CN110909821A - Method for carrying out high-space-time resolution vegetation index data fusion based on crop reference curve - Google Patents
Method for carrying out high-space-time resolution vegetation index data fusion based on crop reference curve Download PDFInfo
- Publication number
- CN110909821A CN110909821A CN201911217214.3A CN201911217214A CN110909821A CN 110909821 A CN110909821 A CN 110909821A CN 201911217214 A CN201911217214 A CN 201911217214A CN 110909821 A CN110909821 A CN 110909821A
- Authority
- CN
- China
- Prior art keywords
- curve
- modis
- landsat
- crop
- fitting
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/251—Fusion techniques of input or preprocessed data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/188—Vegetation
Abstract
The invention discloses a high-space-time resolution vegetation index data set reconstruction method based on a crop reference curve, which comprises the following steps: the method comprises the following steps: constructing a crop reference curve sample library; obtaining a plurality of MODIS reference curves of different crops based on pure pixels by utilizing the crop category information and the spatial position of a crop classification chart CDL, forming an MODIS crop reference curve sample library, and simultaneously combining MODIS local curves to jointly serve as MODIS candidate curves to participate in curve matching; step two: determining an initial fitting curve; step three: determining a final fitting curve; and aiming at the local difference between the Landsat initial fitting curve and the original Landsat time sequence NDVI image, obtaining a Landsat final fitting curve after quadratic fitting smoothing, thereby simultaneously obtaining vegetation index data with high spatial and temporal resolution, namely MODIS time resolution and Landsat spatial resolution.
Description
Technical Field
The invention relates to the technical field of agricultural remote sensing, in particular to a method for reconstructing a vegetation index with high space-time resolution based on a crop reference curve, which is suitable for agricultural remote sensing monitoring research of different farmland systems.
Background
When the earth system in a monitoring area or a global scale dynamically changes, such as vegetation coverage change, land utilization change and the like, the earth observation satellite can provide rich data and a monitoring method to obtain earth surface information, so that scientific basis is provided for resource management and policy making. However, remote sensing data acquired by earth observation satellites have a problem of mutual restriction in terms of high temporal resolution and high spatial resolution, and remote sensing monitoring represented by a field-scale study area has high requirements for both temporal resolution and spatial resolution, and therefore, a specific method is required to generate remote sensing data with high temporal and spatial resolution. Taking Landsat and MODIS images as examples, the Landsat image has a medium-high spatial resolution of 30m and can better meet the requirement of field scale remote sensing monitoring, but the 16-day reentry period and the more serious cloud coverage problem greatly limit the capability of the Landsat image for monitoring the rapid change of the earth surface; in contrast, the MODIS image has the ability to acquire data every day, but its low spatial resolution (250-1000m) makes it difficult to achieve accurate monitoring of a complex farmland system.
In order to meet the requirement of the remote sensing monitoring on the high-spatial-temporal-resolution data, if the requirement is simply considered from the perspective of a data source, more earth observation satellites with different application characteristics can be emitted, so that enough high-spatial-temporal-resolution images can be obtained, but the method needs more advanced software and hardware technologies, huge research and development production cost and the like, and has no practical operability. Therefore, based on the existing multi-source remote sensing image, the method for obtaining the high-space-time resolution image by utilizing the space-time fusion technology is concerned from the beginning of the proposal, and has great development and application. The technology well combines the time, space and spectral characteristics of different sensors, simulates the corresponding relation between the high-time low-time resolution data and the high-time low-space resolution data acquired by different satellites in the same region through an algorithm, generates an image with high time-space resolution by fusion, and well solves the problem of limitation of inconsistent time-space resolution on the remote sensing monitoring efficiency and precision. In the development process of recent decades, a space-time fusion classical algorithm STARFM has a good application effect in a plurality of fields such as vegetation dynamic change monitoring, crop phenological information extraction, water resource monitoring, surface temperature monitoring and the like, and a series of improved algorithms such as STAARCH, ESTARFM, FSDAF and the like are generated at the same time. The theoretical basis of such a spatio-temporal fusion algorithm is that, for high and low spatial resolution images in the same time period in the same region, the image features of the high and low spatial resolution images in the same time phase are in one-to-one correspondence, so that continuous time phase change features on the low spatial resolution image can be mapped onto the high spatial resolution image, and the defect of lack of time phase change information caused by a long acquisition period of the high spatial resolution image is overcome, so that spatial structure information provided by the high spatial resolution image and time dimension information provided by the low spatial resolution image are well combined, and image data with high spatio-temporal resolution are generated. However, the space-time fusion method has some limitations, and a low-spatial-resolution image which provides time dimensional information at first has a more serious mixed pixel phenomenon, so that the obtained space-time fusion product has higher space-time resolution, but the accuracy is often not high when ground feature identification and classification are carried out, misjudgment is easy to occur, and the algorithm is particularly suitable for remote sensing monitoring and identification of a spatial homogeneous region, but is not suitable for a region with larger spatial heterogeneity, and the classification identification of linear ground features, small ground features or irregular ground features is greatly influenced. Secondly, the space-time fusion algorithm needs to match and fit a plurality of high-space low-time-resolution images and low-space high-time-resolution images acquired under a clear air condition, the low-space-resolution images have higher time resolution and can meet the requirement of the number of available images under the clear air condition, but the high-space-resolution images are easily influenced by factors such as cloud coverage due to longer acquisition period, so that the number of the images available for matching is limited. Especially in areas with cloudy, rainy weather, the available high spatial resolution images are often fewer, and the error of the fusion result is larger.
