Winter wheat photosynthetically active radiation component estimation method

Abstract

The invention discloses a method for estimating a photosynthetically active radiation component of winter wheat, which comprises the following steps of: simulating common 21 vegetation indexes of Sentinel-2 on the basis of reflectivity data measured on the ground, comparing the inversion capability of the indexes on fPAR of the winter wheat, testing the stability of the vegetation index with the highest correlation in fPAR inversion under the stress of different varieties, illumination and water, and calculating the photosynthetically active radiation component of the winter wheat. And then the estimation model is verified at a satellite scale, and the constructed estimation model is verified to still have relatively high stability and precision for fPAR estimation under different scales and different environments. The constructed estimation model can be used in large-scale and near-real-time winter wheat fPAR estimation, provides reference for agricultural condition monitoring and winter wheat yield estimation, and is suitable for performing vegetation index-based winter wheat fPAR estimation in a field scale and a satellite scale.

Description

[0001] The present application is a divisional application, the original application is entitled "Method for constructing winter wheat photosynthetically active radiation component estimation model", the application number is 202311819429.9, and the application date is December 27, 2023. TECHNICAL FIELD

[0002] The present application relates to the field of agricultural remote sensing technology, and particularly relates to a method for constructing a winter wheat photosynthetically active radiation component estimation model. BACKGROUND

[0003] Winter wheat is one of the main food crops in China. By monitoring and estimating the yield of winter wheat, the growth of wheat in farmland can be timely grasped, which provides an important reference for the supply of the grain market and ensures the stable supply of the market. With the advancement of satellite technology and image processing algorithms, remote sensing technology can quickly obtain large-scale and high-resolution image data of farmland, thereby realizing near-real-time monitoring and prediction of crop growth conditions and yield. The principle of remote sensing monitoring of crop yield is to observe and analyze the characteristics of electromagnetic energy reflected, radiated or absorbed by crops in different spectral bands, establish a model related to crop yield, and predict the growth conditions and yield of crops. To achieve near-real-time prediction of crop yield based on remote sensing technology, it is necessary to quickly obtain high-precision and large-scale parameter data required by the model. Therefore, how to quickly obtain the parameters of the large-scale yield model based on remote sensing data has become a necessary problem to be solved.

[0004] Light use efficiency (LUE) models have been proven to accurately and widely estimate the yield of winter wheat. Fraction of absorbed Photosynthetically Active Radiation (fPAR) is an essential variable in the process of calculating Gross Primary Productivity (GPP) based on light use efficiency models, and is one of the important parameters in ecological balance and carbon sink estimation. It is defined as the proportion of photosynthetically active radiation (PAR) light energy absorbed under crop or vegetation coverage to the total incident photosynthetically active radiation, which represents the ability of vegetation to perform photosynthesis, participate in carbon cycle and energy balance. At the same time, fPAR is also an important parameter in many ecological system production, crop yield estimation models and net primary productivity estimation. Rapid and accurate estimation of fPAR can improve the accuracy of crop yield estimation, which has a positive significance for crop field management and food security. SUMMARY

[0005] The method for constructing a winter wheat photosynthetically active radiation component estimation model provided by the present application realizes rapid estimation of the winter wheat photosynthetically active radiation component.

[0006] In order to achieve the above-mentioned purposes, the technical scheme adopted by the present application is as follows:

[0007] The method for constructing a winter wheat photosynthetically active radiation component estimation model comprises the following steps:

[0008] S1, collecting canopy reflectance data of winter wheat, and converting the canopy reflectance data of winter wheat into reflectance of satellite wave bands based on a spectral response function of a satellite to obtain a conversion result;

[0009] S2, measuring the winter wheat canopy incident photosynthetically active radiation, the winter wheat canopy reflected photosynthetically active radiation, the soil incident photosynthetically active radiation, and the soil reflected photosynthetically active radiation three times respectively, and calculating the winter wheat photosynthetically active radiation component after averaging the three times;

[0010] S3, calculating a vegetation index according to the conversion result, and fitting the vegetation index with the winter wheat photosynthetically active radiation component to obtain an initial estimation model corresponding to different vegetation indexes and the winter wheat photosynthetically active radiation component;

[0011] S4, verifying the accuracy of the different initial estimation models constructed from the aspects of water stress, light conditions, and winter wheat varieties, and selecting the initial estimation model with the highest determination coefficient as the winter wheat photosynthetically active radiation component estimation model.

