Disclosure of Invention
In order to solve at least one technical problem in the prior art, the application provides an agricultural drought grading method based on information diffusion.
The application discloses an agricultural drought grade classification method based on information diffusion, which comprises the following steps:
step 101, acquiring remote sensing data and drought form data of a monitoring target in a preset time period, wherein the drought form data comprises position information of the monitoring target and drought grade information judged manually;
102, selecting a remote sensing index of a preset type as the drought index of the monitoring target, and calculating a corresponding drought index value according to the remote sensing data;
step 103, converting the drought shape data into fuzzy subsets by using fuzzy set theory;
step 104, forming a sample set by the drought index value of the step 102 and the fuzzy subset of the step 103;
105, constructing a fuzzy information matrix by using the sample set as a parameter and using an information diffusion method, so as to obtain a mapping relation between the drought index value and the drought grade expressed by the drought form data;
and 106, evaluating the drought grade of the monitoring target based on the mapping relation obtained in the step 105.
According to at least one embodiment of the present application, the drought level in the drought level information includes no drought, light drought, medium drought, heavy drought, and extreme drought.
According to at least one embodiment of the present application, the step 103 specifically includes:
step 1031, setting a given argument U, U to the closed interval [0,1]]Is any one of the maps mu B The method comprises the following steps:
μ B :U→[0,1],
wherein U is the argument of fuzzy set B, then B is called a fuzzy subset on U, and the value mu B (x) For x belonging to the membership of B, function μ B Is a membership function of B;
step 1032, assuming that domain U is a finite set, express a fuzzy subset B of U with zade notation:
B=μ B (x 1 )/x 1 +μ B (x 2 )/x 2 …+μ B (x n )/x n ;
wherein x corresponds to the drought grade, mu in the drought grade information B (x) Is the possibility corresponding to each drought grade.
According to at least one embodiment of the present application, in the step 102, the predetermined type of telemetry index is a temperature vegetation drought index.
According to at least one embodiment of the present application, the temperature vegetation drought index TVDI is obtained by the following formula:
wherein ,T
S Representing the surface temperature of any pixel;
representing the lowest surface temperature corresponding to the same vegetation index of a certain land; />
Representing the highest surface temperature corresponding to the same vegetation index of a certain place;
in addition, in the case of the optical fiber,
the specific calculation formula is as follows:
in the formula :a1 ,b 1 Fitting coefficients representing wet edge equations; a, a 2 ,b 2 Fitting coefficients representing the dry-edge equation; VI represents a vegetation index.
According to at least one embodiment of the present application, in step 102, before calculating the corresponding drought index value according to the remote sensing data, the method further includes the following steps:
and carrying out remote sensing image preprocessing, cloud removal processing and shadow removal processing on the remote sensing data in sequence.
According to at least one embodiment of the present application, the remote sensing image preprocessing includes at least one of the following:
radiometric calibration processing, atmospheric correction processing, orthographic correction processing, geometric correction processing and image mosaic processing.
According to at least one embodiment of the present application, in said step 103, before converting said drought morphology data into a fuzzy subset, further comprising the steps of:
and preprocessing the drought form data to remove invalid data and interference data.
In step 104, according to at least one embodiment of the present application, a sample set is constructed with geographic location as a tie, drought index values as inputs, and fuzzy subsets as outputs.
According to at least one embodiment of the present application, in step 106, the drought grade distribution map of the monitoring target is directly obtained by using the drought index value of the monitoring target based on the mapping relationship, so as to evaluate the drought grade of the monitoring target.
The application has at least the following beneficial technical effects:
aiming at the uncertainty existing in the current dynamic determination process of the drought grade of the agricultural drought remote sensing index, the method for dividing the agricultural drought grade based on information diffusion provides a dynamic drought grade determination method taking qualitative information as a reference, specifically, qualitative information of easy-to-acquire drought grade judgment is converted into a fuzzy subset by utilizing a fuzzy set theory, and then a fuzzy relation matrix between the remote sensing drought index and the agricultural drought grade is constructed by adopting an information diffusion method so as to describe the fuzzy relation between the drought index and the drought grade, thereby carrying out regional agricultural drought monitoring.