Through the existing research, 84 percent of the areas of the global agricultural production units are all less than 0.02km2(approximately 140 x 140m,) while there are some regional differences in different regions, such as U.S. agricultureThe area of the industrial production unit is relatively large, the average value and the median value can reach 0.193km2 and 0.278km2, and the area of more than half of the field in part of African countries is less than 0.004km2(approximately 63 x 63m), while such smaller area fields typically have higher spatial heterogeneity. Therefore, when agricultural production conditions of field scales are monitored by using a remote sensing means, a common space-time fusion algorithm has great limitation and is not ideal in application effect. Meanwhile, considering that the agricultural production units of the field scale are widely distributed in the global scope, the remote sensing method for realizing the accurate monitoring of the agricultural production activities of the field scale has very important significance for reducing global poverty and guaranteeing the grain safety.
Disclosure of Invention
The technical problem to be solved by the invention is to provide a method for carrying out high-space-time resolution vegetation index data fusion based on a crop reference curve aiming at the defects of the prior art, and carry out space-time fusion reconstruction based on a multi-source remote sensing image to obtain high-space-time resolution vegetation index data which are suitable for remote sensing monitoring of a complex farmland system with high spatial heterogeneity.
In order to solve the technical problem, the invention provides a method for carrying out high-space-time resolution vegetation index data fusion based on a crop reference curve.
A method for high spatial and temporal resolution vegetation index data fusion based on crop reference curves comprises the following steps: the method comprises the following steps: constructing a crop reference curve sample library; obtaining a plurality of MODIS reference curves of different crops based on pure pixels by utilizing the crop category information and the spatial position of a crop classification chart CDL, forming an MODIS crop reference curve sample library, and simultaneously combining MODIS local curves to jointly serve as MODIS candidate curves to participate in curve matching; step two: determining an initial fitting curve; fitting and matching the pixel values of the Landsat time sequence NDVI images with the image pixel values of the corresponding positions of all the curves in the MODIS candidate curve, and finding out the candidate curve with the highest fitting degree as the optimal reference curve based on the MODIS crop reference curve; fitting the Landsat time sequence NDVI image by using the fitting parameters of the optimal reference curve to obtain a Landsat initial fitting curve; step three: determining a final fitting curve; and aiming at the local difference between the Landsat initial fitting curve and the original Landsat time sequence NDVI image, obtaining a Landsat final fitting curve after quadratic fitting smoothing, thereby simultaneously obtaining vegetation index data with high spatial and temporal resolution, namely MODIS time resolution and Landsat spatial resolution.
The method comprises the following steps:
step A: setting a moving window with the size of 25 × 25 pixels by utilizing crop category information in a crop classification Chart (CDL), and when more than 95% of pixels in the window belong to a single crop category in the crop classification chart through the moving pixel window, considering the window pixel as a pure pixel to acquire position information of the pure pixel window of the CDL;
and B: resampling and reprojection preprocessing are carried out on the MODIS data with low spatial resolution, so that the MODIS data with high spatial resolution and the CDL data with crop classification map have the same spatial resolution and projection information, geometric registration is carried out on the MODIS data and the Landsat data, and then MODIS time sequence vegetation index NDVI data are obtained through calculation; meanwhile, interpolation processing is carried out on partial Landsat7 images in the high-spatial-resolution Landsat data to obtain data of a complete coverage research area, and Landsat time sequence NDVI data are obtained through calculation;
and C: b, selecting a position corresponding to a part of pixel areas in the middle of the window on the MODIS time sequence vegetation index NDVI data by using the position information of the CDL pure pixel window obtained in the step A, and extracting to obtain an NDVI average value of pixels contained in the position to obtain a plurality of MODIS time sequence NDVI curves of different types of crops; and (3) eliminating the abnormal values of the partial curves by adopting a linear interpolation method, and constructing a sample library by using the curves as MODIS reference curves of different ground objects.
The method comprises the following steps:
step D: obtaining a time-series vegetation index NDVI curve of each pixel of the MODIS image by using the original value of the MODIS time-series vegetation index NDVI data obtained in the step B, referring to the MODIS local curve of the pixel, then comprehensively considering the MODIS local curve and the MODIS reference curve sample library obtained in the step C, and combining the MODIS local curve and the MODIS reference curve sample library together to be used as an MODIS candidate curve to participate in curve matching;
step E: and C, analyzing the correlation indexes R of all image pixels of the data and all curves in the MODIS crop reference curve sample library obtained in the step C by utilizing the Landsat time sequence NDVI data obtained in the step B2Size; and all image pixels of the Landsat time sequence NDVI data and the correlation index R of the MODIS local curve obtained in the step D2Size, correlation index R2And (3) taking the maximum curve as an MODIS optimal reference curve of the image pixel of the Landsat time sequence NDVI data, and then further fitting the Landsat time sequence NDVI data and the MODIS optimal reference curve in the step (B) through a formula (1) to obtain a Landsat fitting curve.