[0012] Further, the specific method for collecting the canopy reflectance data of winter wheat in step S1 is as follows:

[0013] Between 10 o'clock and 14 o'clock, the canopy hyperspectral data of winter wheat are collected at a position 0.5 meters above the canopy of winter wheat by a spectral radiometer, 10 canopy hyperspectral data of winter wheat are obtained at the same position each time, the hyperspectral data are converted into canopy reflectance data of winter wheat by ViewSpecPro software, and the average value is taken; wherein, before collecting the canopy hyperspectral data of winter wheat, the spectral radiometer is corrected by a white board.

[0014] Further, the specific method of step S2 is as follows:

[0015] The winter wheat canopy incident photosynthetically active radiation, the winter wheat canopy reflected photosynthetically active radiation, the soil incident photosynthetically active radiation, and the soil reflected photosynthetically active radiation are measured three times respectively by a SunScan plant canopy analyzer, and the average value is taken; the horizontal included angle of adjacent two measurements is 45°.

[0016] According to the formula:

[0017]

[0018] obtaining a winter wheat photosynthetically active radiation component fPAR; wherein PAR in-can is the average of the 3 winter wheat canopy incident photosynthetically active radiation; PAR re-can is the average of the 3 winter wheat canopy reflected photosynthetically active radiation; PAR in-soil is the average of the 3 soil incident photosynthetically active radiation; PAR re-soil is the average of the 3 soil reflected photosynthetically active radiation.

[0019] Further, the vegetation index in step S3 comprises: NDVI, EVI, EVI2, NDPI, GCVI, RVI, DVI, LSWI-b8b11, LSWI-b8b12, LSWI-b8Ab11, LSWI-b8Ab12, MNDVI, SAVI, OSAVI, CIG, CIR, MNDWI, NDBI, GNDVI, NIRV and MTCI; wherein:

[0020]

[0021]

[0022]

[0023]

[0024]

[0025]

[0026]

[0027]

[0028]

[0029]

[0030]

[0031] B3 is the reflectance data of the Green band of the Sentinel 2 satellite corresponding to the canopy reflectance data of winter wheat when the ground spectral range is 542-577 nm; B4 is the reflectance data of the Red band of the Sentinel 2 satellite corresponding to the canopy reflectance data of winter wheat when the ground spectral range is 649-680; B5 is the reflectance data of the RE-1 band of the Sentinel 2 satellite corresponding to the canopy reflectance data of winter wheat when the ground spectral range is 697-711; B6 is the reflectance data of the RE-2 band of the Sentinel 2 satellite corresponding to the canopy reflectance data of winter wheat when the ground spectral range is 733-747; B7 is the reflectance data of the RE-3 band of the Sentinel 2 satellite corresponding to the canopy reflectance data of winter wheat when the ground spectral range is 772-791; and B8 is the reflectance data of the NIR band of the Sentinel 2 satellite corresponding to the canopy reflectance data of winter wheat when the ground spectral range is 781-885.

[0032] Further, the initial estimation model in step S3 includes a linear model, an exponential model and a logarithmic model.

[0033] The expression of the linear model is: fPAR = a x VI + b

[0034] The expression of the exponential model is: fPAR = a x e b×VI

[0035] The expression of the logarithmic model is: fPAR = a + b x ln(VI)

[0036] Wherein fPAR is the photosynthetically active radiation component of winter wheat, VI represents a single vegetation index; a and b are to-be-fitted parameters; e is a natural constant; and ln(.) represents an index with a natural constant e as a base.

[0037] Further, the expression of the determination coefficient in step S4 is:

[0038]

[0039] Wherein R 2 is the determination coefficient; is the estimated value of the photosynthetically active radiation component model; is the mean value of the real measurement value of the photosynthetically active radiation component of winter wheat; fPAR k is the real measurement value of the photosynthetically active radiation component of winter wheat; and n represents the number of all plots participating in the precision verification, and k represents the kth ground plot.

[0040] Further, the expression of the estimation model of the photosynthetically active radiation component of winter wheat in step S4 is:

[0041]

[0042] Wherein fPAR is the winter wheat photosynthetically active radiation component, VI represents a single vegetation index; e is a natural constant.

[0043] The method is based on ground-measured reflectance data, simulates 21 common vegetation indices of Sentinel-2 and compares their inversion capabilities for winter wheat fPAR, tests the stability of the vegetation index with the highest correlation in the inversion of fPAR under different varieties, light and water stress, and then verifies the estimation model at the satellite scale, and the constructed estimation model still has high stability and precision for fPAR estimation under different scales and different environments. The constructed estimation model can be used for large-scale, near-real-time winter wheat fPAR estimation, provides reference for crop monitoring and winter wheat yield estimation, and is suitable for winter wheat fPAR estimation based on vegetation index at field scale and satellite scale. BRIEF DESCRIPTION OF DRAWINGS

[0044] Figure 1 The flowchart of the method is shown in the figure;

[0045] Figure 2 The precision verification results of the estimation model on different varieties are shown in the figure;

[0046] Figure 3 The precision verification results of the estimation model on different dates are shown in the figure;

[0047] Figure 4 The precision verification results of the estimation model under different irrigation schemes are shown in the figure;

[0048] Figure 5 The estimation results of the estimation model at the satellite scale are compared with the estimation results of the physical model. DETAILED DESCRIPTION

[0049] The specific embodiments of the present application are described below to facilitate the understanding of the present application by those skilled in the art, but it should be clear that the present application is not limited to the scope of the specific embodiments, and for those skilled in the art, it is obvious that various changes are within the spirit and scope of the present application defined and determined by the appended claims, and all the inventions utilizing the concept of the present application are within the scope of protection.

[0050] As shown in the figure, the construction method of the winter wheat photosynthetically active radiation component estimation model comprises the following steps: Figure 1

[0051] S1, collecting the canopy reflectance data of winter wheat, and converting the canopy reflectance data of winter wheat into reflectance of satellite bands based on the spectral response function of the satellite to obtain a conversion result;

[0052] ​S2, respectively measuring 3 times of winter wheat canopy incident photosynthetically active radiation, 3 times of winter wheat canopy reflected photosynthetically active radiation, 3 times of soil incident photosynthetically active radiation, 3 times of soil reflected photosynthetically active radiation, and calculating winter wheat photosynthetically active radiation components after averaging three times;

[0053] S3, calculating vegetation index according to the conversion result, and fitting the vegetation index with the winter wheat photosynthetically active radiation components to obtain an initial estimation model corresponding to different vegetation indexes and winter wheat photosynthetically active radiation components;

[0054] S4, verifying the accuracy of different initial estimation models from the aspects of water stress, light conditions and winter wheat varieties, and selecting the initial estimation model with the highest determination coefficient as the winter wheat photosynthetically active radiation component estimation model. The winter wheat photosynthetically active radiation component estimation model (i.e. estimation model) can be used to estimate the winter wheat photosynthetically active radiation component.

[0055] In an embodiment of the present application, 11 common winter wheat varieties in North China Plain are set in the ground test: T1 (Chang8744), T2 (Shimai22), T3 (Luyuan472), T4 (Shimai15), T5 (HengH1603), T6 (Xinmai28), T7 (Jimai418), T8 (Shannong28), T9 (Nongda212), T10 (Heng4399), and T11 (Jimai22). Seven different irrigation schemes of each variety are set to simulate the growth differences of winter wheat under different degrees of water stress, and each scheme is repeated three times. The irrigation schemes are shown in Table 1. Therefore, 231 ground plots are set, and the size of each plot is 1.5x6m.

[0056] Table 1: Specific irrigation scheme

[0057]

[0058] The spectral radiometer (FieldSpec4, wavelength range: 350-2500nm, spectral resolution: 1nm) produced by the American ASD (Analytical Spectral Devices Inc.) company is used to collect winter wheat canopy hyperspectral data. The measurement is also carried out under relatively good weather conditions and sufficient light, and the observation time is from 10am to 14pm. The white board is used for correction before measurement, and the measurement position is about 0.5m above the wheat canopy. Ten canopy spectral curves are obtained at the same position of fPAR measurement each time, and the average value is taken as the spectral reflectance curve of the plot.

[0059] The ground spectral data were converted to reflectance in the satellite bands using the spectral response functions of the satellite, so that the ground reflectance corresponded to the Sentinel-2 reflectance. Since Sentinel-2 is a dual satellite, the corresponding relationship between the ASD spectrometer and the Sentinel-2 satellite is shown in Table 2 after integrating the ranges of different bands of the two satellites.

[0060] Table 2: Corresponding relationship between ground reflectance data and Sentinel-2 satellite reflectance

[0061]

[0062] 21 common vegetation indices were used in the correlation analysis with the ground measured fPAR. These vegetation indices were

[0063] The 21 common vegetation indices were used to estimate the fPAR of different vegetation, and showed different degrees of estimation ability. Therefore, the goal of this study was to evaluate their ability to estimate the fPAR of winter wheat at key growth stages, and their definitions and calculation methods are shown in Table 3.

[0064] Table 3: Commonly used vegetation indices based on Sentinel-2 data reflectance

[0065]

[0066] The SunScan plant canopy analyzer was used to measure fPAR. SunScan includes a 1-meter probe and a handheld PDA. A total of 64 quantum sensors are embedded in the probe, and an RS-232 cable is used to connect the probe and the PDA. When measuring the fPAR of each plot, first measure the observation incident photosynthetically active radiation 3 times for the average value of canopy incident (PARin-can), 3 times for the average value of canopy reflection (PARre-can), 3 times for the average value of soil incident (PARin-soil), and 3 times for the average value of soil reflection (PARre-soil). The fPAR of the plot is obtained by averaging the three measurements. The spatial relationship between the probe and the winter wheat plot during the three measurements is shown in Figure 2 To ensure that all measurement work is completed between 10 am and 2 pm, only one repeated test of each water treatment was selected for ground data collection during measurement, a total of 77 ground plot data were measured each time. The horizontal angle between the two adjacent measurements is 45°, and the fPAR of each plot is defined as:

[0067] The SunScan plant canopy analyzer was used to measure fPAR. SunScan includes a 1-meter probe and a handheld

[0068] Formula PDA. A total of 64 quantum sensors are embedded in the probe, and an RS-232 cable is used to connect the probe and the PDA. When measuring the fPAR of each plot, the observed incident photosynthetically active radiation is first measured for 3 times (PAR in-can ), 3 times of canopy reflectance average (PAR re-can ), 3 times of soil incident average (PAR in-soil ), 3 times of soil reflectance average (PAR re-soil ), and the fPAR of the plot is obtained by averaging the three times. The spatial relationship between the probe and the winter wheat plot during the three measurements is shown in FIG. 4. Figure 2 Since the observation of fPAR consumes a lot of time, in order to ensure that all the measurement work is completed between 10 am and 2 pm, only one repeated test of each water treatment is selected for ground data collection during measurement, and a total of 77 ground plots are measured each time. The horizontal angle between the adjacent two measurements is 45°, and the fPAR of each plot is defined as:

[0069]

[0070] In order to find more stable and strongly correlated vegetation indices, in this embodiment, the near-surface reflectance of all measurements in 11 days is simulated as Sentinel-2 vegetation index, and three correlation relationships are established with the fPAR measured by SunScan. At the same time, the determination coefficient R 2 is calculated, and the RMSE is calculated using all the measured data for verification. The fitting results are shown in Table 4. The determination coefficient R 2 and the root mean square error RMSE are defined as follows:

[0071]

[0072]

[0073] wherein is the estimated value of the photosynthetically active radiation component model; is the mean value of the true measurement value of the photosynthetically active radiation component of winter wheat; fPAR k is the true measurement value of the photosynthetically active radiation component of winter wheat; n represents the number of plots participating in the accuracy verification, and k represents the kth ground plot.

[0074] Table 4: Accuracy comparison of different vegetation correlation fitting

[0075]

[0076]

[0077] Where MNDWI, NDBI and NDWI values are less than 0, so there is no logarithmic correlation with fPAR. From the fitting results, we can see that there are large differences in the fitting of fPAR for different vegetation indices, and the linear, exponential and logarithmic fitting results are also different. The exponential model based on MNDVI has the highest determination coefficient, and the model established by it is suitable for the estimation of winter wheat fPAR, and the model is as follows:

[0078] fPAR = 0.0772e 2.7526VI

[0079] Excluding T3 and T5 varieties due to insufficient emergence rate to complete vegetation canopy reflectance measurement and fPAR, and on April 11 and April 18 due to irrigation F and G cannot be measured, all measurement values participate in the accuracy verification. Except for variety T10, the rest of the varieties show good correlation and low RMSE. Among them, the determination coefficient of T11 variety is the highest, reaching 0.911, and the RMSE of T1 variety is the lowest, which is 0.074. It shows that MNDVI can maintain high stability between different varieties, and MNDVI has high application potential in yield estimation in these study areas. The verification results are shown in Figure 2 .

[0080] When estimating the vegetation index fPAR, with the growth of crops, the light conditions are constantly changing, and whether the correlation between vegetation index and fPAR is stable under different light conditions is an important factor to judge the universality of the model. These light conditions include solar elevation angle and light intensity under different weather conditions. They may affect the correlation between fPAR and other indicators. Therefore, when using MNDVI to estimate fPAR, we need to verify the accuracy of the index on different observation dates. In order to verify the estimation accuracy of MNDVI on different observation dates, we calculate by using the estimation model fPAR = 0.0772e 2.7526VI , and get the corresponding estimated values. These estimated values represent the fPAR on different observation dates. Our research aims to prove that using MNDVI to estimate fPAR has high stability under different phenological stages and light conditions. The verification results are shown in Figure 3 .

[0081] The irrigation scheme caused different degrees of water stress to winter wheat, and the spatial range of the study area was wide, and the rainfall distribution was uneven. To verify the stability of MNDVI in estimating fPAR under different water stress caused by different growth conditions, and to ensure that the estimation accuracy of the related model would not be affected by the different growth states of crops when applied to large-scale areas. By comparing the estimated results with the actual measurement results, the determination coefficient of the C irrigation scheme was relatively low, but the accuracy of the remaining different water stress could meet the actual application requirements. The longness of almost all plots in the C scheme was good, and the vegetation index and fPAR value were high, so it was difficult to get a higher R 2 , but with the lowest RMSE. It can be seen that MNDVI can stably and accurately estimate fPAR when receiving different degrees of water stress. The verification results are shown in Figure 4 .

[0082] Three cloud-free 10m Sentinel-2 data were selected in the key growth period of winter wheat in a certain area for precision verification. The size of the verification area was selected as 1000x1000 pixels, and the non-winter wheat area was masked. All winter wheat pixels in the area were used for verification. The fPAR calculated by the Biophysical Processor tool of SNAP software was used as the true data to verify the reliability of the data estimated by the model, and the results are shown in Figure 5 . The determination coefficients of the three periods were 0.718, 0.550, and 0.578, respectively, which proved that the photosynthetically active radiation component estimation model established by this method still had high accuracy on the satellite scale.

[0083] In summary, this method is based on ground-measured reflectance data, simulates the common 21 types of vegetation index of Sentinel-2, and compares their inversion ability for winter wheat fPAR. The highest correlation vegetation index is tested for the stability of fPAR inversion under different varieties, light and water stress, and then the estimation model is verified on the satellite scale. The constructed estimation model still has high stability and accuracy for fPAR estimation under different scales and different environments. The constructed estimation model can be used for large-scale, near-real-time winter wheat fPAR estimation, and provides reference for crop monitoring and winter wheat yield estimation, and is suitable for winter wheat fPAR estimation based on vegetation index on the field scale and satellite scale.

Claims

1. A method for estimating a photosynthetically active radiation component of winter wheat, characterized by, The method comprises the following steps: S1, collecting the canopy reflectance data of winter wheat, converting the canopy reflectance data into reflectance of Sentinel-2 satellite bands based on the spectral response function of Sentinel-2 satellite, and obtaining a conversion result; S2, measuring the incident photosynthetically active radiation of the winter wheat canopy, the reflected photosynthetically active radiation of the winter wheat canopy, the incident photosynthetically active radiation of the soil, and the reflected photosynthetically active radiation of the soil, and calculating the winter wheat photosynthetically active radiation component based on the test data, and the specific calculation process is: where fPAR is the fraction of photosynthetically active radiation component of winter wheat, PAR in-can is the average of 3 times of winter wheat canopy incident photosynthetically active radiation; PAR re-can is the average of 3 times of winter wheat canopy reflected photosynthetically active radiation; PAR in-soil is the average of 3 times of soil incident photosynthetically active radiation; PAR re-soil is the average of 3 times of soil reflected photosynthetically active radiation; S3, calculating the vegetation index MNDVI according to the conversion result, wherein the calculation formula of MNDVI is: In the formula, B4 is the reflectance data of the Red band of Sentinel-2 satellite corresponding to the canopy reflectance data of winter wheat when the ground spectral range is 649-680; B8 is the reflectance data of the NIR band of Sentinel-2 satellite corresponding to the canopy reflectance data of winter wheat when the ground spectral range is 781-885; Correlating the calculated vegetation index MNDVI and the winter wheat photosynthetically active radiation component to obtain an index model based on MNDVI: In the formula, e is a natural constant; S4, using the index model based on MNDVI as a winter wheat photosynthetically active radiation component estimation model to perform a winter wheat photosynthetically active radiation component estimation task.

2. The method of claim 1, wherein, The specific method for collecting the canopy reflectance data of winter wheat in the step S1 is: Between 10 o'clock and 14 o'clock, the winter wheat canopy hyperspectral data is collected at a position 0.5 meters above the winter wheat canopy by a spectral radiometer, 10 pieces of winter wheat canopy hyperspectral data are obtained at the same position each time, the hyperspectral data is converted into winter wheat canopy reflectance data by ViewSpecPro software, and the average value is taken; wherein before collecting the winter wheat canopy hyperspectral data, the spectral radiometer is corrected by a white board.

3. The method of claim 1, wherein, The selection process of the vegetation index MNDVI specifically comprises: Based on the measured reflectance data, different vegetation indices corresponding to Sentinel-2 are simulated, the inversion ability of each vegetation index for winter wheat fPAR is compared, and the accuracy of different initial estimation models constructed from water stress, light conditions and winter wheat varieties is verified, and the initial estimation model with the highest determination coefficient is selected as the winter wheat photosynthetically active radiation component estimation model; wherein the vegetation indices include NDVI, EVI, EVI2, NDPI, GCVI, RVI, DVI, LSWI-b8b11, LSWI-b8b12, LSWI-b8Ab11, LSWI-b8Ab12, MNDVI, SAVI, OSAVI, CIG, CIR, MNDWI, NDBI, GNDVI, NIRV and MTCI.

4. The method according to claim 3, wherein the determination coefficient is: The winter wheat photosynthetically active radiation component estimation model is specifically: where R 2 is the coefficient of determination; is the estimated value of the PAR component model; is the mean of the true measurements of the PAR component for winter wheat; fPAR k is the true measurement of the PAR component for winter wheat; n indicates the number of all plots participating in the accuracy verification, and k represents the kth ground plot. In the formula, fPAR is the winter wheat photosynthetically active radiation component, VI represents a single vegetation index, and e is a natural constant. ​

Images