Detailed Description
In order to make the purposes, technical solutions and advantages of the implementation of the present application more clear, the technical solutions in the embodiments of the present application will be described in more detail below with reference to the accompanying drawings in the embodiments of the present application. The described embodiments are some, but not all, of the embodiments of the present application. The embodiments described below by referring to the drawings are exemplary and intended for the purpose of explaining the present application and are not to be construed as limiting the present application.
Example 1
Taking the example of monitoring agricultural drought in the third northeast province in 8 th month of 2009 (application area is shown in fig. 3).
Specifically, referring to fig. 1, the method for classifying agricultural drought grades based on information diffusion in this embodiment includes the following steps:
step 101, obtaining MODIS remote sensing data and drought state data of a monitoring target (namely crops in the region of three provinces in northeast) in a preset time period (namely 8 th and the last ten days in 2009), wherein the drought state data comprises position information of the monitoring target and drought grade information of human judgment (such as judgment of a planter and an agricultural field expert).
The type of the remote sensing data can be selected appropriately according to the needs, and in the embodiment, MODIS remote sensing data are adopted; likewise, a suitable drought index, in this embodiment a TVDI index, is selected according to the specific application scenario (which may be determined according to the monitoring objective), such as the monitored area, the crop growth period, the monitored spatio-temporal scale, the availability of data, etc. In addition, in the present embodiment, the drought level in the drought level information includes no drought, light drought, medium drought, heavy drought, and extreme drought.
Step 102, selecting a remote sensing index of a preset type as a drought index of a monitoring target, and calculating a corresponding drought index value according to MODIS remote sensing data.
It should be noted that, in order to obtain the agricultural drought condition of the area, the agricultural drought index may be calculated by using remote sensing data; the drought index may be, for example, ATI, PDI, MPDI, TVDI, TCI, VCI, etc., and may be specifically selected according to the monitoring target; in this embodiment, since the observation time is 8 late in 2009, the crop is in the middle and late stage of growth, the vegetation coverage is high, and the Temperature Vegetation Drought Index (TVDI) is preferentially selected as the drought monitoring index in the case where the vegetation index VI and the land surface temperature LST are available.
In addition, in this embodiment, before calculating the corresponding drought index value, preprocessing is further required to be performed on the MODIS remote sensing data, which specifically includes sequentially performing quality control processes such as remote sensing image preprocessing, cloud removal processing, shadow removal processing, and the like, where the remote sensing image preprocessing includes at least one of radiation calibration processing, atmospheric correction processing, orthographic correction processing, geometric correction processing, and image mosaic processing.
Further, in this embodiment, the temperature vegetation drought index TVDI is obtained by the following formula:
wherein ,T
S Representing the surface temperature of any pixel;
representing the lowest surface temperature corresponding to the same vegetation index of a certain land; />
Representing the highest surface temperature corresponding to the same vegetation index of a certain place;
in addition, in the case of the optical fiber,
the specific calculation formula is as follows:
in the formula :a1 ,b 1 Fitting coefficients representing wet edge equations; a, a 2 ,b 2 Fitting coefficients representing the dry-edge equation; VI represents a vegetation index; the vegetation index VI may be calculated using the reflectance data, where EVI may also be selected as the vegetation index. The larger the TVDI, the more arid.
It should be further noted that, because the grading standards of TVDI are many and are confusing, and the grading standards of TVDI in different areas and different times have great differences, the drought difference between the same research area and the evaluation of different TVDI grading standards is great, so it is difficult to determine a reasonable TVDI drought grade. For this purpose, the ground data should be used as a reference, and the drought level should be dynamically determined by specific analysis. In consideration of time and labor consumption and relative lack of data when measuring the relative humidity of soil by using an instrument, in the subsequent steps, the observed drought form data of farmlands and crops are used as ground reference data, are converted into quantitative information in a certain mode, and are used as output parameters of a model to establish a mapping relation between drought indexes and drought grades, so that establishment from the drought indexes to the drought grades is completed.