L(x)＝a*M(x+x0)+b (1)
Wherein x represents a certain day of the year, L (x) is a Landasat time series NDVI curve function, M (x) is an MODIS optimal reference curve function, x0 represents the offset in time possibly existing between two curves, the range is +/-30 days, and a and b are fitting parameters obtained by using a least square method;
step F: calculating a correlation index R between the Landsat time sequence NDVI image pixel value in the step B and all candidate curve pixel values of the corresponding position MODIS2Analysis of the correlation index R between the two2Selecting an optimal MODIS reference curve;
step G: and E, confirming an optimal reference curve by using the steps E and F, wherein the optimal time offset x0 is 0, obtaining corresponding fitting parameters a and b simultaneously, and then obtaining a Landsat initial fitting curve by using the MODIS optimal reference curve and the corresponding fitting parameters.
The method comprises the following steps: step H: and D, obtaining a difference value between the NDVI fitting values of corresponding pixels on the Landsat initial fitting curve by calculating the pixel value (clear sky pixel) of the NDVI of the original Landsat time sequence and the step G, fitting the linear interpolation result to the initial fitting curve by adopting a linear interpolation method to reduce the difference between the NDVI fitting values and the initial fitting curve, and finally smoothing the optimized fitting curve by utilizing a Gaussian function in an IDL programming platform to ensure that the curve contains all available NDVI values of the original Landsat time sequence as much as possible, and calling the curve as a final fitting curve.
The method, step F, has 60 correlation indexes R for each MODIS crop reference curve2And the value range is concentrated between 0.11 and 0.92, then a candidate curve with the largest correlation index is selected as a curve to be matched of the MODIS crop reference curve, the correlation index between the curve to be matched corresponding to all the MODIS crop reference curves and the Landsat time series NDVI data is compared, and the value range of the correlation index between the MODIS local curve and the Landsat time series NDVI data is concentrated between 0.78 and 0.92. If MODIS local curve correlation index R2If the value is maximum, directly taking the MODIS local curve as the optimal reference curve; if the correlation index R2If the maximum value appears in the curve to be matched of the MODIS crop reference curve, calculating all the candidate curve correlation indexes R2Is selected to be higher than R2And averaging the candidate curves of the average value again, and taking the finally obtained average curve as an optimal reference curve.
In the method, MODIS reference curves of different ground features are obtained by utilizing a pure pixel region selected by a result with higher classification precision in known crop classification chart CDL data, and the specific setting parameters in the step A are empirical parameters according to multiple experiments.
Has the advantages that:
the invention provides a data space-time fusion method based on a crop reference curve according to the characteristic that vegetation index data with different spatial resolutions of different sensor images have a linear relation, and is suitable for remote sensing monitoring of a complex farmland system with higher spatial heterogeneity. Setting a moving window by using the CDL data of the known crop classification layer to determine pure pixel areas of different types of crops, and constructing an MODIS crop reference curve sample library based on the pure pixel areas; establishing an optimal reference curve selection method by comparing the correlation between the time sequence NDVI data of the Landsat image pixels and an MODIS crop reference curve and the local curve of the original NDVI data of pixels corresponding to the MODIS, and obtaining an initial fitting curve based on the fitting parameters of the optimal reference curve and the Landsat time sequence NDVI original data; the function of the Landsat original effective value in curve fitting is retained to the maximum extent through methods such as linear interpolation and Gaussian function fitting, the phenomenon of initial overfitting is further reduced, a final fitting curve is obtained, and therefore a vegetation index data set with high time and high spatial resolution is generated. The method comprehensively considers two problems of serious mixed pixel of the low-spatial-resolution image and less matched data of the traditional fusion method, reduces the influence of the mixed pixel by utilizing known and reliable classified data, avoids the problem that the matchable number of the high-spatial-resolution image is limited due to cloud coverage by transmitting the time dimension information of the optimal reference curve obtained based on the high-temporal-resolution image, can utilize all available original image information to the maximum extent, and can well apply the generated vegetation index data with high temporal and spatial resolution to the remote sensing monitoring of a complex farmland system.
Drawings
FIG. 1 is a MODIS reference graph of four typical features; the upper left is a forest, the upper right is a double-season crop, the lower left is corn, and the lower right is soybean;
FIG. 2 shows MODIS reference curves, best reference curves, initial fit curves, final fit curves, and Landsat time series NDVI values for four representative species; corn on the upper left, soybean on the upper right, double-season crop on the lower left, and forest on the lower right; (1) a MODIS reference curve, (2) an optimal reference curve, (3) an initial fitting curve, (4) a final fitting curve, and dots represent Landsat time series NDVI values;
FIG. 3 is a comparison of the NDVI fit curves on three crop types based on crop reference curves and on STARFM in 2013; the graph (a) is a crop classification graph, the graphs (b), (c) and (d) are graphs of corn, soybean and double-season crops corresponding to the positions of dots in the graph (a), the small dots in the graphs (b), (c) and (d) are time-series Landsat images, the large dots are STARFM paired images, (1) a fitting curve based on a crop reference curve method, (2) a STARFM method fitting curve, and (3) an MODIS reference curve;
Detailed Description
The present invention will be described in detail with reference to specific examples.
