CN-119964033-B - Moisture prediction method and system based on airborne multispectral sensitive wave band combination

CN119964033BCN 119964033 BCN119964033 BCN 119964033BCN-119964033-B

Abstract

The invention discloses a moisture prediction method and a system based on airborne multispectral sensitive wave band combination, which relate to the technical field of agricultural remote sensing and image processing and comprise the following steps of determining the bandwidth of all center wavelengths based on the response characteristics of soil and plant canopy reflection spectrum to moisture, combining the physical mechanism of O-H bond vibration in water molecules and considering the influence of atmospheric vapor on near infrared spectrum; the method comprises the steps of generating a reflection spectrum of soil and vegetation canopy, generating a spectral response function by utilizing the relation between central wavelength and bandwidth, converting an analog spectrum into a moisture sensitive wave band, calculating the reflectivity and spectral index, collecting the reflection spectrum images of the soil and vegetation canopy at the selected position, inputting the reflection spectrum images into a moisture prediction model, and predicting moisture. The invention can effectively solve the problems of insufficient band selection pertinence and low monitoring precision of the existing airborne multispectral sensor when the existing airborne multispectral sensor executes the water monitoring task by optimizing the band selection and the bandwidth allocation.

Inventors

ZHANG ZHITAO
LIU YANFU
CHEN JUNYING
SUN SHIKUN
BIAN JIANG
YANG XIAOFEI
LIU HAO
LU XIAOHAN

Assignees

西北农林科技大学

Dates

Publication Date: 20260512
Application Date: 20250108

Claims (10)

1. The moisture prediction method based on the airborne multispectral sensitive wave band combination is characterized by comprising the following steps of: Based on the response characteristics of soil and plant canopy reflection spectrum to moisture, selecting the center wavelengths of a plurality of moisture sensitive wave bands, increasing and selecting the center wavelengths of a plurality of different moisture sensitive wave bands according to the physical mechanism of O-H bond vibration in water molecules, and determining the bandwidths of all the center wavelengths; simulating to generate reflection spectrums of soil and vegetation canopy, generating a spectral response function by utilizing the relation between the central wavelengths and bandwidths of the plurality of different moisture sensitive wave bands, converting the simulated reflection spectrums into the plurality of moisture sensitive wave bands by the spectral response function, obtaining the reflectivities and the spectral indexes of the plurality of moisture sensitive wave bands, and constructing a moisture prediction model by utilizing the reflectivities and the spectral indexes; And combining all moisture sensitive wave bands with determined bandwidths, integrating the combined moisture sensitive wave bands onto an airborne multispectral sensor, collecting reflection spectrum images of soil and vegetation canopy at the selected position, inputting the reflection spectrum images into a moisture prediction model, and predicting the moisture of the soil and vegetation canopy at the selected position.
2. The method for predicting moisture based on airborne multispectral sensitive band combination of claim 1, wherein the simulating to generate the reflectance spectrum of soil and vegetation canopy comprises: using MARMIT2 radiation transmission model to simulate the reflection spectrum of soil in the solar domain, and normalizing the soil moisture content SMC of the simulated soil reflection spectrum as dimensionless soil factor psoil; taking the soil reflection spectrum with the lowest water content and the highest water content as the input of a PROSAIL radiation transmission model, and simulating the reflection spectrum of a vegetation canopy; After the MARMIT radiation transmission model simulates the reflection spectrum of soil in the solar domain, a sigmoid function is used for fitting and representing the soil moisture content SMC of the soil reflection spectrum according to the product of the thickness of an input water layer and the water coverage rate of the surface, and the specific expression is as follows: BRF Soil (λ)＝f MARMIT2 (λ,L,ε,δ); Wherein BRF Soil (lambda) is the soil reflectivity, the soil moisture content SMC of the characteristic soil reflection spectrum, f MARMIT2 (·) is MARMIT2 radiation transmission model operation function, lambda is the wavelength, L is the water layer thickness, epsilon is the surface water coverage rate, delta is the soil particle volume fraction; In the PROSAIL radiation transmission model, the equivalent water thickness is obtained, and the equivalent water thickness and psoil are used for representing the equivalent water thickness of a vegetation canopy and the normalized soil moisture content, wherein the specific expression is as follows: CEWT=LAI×C w ; DHR Leaf (λ)＝f PROSPECT (λ,θ DHR ); BRF Canopy (λ)＝f SAIL (λ,DHR(λ),θ BRF ); Wherein CEWT is the canopy equivalent water thickness, LAI is the leaf area index, C w is the equivalent water thickness, DHR Leaf (λ) is the directional hemispherical reflectivity of the leaf when the wavelength simulated by f PROSPECT using PROSPECT model is λ, θ DHR is the input parameter of PROSPECT model, BRF Canopy (λ) is the input parameter of 4SAIL model when DHR Leaf (λ) is coupled by f SAIL , and θ BRF is the input parameter of 4SAIL model when the wavelength is λ.
3. The method for predicting moisture based on airborne multispectral sensitive wave band combination as set forth in claim 2, wherein generating a spectral response function using a relationship between center wavelengths and bandwidths of the plurality of different moisture sensitive wave bands, converting the simulated reflection spectrum to the plurality of moisture sensitive wave bands by the spectral response function, and obtaining the reflectivities and the spectral indexes of the plurality of moisture sensitive wave bands, comprises: according to the central wavelength and bandwidth of the selected moisture sensitive wave band, a Gaussian function is used for approximating a spectral response function, and the specific expression is as follows: wherein, the SRF (lambda) wavelength is the spectrum index at lambda, lambda 0 is the center wavelength, sigma is the standard deviation, and the width of the spectrum response function curve is controlled by calculating the bandwidth FWMH; The soil reflection spectrum and the canopy reflection spectrum are converted to obtain a plurality of selected moisture sensitive wave band combinations and the reflectivities of the moisture sensitive wave band combinations, wherein the specific expression is as follows: Where R band is the resampled reflectivity, λ 1 and λ 2 are the wavelength range of the band, and BRF (λ) is the reflectivity at the analog wavelength λ; And combining a plurality of moisture sensitive wave bands, and inputting the combined moisture sensitive wave bands into a spectral response function to obtain a spectral index.
4. The method for predicting moisture based on airborne multispectral sensitive band combination of claim 1, wherein before inputting the reflectance spectral image into the moisture prediction model, further comprises: collecting real soil and vegetation canopy reflection spectrum data, converting the real soil and vegetation canopy reflection spectrum data into a plurality of moisture sensitive wave bands by utilizing a spectral response function, and obtaining the reflectivity and the spectral index of the moisture sensitive wave bands under a plurality of real conditions; And carrying out nested cross validation on the water forecast model forecast result by using the reflectivity and the spectrum index of the water sensitive wave bands under a plurality of real conditions, and optimizing the water forecast model by using the nested cross validation result to obtain the optimal water forecast model.
5. The method for predicting moisture based on airborne multispectral sensitive band combination as recited in claim 4, wherein the nested cross-validation of the moisture prediction model prediction results using the reflectivity and the spectral index of the moisture sensitive band under a plurality of real conditions comprises: the outer layer cross validation of nested cross validation divides data into 10 folds randomly, 1 fold is reserved each time as a test set, the rest 9 folds are used as training sets, the optimal model selected in the inner layer cross validation is evaluated, the inner layer cross validation divides the training set of the outer layer cross validation into 10 folds, and super parameters of the moisture prediction model are optimized and selected.
6. The method for predicting moisture based on airborne multispectral sensitive wave band combination as set forth in claim 1, wherein the selecting the center wavelength of a plurality of moisture sensitive wave bands based on the response characteristic of soil and plant canopy reflection spectrum to moisture, increasing and selecting the center wavelength of a plurality of different moisture sensitive wave bands based on the physical mechanism of O-H bond vibration in water molecules, comprises: based on the response characteristics of the soil and plant canopy reflection spectrum to moisture, 560nm, 705nm, 750nm, 800nm, 900nm, 970nm, 1100nm, 1200nm, 1650nm and 2200nm are selected as the center wavelengths of moisture sensitive bands; Based on the physical mechanism of O-H bond vibration in water molecule, 450m and 660nm are selected as the central wavelength of moisture sensitive wave band.
7. The method for moisture prediction based on-board multispectral sensitive band combination of claim 1, wherein the determining the bandwidths of all the center wavelengths comprises: the spectral information and the spatial-temporal resolution are balanced by a bandwidth in a selected moisture sensitive band, wherein the bandwidth in the selected moisture sensitive band is 30nm, 10nm, 15nm, 40nm, 10nm, 50nm, and 100nm, respectively.
8. A moisture prediction system based on an airborne multispectral sensitive band combination, comprising: The selection module is used for selecting the center wavelengths of a plurality of moisture sensitive wave bands based on the response characteristics of the soil and plant canopy reflection spectrum to moisture, increasing and selecting the center wavelengths of a plurality of different moisture sensitive wave bands according to the physical mechanism of O-H bond vibration in water molecules, and determining the bandwidths of all the center wavelengths; the model construction module is used for simulating and generating reflection spectrums of soil and vegetation canopy, generating a spectral response function by utilizing the relation between the central wavelengths and bandwidths of the plurality of different moisture sensitive wave bands, converting the simulated reflection spectrums into the plurality of moisture sensitive wave bands through the spectral response function, acquiring the reflectivities and the spectral indexes of the plurality of moisture sensitive wave bands, and constructing a moisture prediction model by utilizing the reflectivities and the spectral indexes; And the prediction module is used for combining all moisture sensitive wave bands with determined bandwidths, integrating the combined moisture sensitive wave bands onto an airborne multispectral sensor, collecting reflection spectrum images of soil and vegetation canopy at the selected position, inputting the reflection spectrum images into a moisture prediction model, and predicting the moisture of the soil and vegetation canopy at the selected position.
9. A computer device comprising a memory and a processor, the memory having stored therein a program which, when executed by the processor, causes the processor to perform the steps of a method of moisture prediction based on an on-board multispectral sensitive band combination as claimed in any one of claims 1 to 7.
10. A storage medium having stored thereon a computer program, which when executed by a processor, implements the steps of a method for moisture prediction based on-board multispectral sensitive band combination of any one of claims 1 to 7.