Step 103, converting drought shape data into fuzzy subsets by utilizing fuzzy set theory.
It should be noted that, drought state data is the drought degree of crops which can be judged by growers and agricultural field experts through experience, and meanwhile, the position information of the crops is recorded, and the GB_T 32136-2015 "agricultural drought grade" divides drought conditions into 4 grades: the drought shape index of farmland and crops with various drought grades is described in detail. However, some interference factors inevitably exist in the drought state information of the ground farmland and crops provided by the growers and the experts, and data preprocessing is needed before the information is utilized, mainly including that invalid values are removed through screening and interference data are removed by combining corresponding remote sensing index values, so that the reliability of samples is guaranteed to the greatest extent possible.
Further, the filtered drought morphology information of the farmland and crops is converted into a fuzzy subset.
According to the cognition law of human beings, the cognition and judgment of the brain on a certain object are fuzzy, for example, the 'withering in daytime due to drought leaves' is defined as medium drought, the 'dead seedlings, withering of leaves and falling off of fruits due to drought' corresponds to heavy drought, and the 'withering' or the 'withering' is difficult to absolutely judge aiming at a certain phenomenon in the actual judging process. And more climates are a comprehensive assessment of drought conditions in a certain crop population rather than in individual plants, which makes the situation more complex. Often humans use the word "more prone to … …" or the like to express views, more specifically, knowledge of drought patterns in farms and crops over a certain spatial range can be represented by figure 2. At this time, the possibility belonging to no drought, light drought, medium drought, heavy drought and extreme drought is considered to be 0, 0.5, 0.8, 0.7, 0.2, respectively, with the highest possibility of medium drought. Thus, the expression can acquire the cognitive judgment information of drought forms of farmlands and crops as much as possible.
Specifically, the steps of converting drought morphology information into fuzzy subsets using fuzzy set theory are as follows:
setting χ if x ε B
B (x) =1, if
Then χ is
B (x) =0, i.e.:
the absolute membership in the classical set theory is activated, i.e. the membership of an element to a set is no longer this is the case, but only 0 or 1, but any value between 0,1 can be taken.
Let U be the argument U, U to the closed interval [0,1]]Is any one of the maps mu B :
A fuzzy subset B of U may be determined.
The fuzzy subset is generally abbreviated as fuzzy set, if U is the argument of fuzzy set B, then B is called as a fuzzy subset on U, B is abbreviated as fuzzy set of U, and the value mu B (x) For x belonging to the membership of B, function μ B Is a membership function of B.
If the universe U is a finite set, the zade notation can be used to express:
B=μ B (x 1 )/x 1 +μ B (x 2 )/x 2 …+μ B (x n )/x n (5);
the zade notation of the above formula is not a partial sum, but is a notation. The denominator is an element of the domain U (namely, the drought grade corresponding to the drought grade information), and the numerator is the membership degree corresponding to the element (the possibility corresponding to each drought grade). When the membership degree of an element is 0, the corresponding item may not be written.
For example, agricultural drought conditions are classified into five according to drought disaster observation record dataThe following grades: no drought, light drought, moderate drought, heavy drought and extreme drought, respectively designated as upsilon 1 ,υ 2 ,υ 3 ,υ 4 ,υ 5 (i.e., corresponding to x in the above formula (5)).
Step 104, the drought index value of step 102 and the fuzzy subset of step 103 are combined to form a sample set.
Specifically, the step takes the geographic position as a tie, takes the drought index value as input and takes the fuzzy subset as output to construct a sample set, and the obtained sample set is shown in table 1:
TABLE 1 parameters for constructing two-dimensional information diffusion model with qualitative information as reference
Taking sample number 36 in the table as an example, the TVDI value corresponding to the sample number is 0.90, and the fuzzy subset generated by the cognitive judgment of the drought state of the farmland and the crops can be expressed as follows:
J=0.0/υ 1 +0.0/υ 2 +0.1/υ 3 +0.6/υ 4 +0.4/υ 5 。
and 105, constructing a fuzzy information matrix by using the sample set as a parameter and using an information diffusion method, so as to obtain a mapping relation between the drought index value and the drought grade expressed by the drought form data.
When the information diffusion technology is applied to perform actual calculation, the construction of the monitoring interval (also called a monitoring point sequence) Z is generally based on the original sample data X itself (i.e. TVDI value), and the construction process is as shown in formula (6):
in the formula ,vi Represents the ith monitoring point in the sequence Z, and k representsShowing the number of monitoring points to be constructed in Z, wherein Deltaz represents the step length of the monitoring point sequence.
The estimation function of normal information diffusion is as follows:
where n represents the sample volume and h represents the diffusion coefficient. The diffusion coefficient h is calculated by the formula (8),
when the sample data is estimated by using the information diffusion technique, the following steps are followed:
1) Constructing a monitoring point sequence Z by using sample data X according to a relation formula (6);
2) Calculating a diffusion coefficient h according to the relation (8);
3) According to the relation (7), the monitoring point sequence Z in Z
i Substituting information diffusion function as value of variable x
Calculating to obtain information quantity p (z) of sample data diffused to each monitoring point
i );
4) Normalizing the information quantity sequences of all the monitoring points according to the following formula (9) to obtain a fuzzy relation matrix;
5) Substituting the following gravity center formula (10), calculating the gravity center value of the monitoring point as an estimated value z' of the sample data:
wherein z' represents the calculated gravity center value of the monitoring point, z i Represents the ith monitoring point, p (z i ) ' represents the information amount of the i-th monitoring point after normalization.
In this embodiment, an information diffusion method is used to construct an input-output relationship between the remote sensing drought index TVDI and the drought level, and a monitoring space is constructed as follows (where U corresponds to the monitoring point sequence Z, U in the above formula (6) 1 –u 73 I.e. to the monitoring point Z in the above mentioned Z 1 -z k ):
U={u 1 ,u 2 ,u 3 ,u 4 ,…,u 71 ,u 72 ,u 73 }={0.23,0.24,0.25,0.26,…,0.93,0.94,0.95}
Wherein, the drought class set is expressed as:
V={v 1 ,v 2 ,v 3 ,v 4 ,v 5 "Dry land
And then generating a fuzzy relation matrix based on an information diffusion method by utilizing a fuzzy subset generated by cognition judgment of farmland and crop drought forms in the sample:
and 106, directly obtaining a drought grade distribution map of the monitoring target by using the drought index value of the monitoring target based on the mapping relation obtained in the step 105, so as to evaluate the drought grade of the monitoring target.
Specifically, a fuzzy relation between TVDI and drought grade can be obtained according to the fuzzy relation matrix. And (3) performing approximate reasoning by using the fuzzy relation matrix, and inverting by using the agricultural drought remote sensing monitoring index to obtain a corresponding agricultural drought grade, wherein a measure space between the two can be established according to the corresponding agricultural drought grade. The TVDI of fig. 4 was approximated using the fuzzy relationship between TVDI and drought rating, and the results are shown in fig. 5.
As can be seen from fig. 4, the drought is mainly distributed in western areas of the Liaoning, jilin and Heilongjiang provinces, especially western areas of the Liaoning and Jilin provinces, and the drought grade distribution obtained from the data is relatively similar, which indicates that if the cognitive judgment on the drought forms of farmlands and crops can be acquired under the condition of lack of the site-based soil relative humidity data, the obtained drought grade distribution is relatively more in accordance with the actual drought grade distribution based on the information diffusion technology. It is worth noting that when drought level monitoring is performed on drought degree of a certain space-time scale based on qualitative information, cognitive judgment information on drought forms of farmlands and crops needs to be related to all levels appearing in the period as much as possible.
The foregoing is merely specific embodiments of the present application, but the scope of the present application is not limited thereto, and any changes or substitutions easily conceivable by those skilled in the art within the technical scope of the present application should be covered in the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.