The invention provides a method for carrying out high-space-time resolution vegetation index data fusion based on a crop reference curve, which comprises the following steps:
s1: the study area of this example was located in the Choptank river basin on the east coast of the united states, including parts of the maryland, terahua, and new jersey areas where land utilization was predominantly 58% of agricultural land, 33% of forests, and only 9% of urban construction land, with the most common crop planting regimes including corn-soybean rotation, and dual-season soybean-winter wheat planting. Taking the CDL data of the crop classification diagrams in 2013 and 2014 as reliable crop distribution data, then setting a moving window with the size of 25 × 25 pixels, and by moving the pixel window, when more than 95% of pixels in the window belong to a single crop category in the CDL, considering the window as a pure pixel window, and simultaneously acquiring the position information of the pure pixel window of the CDL, wherein the pure pixel window of four types of ground objects such as corn, soybean, double-season crops (winter wheat and soybean) and forest is mainly selected in the research example;
s2: utilizing MRT (MODIS reproduction tool) tool to carry out preprocessing such as resampling and Reprojection on MODIS data (MCD43A4 collection 6, the resolution is 500m, the time range is 1/2013-12/2014-31/2014, the image row and column is h12v05) with low spatial resolution, so that the spatial resolution and the projection information are the same as those of Landsat data with high spatial resolution and crop classification map CDL data, the resampling method is bilinear interpolation, the Reprojection is UTM (Universal TRANSVERSE Mercury grid system), automatic geometric registration (automatic mapping method) is carried out on the MODIS data and the Landsat data, and MODIS time series NDVI data are obtained through calculation; meanwhile, the high spatial resolution Landsat data used in the research comprise 16-stage Landsat7 data and 18-stage Landsat8 data which are available in 2013 and 2014, and the row numbers are 33 and 14; because the Landsat7 images in certain periods have obvious banding problems, a GNSPI (geographic information System medical Pixel Interpolator) interpolation method is used for carrying out interpolation processing on the Landsat7 images with banding gaps to obtain data of a complete coverage research area, and finally Landsat time sequence NDVI data are obtained through calculation;
s3: and (4) selecting the position corresponding to the partial pixel area (16 × 16 pixels) in the middle of the CDL pure pixel window on the MODIS time sequence vegetation index NDVI data by using the position information of the CDL pure pixel window obtained in the step (S1), and extracting to obtain an NDVI average value of the pixels contained in the position, so that a plurality of MODIS time sequence NDVI curves of different types of crops can be obtained according to the number of the pure pixel windows of the different types of crops. In the present study example, 23 corn curves, 1 soybean curve, 5 double-season crop (winter wheat and soybean) curves and 26 forest curves were obtained in 2013, and 24 corn curves, 3 soybean curves, 10 double-season crop (winter wheat and soybean) curves and 23 forest curves were obtained in 2014; because some curves still have abnormal values, the abnormal values are eliminated by adopting a linear interpolation method in the example, and the curves are used as MODIS crop reference curves of different crops to construct an MODIS crop reference curve sample library, wherein FIG. 1 is an MODIS reference curve graph of four typical objects in 2014;
s4: and (4) obtaining a time-series vegetation index NDVI curve of each pixel of the MODIS image by using the original value of the MODIS time-series vegetation index NDVI data obtained by calculation in the step (S2), which is called an MODIS local curve of the pixel. Because the Landsat image pixel and the corresponding MDOSI image pixel may both be pure pixels, the NDVI curves of the Landsat image pixel and the MDOSI image pixel have good correlation, or both are mixed pixels but still have good correlation, the research comprehensively considers the MODIS local curve and the MODIS crop reference curve sample library obtained in the step S3, and combines the two to be used as a candidate curve to participate in curve matching;
s5: analyzing the correlation index R of all image pixels of the data and all curves in the MODIS crop reference curve sample library obtained in the step S3 and the MODIS local curve obtained in the step S4 by using the Landsat time series NDVI data obtained in the step S22Size, correlation index R2The maximum curve is used as an MODIS optimal reference curve of the image pixels of the Landsat time series NDVI data, and then the Landsat time series NDVI data and the MODIS optimal reference curve in step S2 are further fitted by using a formula (1), so as to obtain a Landsat fitting curve.
