Corn phenology remote sensing monitoring method based on NIRv and crop progress report

Abstract

The invention provides a corn phenology remote sensing monitoring method based on NIRv and a crop progress report. Comprising the following steps: constructing an NIRv time sequence curve by utilizing remote sensing data; constructing a corn growth model in combination with the open crop progress report and the cultivated land utilization data; calibrating phenological parameters by using data in previous years and combining with the growth model to obtain a shape model; a traditional SMF method is optimized, a shape model and an annual vegetation index time sequence curve are fitted, optimal parameters are calculated, and five main phenological periods of corn are monitored; pixel-level corn phenology monitoring information is obtained, regional phenology mapping is carried out, and quantitative evaluation is carried out on regional phenology. According to the method, the modeling problem that accurate estimation of the phenological period of the crop can be achieved without depending on long-term observation data on the ground is solved, the application range of shape model fitting in phenological monitoring is expanded, the phenological monitoring capacity of the NIRv index is evaluated, and the method is suitable for research in a large area. The remote sensing monitoring precision of crop phenology is further improved, and governments and farmers are assisted in accurately and scientifically managing crops.

Description

Technical Field

[0001] The invention relates to the field of agricultural remote sensing, and in particular to a method for conducting crop phenological period remote sensing monitoring research based on a shape fitting model, in particular to a corn phenological remote sensing monitoring method based on NIRv and crop progress reports. Background Art

[0002] Crops have relatively stable physiological process characteristics (such as photosynthesis and dry matter distribution) in different phenological periods. Crop phenology, as a calendar of crop seasonal growth dynamics, provides important information for agricultural activities such as fertilizer management, irrigation scheduling, disease prevention and crop type classification. In addition, when estimating crop yield based on remote sensing vegetation index (VI) data, the accurate time of the key growth stage of crops is also a key input information. There are many methods for crop phenology monitoring at present. The most traditional method is mainly obtained through field observation. The results obtained by this method are relatively accurate, but it is mixed with subjective factors, and it also has the disadvantages of being time-consuming and labor-intensive, with a long observation period and a small coverage area, making it difficult to obtain phenological information over a large range. The emergence of remote sensing technology has provided a major opportunity for large-scale spatial continuous crop phenology observation. It has the characteristics of large coverage, short detection period, strong timeliness and low cost. It can monitor the dynamic changes of crops at regional and even global scales. Therefore, the crop phenology detection method based on remote sensing is a key technology for evaluating the impact of environmental stress on crops.

[0003] The shape model fitting method (SMF) is currently a commonly used method in remote sensing-based phenological monitoring. However, many related studies require years of field observation data due to the influence of the prior reference shape and reference feature date. It is heavily dependent on ground observation conditions, which leads to challenges in large-scale crop phenological monitoring. At the same time, the contribution of the vegetation near-infrared reflectance index (NIRv) developed in recent years to the shape model fitting method is not yet clear.

[0004] Therefore, this method solves the problem of dependence on long-term ground observation data in traditional methods of remote sensing monitoring of crop phenology, and invents a new method that can accurately monitor crop phenology by relying only on remote sensing data and public crop progress reports, which expands the application scope of remote sensing phenological monitoring and reduces the prerequisites for remote sensing phenological monitoring, which will help remote sensing phenological monitoring to be convenient and accurate. At the same time, the present invention further explores the monitoring ability of the NIRv index on crop phenology, which will help to further improve the accuracy of crop phenology remote sensing monitoring, and provide some valuable references and methods for the study of compound drought that takes into account the spatiotemporal heterogeneity of ground crops. Summary of the invention

[0005] The purpose of this method is to propose a corn phenology remote sensing monitoring method based on NIRv and crop progress report, in view of the dependence of the shape fitting model method on long-term ground observation data and the limitations of the method itself. The method is based on the improved SMF-s method and specifically includes the following steps:

[0006] S1: Collect public crop data and surface reflectance data;

[0007] S2: Use the surface reflectance data collected in S1 to calculate the NIRv for several years;

[0008] S3: Use the remote sensing data of the most densely planted corn area obtained by moving window search to calculate the vegetation near infrared reflectivity index NIRv of the year, and then smooth and average the vegetation near infrared reflectivity index NIRv of the year to obtain the vegetation index time series curve, and then construct the typical growth curve of corn;

[0009] S4: Under the initial phenological parameters, the phenological parameters are updated with a preset time window and a preset step size. The optimal reference phenology of each stage is inverted by minimizing the difference between the phenological estimate and the regional scale phenological report of the public crop data in S1, and the crop shape model is constructed in combination with the typical maize growth curve in S3.

[0010] S5: According to the NIRv calculated in step S2, the vegetation index time series curve of the study area for several years is obtained, and the crop shape model is fitted with the vegetation index time series curve of each year through an iterative algorithm to calculate the optimal phenological parameters, realize phenological monitoring, and obtain pixel-level phenological monitoring information;

[0011] S6: Conduct regional phenological mapping of key phenological periods of maize based on pixel-level phenological monitoring information to reveal the spatiotemporal pattern of maize phenology in the study area, analyze the regional maize phenological distribution, produce regional crop growth rate maps, and analyze the spatial variation of maize growth rate in the study area. The key phenological periods include the emergence stage, silking stage, milky stage, waxy stage, and maturity stage.

[0012] A storage device stores instructions and data for implementing a corn phenology remote sensing monitoring method based on NIRv and crop progress reports.

[0013] A corn phenology remote sensing monitoring device based on NIRv and crop progress report comprises: a processor and a storage device; the processor loads and executes instructions and data in the storage device to implement a corn phenology remote sensing monitoring method based on NIRv and crop progress report.

[0014] Compared with the prior art, the beneficial effects of the present invention include:

[0015] (1) Previous studies mostly relied on long-term ground observation data, which had certain limitations in application. The present invention only requires input from freely accessible public data (crop progress reports, cultivated land coverage data) and satellite data, and is unique in terms of application scope and convenience.

[0016] (2) Previous studies mostly used the WDRVI index and the EVI index. This paper attempts to use the NIRv index for the first time to explore the effect of the NIRv index in crop monitoring with shape model fitting, which helps to accurately monitor crop phenology.

[0017] (3) Based on regional phenological mapping, phenological time characteristics and crop growth rates can be analyzed at the pixel level, which can reveal the regional crop phenological spatiotemporal characteristics and phenological spatiotemporal changes in more detail, providing other agricultural-related research with the possibility of considering the spatiotemporal heterogeneity of ground crops. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solution in this embodiment, the drawings required for use in the description of the embodiment will be briefly introduced below. Obviously, the drawings described below are an embodiment of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0019] Figure 1 A flowchart of a method for remote sensing monitoring of corn phenology based on NIRv and crop progress report provided in an embodiment of the present invention;

[0020] Figure 2 A schematic diagram of a framework of a corn phenology remote sensing monitoring method based on NIRv and crop progress report provided in an embodiment of the present invention;

[0021] Figure 3 is a flow chart of shape model construction in an example of the present invention;

[0022] Figure 4 It is a flow chart of phenological parameter calibration in an example of the present invention;

[0023] Figure 5 It is a calibration effect diagram of phenological parameters in an example of the present invention;

[0024] Figure 6 It is a phenological monitoring flow chart of the present invention;

[0025] Figure 7 It is a summary diagram of phenological monitoring accuracy in the example of the present invention;

[0026] Figure 8 It is a phenological result evaluation diagram in the example of the present invention in units of years;

[0027] Fig. 9 It is a line chart comparing the phenological monitoring results and CPR data in the example of the present invention;

[0028] Fig.10 This is a phenological regional mapping based on 2014 as an example in the present invention;

[0029] Fig.11 It is a regional crop growth rate diagram taking 2014 as an example in the example of the present invention;

[0030] Fig.12 It is a schematic diagram of the operation of the hardware device in the embodiment of the present invention. DETAILED DESCRIPTION

[0031] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0032] like Figure 1-2 As shown, the present invention proposes a corn phenology remote sensing monitoring method based on NIRv and crop progress report, which specifically includes the following steps:

[0033] S1: Collect public crop data and surface reflectance data;

[0034] S2: Use the surface reflectance data collected in S1 to calculate the NIRv for several years;

[0035] S3: Use the remote sensing data of the most densely planted corn area obtained by moving window search to calculate the vegetation near infrared reflectivity index NIRv of the year, and then smooth and average the vegetation near infrared reflectivity index NIRv of the year to obtain the vegetation index time series curve, and then construct the typical growth curve of corn;

[0036] S4: Under the initial phenological parameters, the phenological parameters are updated with a preset time window and a preset step size. The optimal reference phenology of each stage is inverted by minimizing the difference between the phenological estimate and the regional scale phenological report of the public crop data in S1, and the crop shape model is constructed in combination with the typical maize growth curve in S3.

[0037] S5: According to the NIRv calculated in step S2, the vegetation index time series curve of the study area for several years is obtained, and the crop shape model is fitted with the vegetation index time series curve of each year through an iterative algorithm to calculate the optimal phenological parameters, realize phenological monitoring, and obtain pixel-level phenological monitoring information;

[0038] S6: Conduct regional phenological mapping of key phenological periods of maize based on pixel-level phenological monitoring information to reveal the spatiotemporal pattern of maize phenology in the study area, analyze the regional maize phenological distribution, produce regional crop growth rate maps, and analyze the spatial variation of maize growth rate in the study area. The key phenological periods include the emergence stage, silking stage, milky stage, waxy stage, and maturity stage.

[0039] Taking the corn planting in Illinois, USA from 2009 to 2020 as an example, the details are as follows:

[0040] (1) Collection of basic data: Collection of public crop data and surface reflectance data;

[0041] In step S1, the surface reflectance data downloaded are the reflectance data of band 1 (620nm-670nm) and band 2 (841nm-876nm) of the MODIS surface reflectance product MOD09Q1, with a temporal resolution of 8 days and a spatial resolution of 231.7m. The covered grid locations are h10v04, h10v05, h11v04, and h11v05. The public crop data includes crop progress data and surface classification data. The Crop Progress Report (CPR) is data provided by the Agricultural Crop Progress Survey Project of the U.S. Department of Agriculture (USDA) Agricultural Statistics Service (NASS for short). Crop progress data are collected through regular surveys and observations of the growth and development of major crops in farmland. These data provide information on key stages such as crop sowing, germination, growth, and harvesting, as well as progress compared with the same period in previous years. It is recorded spatially in states, and crop phenology progress is released once a week. The growth stages of corn in NASS-CPR are recorded as sowing, emergence, silking, milky, waxy, mature and harvest. The download time range of CPR data is 2008-2020. The surface classification data, namely the Cultivated Land Data Layer (CDL), is a specific crop land use created by the National Agricultural Statistics Service (NASS) of the United States Department of Agriculture based on medium-resolution optical satellite sensor data. It can be obtained free of charge from the USDA website and is used for surface crop classification. Its resolution is refined to 30m. The download time range of CPR and CDL data is 2008-2020. In order to select MODIS pixels covered by crops, it is necessary to convert the map projection of CDL data from Universal Transverse Mercator to MODIS Sine Wave projection using the nearest neighbor method.

[0042] (2) Calculation of vegetation near-infrared reflectivity index: Calculate the near-infrared reflectivity index (NIRv) using the surface reflectivity data collected in S1;

[0043] In order to explore the impact of different vegetation indices on the accuracy of phenological monitoring in the method of the present invention, the present invention selected a vegetation index developed in recent years, vegetation near-infrared reflectance (NIRv), for comparative experiments. In this example, MODIS data was used to calculate the NIRv vegetation index in Illinois from 2009 to 2020. The calculation formula of NIRv is as follows:

[0044]

[0045] Among them, ρnir and ρred are the near-infrared band (MOD09Q1 band 2) and the red light band (MOD09Q1 band 1), and C is a set constant. In this embodiment, C is set to 0.08.

[0046] In order to reflect the advantages of the technical method of the present invention, this example also uses the commonly used WDRVI index as a reference for comparison. The calculation formula of the WDRVI index is as follows:

[0047] WDR VI=(αρ nir -ρ red ) / (αρ nir +ρ red )

[0048] Among them, ρ nir and ρ red is the near infrared band (MOD09Q1 band 2) and the red band (MOD09Q1 band 1), and α is the weighting coefficient. When α = 0.1, WDRVI has a strong linear correlation with maize green leaf LAI.

[0049] (3) Establishment of crop shape model that does not rely on long-term ground data: Using a moving window to search for remote sensing data of the most densely planted corn areas in 2008, the vegetation near-infrared reflectivity index is calculated and the vegetation index time series curve is obtained after smoothing and averaging, and the typical growth curve of corn is constructed;

[0050] In this embodiment, in order to get rid of the limitations of the surface coverage changes and the scope of the research area, the method for establishing the shape model is as follows: Figure 3 As shown in Figure 1, we first need to extract the pixels of the crop type corn in the CDL data after projection transformation in 2008, resample them to 231.7 m, the same resolution as the MODIS remote sensing data, and use a 100*100 pixel moving window (about 537 km 2) Search for the location with the highest density of maize pixels in the study area. The center of the highest density is 41°53.0′N, 88°52.60′W, located in the northern part of the study area. Different from the previous studies based on shape models that used the average of many years of data from experimental fields or training stations to obtain shape models, this example uses MODIS remote sensing data of some of the most densely planted pixels in the area of ​​long-term crop phenology observation to calculate the NIRv and WDRVI time series curves pixel by pixel, and uses SG filtering to smooth the time series curves on this basis, and finally calculates the average NIRv and WDRVI time series curves of each pixel in the 100*100 window as the maize shape model of the present invention.

[0051] (4) Calibration of phenological parameters of crop shape model: Under the initial phenological parameters, the phenological parameters are updated with a time window of 10 days and a step size of 2 days. The optimal reference phenology of each stage is inverted by minimizing the difference between the phenological estimates in 2009, 2019, and 2020 and the regional-scale phenological reports of public crop data in S1, thereby obtaining the crop shape model;

[0052] In this embodiment, the phenological parameter calibration process is as follows: Figure 4 As shown in the figure, the specific steps are as follows: First, select the CPR data of Illinois in 2008 as the basic phenological parameter X ref0 , then X ref0 As a benchmark, in the time window w, with a step size of b, adjust each X back and forth ref0i (i is the phenological stage from 1 to 7), and the phenological parameter X of the nth calibration is obtained ref0i n (n is the number of calibrations), based on the 2009 MODIS data and NASS-CPR statistical data, X ref0 Calibrate and calculate the reference phenology as X by shape model fitting method ref0 n Estimated phenological periods for Illinois in 2009 under X est n , X est n Indicates that the basic phenological parameter X ref0 The estimated phenological period after n calibrations when is the initial value. By calculating the difference d with the 2009 NASS-CPR statistical data, the X that minimizes the difference d is obtained. ref0 n Denoted as X ref1 Then set X ref1 As a reference phenology, X ref1 As the new initial value, a new round of calibration iteration is carried out, and the data of 2019 and 2020 are used to continue the phenological calibration, and finally the best reference phenological X is obtained. ref .

[0053]

[0054] Where n is the nth calibration, i represents each phenological stage, i = 1 to 7, and I is the total number of crop phenological stages; Xcpr i n is the phenological period of the i-th phenological stage in the CPR statistical data at the time of the n-th calibration, Xest i n Xest is the estimated phenological period at the nth calibration of the i-th phenological stage. i n To use the phenological parameter vector set X ref0 The elements of the estimated phenology of the i-th phenological stage calculated when X is used as the reference phenology. For example, when the iteration starts, the basic phenological parameter is X ref0 The seven phenological stages are X ref01 , X ref02 , X ref03 , X ref04 , X ref05 , X ref06 , X ref07 , each phenological stage has n calibrations, Xest1 n It is the phenological period estimated at the nth calibration of the first phenological stage. In the second iteration (the 2019 data is used in this example), the reference phenological parameter is X ref1 , the 7 phenological stages are X ref11 , X ref12 , X ref13 , X ref14 , X ref15 , X ref16 , X ref17 , Xest1 n It is the phenological period estimated at the nth calibration of the first phenological stage in the second iteration. Similarly, until the last iteration, the optimal reference phenology X is obtained. ref .

[0055] After calculation, it is found that the time window w is 10 days and the step length b is 2 days for phenological calibration, which is the most convenient for calculation. Figure 5 As shown in the figure, this example uses the results of phenological monitoring using the WDRVI index in 2011 and 2012 to demonstrate the effect of phenological parameter calibration.

[0056] (5) Monitoring of major crop phenological periods based on shape fitting;

[0057] In this embodiment, the vegetation index time series curve of the study area from 2009 to 2020 is obtained, and the shape model is fitted with the vegetation index time series curve of each year through an iterative algorithm to calculate the optimal parameters and realize phenological monitoring. Figure 6 The specific description is as follows:

[0058] First, it is necessary to establish the VI time series growth curve of corn pixels after smoothing based on remote sensing data; secondly, by introducing the X-axis and the stretching and translation factors xscale and tshift on the time axis, the time series growth curve of each stage of the pixel is fitted and matched with the shape model of each stage. The formula is as follows:

[0059]

[0060] Function g(x) is the shape model; X i ref is the phenological parameter; xscale i is the stretch factor, tshift i is the translation factor.

[0061] Through iterative algorithms, the translation factor and x-axis stretch factor between the two curves are optimized successively to calculate the optimal geometric parameters. The criterion for determining the optimal parameters is the local VI time series segment after the shape model is fitted and matched. The correlation coefficient with the time series curve segment h(x) of the phenological period to be estimated is the largest, and the formula is as follows:

[0062]

[0063] At initial xscale i When the value is 1.0, we first shift t i The range is -45 days to 45 days, with a step size of 1 day, and determines the optimal tshifti value, that is, the tshift when the objective function in the figure is maximized i Next, fix the determined tshift i value, and in the same way by changing xscale i value (0.8~1.2), with a step size of 0.01 to estimate the optimal xscale i Repeat this process, and when the two factors no longer change, the iteration ends. est It can be calculated by referring to the phenological period X0 and the stretching factor and time shift factor of the X-axis. The formula is as follows:

[0064]

[0065] Among them, if the correlation coefficient R>=0.8, the result is output, and if the correlation coefficient R<0.8, the corresponding final phenological estimation is deleted.

[0066] (6) Horizontal evaluation of phenological monitoring results based on phenological periods: After the pixel-level phenological monitoring results obtained in step (5) are analyzed, regional phenological mapping is performed for the five key phenological periods of corn (seedling emergence, silking, milky stage, waxy stage, and maturity) to reveal the spatiotemporal pattern of corn phenology in the study area, produce regional crop growth rate maps, and analyze the spatial variation of corn growth rate in the study area.

[0067] In this example, the RMSE between the phenological period derived from the present invention and the statistical data of the CPR of the phenological period in each corn growth stage from 2009 to 2020 is calculated in turn. i With R i 2 By comparing the monitoring accuracy of each phenological period horizontally, the accuracy of phenological monitoring in the present invention and the differences in monitoring of different phenological periods are quantitatively analyzed. In this way, it is known whether the present invention is feasible in monitoring the phenological period of corn in a large area without relying on ground measured data, and in which phenological period the monitoring is most accurate. It is also convenient to compare the accuracy of the method in the previous study.

[0068]

[0069] T represents the total number of years. In this embodiment, T=12, t is the tth year, and t=1-12. This formula can calculate the estimation accuracy of the i-th phenological period in the past T years.

[0070]

[0071] Among them, Xcpr t i represents the CPR date of the i-th phenological period in the t-th year, Xest t i represents the estimated date of the i-th phenological period in year t, RMSE i With R i 2 Root mean square error and determination coefficient of phenological monitoring in the i-th phenological period, respectively.

[0072] The overall monitoring result accuracy is as follows Figure 7 As shown, the overall effect of using the present invention to monitor the five key phenological periods of corn is ideal, and the monitoring accuracy of the two indexes is less than 5.5 days, among which the monitoring of the silking period (Silking) and the dough period (Dough) is the most accurate. In the scatter plot comparing the monitoring results of the present invention with the NASS-CPR statistical values, the slope of the regression model of the monitoring results and the statistical data scatter points in the figure is close to 1, showing a high degree of correlation overall.

[0073] (7) Longitudinal evaluation of phenological monitoring results on an annual basis;

[0074] In this example, in order to verify the estimation accuracy of the present invention in each phenological period, the RMSE of the estimated phenological period of each study area and the phenological period in the CPR phenological report of that year were calculated in units of years. t With R t 2 By longitudinally observing the monitoring accuracy of each phenological period in a long time series on the time axis, it is possible to observe which years have poor monitoring accuracy and which years have good monitoring effects of which phenology. The differences in phenological monitoring between years are analyzed, and combined with the environmental stress conditions of each year, whether the phenological monitoring effect of the present invention is related to environmental stress or crop planting habit adjustment is studied.

[0075]

[0076] I represents the total number of phenological stages. In this embodiment, I=7. i represents each phenological stage, i=1-7. This formula can calculate the estimation accuracy of crop I phenology in the tth year.

[0077]

[0078] The unit of year is the accuracy of phenological monitoring. Figure 8 As shown in the figure, the average accuracy of the WDRVI index is RMSE 4.9 days, and the average accuracy of the NIRv index is RMSE 4.6 days. Overall, this method has an ideal monitoring effect on state-level corn on a twelve-year time scale, and the predicted phenological time is highly consistent with the phenological time statistically calculated by NASS-CPR. 2 Both are close to 1, indicating that the technical invention can explain about 99% of the variability in actual phenological statistical data, that is, the degree of fit between the predicted phenological events and the actual observations is very high. This shows that the present invention has a strong explanatory ability and fitting effect in the maize phenology monitoring in the study area. Among them, the phenological monitoring accuracy using the two vegetation indices in 2013, 2014, 2015, 2016, 2018, and 2020 was within 5 days, and the monitoring effect was good. The monitoring accuracy in 2012 was relatively poor, but the accuracy of monitoring using the NIRv index was better than that using the WDRVI index.

[0079] Fig. 9 This is a line chart of the five key phenological periods in the corn growth process monitored by this technical invention and the NASS-CPR statistical data (see Fig. 9ae), the interannual variation patterns of the corn phenological period monitored by the two vegetation indices using the present invention (SMF-wg (WDRVI) line with circular marks, SMF-wa (NIRV) line with triangular marks) are consistent with the NASS-CPR statistical data (gray line with squares). This confirms that the phenology monitored by the present invention presents a certain linear relationship with the statistical data, and the monitoring of the early or delayed phenological period caused by interannual planting time changes or environmental changes is also relatively accurate. Fig. 9 The medium-light grey bar graph represents the average annual precipitation. It can be seen from the figure that the highest precipitation was in 2009 and 2019, and the timing of each phenological period was delayed. The least precipitation was in 2012, as shown by the dark grey bar, and the timing of the phenological period was advanced. Figure 8 It can be seen that the error between the phenological monitoring results in 2012 and the CPR statistical data is the largest, but the monitoring effect of using the NIRv index in each phenological period is significantly better than that of using the WDRVI index. The average absolute error is reduced from 12.2d (WDRVI) to 6.8d (NIRv). This to a certain extent shows that NIRv has a certain stability in phenological monitoring in the face of major drought events, especially in the silking period, milky period, and waxy period. The absolute error is significantly reduced.

[0080] (8) Statistics of phenological monitoring results that do not rely on long-term ground observation data;

[0081] In this embodiment, the research scope, research time, whether long-term ground observation data are required, and phenological monitoring results are compared with other related methods to analyze the differences in methods.

[0082] In the early studies related to the SMF method, the establishment of shape models required long-term observation of ground stations, collection of crop phenology, height and other information, and most of the observation data was used as the standard for accuracy evaluation. The rSMF method proposed by Sakamoto et al. in 2018 got rid of the limitations of ground observation data. The use of publicly available CPRs to calibrate the reference date not only overcame the challenge of collecting representative field crop phenology observation data, but also promoted large-scale crop phenology retrieval in a wide geographical area. Among them, rSMF (local) has the best monitoring effect, and its overall accuracy is comparable to that of this example. In addition, the method proposed in the present invention has higher accuracy in monitoring the silking period, milk maturity period, and wax maturity period. The hybrid phenological matching model proposed by Diao et al. in 2021 also reduces the dependence on ground observation data by using public crop data. Among the four scenarios proposed by them, scenario one has the highest accuracy, but because scenario one pairs each year (16 years) with each ASD (9 ASDs), a total of 144 combinations are generated. Each combination of year and ASD has its own reference shape and reference date, so it is only applicable to fine spatial scales. Only scenario 4 is similar to the situation in this example. The crop growth reference model can be shared between different years and ASDs, and has the widest application range. In comparison, the accuracy of the present invention is slightly higher. After comparison with previous related studies, the method proposed in the present invention does not rely on long-term observation data, and realizes phenological monitoring in state units on the basis of improving the original method. The accuracy is equivalent to or slightly higher than that of previous studies, and can effectively monitor corn phenology.

[0083] (9) Conduct phenological regional mapping based on the crop phenological period monitoring results in Illinois in 2014;

[0084] In this embodiment, the monitoring results of the NIRv index in Illinois in 2014 are used as an example to map the phenological results of the five key phenological stages of corn (seedling stage, silking stage, milk stage, waxy stage, and maturity stage) ( Fig.10 ), revealing the spatiotemporal pattern of maize phenology in the study area and producing regional crop growth rate maps ( Fig.11 ), and analyzed the spatial variation of maize growth rate in the study area.

[0085] Fig.10 Each sub-graph is graded based on a 5-day time scale, and the pie chart shows the area ratio of pixels in each of the two levels. Fig.10 The pie chart of ae in the figure shows that the proportion of the area with early phenological time in the study area is getting smaller and smaller from the seedling stage to the maturity stage, while the proportion of the area with medium and late phenological time is gradually increasing. The spatial distribution map of phenology can illustrate the spatiotemporal pattern of corn phenology in Illinois. Fig.10As shown in a in the figure, the corn emergence time in the central and western regions (ASD6) of the study area is earlier (shown in red), and gradually becomes later towards the west and north, and the number of areas with late emergence time (shown in blue) at the north and south ends increases. Fig. 9 It can also be seen from the be in that the phenological time of corn in the central and western parts of the study area (ASD6) is the earliest as a whole, and it is relatively late in the middle and eastern regions (ASD4, 5, 7). The phenological time of each stage is gradually delayed from the central to the north, while the phenological period in the south (ASD8, 9) is the most delayed. The spatial distribution trends of the monitoring results of each phenological period are roughly the same.

[0086] Observing phenological mapping from a temporal dimension can reveal regional differences in plant growth rates. Fig.10 Each sub-graph ae is graded based on a 5-day time scale. By comparing the color changes in multiple sub-graphs, we can also know that the phenological time from the seedling stage to the milky stage is the earliest in the central and western parts of the study area (ASD6), and the areas with the earliest phenological time from the milky stage to the mature stage are gradually decreasing. It can be seen that the growth rate of corn in this region gradually slows down from the reproductive growth stage; corn planting in the southern part of the study area is relatively scattered (ASD8,9), and the phenological time from the seedling stage to the silking stage is gradually advanced, and then the phenological time is gradually delayed. Fig.11 is the distribution map of corn growth rate in the study area in 2014, and af is the distribution map of corn growth rate in different study areas. Fig.11 It can also be seen that the overall growth rate of corn in the ASD7 region is slower ( Fig.11 f) is mainly due to the slow growth rate during the vegetative growth stage ( Fig.11 d); the main reason for the faster overall growth rate of corn in the eastern region (ASD5) is that its reproductive growth stage grows faster; the growth rates of corn in the southern region (ASD8, 9) are relatively fast in both stages. In general, during the entire growth process of corn in Illinois in 2014, the phenological times in the central and western regions were relatively advanced, but the growth rates were relatively average, and the overall growth rates in the north and south ends and the eastern region were relatively fast. The present invention can monitor crop phenology accurately to the pixels of each region, and the spatial distribution characteristics of crop phenology in the study area and the crop growth rates in each region can be obtained through phenological regional mapping and crop growth rate mapping.

[0087] The method of the present invention has the following effects:

[0088] (1) Most previous studies relied on long-term ground observation data, which had certain limitations in application. This study proposed a phenological monitoring method that does not rely on ground-based measured data to address the problems of limited study area selection and reliance on ground-based measured data. It only used the MODIS surface reflectance product MOD09Q1 and freely accessible public data. It focused on calibrating state-level phenological parameters through public crop phenological information modeling, while calculating the vegetation index time series and performing data smoothing. Finally, it used the shape model fitting method to monitor the main phenological period of corn.

[0089] (2) The present invention effectively monitors the five key phenological stages of corn (seedling emergence, silking, milky stage, waxy stage, and maturity) on a state basis, with the silking and milky stages having the best monitoring effects (RMSE < 4d).

[0090] (3) The present invention uses the NIRv index for the first time to explore the effectiveness of the NIRv index in the shape fitting phenological monitoring method. The results show that NIRv can better estimate phenological characteristics than WDRVI when monitoring corn phenology at the state scale, among which the monitoring accuracy of the waxy stage is most significantly improved;

[0091] (4) In the present invention, the advance or delay of the annual phenological period can be monitored more accurately during phenological monitoring. In major drought events, the use of the NIRv index is more stable than the WDRVI index in phenological monitoring.

[0092] (5) The result mapping of the present invention can reflect the spatial variation of phenology and regional growth rate in the study area at the pixel scale.

[0093] In general, the present invention achieves the purpose of calibrating crop-specific phenological parameters in the model and determining the crop shape model using only freely accessible public data, thereby more accurately monitoring the phenology of state-level crops through shape model fitting, expanding the scope of application, reducing application costs, and providing convenience for integrating crop phenological data for more research.

[0094] See also Fig.12 , Fig.12 4 is a schematic diagram of the working of the hardware device of an embodiment of the present invention, wherein the hardware device specifically comprises: a corn phenology remote sensing monitoring device 401 based on NIRv and crop progress report, a processor 402 and a storage device 403.

[0095] A corn phenology remote sensing monitoring device 401 based on NIRv and crop progress report: The A device 401 implements the corn phenology remote sensing monitoring method based on NIRv and crop progress report.

[0096] Processor 402: The processor 402 loads and executes the instructions and data in the storage device 403 to implement the corn phenology remote sensing monitoring method based on NIRv and crop progress report.

[0097] Storage device 403: The storage device 403 stores instructions and data; the storage device 403 is used to implement the corn phenology remote sensing monitoring method based on NIRv and crop progress report.

[0098] The above specific implementations of the present invention do not constitute a limitation on the protection scope of the present invention. Any other corresponding changes and modifications made based on the technical concept of the present invention should be included in the protection scope of the claims of the present invention.

Claims

1. A method for remote sensing monitoring of corn phenology based on NIRv and crop progress report, characterized in that: The following steps are involved: S1: Collect public crop data and surface reflectance data; S2: Use the surface reflectance data collected in S1 to calculate the NIRv for several years; S3: Use the remote sensing data of the most densely planted corn area obtained by moving window search to calculate the vegetation near infrared reflectivity index NIRv of the year, and then smooth and average the vegetation near infrared reflectivity index NIRv of the year to obtain the vegetation index time series curve, and then construct the typical growth curve of corn; S4: Under the initial phenological parameters, the phenological parameters are updated with a preset time window and a preset step size. The optimal reference phenology of each stage is inverted by minimizing the difference between the phenological estimate and the regional scale phenological report of the public crop data in S1, and the crop shape model is constructed in combination with the typical maize growth curve in S3. S5: According to the NIRv calculated in step S2, the vegetation index time series curve of the study area for several years is obtained, and the crop shape model is fitted with the vegetation index time series curve of each year through an iterative algorithm to calculate the optimal phenological parameters, realize phenological monitoring, and obtain pixel-level phenological monitoring information; S6: Conduct regional phenological mapping of key phenological periods of maize based on pixel-level phenological monitoring information to reveal the spatiotemporal pattern of maize phenology in the study area, analyze the regional maize phenological distribution, produce regional crop growth rate maps, and analyze the spatial variation of maize growth rate in the study area. The key phenological periods include the emergence stage, silking stage, milky stage, waxy stage, and maturity stage.

2. A method for remote sensing monitoring of corn phenology based on NIRv and crop progress report as claimed in claim 1, characterized in that: In step S1, the surface reflectance data are the reflectance data of the red light band and the near infrared band of the MODIS surface reflectance product MOD09Q1; the public crop data include crop progress data and surface classification data.

3. A method for remote sensing monitoring of corn phenology based on NIRv and crop progress report as claimed in claim 1, characterized in that: In step S2, the calculation formula of NIRv is as follows: Among them, ρ nir and ρ red are near infrared band and red light band, and C is a set constant.

4. A method for remote sensing monitoring of corn phenology based on NIRv and crop progress report as claimed in claim 1, characterized in that: In step S3, a moving window of a preset size is used to search for the area with the maximum corn planting density in a certain year, and a vegetation index time series curve of the area is obtained. The vegetation index time series curve is smoothed to obtain a typical corn growth curve.

5. The method for remote sensing monitoring of corn phenology based on NIRv and crop progress report as claimed in claim 1, characterized in that: In step S4, the formula for minimizing the difference d between the phenological estimate and the regional-scale phenological report is as follows: Where n is the nth calibration, i represents each phenological stage, i = 1 to 7, and I is the total number of crop phenological stages; Xcpr i n Xest is the phenological period at the nth calibration of the i-th phenological stage in the CPR statistical data; i n is the estimated phenological period at the nth calibration of the i-th phenological stage.

6. A method for remote sensing monitoring of corn phenology based on NIRv and crop progress report as claimed in claim 1, characterized in that: In step S5, the fitting formula is as follows: in, is the time series segment after fitting, g(x) is the crop shape model; X i ref is the phenological parameter; xscale i is the stretch factor, tshift i is the translation factor; Through iterative algorithms, the translation factor and x-axis stretch factor between the shape model and the annual vegetation index time series curve are optimized to calculate the optimal geometric parameters. The criterion for determining the optimal parameters is the local VI time series segment after the shape model is fitted and matched. The correlation coefficient with the time series curve segment h(x) of the phenological period to be estimated is the largest, and the formula is as follows: Among them, R() represents the correlation coefficient between the shape model and the annual vegetation index time series curve; Estimated phenological period X i est It is calculated by referring to the phenological period X0 and the stretching factor and translation factor of the X-axis. The formula is as follows: Among them, X i est represents all estimated phenologies in the i-th phenological stage; If the correlation coefficient R is greater than or equal to the preset coefficient, the final estimated phenological period is output; if the correlation coefficient R is less than the preset coefficient, the corresponding final phenological estimation is deleted.

7. A storage device, characterized in that: The storage device stores instructions and data for implementing the corn phenology remote sensing monitoring method based on NIRv and crop progress report as described in any one of claims 1 to 6.

8. A corn phenology remote sensing monitoring device based on NIRv and crop progress report, characterized in that: include: A processor and a storage device; the processor loads and executes instructions and data in the storage device to implement the corn phenology remote sensing monitoring method based on NIRv and crop progress report as described in any one of claims 1 to 6.

Images