Description

Moisture prediction method and system based on airborne multispectral sensitive wave band combination Technical Field The invention relates to the technical field of agricultural remote sensing and image processing, in particular to a moisture prediction method and system based on airborne multispectral sensitive wave band combination. Background With the increasing global climate change and the increasing shortage of water resources, sustainable management of agricultural production and ecological environments faces unprecedented challenges. The water is taken as a key factor affecting plant growth and soil health, and the accurate monitoring of the water has important significance for improving agricultural yield, optimizing water resource utilization efficiency, coping with extreme weather such as drought and the like and protecting ecological environment. However, the traditional water monitoring methods mainly depend on ground sampling and laboratory analysis, and the methods are time-consuming and labor-consuming, and have the limitations of limited space coverage, insufficient real-time performance and the like. This limitation significantly affects the efficiency and accuracy of agricultural management and environmental monitoring, especially in large-area farms or ecosystems. Currently, moisture monitoring methods based on remote sensing technology have been widely used. With the rapid development of remote sensing technology, a moisture monitoring method based on an onboard sensor gradually becomes an efficient alternative. Compared with ground monitoring, the airborne sensing system has the advantages of wide coverage, high acquisition speed, high flexibility and the like. Especially, the multispectral sensor carried on the unmanned aerial vehicle platform can acquire reflectivity data of a large area in a short time, and can realize rapid monitoring of soil and vegetation moisture information. In addition, advances in modern multispectral imaging technology, particularly the refinement of spectral band selection and the improvement of sensor resolution, have enabled the fine monitoring of soil and vegetation moisture status. Under drought stress, the structure and moisture content of plant leaves can change significantly, resulting in a change in response characteristics of the plant canopy reflectance spectrum. For example, when plants lack water, synthesis of chlorophyll decreases or decomposition accelerates, resulting in an increase in reflectance in the visible light region, accompanied by the phenomenon of red-edge blue shift. The lack of water can also result in reduced blade thickness, altered cell structure, and closed pores, thereby affecting the reflection and absorption characteristics of the blade in the near infrared region. In addition, light passing through the canopy can undergo multiple scattering and reflection at the leaf and soil surface, which allows the canopy reflectance spectrum to contain not only information about the plant itself, but also the moisture content of the root zone soil can be predicted by resolving the canopy spectrum. For bare soil surfaces, the spectral reflectance characteristics of bare soil are mainly affected by factors such as soil particle structure, moisture content, and organic content. Variations in moisture content can particularly significantly affect the reflectivity of soil in the mid-infrared and short-wave infrared regions. Therefore, by combining the canopy spectrum and bare soil spectral response characteristics, comprehensive predictions of root zone and bare soil moisture content can be achieved. However, existing on-board multispectral sensors are mostly focused in the visible and near infrared bands. These bands, while providing some moisture information, have limited pertinence and accuracy, lack sufficient spectral resolution and sensitivity, and are not able to accurately capture the absorption characteristics of moisture, resulting in limited effectiveness in moisture monitoring applications. Disclosure of Invention The invention aims to overcome the defects of the prior art, and provides a moisture prediction method and a system based on airborne multispectral sensitive wave band combination, so as to solve the problem that the prior art lacks enough spectral resolution and sensitivity, cannot accurately capture the absorption characteristics of moisture, and has limited effect in moisture monitoring application. The invention specifically provides the following technical scheme: a moisture prediction method based on airborne multispectral sensitive wave band combination comprises the following steps: Based on the response characteristics of soil and plant canopy reflection spectrum to moisture, selecting the center wavelengths of a plurality of moisture sensitive wave bands, increasing and selecting the center wavelengths of a plurality of different moisture sensitive wave bands according to the physical mechanism of O-H bond