L(x)＝a*M(x+x0)+b (1)
Wherein x represents a certain day of the year, L (x) is a Landasat time series NDVI curve function, M (x) is an MODIS optimal reference curve function, x0 represents the offset in time possibly existing between two curves, the range is +/-30 days, and a and b are fitting parameters obtained by using a least square method;
s6: calculating the correlation index R between the Landsat time sequence NDVI image pixel value in the step S2 and all candidate curve pixel values of the corresponding position MODIS2Analysis of the correlation index R between the two2And (5) sizing and selecting an optimal MODIS reference curve. For each MODIS crop reference curve, there are 60 correlation indexes R2And the value range is concentrated between 0.11 and 0.92, then a candidate curve with the largest correlation index is selected as a curve to be matched of the MODIS crop reference curve, the correlation index between the curve to be matched corresponding to all the MODIS crop reference curves and the Landsat time series NDVI data is compared, and the value range of the correlation index between the MODIS local curve and the Landsat time series NDVI data is concentrated between 0.78 and 0.92. If MODIS local curve correlation index R2If the value is maximum, directly taking the MODIS local curve as the optimal reference curve; if the correlation index R2If the maximum value appears in the curve to be matched of the MODIS crop reference curve, calculating all the candidate curve correlation indexes R2Is selected to be higher than R2The candidate curves for the average are averaged again and,and taking the finally obtained average curve as an optimal reference curve. As shown in fig. 2, the black dots are Landsat time series NDVI values calculated in step S2, the curve (1) is the MODIS reference curve obtained in step S3, and the curve (2) is the optimal MODIS reference curve obtained by fitting in steps S5 and S6;
s7: the optimal reference curve can be confirmed by steps S5 and S6, where the optimal time offset x0 is 0, and the corresponding fitting parameters a and b are obtained at the same time, and then the Landsat initial fitting curve is obtained by using the MODIS optimal reference curve and the corresponding fitting parameters. As shown in fig. 2, the curve (3) is an initial fitting curve, and at this time, the initial fitting curve and the Landsat time series NDVI value of the target pixel region have a better fitting relationship, but still some Landsat original NDVI values cannot be well fitted to the initial fitting curve;
s8: and (3) calculating the difference value between the pixel value (clear sky pixel) of the NDVI of the original Landsat time sequence and the corresponding pixel NDVI fitting value on the Landsat initial fitting curve obtained in the step (S7), fitting the linear interpolation result to the initial fitting curve by adopting a linear interpolation method to reduce the difference between the two fitting values, and finally smoothing the optimized fitting curve by utilizing a Gaussian function in an IDL programming platform to ensure that the curve contains all available NDVI of the original Landsat time sequence as much as possible, so that the curve is called a final fitting curve, as shown in FIG. 2, the curve (4) is the final fitting curve of Landsat, and at the moment, the NDVI of the Landsat time sequence and the final fitting curve of the Landsat have a good fitting relationship.
In order to compare the effects of the data reconstruction method based on the crop reference curve in the research, a comparison experiment is performed by using a space-time fusion classical algorithm STARFM, and the algorithm needs to fit more paired images acquired under a clear air condition. Fig. 3 shows NDVI fitted curves of three different crops obtained in 2013 based on a crop reference curve and based on the STARFM algorithm, where the curve (3) is a MODIS reference curve, the large black dots are cloud-free paired images available for the STARFM algorithm, the curve (2) is a fitted curve obtained based on the STARFM algorithm, the small black dots are all Landsat images available for the crop reference curve reconstruction method adopted in the present study, and the curve (1) is a fitted curve obtained based on the crop reference curve reconstruction method. It can be found that the coarse resolution MODIS image is affected by the resolution and the mixed pixel, the classical algorithm STARFM is affected by cloud coverage and the like, so that an available paired image is lacked, the data reconstruction effect is not ideal, and the climate change characteristics of different crops in different growth periods cannot be reflected. The data reconstruction method adopted in the research greatly avoids the influence of cloud coverage, and meanwhile, the obtained high-space-time resolution data can well reflect the key phenological change characteristics of different crops.
In conclusion, the research sets a pure pixel area through known crop classification data, obtains a crop reference curve based on low spatial resolution by using a pure pixel concept, then performs fitting conversion and secondary optimization adjustment on the high spatial resolution image and the crop reference curve through fitting modes such as translation and stretching, well converts time dimension information of the low spatial resolution crop reference curve to the high spatial resolution image, realizes high spatial and temporal resolution vegetation index data fusion reconstruction based on the crop reference curve, can well solve the problems of interference and limitation of factors such as mixed pixels and cloud coverage on data fusion of areas with high spatial heterogeneity by an original data space-time fusion algorithm, and has important significance for remote sensing monitoring of farmland systems of fields with different complexity degrees.
It will be understood that modifications and variations can be made by persons skilled in the art in light of the above teachings and all such modifications and variations are intended to be included within the scope of the invention as defined in the appended claims.
Claims (6)
1. A method for carrying out high spatial and temporal resolution vegetation index data fusion based on a crop reference curve is characterized by comprising the following steps: the method comprises the following steps: constructing a crop reference curve sample library; obtaining a plurality of MODIS reference curves of different crops based on pure pixels by utilizing the crop category information and the spatial position of a crop classification chart CDL, forming an MODIS crop reference curve sample library, and simultaneously combining MODIS local curves to jointly serve as MODIS candidate curves to participate in curve matching; step two: determining an initial fitting curve; fitting and matching the pixel values of the Landsat time sequence NDVI images with the image pixel values of the corresponding positions of all the curves in the MODIS candidate curve, and finding out the candidate curve with the highest fitting degree as the optimal reference curve based on the MODIS crop reference curve; fitting the Landsat time sequence NDVI image by using the fitting parameters of the optimal reference curve to obtain a Landsat initial fitting curve; step three: determining a final fitting curve; and aiming at the local difference between the Landsat initial fitting curve and the original Landsat time sequence NDVI image, obtaining a Landsat final fitting curve after quadratic fitting smoothing, thereby simultaneously obtaining vegetation index data with high spatial and temporal resolution, namely MODIS time resolution and Landsat spatial resolution.
2. The method of claim 1, wherein the first step comprises the steps of:
step A: setting a moving window with the size of 25 × 25 pixels by utilizing crop category information in a crop classification Chart (CDL), and when more than 95% of pixels in the window belong to a single crop category in the crop classification chart through the moving pixel window, considering the window pixel as a pure pixel to acquire position information of the pure pixel window of the CDL;
and B: resampling and reprojection preprocessing are carried out on the MODIS data with low spatial resolution, so that the MODIS data with high spatial resolution and the CDL data with crop classification map have the same spatial resolution and projection information, geometric registration is carried out on the MODIS data and the Landsat data, and then MODIS time sequence vegetation index NDVI data are obtained through calculation; meanwhile, interpolation processing is carried out on partial Landsat7 images in the high-spatial-resolution Landsat data to obtain data of a complete coverage research area, and Landsat time sequence NDVI data are obtained through calculation;
and C: b, selecting a position corresponding to a part of pixel areas in the middle of the window on the MODIS time sequence vegetation index NDVI data by using the position information of the CDL pure pixel window obtained in the step A, and extracting to obtain an NDVI average value of pixels contained in the position to obtain a plurality of MODIS time sequence NDVI curves of different types of crops; and (3) eliminating the abnormal values of the partial curves by adopting a linear interpolation method, and constructing a sample library by using the curves as MODIS reference curves of different ground objects.
3. The method according to claim 2, wherein the second step comprises the steps of:
step D: obtaining a time-series vegetation index NDVI curve of each pixel of the MODIS image by using the original value of the MODIS time-series vegetation index NDVI data obtained in the step B, referring to the MODIS local curve of the pixel, then comprehensively considering the MODIS local curve and the MODIS reference curve sample library obtained in the step C, and combining the MODIS local curve and the MODIS reference curve sample library together to be used as an MODIS candidate curve to participate in curve matching;
step E: and C, analyzing the correlation indexes R of all image pixels of the data and all curves in the MODIS crop reference curve sample library obtained in the step C by utilizing the Landsat time sequence NDVI data obtained in the step B2Size; and all image pixels of the Landsat time sequence NDVI data and the correlation index R of the MODIS local curve obtained in the step D2Size, correlation index R2And (3) taking the maximum curve as an MODIS optimal reference curve of the image pixel of the Landsat time sequence NDVI data, and then further fitting the Landsat time sequence NDVI data and the MODIS optimal reference curve in the step (B) through a formula (1) to obtain a Landsat fitting curve.
L(x)＝a*M(x+x0)+b (1)
Wherein x represents a certain day of the year, L (x) is a Landasat time series NDVI curve function, M (x) is an MODIS optimal reference curve function, x0 represents the offset in time possibly existing between two curves, the range is +/-30 days, and a and b are fitting parameters obtained by using a least square method;
step F: calculating a correlation index R between the Landsat time sequence NDVI image pixel value in the step B and all candidate curve pixel values of the corresponding position MODIS2Analysis of the correlation index R between the two2Selecting optimal MODIS parametersExamining a curve;
step G: and E, confirming an optimal reference curve by using the steps E and F, wherein the optimal time offset x0 is 0, obtaining corresponding fitting parameters a and b simultaneously, and then obtaining a Landsat initial fitting curve by using the MODIS optimal reference curve and the corresponding fitting parameters.
4. The method of claim 3, wherein step three comprises the steps of: step H: and D, obtaining a difference value between the NDVI fitting values of corresponding pixels on the Landsat initial fitting curve by calculating the pixel value (clear sky pixel) of the NDVI of the original Landsat time sequence and the step G, fitting the linear interpolation result to the initial fitting curve by adopting a linear interpolation method to reduce the difference between the NDVI fitting values and the initial fitting curve, and finally smoothing the optimized fitting curve by utilizing a Gaussian function in an IDL programming platform to ensure that the curve contains all available NDVI values of the original Landsat time sequence as much as possible, and calling the curve as a final fitting curve.
5. The method according to claim 4, wherein in step F, there are 60 correlation indexes R for each MODIS crop reference curve2And the value range is concentrated between 0.11 and 0.92, then a candidate curve with the largest correlation index is selected as a curve to be matched of the MODIS crop reference curve, the correlation index between the curve to be matched corresponding to all the MODIS crop reference curves and the Landsat time series NDVI data is compared, and the value range of the correlation index between the MODIS local curve and the Landsat time series NDVI data is concentrated between 0.78 and 0.92. If MODIS local curve correlation index R2If the value is maximum, directly taking the MODIS local curve as the optimal reference curve; if the correlation index R2If the maximum value appears in the curve to be matched of the MODIS crop reference curve, calculating all the candidate curve correlation indexes R2Is selected to be higher than R2And averaging the candidate curves of the average value again, and taking the finally obtained average curve as an optimal reference curve.
6. The method as claimed in claim 2, wherein the MODIS reference curves of different features are obtained by using the selected pure pixel region with higher classification precision result in the known CDL data of the crop classification map, and the specific setting parameters in the step A are empirical parameters according to a plurality of experiments.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911217214.3A CN110909821B (en) | 2019-12-03 | 2019-12-03 | Method for carrying out high-space-time resolution vegetation index data fusion based on crop reference curve |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911217214.3A CN110909821B (en) | 2019-12-03 | 2019-12-03 | Method for carrying out high-space-time resolution vegetation index data fusion based on crop reference curve |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110909821A true CN110909821A (en) | 2020-03-24 |
| CN110909821B CN110909821B (en) | 2020-07-28 |
Family
ID=69821743
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911217214.3A Expired - Fee Related CN110909821B (en) | 2019-12-03 | 2019-12-03 | Method for carrying out high-space-time resolution vegetation index data fusion based on crop reference curve |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110909821B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113643409A (en) * | 2021-08-24 | 2021-11-12 | 中国农业大学 | Method and device for representing vegetation production rate and storage medium |
| CN115062863A (en) * | 2022-07-13 | 2022-09-16 | 中国农业科学院农业资源与农业区划研究所 | Apple flowering phase prediction method based on crop reference curve and accumulated temperature correction |
Citations (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102831310A (en) * | 2012-08-17 | 2012-12-19 | 北京师范大学 | Method for building high-spatial resolution NDVI (normalized difference vegetation index) time series data |
| CN103678914A (en) * | 2013-12-16 | 2014-03-26 | 中国科学院遥感与数字地球研究所 | Alpine grassland soil respiration estimation method based on satellite remote sensing data |
| CN103954957A (en) * | 2014-05-22 | 2014-07-30 | 北京空间机电研究所 | Remote-sensing image atmosphere proximity effect correction method based on atmosphere separation MTF |
| US20150146928A1 (en) * | 2013-11-27 | 2015-05-28 | Electronics And Telecommunications Research Institute | Apparatus and method for tracking motion based on hybrid camera |
| CN105046648A (en) * | 2015-06-25 | 2015-11-11 | 北京师范大学 | Method for constructing high temporal-spatial remote sensing data |
| US20170011089A1 (en) * | 2015-07-07 | 2017-01-12 | International Business Machines Corporation | Parallel querying of adjustable resolution geospatial database |
| WO2017129940A1 (en) * | 2016-01-29 | 2017-08-03 | Global Surface Intelligence Limited | System and method for earth observation and analysis |
| CN107103584A (en) * | 2017-04-11 | 2017-08-29 | 北京师范大学 | A kind of production high-spatial and temporal resolution NDVI weighted based on space-time method |
| CN107783106A (en) * | 2016-08-25 | 2018-03-09 | 大连楼兰科技股份有限公司 | Data fusion method between unmanned plane and barrier |
| CN108241871A (en) * | 2017-12-27 | 2018-07-03 | 华北水利水电大学 | Laser point cloud and visual fusion data classification method based on multiple features |
| CN108613933A (en) * | 2018-06-13 | 2018-10-02 | 中南林业科技大学 | Forest land arid space-time dynamic monitoring method based on multi-sources RS data fusion |
| CN108982369A (en) * | 2018-04-28 | 2018-12-11 | 中国农业大学 | Merge the plot scale crop condition monitoring method of GF-1WFV and MODIS data |
| CN109358162A (en) * | 2018-11-06 | 2019-02-19 | 福州大学 | A kind of construction method of the novel remote sensing ecology index based on space geometry principle |
| CN109543630A (en) * | 2018-11-28 | 2019-03-29 | 苏州中科天启遥感科技有限公司 | Remote sensing image forest land extracting method and system, storage medium, electronic equipment based on deep learning |
| CN110363246A (en) * | 2019-07-18 | 2019-10-22 | 滨州学院 | A kind of fusion method of high-spatial and temporal resolution vegetation index NDVI |
| US10489689B1 (en) * | 2016-03-30 | 2019-11-26 | Descartes Labs, Inc. | Iterative relabeling using spectral neighborhoods |
| CN110503137A (en) * | 2019-07-29 | 2019-11-26 | 电子科技大学 | Based on the determination method of the remote sensing image temporal-spatial fusion base image pair of mixing together |
-
2019
- 2019-12-03 CN CN201911217214.3A patent/CN110909821B/en not_active Expired - Fee Related
Patent Citations (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102831310A (en) * | 2012-08-17 | 2012-12-19 | 北京师范大学 | Method for building high-spatial resolution NDVI (normalized difference vegetation index) time series data |
| US20150146928A1 (en) * | 2013-11-27 | 2015-05-28 | Electronics And Telecommunications Research Institute | Apparatus and method for tracking motion based on hybrid camera |
| CN103678914A (en) * | 2013-12-16 | 2014-03-26 | 中国科学院遥感与数字地球研究所 | Alpine grassland soil respiration estimation method based on satellite remote sensing data |
| CN103954957A (en) * | 2014-05-22 | 2014-07-30 | 北京空间机电研究所 | Remote-sensing image atmosphere proximity effect correction method based on atmosphere separation MTF |
| CN105046648A (en) * | 2015-06-25 | 2015-11-11 | 北京师范大学 | Method for constructing high temporal-spatial remote sensing data |
| US20170011089A1 (en) * | 2015-07-07 | 2017-01-12 | International Business Machines Corporation | Parallel querying of adjustable resolution geospatial database |
| WO2017129940A1 (en) * | 2016-01-29 | 2017-08-03 | Global Surface Intelligence Limited | System and method for earth observation and analysis |
| US10489689B1 (en) * | 2016-03-30 | 2019-11-26 | Descartes Labs, Inc. | Iterative relabeling using spectral neighborhoods |
| CN107783106A (en) * | 2016-08-25 | 2018-03-09 | 大连楼兰科技股份有限公司 | Data fusion method between unmanned plane and barrier |
| CN107103584A (en) * | 2017-04-11 | 2017-08-29 | 北京师范大学 | A kind of production high-spatial and temporal resolution NDVI weighted based on space-time method |
| CN108241871A (en) * | 2017-12-27 | 2018-07-03 | 华北水利水电大学 | Laser point cloud and visual fusion data classification method based on multiple features |
| CN108982369A (en) * | 2018-04-28 | 2018-12-11 | 中国农业大学 | Merge the plot scale crop condition monitoring method of GF-1WFV and MODIS data |
| CN108613933A (en) * | 2018-06-13 | 2018-10-02 | 中南林业科技大学 | Forest land arid space-time dynamic monitoring method based on multi-sources RS data fusion |
| CN109358162A (en) * | 2018-11-06 | 2019-02-19 | 福州大学 | A kind of construction method of the novel remote sensing ecology index based on space geometry principle |
| CN109543630A (en) * | 2018-11-28 | 2019-03-29 | 苏州中科天启遥感科技有限公司 | Remote sensing image forest land extracting method and system, storage medium, electronic equipment based on deep learning |
| CN110363246A (en) * | 2019-07-18 | 2019-10-22 | 滨州学院 | A kind of fusion method of high-spatial and temporal resolution vegetation index NDVI |
| CN110503137A (en) * | 2019-07-29 | 2019-11-26 | 电子科技大学 | Based on the determination method of the remote sensing image temporal-spatial fusion base image pair of mixing together |
Non-Patent Citations (3)
| Title |
|---|
| HILKER,T.,WULDER: "A new data fusion method based on Landsat and MODIS NDVI data for high-spatial-resolution forest mapping", 《IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING》 * |
| 任建强,陈仲新: "MODIS植被指数的美国玉米单产遥感估测", 《遥感学报》 * |
| 谢登峰: "Landsat8和MODIS融合构建高时空分辨率", 《遥感学报》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113643409A (en) * | 2021-08-24 | 2021-11-12 | 中国农业大学 | Method and device for representing vegetation production rate and storage medium |
| CN115062863A (en) * | 2022-07-13 | 2022-09-16 | 中国农业科学院农业资源与农业区划研究所 | Apple flowering phase prediction method based on crop reference curve and accumulated temperature correction |
| CN115062863B (en) * | 2022-07-13 | 2023-08-25 | 中国农业科学院农业资源与农业区划研究所 | Apple flowering phase prediction method based on crop reference curve and accumulated temperature correction |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110909821B (en) | 2020-07-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109884664B (en) | Optical microwave collaborative inversion method and system for urban overground biomass | |
| Lasko et al. | Mapping double and single crop paddy rice with Sentinel-1A at varying spatial scales and polarizations in Hanoi, Vietnam | |
| CN107909607B (en) | A kind of year regional vegetation coverage computational methods | |
| Lenney et al. | The status of agricultural lands in Egypt: the use of multitemporal NDVI features derived from Landsat TM | |
| CN105046648B (en) | A method of constructing high space-time remotely-sensed data | |
| CN113128134B (en) | Mining area ecological environment evolution driving factor weight quantitative analysis method | |
| CN106918816B (en) | Crop growth monitoring system and its data processing method and module | |
| CN102194127B (en) | Multi-frequency synthetic aperture radar (SAR) data crop sensing classification method | |
| CN111832518B (en) | Space-time fusion-based TSA remote sensing image land utilization method | |
| CN112884672B (en) | Multi-frame unmanned aerial vehicle image relative radiation correction method based on contemporaneous satellite images | |
| CN114092835B (en) | Normalized vegetation index data space-time fusion method based on different space-time resolutions | |
| CN111337434A (en) | Mining area reclamation vegetation biomass estimation method and system | |
| CN103544477A (en) | Improved linear spectral mixture model based vegetation coverage estimation method | |
| CN103489171A (en) | Wide-range remote-sensing image automatic dodging and color uniformizing method based on standard color library | |
| CN112101251B (en) | SAR automatic target recognition method based on variable convolutional neural network | |
| CN110909821A (en) | Method for carrying out high-space-time resolution vegetation index data fusion based on crop reference curve | |
| Wald | Data fusion: a conceptual approach for an efficient exploitation of remote sensing images | |
| CN108985154B (en) | Small-size ground object sub-pixel positioning method and system based on image concentration | |
| CN116522090A (en) | Soil moisture downscaling method considering vegetation memory and fine-scale vegetation index | |
| CN112364289B (en) | Method for extracting water body information through data fusion | |
| CN115830464A (en) | Plateau mountain agricultural greenhouse automatic extraction method based on multi-source data | |
| CN114882169A (en) | Three-dimensional data-based power grid engineering big data intelligent analysis system and method | |
| Nelson et al. | Combining moderate-resolution time-series RS data from SAR and optical sources for rice crop characterisation: examples from Bangladesh | |
| Colditz | Time series generation and classification of MODIS data for land cover mapping | |
| RU2705549C1 (en) | Method of agrochemical survey of agricultural lands |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20200728 Termination date: 20201203 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |