a soil taking and drying method, wherein the sampling depths are 0-20cm, 20cm-40cm, 40cm-60cm, 60cm-80cm and 80cm-100cm respectively.
S102, laying soil moisture sensors by adopting a geostatistical analysis method, and monitoring the change condition of soil moisture in real time, wherein the laying depth is 0-20cm, 20cm-40cm, 40cm-60cm, 60cm-80cm and 80cm-100cm respectively. The geostatistical analysis method is based on a regionalized variable theory and takes a variation function as a main tool to research the distribution condition of soil moisture in space.
S103, collecting canopy spectrum data from the jointing stage to the milk stage of the winter wheat by using a Field Spec Handheld ground feature spectrometer (product of ASD company, spectrum range 350 nm-1075 nm, resolution 3.5nm, sampling interval 1.6nm and viewing angle 25 degrees). The spectrum acquisition time was arranged at 10: 00-12: 00, selecting weather which is clear and cloudless and has wind power less than 3 grade; the worker wears dark clothing and stands behind the target area facing the sun; the sensor adopts a 25-degree field angle probe, is arranged at a position 1.0m above the canopy and is vertical to the canopy surface; the measurement is repeated 15 times for each treatment, and the instrument is corrected once by using a reference plate every 0.5 h;
s104, preprocessing the acquired spectrum data by using ViewSpec software, outputting the result to an Excel document, and calculating the vegetation index formed by combining the original spectra of any wave band by adopting formulas (1) to (8).
OSAVI＝1.16×(R800-R670)/(R800+R670+0.16) (1)
EVI2＝2.5×(R800-R660)/(1+R800+2.4×R660) (2)
NDVI＝(R810-R680)/(R810+R680) (3)
mNDVI＝(R750-R705)/(R750+R705-2R445) (4)
MSRI＝(R750-R445)/(R750+R445) (5)
RVI＝R810/R680 (6)
NVI＝(R777-R747)/R673 (7)
WI＝R900/R970 (8)
The correlation between the 8 vegetation index and the soil moisture of 1m soil layer (data obtained by artificial soil collection) (n ═ 135) was analyzed, and the results are shown in table 1.
TABLE 1 correlation coefficient of vegetation index and soil moisture
Table 1 Correlation coefficient between vegetation index and soil moisture
Note:**and*indicating significance at the 0.01 and 0.05 levels, respectively.
A hyperspectral winter wheat soil moisture estimation model (a linear model and a polynomial model) is established by using the screened vegetation index NVI as an independent variable x and soil moisture of a 1m soil layer as a dependent variable y. In order to ensure the validity of the built model, eliminate the influence of random factors on the result and ensure that the verification process can be completely repeated, a leave-one-out cross verification method is adopted to verify the simulation precision of the model. To measure the performance of the model, the Mean Relative Error (MRE), the Root Mean Square Error (RMSE), and the coefficient of determination (R) were used2) And carrying out statistical test on the estimation model.
In the formula, xiThe simulation value of soil moisture is obtained; y isiThe measured value of the soil moisture is obtained; i is a sample number, i ═ 1, 2.., n;
respectively are the mean values of the soil moisture simulation value and the measured value; n is the number of samples.
As shown in table 2, the average relative error MRE and the root mean square error RMSE of the constructed linear model were 16.73% and 0.0478, respectively, which were lower than those of the polynomial model (MRE ═ 19.28, RMSE ═ 0.0525). Therefore, a linear model was chosen for soil moisture simulation.
TABLE 2 fitting and validation of winter wheat soil moisture estimation model
Table 2 Fitting and validation of soil moisture estimation model of winter wheat
S105, establishing a hyperspectral winter wheat soil moisture estimation model (linear model) by using winter wheat monitoring data of 2014-2016 two years, using the vegetation index NVI as an independent variable x and the soil moisture as a dependent variable y, and verifying by using actual measurement data of 2016-2017 years, wherein the obtained soil moisture inversion model is shown as a formula (12).
y＝0.0373x+0.1264 (12)
According to verification, the average relative error MRE and the root mean square error RMSE of the constructed hyperspectral-based soil moisture estimation model are respectively 18.26% and 0.0466.
Determining the expression form of the soft data and the hard data, and fusing the soil moisture by adopting Bayes maximum entropy. The Bayesian maximum entropy is based on actually measured hard data (soil moisture data obtained by soil sampling), and data from various sources such as 'soft data' with certain error and uncertainty and other prior information are blended to analyze the variation trend of the soil moisture. Wherein the hard data are soil moisture data obtained by manually taking soil, and the soft data comprise three forms;
the first soft data form is that the soil moisture ET-5 monitored by the sensor in real time is obtained by the laid soil moisture sensor in real time;
s106, converting soil moisture obtained by hyperspectral inversion into interval soft data in a second soft data form, obtaining a simulated soil moisture value according to the established hyperspectral vegetation index soil moisture inversion model, establishing a linear relation between the simulated soil moisture value and the simulated soil moisture value by taking actual measured soil moisture data as a dependent variable and inverted soil moisture as an independent variable, and calculating the standard variance of the residual error of a linear fitting equation
Calculating a soil moisture fitting value through linear regression, converting the inverted soil moisture data into interval data, and respectively calculating an upper limit value SM and a lower limit value SM of the interval data by adopting a formula (13) and a formula (14)uAnd SML。
SMu＝SMestimation-σε (13)
SML＝SMestimation+σε (14)
In equations (13) and (14), SMuAnd SMLRespectively upper and lower limits of interval data, SMestimationFor estimated soil moisture value, σεIs the square root of the standard deviation of the residual.
S106, converting the soil moisture subjected to hyperspectral inversion into Gaussian distribution soft data, namely converting the soil moisture into corresponding mean values and variances;
s107, fusing by using the basic principle of Bayes maximum entropy, constructing a covariance model, and calculating by using a formula (15) and a formula (16);
in equations (15) and (16), C (r) is covariance, N is the number of covariance structures, and c0iThe variance contribution rate of the ith covariance Structure, cri(r) represents a spatial function, the fusion process uses an exponential function, ariIs a spatial range.
S108, fusing three different combination forms, namely, fusing the actually measured soil moisture data as hard data and the monitoring data of the EC-5 sensor as soft data, and recording as BME1 (soil sampling + EC-5); secondly, the actually measured soil moisture data is used as hard data, the EC-5 sensor monitoring data is used as soft data, the soil moisture data inverted by the vegetation index is converted into interval soft data, and the interval soft data is recorded as BME2 (soil sampling + EC-5+ inverted soil moisture is converted into interval soft data); and thirdly, taking actually measured soil moisture data as hard data, taking EC-5 sensor monitoring data as soft data, taking vegetation index inverted soil moisture data as probability soft data of Gaussian distribution, and recording as BME3 (Gaussian distribution probability soft data converted from soil and EC-5+ inverted soil moisture).
S109, the fusion result is shown in figures 2-4, and it can be seen from figures 2-4 that the variation ranges of the standard deviation are respectively 0.003-0.02, 0.06-0.4 and 0.0007-0.0021, and compared with BME1 and BME2, the standard deviation after BME3 fusion is the smallest, which shows that the fusion is performed by Gaussian distribution probability soft data converted from soil sampling + EC-5+ inverted soil moisture, and the effect is the best. Meanwhile, the change ranges of the average values obtained by the three fusion modes are-0.23-1.0, 0.11-0.35 and 0.18-0.25 respectively, and compared with measured values, the fluctuation of the soil moisture predicted value obtained by adopting a BME1 combination mode is larger, the combination mode of BME2 and BME3 is adopted for fusion, the fusion result is closer to the actual measurement, and the further explanation is that soft data with certain uncertainty and ambiguity are involved in the fusion process, so that the fusion result is more reasonable while the number of samples is increased. By comparing the mean values of BME2 and BME3, the BME3 is found to have better fusion effect than BME2, which shows that the soft data forms participating in the fusion are different, and the fusion effect is also influenced, namely, the inverted soil moisture is converted into a probability soft data form of Gaussian distribution, and the fusion result obtained by converting the inversion into interval soft data is more similar to the actual measurement result.
And S110, evaluating the uncertainty of the fusion result, wherein the evaluation indexes are standard square error (SME), absolute error mean value (ABSMPE), mean square error (AVAR), Root Mean Square Error (RMSE) and root mean square error (RMSSE) of the standard error.
In the above 5 formulae, xiAnd yiRespectively the measured value and the predicted value of the soil moisture, sigma2Is the variance, and n is the number of samples. The fusion results are shown in table 3.
TABLE 3 uncertainty evaluation of fusion results
Table 3 Uncertainty evaluation of the fusion results
By comparing the three different fusion forms, compared with the evaluation results of BME1 and BME2, the ABSMPE value obtained by adopting the BME3 fusion form is closer to 0, AVAR and RMSE are minimum, and RMSSE is closer to 1, which shows that the fusion effect of BME3 is closer to the actual value, and the fusion effect is optimal. Comparing the evaluation results of BME2 and BME3 with soft data, it was found that ABSMPE, AVAR, and RMSE were all greater for BME2 than for BME3, and that RMSSE for BME2 deviated from 1. Therefore, it can be concluded that the soil moisture data of the hyperspectral inversion is converted into an interval soft data form, and the fusion effect of the soil moisture data is not ideal for converting the soil moisture data into a soft data form of Gaussian distribution. Comparing the results of the evaluation of BME1 and BME2 and BME3 fused with soft data, BME2 and BME3 found that ABSMPE, RMSE, RMSSE and SME were all smaller in absolute value than BME1, further indicating that the fusion results were closer to actual values due to the addition of soft data.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (6)
1. A soil moisture fusion method based on Bayes maximum entropy is characterized by comprising the following specific steps:
the method comprises the following steps: collecting soil moisture information once every 7 days in the winter wheat growth period to obtain soil moisture content, wherein the sampling depth is 0-20cm, 20-40 cm, 40-60 cm, 60-80 cm and 80-100 cm;
step two: laying soil moisture sensors by adopting a land statistical analysis method, and monitoring the change condition of soil moisture in real time, wherein the laying depth is 0-20cm, 20-40 cm, 40-60 cm, 60-80 cm and 80-100 cm respectively;
step three: collecting canopy hyperspectral data from the elongation stage to the maturity stage of the winter wheat by using a surface feature spectrometer, and constructing a hyperspectral-based soil moisture inversion model by using a vegetation index as an independent variable and soil moisture as a dependent variable;
step four: fusing the soil moisture data obtained in the first step, the second step and the third step by adopting a Bayesian maximum entropy theory, and performing uncertainty evaluation on a fusion result;
the fourth step comprises:
1) the Bayesian maximum entropy is based on actually measured hard data, soft data with certain error and uncertainty and data from other prior information and other various sources are blended to analyze the variation trend of soil moisture, the soft data comprises three forms,
firstly, soil moisture ET-5 is monitored by a sensor in real time;
secondly, converting soil moisture obtained by hyperspectral inversion into interval soft data, wherein the conversion process is to calculate and obtain a simulated soil moisture value according to an established hyperspectral vegetation index soil moisture inversion model, establish a linear relation between the simulated soil moisture value and the interval soft data by taking actual measured soil moisture data as a dependent variable and inverted soil moisture as an independent variable, and calculate a standard variance of a linear fitting equation residual error
Calculating a soil moisture fitting value through linear regression, converting the inverted soil moisture data into interval data, and respectively calculating an upper limit value SM and a lower limit value SM of the interval data by adopting a formula (9) and a formula (10)uAnd SML，
SMu＝SMestimation-σε (9)
SML＝SMestimation+σε (10)
In formulae (9) and (10), SMuAnd SMLRespectively upper and lower limits of interval data, SMestimationFor estimated soil moisture value, σεIs the square root of the standard deviation of the residual;
thirdly, converting soil moisture subjected to hyperspectral inversion into Gaussian distribution soft data, namely converting the soil moisture into corresponding mean values and variances;
2) fusing by using a Bayes maximum entropy basic principle, constructing a covariance model, and calculating by using a formula (11) and a formula (12);
in the formulae (11) and (12), C (r) is covariance, N is the number of covariance structures, and c0iThe variance contribution rate of the ith covariance Structure, cri(r) represents a spatial function, the fusion process uses an exponential function, ariIs a spatial range;
3) three different combination forms are adopted for fusion,
firstly, actually measured soil moisture data is used as hard data, EC-5 sensor monitoring data is used as soft data to be fused and recorded as BME1, and soil is taken and EC-5 is added;
secondly, the actually measured soil moisture data is used as hard data, the EC-5 sensor monitoring data is used as soft data, the soil moisture data inverted by the vegetation index is converted into interval soft data which is recorded as BME2, and the interval soft data is converted from soil and EC-5+ inverted soil moisture;
and thirdly, taking actually measured soil moisture data as hard data, taking EC-5 sensor monitoring data as soft data, taking vegetation index inverted soil moisture data as Gaussian distribution probability soft data, recording the probability soft data as BME3, and converting soil moisture inverted by soil + EC-5+ into Gaussian distribution probability soft data.
2. The Bayesian maximum entropy-based soil moisture fusion method according to claim 1, characterized in that: the second step comprises the following steps: a statistical analysis method is adopted to carry out optimal layout of soil moisture content monitoring points, the method is based on a half variance function and an interpolation method, the reasonable spatial distribution positions of the soil moisture content monitoring points are determined, and soil moisture sensors are laid according to the determined positions.
3. The Bayesian maximum entropy-based soil moisture fusion method according to claim 1, wherein: the third step comprises:
1) preprocessing the collected hyperspectral data by using ViewSpec software, and calculating a vegetation index formed by combining original spectrums of any wave band by using formulas (1) to (8), wherein OSAVI is used for adjusting the vegetation index for optimizing soil, EVI2 is an enhanced vegetation index II, NDVI is a normalized vegetation index, mNDVI is a corrected normalized difference vegetation index, MSRI is a corrected simple ratio vegetation index, RVI is a ratio vegetation index, NVI is a novel vegetation index, and WI is a moisture vegetation index;
OSAVI＝1.16×(R800-R670)/(R800+R670+0.16) (1)，
EVI2＝2.5×(R800-R660)/(1+R800+2.4×R660) (2)，
NDVI＝(R810-R680)/(R810+R680) (3)，
mNDVI＝(R750-R705)/(R750+R705-2R445) (4)，
MSRI＝(R750-R445)/(R750+R445) (5)，
RVI＝R810/R680 (6)，
NVI＝(R777-R747)/R673 (7)，
WI＝R900/R970 (8)；
2) analyzing 8 the correlation between the vegetation index and the soil moisture, screening out a vegetation index with a higher correlation coefficient with the soil moisture, and performing inversion on the soil moisture;
3) and establishing a hyperspectral winter wheat soil moisture estimation model by using the screened vegetation index as an independent variable x and soil moisture of a 1m soil layer as a dependent variable y.
4. The Bayesian maximum entropy-based soil moisture fusion method according to claim 3, wherein: the winter wheat soil moisture estimation model adopts a leave-one-out cross verification method to test the simulation precision of the model.
5. The Bayesian maximum entropy-based soil moisture fusion method according to claim 1, wherein: the fourth step comprises:
carrying out uncertainty evaluation on the quality of the fusion result, wherein the evaluation process adopts a mode of modeling by adopting one group of data and verifying the other group of data, and the method specifically comprises the steps of selecting m samples of n samples in total for modeling and verifying the rest n-m samples; evaluating the fusion result by adopting the following 5 evaluation index formulas 13 to 17, namely standard square error SME, average value ABSMPE of absolute value of error, average variance AVAR, root mean square error RMSE and root mean square RMSSE of standard error;
in the above 5 formulae, xiAnd yiRespectively the measured value and the predicted value of the soil moisture, sigma2Is the variance, and n is the number of samples.
6. The Bayesian maximum entropy-based soil moisture fusion method according to claim 1, characterized in that: the method for obtaining the water content of the soil is a soil-taking and drying method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811316588.6A CN109460789B (en) | 2018-11-07 | 2018-11-07 | Soil moisture fusion method based on Bayes maximum entropy |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811316588.6A CN109460789B (en) | 2018-11-07 | 2018-11-07 | Soil moisture fusion method based on Bayes maximum entropy |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109460789A CN109460789A (en) | 2019-03-12 |
| CN109460789B true CN109460789B (en) | 2021-07-27 |
Family
ID=65609586
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811316588.6A Active CN109460789B (en) | 2018-11-07 | 2018-11-07 | Soil moisture fusion method based on Bayes maximum entropy |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109460789B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110427995B (en) * | 2019-07-24 | 2023-11-17 | 中国科学院空天信息创新研究院 | Bayesian soil moisture estimation method based on multi-source remote sensing data |
| CN111507312A (en) * | 2020-05-26 | 2020-08-07 | 长江勘测规划设计研究有限责任公司 | Soil moisture extraction method based on hyperspectral data |
| CN112528555B (en) * | 2020-11-23 | 2023-02-21 | 中国科学院空天信息创新研究院 | Soil moisture map drawing method and device |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102737155A (en) * | 2011-04-12 | 2012-10-17 | 中国科学院寒区旱区环境与工程研究所 | Bayesian fitering-based general data assimilation method |
| CN102819673A (en) * | 2012-07-26 | 2012-12-12 | 中国农业科学院农田灌溉研究所 | Multisource irrigation information fusion method and device |
| CN104573393A (en) * | 2015-01-28 | 2015-04-29 | 北京师范大学 | Soil moisture site data upscaling method based on Bayesian theory |
| CN106446444A (en) * | 2016-10-14 | 2017-02-22 | 中国科学院遥感与数字地球研究所 | Soil moisture spatial predication research based on Bayes maximum entropy and priori knowledge |
| CN108663330A (en) * | 2018-04-19 | 2018-10-16 | 中国国土资源航空物探遥感中心 | A kind of vegetation-covered area soil copper inversion method based on blade measured spectra |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2016070195A1 (en) * | 2014-10-31 | 2016-05-06 | Purdue Research Foundation | Moisture management & perennial crop sustainability decision system |
-
2018
- 2018-11-07 CN CN201811316588.6A patent/CN109460789B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102737155A (en) * | 2011-04-12 | 2012-10-17 | 中国科学院寒区旱区环境与工程研究所 | Bayesian fitering-based general data assimilation method |
| CN102819673A (en) * | 2012-07-26 | 2012-12-12 | 中国农业科学院农田灌溉研究所 | Multisource irrigation information fusion method and device |
| CN104573393A (en) * | 2015-01-28 | 2015-04-29 | 北京师范大学 | Soil moisture site data upscaling method based on Bayesian theory |
| CN106446444A (en) * | 2016-10-14 | 2017-02-22 | 中国科学院遥感与数字地球研究所 | Soil moisture spatial predication research based on Bayes maximum entropy and priori knowledge |
| CN108663330A (en) * | 2018-04-19 | 2018-10-16 | 中国国土资源航空物探遥感中心 | A kind of vegetation-covered area soil copper inversion method based on blade measured spectra |
Non-Patent Citations (2)
| Title |
|---|
| 《基于土壤水分异质性的小麦滴灌试验小区布设方法》;王鹏翔等;《农业工程学报》;20171130;全文 * |
| 《河南省冬小麦需水量的时空变化及影响因素》;宋妮等;《应用生态学报》;20140630;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109460789A (en) | 2019-03-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109460789B (en) | Soil moisture fusion method based on Bayes maximum entropy | |
| Pascale et al. | Analysis of rainfall seasonality from observations and climate models | |
| Giustolisi et al. | Advances in data-driven analyses and modelling using EPR-MOGA | |
| Zhang et al. | Allocate soil individuals to soil classes with topsoil spectral characteristics and decision trees | |
| Baik et al. | Assessment of satellite-and reanalysis-based evapotranspiration products with two blending approaches over the complex landscapes and climates of Australia | |
| CN111665575B (en) | Medium-and-long-term rainfall grading coupling forecasting method and system based on statistical power | |
| Li et al. | Effects of spatial resolution of remotely sensed data on estimating urban impervious surfaces | |
| Esfanjani et al. | MaxEnt modeling for predicting impacts of environmental factors on the potential distribution of Artemisia aucheri and Bromus tomentellus-Festuca ovina in Iran. | |
| CN113252583B (en) | Method for calculating alpine hay coverage based on hay vegetation index | |
| Banerjee et al. | Spatial relationships between leaf area index and topographic factors in a semiarid grassland: Joint multifractal analysis | |
| Kim et al. | Uncertainties in calculating precipitation climatology in East Asia | |
| Nowosad et al. | Statistical techniques for modeling of Corylus, Alnus, and Betula pollen concentration in the air | |
| Tian et al. | Searching for an optimized single-objective function matching multiple objectives with automatic calibration of hydrological models | |
| CN114357737B (en) | Agent optimization calibration method for time-varying parameters of large-scale hydrologic model | |
| Zhao et al. | Comparison and correction of IDW based wind speed interpolation methods in urbanized Shenzhen | |
| Achite et al. | Analysis of monthly average precipitation of Wadi Ouahrane basin in Algeria by using the ITRA, ITPAM, and TPS methods | |
| Bora et al. | Regional analysis of maximum rainfall using L-moment and TL-moment: a comparative case study for the north East India | |
| Cho et al. | Profile soil property estimation using a VIS-NIR-EC-force probe | |
| Vafakhah et al. | Spatial variations of runoff generation at watershed scale | |
| Bhat et al. | Spatial Interpolation Techniques for Analysing Precipitation Trends in Over Kashmir Valley, Jammu and Kashmir India | |
| Quets et al. | Comparison of spectral and time domain calibration methods for precipitation‐discharge processes | |
| Zhang et al. | Calibration and uncertainty analysis of a hydrological model based on cuckoo search and the M-GLUE method | |
| Larabi et al. | A New Concept to Calibrate and Evaluate a Hydrological Model Based on Functional Data Analysis. | |
| Abdullahia et al. | Linear and non-linear ensemble concepts for pan evaporation modeling | |
| Tang | Reliability-based calibration of design wind load, snow load and companion load factors considering climate change effect |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |