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

Abstract

The present application provides a kind of corn phenology remote sensing monitoring method based on NIRv and crop progress report. Including: utilize remote sensing data to construct NIRv time curve;Combining public crop progress report and cultivated land use data to construct corn growth model;Utilize last year data to calibrate phenology parameter and growth model to obtain shape model;Optimize traditional SMF method, fit shape model and each year vegetation index time curve, calculate optimal parameter, monitor five main phenological stages of corn;Obtain pixel-level corn phenology monitoring information and carry out regional phenology mapping, quantitatively evaluate regional phenology.The present application solves the modeling problem that crop phenological stage can be accurately estimated without relying on long-term ground observation data, expands the application range of shape model fitting in phenology monitoring, evaluates the phenology monitoring ability of NIRv index, and is suitable for larger area research.Further improve the precision of crop phenology remote sensing monitoring, assist government and farmer to accurately carry out crop scientific management.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of agricultural remote sensing, in particular to a method for remote sensing monitoring of crop phenology based on shape fitting model, and more particularly to a method for remote sensing monitoring of corn phenology based on NIRv and crop progress report. BACKGROUND

[0002] Crops have relatively stable physiological process characteristics (such as photosynthesis, dry matter distribution) at different phenological stages, and 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 stage of crop growth is also a key input information. There are many current methods for monitoring crop phenology, and the most traditional method is mainly obtained by field observation. The results obtained by this method are relatively accurate, but are mixed with subjective factors, have the disadvantages of time-consuming, labor-intensive, long observation period, small coverage area, and are difficult to obtain large-scale phenology information. The emergence of remote sensing technology provides a major opportunity for large-area spatially continuous crop phenology observation, with the characteristics of large coverage, short detection period, strong timeliness, low cost, etc., and can monitor regional or even global scale crop dynamics. Therefore, the remote sensing-based crop phenology detection method is a key technology for evaluating the impact of environmental stress on crops.

[0003] The current shape model fitting method (SMF) is a commonly used method for remote sensing-based phenology monitoring, but in many related studies, the SMF method is affected by the prior reference shape and reference feature date, and requires years of field observation data, which has a serious dependence on ground observation conditions, leading to challenges in large-scale crop phenology monitoring. At the same time, the contribution of the developed vegetation near-infrared reflectance index (NIRv) in the shape model fitting method is not clear.

[0004] Therefore, the present method solves the problem of dependence on long-term ground observation data in the traditional method of remote sensing monitoring of crop phenology, and invents a new method that can accurately monitor crop phenology only by relying on remote sensing data and public crop progress reports, which expands the application range of remote sensing phenology monitoring and reduces the prerequisite for remote sensing phenology monitoring, which will help remote sensing phenology monitoring to be convenient and accurate. At the same time, the present application further explores the monitoring ability of the NIRv index for crop phenology, which will help to further improve the accuracy of remote sensing monitoring of crop phenology, and provide some valuable reference basis and method for considering the spatial and temporal heterogeneity of ground crops and drought research. SUMMARY

[0005] The method aims at the dependence of the shape fitting model method on long-term ground observation data and the limitations of the method itself, and proposes a corn phenology remote sensing monitoring method based on NIRv and crop progress report.

[0006] S1: collecting public crop data and ground reflectivity data;

[0007] S2: calculating NIRv of several years using the ground reflectivity data collected in S1;

[0008] S3: using 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 current year, and performing smoothing and averaging processing on the vegetation near-infrared reflectivity index NIRv of the current year, then obtaining the vegetation index time curve, and then constructing a typical corn growth curve;

[0009] S4: under the initial phenological parameters, updating the phenological parameters with a preset time window and a preset step, and inversely calculating the optimal reference phenology of each stage by minimizing the difference between the phenological estimation and the regional scale phenology report of the public crop data in S1, and constructing a crop shape model combined with the typical corn growth curve in S3;

[0010] S5: according to the NIRv calculated in step S2, obtaining the vegetation index time curve of the research area in several years, fitting the crop shape model and the vegetation index time curve of each year through an iterative algorithm, calculating the optimal phenological parameters, realizing phenological monitoring, and obtaining pixel-level phenological monitoring information;

[0011] S6: according to the pixel-level phenological monitoring information, regional phenological mapping is carried out on the key phenological stages of corn, the spatio-temporal pattern of corn phenology in the research area is revealed, the regional corn phenology distribution is analyzed, the regional crop growth rate map is made, and the spatial variation of corn growth speed in the research area is analyzed, and the key phenological stages include emergence stage, silk stage, milk stage, dough 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 report.

[0013] A corn phenology remote sensing monitoring device based on NIRv and crop progress report, comprising: a processor and a storage device; the processor loads and executes the 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 application include:

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

[0016] (2) Previous studies have mostly used the WDRVI index and EVI index. This invention is the first to try using the NIRv index 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 rate can be analyzed at the pixel level. This can reveal the spatiotemporal characteristics and changes of regional crop phenology in more detail, providing a possibility for other agricultural-related studies to consider the spatiotemporal heterogeneity of ground crops. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in this embodiment, the accompanying drawings used in the description of the embodiment will be briefly introduced below. Obviously, the accompanying drawings described below are one embodiment of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

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

[0020] Figure 2 A schematic diagram illustrating the framework of a remote sensing monitoring method for maize phenology based on NIRv and crop progress reports, provided for an embodiment of the present invention;

[0021] Figure 3 This is a flowchart of the shape model construction process in an example of the present invention;

[0022] Figure 4 This is a flowchart of phenological parameter calibration in an example of the present invention;

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

[0024] Figure 6 This is a flowchart of the phenological monitoring process of the present invention;

[0025] Figure 7 This is a summary chart of the phenological monitoring accuracy in examples of this invention;

[0026] Figure 8 This is an example of an invention that uses an annual unit to evaluate phenological results.

[0027] Figure 9 This is a line graph comparing phenological monitoring results with CPR data in an example of the present invention;

[0028] Figure 10 This is an example of phenological regional mapping in this invention, using 2014 as an example;

[0029] Figure 11 This is a regional crop growth rate map using 2014 as an example in this invention.

[0030] Figure 12 This is a schematic diagram of the hardware device working in an embodiment of the present invention. Detailed Implementation

[0031] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0032] like Figures 1-2 As shown, this invention proposes a remote sensing monitoring method for maize phenology based on NIRv and crop progress reports, specifically including the following steps:

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

[0034] S2: Calculate the NIRv for several years using the surface reflectance data collected in S1;

[0035] S3: Using remote sensing data of the densest corn planting area obtained by moving window search, calculate the vegetation near-infrared reflectance index (NIRv) for the current year, and smooth and average the vegetation near-infrared reflectance index (NIRv) for the current year. Then, obtain the vegetation index time series curve, and then construct the typical corn growth curve.

[0036] S4: Under the initial phenological parameters, update the phenological parameters with a preset time window and a preset step size. Invert the optimal reference phenology for each stage by minimizing the difference between the phenological estimate and the regional scale phenological report of the public crop data in S1. Combine the typical maize growth curve in S3 to construct a crop shape model.

[0037] S5: Based on the NIRv calculated in step S2, obtain the time series curves of vegetation index in the study area for several years, fit the crop shape model with the time series curves of vegetation index each year through an iterative algorithm, calculate the optimal phenological parameters, realize phenological monitoring, and obtain pixel-level phenological monitoring information.

[0038] S6: Based on pixel-level phenological monitoring information, regional phenological maps are created for key phenological periods of maize to reveal the spatiotemporal pattern of maize phenology in the study area, analyze the regional distribution of maize phenology, create a regional crop growth rate map, and analyze the spatial variation of maize growth rate in the study area. The key phenological periods include seedling emergence, silking, milk stage, waxy stage, and maturity stage.

[0039] The following is a specific example of corn planting in Illinois, USA, from 2009 to 2020:

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

[0041] In step S1, the surface reflectance data is downloaded as reflectance data for 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. Public crop data includes crop progress data and surface classification data. The Crop Progress Report (CPR) is provided by the Agricultural Crop Progress Survey Project of the USDA's Bureau of Agricultural Statistics (NASS). Crop progress data is collected through regular surveys and observations of the growth and development of major crops in farmland. This data provides information on key stages of crop growth, such as sowing, germination, growth, and harvest, as well as progress compared to the same period in previous years. Spatially, it is recorded by state, and temporally, crop phenological progress is released weekly. The NASS-CPR records the growth stages of maize as seven stages: sowing, emergence, silking, milk stage, waxy stage, maturity, and harvest. The download period for CPR data is 2008-2020. Land cover classification data, also known as the Crop Data Layer (CDL), is crop-specific land use data created by the USDA's National Agricultural Statistics Service (NASS) based on medium-resolution optical satellite sensor data. It is freely available from the USDA website and is used for crop classification. Its resolution is refined to 30m. Both CPR and CDL data are available for download from 2008-2020. To select MODIS pixels for crop cover, the CDL data needs to be converted from the Universal Transverse Mercator projection to the MODIS sinusoidal projection using the nearest neighbor method.

[0042] (2) Calculation of near-infrared reflectance index of vegetation: The near-infrared reflectance index (NIRv) was calculated using the surface reflectance data collected in S1.

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

[0044]

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

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

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

[0048] Where, ρ nir and ρ red The near-infrared band (MOD09Q1 band 2) and the red band (MOD09Q1 band 1) are used, with α being the weighting coefficient. When α = 0.1, WDRVI shows a strong linear correlation with maize green leaf LAI.

[0049] (3) Crop shape model establishment without relying on long-term ground data: The vegetation near-infrared reflectance index was calculated by using the remote sensing data of the densest corn planting area in 2008 through a moving window search, and the vegetation index time series curve was obtained after smoothing and averaging, and a typical corn growth curve was constructed.

[0050] In this embodiment, in order to overcome the limitations of land cover change and the scope of the study area, the method for establishing the shape model is as follows: Figure 3 As shown, the first step is to extract pixels with the crop type of maize from the CDL data after projection transformation in 2008. These pixels are then resampled to 231.7m, the same resolution as the MODIS remote sensing data, using a 100*100 pixel moving window (approximately 537km). 2The location of the densest maize pixels in the study area was searched. The center of the densest area was located at 41°53.0′N, 88°52.60′W, in the northern part of the study area. Unlike previous studies based on shape models that used multi-year 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 from long-term crop phenological observations to calculate the NIRv and WDRVI time-series curves pixel by pixel. Based on this, SG filtering was used to smooth the time-series curves. Finally, the average NIRv and WDRVI time-series curves of each pixel in a 100*100 window were calculated as the maize shape model of this invention.

[0051] (4) Phenological parameter calibration 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 for each stage is obtained by minimizing the difference between the phenological estimates of 2009, 2019, and 2020 and the regional scale phenological reports of public crop data in S1.

[0052] In this embodiment, the phenological parameter calibration process is as follows: Figure 4 As shown, the specific steps are as follows: First, select the CPR data of Illinois in 2008 as the basic phenological parameter X. ref0 Then with X ref0 Based on this, within the time window w, each X is adjusted sequentially with a step size of b. ref0i (i represents each phenological stage from 1 to 7), obtain the phenological parameters X for the nth calibration. ref0i n (n represents the number of calibrations), based on 2009 MODIS data and NASS-CPR statistics, X ref0 Calibration was performed, and the reference phenology was calculated as X using the shape model fitting method. ref0 n The estimated phenological period X for Illinois in 2009 est n X est n This indicates that the basic phenological parameter X ref0 Given the initial values, the estimated phenological period after n calibrations. The value of X that minimizes the difference d between the estimated phenological period and the 2009 NASS-CPR statistics is obtained. ref0 n Noted as X ref1 Then X ref1 As a reference phenology, that is, X ref1 Using the new initial values, a new round of calibration iterations was conducted, continuing phenological calibration using data from 2019 and 2020 to ultimately obtain the optimal reference phenology X. 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 Xest represents the phenological period at the nth calibration of the i-th phenological stage in the CPR statistics. i n Xest is the estimated phenological period for the nth calibration in the i-th phenological stage. i n To use the phenological parameter vector set X ref0 The elements in the estimated phenological stage of the i-th phenological stage are calculated using the reference phenological model. For example, at the start of the iteration, the basic phenological parameters are 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 This refers to the phenological period estimated during the nth calibration of the first phenological stage. In the second iteration (using 2019 data in this example), the reference phenological parameter is X. ref1 The seven phenological stages are X ref11 X ref12 X ref13 X ref14 X ref15 X ref16 X ref17 Xest1 n This refers to the phenological period estimated during the nth calibration of the first phenological stage in the second iteration. Similarly, this process continues until the optimal reference phenology X is obtained in the final iteration. ref .

[0055] Calculations showed that a time window w of 10 days and a step size b of 2 days were most convenient for phenological calibration. The calibration results for phenological parameters are shown in the following figure. Figure 5 As shown, this example demonstrates the effect of phenological parameter calibration using the results of phenological monitoring using the WDRVI index in 2011 and 2012.

[0056] (5) Monitoring of major phenological stages of crops based on shape fitting;

[0057] In this embodiment, the time-series curves of vegetation index in the study area from 2009 to 2020 are obtained. An iterative algorithm is used to fit a shape model to the annual vegetation index time-series curves, and the optimal parameters are calculated to achieve phenological monitoring. The flowchart is as follows:Figure 6 As shown, the specific description is as follows:

[0058] First, a smoothed VI time-series growth curve for maize pixels needs to be established based on remote sensing data. Second, by introducing the X-axis and stretching and translation factors xscale and tshift on the time axis, the time-series growth curves of each pixel stage are fitted and matched with the shape models of each stage, as shown in the following formula:

[0059]

[0060] The function g(x) is a shape model; X i ref For phenological parameters; xscale i tshift is the stretch factor. i This is the translation factor.

[0061] The optimal geometric parameters are calculated by iteratively optimizing the translation factor and x-axis stretching factor between the two curves. The criterion for determining the optimal parameters is the local VI time series segment after shape model fitting and matching. The time series curve segment h(x) with the phenological period to be estimated has the largest correlation coefficient, as shown in the following formula:

[0062]

[0063] In the initial xscale i With a value of 1.0, we first set tshift. i The range is from -45 days to 45 days, with a step size of 1 day, and the optimal tshifti value is determined, which is the tshift when the objective function is maximized in the figure. i Value. Next, fix the tshift value. i Values, and in the same way, by changing xscale. i Values ​​(0.8–1.2), with a step size of 0.01, are used to estimate the optimal xscale. i Value. Repeat this process until both factors no longer change, at which point the iteration terminates. The final phenological period X to be estimated. est It can be calculated using the reference phenological period X0, the stretching factor of the X-axis, and the time shift factor, as shown in the following formula:

[0064]

[0065] If the correlation coefficient R >= 0.8, the result is output; if the correlation coefficient R < 0.8, the corresponding final phenological estimate is deleted.

[0066] (6) Horizontal evaluation of phenological monitoring results in units of phenological periods: After obtaining the pixel-level phenological monitoring results in step (5), regional phenological maps are made for the five key phenological periods of maize (emergence period, silking period, milk stage, waxy stage, and maturity stage) to reveal the spatiotemporal pattern of maize phenology in the study area, and regional crop growth rate maps are made to analyze the spatial variation of maize growth rate in the study area.

[0067] In this example, we first calculate the RMSE between the phenological period derived from this invention and the statistical data of the CPR of that phenological period over the 12 years from 2009 to 2020, taking phenological periods as the unit. i With R i 2 By comparing the monitoring accuracy of various phenological stages, this study quantitatively analyzes the accuracy of phenological monitoring in this invention and the differences in monitoring different phenological stages. This allows us to determine the feasibility of this invention for large-area monitoring of maize phenological stages without relying on ground-based measured data, and to identify which phenological stages it is most accurate for. Furthermore, it facilitates comparison with the accuracy of methods used in previous studies.

[0068]

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

[0070]

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

[0072] The overall monitoring results accuracy is as follows Figure 7 As shown, the overall effect of using this invention to monitor the five key phenological stages of maize is quite ideal. The monitoring accuracy using both indices is less than 5.5 days, with the monitoring of the silking and dough stages being the most accurate. In the scatter plot comparing the monitoring results of this invention with the NASS-CPR statistical values, the slope of the regression model between the monitoring results and the statistical data scatter points is close to 1, showing a high overall correlation.

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

[0074] In this example, to verify the estimation accuracy of the present invention for each phenological period, the RMSE of the estimated phenological data for each phenological period in the study area each year and the RMSE of the data for each phenological period in the CPR phenological report for that year are calculated on an annual basis during the accuracy evaluation. t With R t 2 By longitudinally observing the monitoring accuracy of each phenological stage over a long time series, it is possible to identify which years had poor monitoring accuracy and which years had good monitoring results for which phenological stage. Analyzing the differences in phenological monitoring between years and combining this with the environmental stress conditions each year, it is possible to study whether the phenological monitoring effect of this invention is related to environmental stress or adjustments in crop planting habits.

[0075]

[0076] I represents the total number of phenological stages. In this embodiment, I = 7, and i represents each phenological stage, i = 1 to 7. This formula can calculate the estimated accuracy of the I phenological stage of the crop in year t.

[0077]

[0078] The accuracy of phenological monitoring results is measured in years. Figure 8 As shown, the average accuracy using the WDRVI index is 4.9 days, and the average accuracy using the NIRv index is 4.6 days. Overall, this method performs well in monitoring state-level maize on a twelve-year timescale, and the predicted phenological timing shows high consistency with the phenological data from NASS-CPR. 2 The values ​​are all close to 1, indicating that this invention can explain approximately 99% of the variability in real-world phenological statistics, meaning that the predicted phenological events fit the actual observed values ​​very well. This demonstrates that this invention has strong explanatory power and fitting effect in maize phenological monitoring in the study area. Specifically, the phenological monitoring accuracy using both vegetation indices was within 5 days in 2013, 2014, 2015, 2016, 2018, and 2020, showing good monitoring results. The monitoring accuracy in 2012 was relatively poor, but the accuracy using the NIRv index was better than that using the WDRVI index.

[0079] Figure 9 Line graphs showing the five key phenological stages of maize growth monitored in this invention and their correlation with NASS-CPR statistics (see details). Figure 9The interannual variation patterns of maize phenological periods monitored using this invention (SMF-wg (WDRVI) line with circular markings and SMF-wa (NIRV) line with triangular markings) are consistent with NASS-CPR statistical data (gray line with squares). This confirms that the phenological and statistical data monitored by this invention exhibit a certain linear relationship, and are also relatively accurate in monitoring the early or late phenological periods caused by annual changes in planting time or environmental changes. Figure 9 The light gray bars represent the annual average precipitation. The graph shows that 2009 and 2019 had the most precipitation, while the timing of each phenological period was delayed. 2012 had the least precipitation, as shown by the dark gray bars, indicating that the phenological period occurred earlier. Figure 8 As can be seen, the phenological monitoring results in 2012 had the largest error compared to the CPR statistics. However, the NIRv index was significantly better than the WDRVI index in monitoring each phenological stage, with the average absolute error decreasing from 12.2 days (WDRVI) to 6.8 days (NIRv). This indicates that NIRv has a certain degree of stability in phenological monitoring in the face of major drought events, especially with a significant reduction in absolute error during the silking, milk, and waxy stages.

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

[0081] In this embodiment, the differences between the methods are analyzed by comparing their research scope, research time, whether long-term ground observation data is required, and phenological monitoring results with those of other related methods.

[0082] Early studies on the SMF method required long-term observation of ground stations to collect information on crop phenology, altitude, etc., and the accuracy evaluation was mostly based on observational data. The rSMF method proposed by Sakamoto et al. in 2018 overcame the limitations of ground observation data. By using publicly available CPRs to calibrate reference dates, it not only overcame the challenge of collecting representative field crop phenological observation data but also facilitated large-scale crop phenological retrieval across a wide geographical area. Among these, rSMF(local) showed the best monitoring effect, with overall accuracy comparable to that of this example, and even higher accuracy in monitoring the silking, milk, and waxy stages. The mixed phenological matching model proposed by Diao et al. in 2021 also reduced reliance on ground observation data by using public crop data. Of the four scenarios they proposed, scenario one had the highest accuracy. However, because scenario one paired each year (16 years) with each ASD (9 ASDs), it generated 144 combinations. Each year and ASD combination has its own reference shape and reference date, thus it is only suitable for fine spatial scales. Only scenario four is similar to this example, where the crop growth reference model can be shared across different years and ASDs, making it the most widely applicable. In comparison, the accuracy of this invention is slightly higher. Compared with previous related studies, the method proposed in this invention neither relies on long-term observation data nor fails to achieve state-level phenological monitoring based on improvements to existing methods. Its accuracy is comparable to or slightly higher than previous studies, and it can effectively monitor maize phenology.

[0083] (9) Phenological regional mapping was carried out using the crop phenological monitoring results in Illinois in 2014 as an example;

[0084] In this embodiment, taking the NIRv index monitoring results in Illinois in 2014 as an example, phenological results for five key phenological stages of maize (emergence, silking, milk, waxy, and maturity) are plotted. Figure 10 This study reveals the spatiotemporal pattern of maize phenology in the study area and produces a regional crop growth rate map. Figure 11 This study analyzed the spatial variation of maize growth rate in the research area.

[0085] Figure 10 Each sub-image is graded on a 5-day time scale, and the pie chart shows the area ratio of pixels in each pair of levels. Figure 10 The pie chart of ae in the diagram macroscopically shows that from emergence to maturity, the area with earlier phenological stages gradually decreases in proportion, while the area with medium and later phenological stages gradually increases in proportion. The spatial distribution map of phenology illustrates the spatiotemporal pattern of maize phenology in Illinois. For example... Figure 10As shown in 'a', maize emergence time is earlier in the central and western regions of the study area (ASD6) (shown in red), gradually decreasing westward and northward, with an increase in areas at both the northern and southern ends where emergence time is later (shown in blue). From Figure 9 The data also shows that the phenological time of maize in the central and western part of the study area (ASD6) is the earliest overall, while that in the middle and eastern regions (ASD4,5,7) is relatively later. The phenological time of each stage is gradually delayed from the central part to the north, while the phenological time in the south (ASD8,9) is the most delayed. The spatial distribution trend of the monitoring results of each phenological stage is roughly the same.

[0086] Observing phenological maps from a time perspective can reveal regional differences in plant growth rates. Figure 10 Each subplot (ae) is graded on a 5-day time scale. Comparing the color changes in multiple subplots also reveals that the western part of the study area (ASD6) has the earliest phenological time from emergence to milk stage. The area with the earliest phenological time from milk stage to maturity gradually decreases, indicating that the maize growth rate in this area gradually slows down from the reproductive growth stage. In the southern part of the study area, maize planting is more scattered (ASD8,9), and the phenological time from emergence to silking stage gradually advances, and then the phenological time gradually delays. Figure 11 The first image shows the distribution of maize growth rate in the study area in 2014, and the second image shows the distribution of maize growth rate in different study areas. Figure 11 It can also be seen that the overall growth rate of corn in the ASD7 region is relatively slow. Figure 11 The main reason for f) is that the growth rate is relatively slow during the vegetative growth stage. Figure 11 (d) In the eastern region (ASD5), the overall growth rate of maize was faster, mainly due to its faster growth rate during the reproductive growth stage; in the southern region (ASD8, 9), the growth rate of maize in both stages was relatively fast. Overall, during the entire growth process of maize in Illinois in 2014, the phenological timing in the Midwest region was relatively earlier, but the growth rate was more even, while the overall growth rate was faster in the northern and southern ends and the eastern region. This invention can monitor crop phenology with pixel-level precision. Through phenological regional mapping and crop growth rate mapping, the spatial distribution characteristics of crop phenology in the study area and the crop growth rate in each region can be obtained.

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

[0088] (1) Previous studies have mostly relied on long-term ground observation data, which has certain limitations in application. This study proposes a phenological monitoring method that does not rely on ground measurement data to address the problem of limited selection of study areas and reliance on ground measurement data. It only uses MODIS surface reflectance product MOD09Q1 and freely accessible public data. The focus is on calibrating state-level phenological parameters by modeling public crop phenological information, while calculating vegetation index time series and performing data smoothing. Finally, the shape model fitting method is used to monitor the main phenological stages of maize.

[0089] (2) This invention effectively monitored five key phenological stages of maize (emergence stage, silking stage, milk stage, waxy stage, and maturity stage) on a state-by-state basis, with the best monitoring effect on the silking stage and milk stage (RMSE<4d).

[0090] (3) This 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 when monitoring maize phenology at the state scale, the NIRv index can better estimate phenological characteristics than the WDRVI index, with the most significant improvement in monitoring accuracy for the waxy ripening stage.

[0091] (4) In phenological monitoring, the early or late phenological period of each year can be monitored more accurately. In major drought events, the NIRv index is more stable than the WDRVI index in phenological monitoring.

[0092] (5) The results mapping of the present invention can reflect the spatial changes of phenology and the growth rate of the region at the pixel scale.

[0093] Overall, this invention enables the calibration of crop-specific phenological parameters in a model and the determination of a crop shape model using only freely accessible public data. This allows for more accurate phenological monitoring of state-level crops through shape model fitting, expanding the scope of application, reducing application costs, and facilitating the integration of crop phenological data for more research.

[0094] Please see Figure 12 , Figure 12 This is a schematic diagram of the hardware device in operation according to an embodiment of the present invention. The hardware device specifically includes: a maize phenological remote sensing monitoring device 401 based on NIRv and crop progress reports, a processor 402, and a storage device 403.

[0095] A maize phenological remote sensing monitoring device 401 based on NIRv and crop progress reports: The device 401 implements the maize phenological remote sensing monitoring method based on NIRv and crop progress reports.

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

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

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

Claims

1. A method for corn phenology remote sensing monitoring based on NIRv and crop progress reporting, characterized by: The method comprises the following steps: S1: collecting public crop data and surface reflectivity data; S2: calculating NIRv of several years using the surface reflectivity data collected in S1; S3: calculating the vegetation near-infrared reflectivity index NIRv of the current year using the remote sensing data of the corn planting densest area obtained by mobile window search, and performing smoothing and averaging processing on the vegetation near-infrared reflectivity index NIRv of the current year, then obtaining a vegetation index time series curve, and further constructing a corn typical growth curve; S4: updating the phenological parameters with a preset time window and a preset step under the initial phenological parameters, and inverting the optimal reference phenology of each stage by minimizing the difference between the phenological estimation and the regional scale phenology report of the public crop data in S1, and constructing a crop shape model in combination with the typical corn growth curve in S3; S5: obtaining the vegetation index time series curve of the research area in several years according to the NIRv calculated in step S2, fitting the crop shape model with the vegetation index time series curve of each year through an iterative algorithm, calculating the optimal phenological parameters, realizing phenological monitoring, and obtaining pixel-level phenological monitoring information; S6: performing regional phenological mapping on the key phenological stages of corn according to the pixel-level phenological monitoring information, revealing the spatiotemporal pattern of corn phenology in the research area, analyzing the regional corn phenology distribution, making a regional crop growth rate map, and analyzing the spatial variation of corn growth rate in the research area, wherein the key phenological stages include the emergence stage, the silk stage, the milk stage, the dough stage, and the maturity stage.

2. The method for corn phenology remote sensing monitoring based on NIRv and crop progress reporting according to claim 1, characterized in that: In step S1, the surface reflectivity data are the reflectivity data of the red light band and the near-infrared band of the MODIS surface reflectivity product MOD09Q1; and the public crop data include crop progress data and surface classification data.

3. The method for corn phenology remote sensing monitoring based on NIRv and crop progress reporting according to claim 1, characterized in that: In step S2, the calculation formula of NIRv is as follows: wherein ρ nir and ρ red are in the near-infrared and red light bands, and C is a set constant.

4. The method for corn phenology remote sensing monitoring based on NIRv and crop progress reporting according to claim 1, characterized in that: In step S3, a mobile window of a preset size is used to search for the corn planting density maximum area in a certain year, and the vegetation index time series curve of the area is obtained, and the corn typical growth curve is obtained after smoothing the vegetation index time series curve.

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

6. The method for corn phenology remote sensing monitoring based on NIRv and crop progress reporting according to claim 1, characterized in that: In step S5, the fitting formula is as follows: wherein, is the fitted time series segment, 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; The optimal geometric parameters are calculated by iterative algorithm, which optimizes the translation factor and x-axis stretching factor between the shape model and the time series curve of vegetation index each year. The optimal parameters are determined by the standard of the shape model fitting the local VI time series segment after matching The correlation coefficient of the time series curve segment h(x) to be estimated and the phenological phase is maximum, and the formula is as follows: Wherein, R() represents the correlation coefficient of the shape model and the vegetation index time series curve of each year; Estimating phenophase X i est By referring to the phenophase X0 and the stretching factor and the translation factor of the X axis, the formula is as follows: wherein X i est represents all the estimated phenophases of the ith phenological phase; If the correlation coefficient R is greater than or equal to a preset coefficient, the final estimated phenological stage is output, and if the correlation coefficient R is less than the preset coefficient, the corresponding final phenological estimation is deleted.

7. A storage device, characterized by: The storage device stores instructions and data for implementing the corn phenological remote sensing monitoring method based on NIRv and crop progress report according to any one of claims 1-6.

8. A corn phenology remote sensing monitoring device based on NIRv and crop progress reporting, characterized by: It comprises: A processor and a storage device; the processor loads and executes the instructions and data in the storage device to implement the corn phenological remote sensing monitoring method based on NIRv and crop progress report according to any one of claims 1-6.

Images