CN113469145B - Vegetation phenology extraction method based on high spatial and temporal resolution remote sensing data - Google Patents

Vegetation phenology extraction method based on high spatial and temporal resolution remote sensing data Download PDF

Info

Publication number
CN113469145B
CN113469145B CN202111018642.0A CN202111018642A CN113469145B CN 113469145 B CN113469145 B CN 113469145B CN 202111018642 A CN202111018642 A CN 202111018642A CN 113469145 B CN113469145 B CN 113469145B
Authority
CN
China
Prior art keywords
ndvi
phenological
curve
pos
vegetation
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.)
Active
Application number
CN202111018642.0A
Other languages
Chinese (zh)
Other versions
CN113469145A (en
Inventor
车向红
孙擎
刘纪平
王勇
徐胜华
罗安
刘慧慧
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Chinese Academy of Surveying and Mapping
Original Assignee
Chinese Academy of Surveying and Mapping
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Chinese Academy of Surveying and Mapping filed Critical Chinese Academy of Surveying and Mapping
Priority to CN202111018642.0A priority Critical patent/CN113469145B/en
Publication of CN113469145A publication Critical patent/CN113469145A/en
Application granted granted Critical
Publication of CN113469145B publication Critical patent/CN113469145B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10032—Satellite or aerial image; Remote sensing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30181—Earth observation
    • G06T2207/30188—Vegetation; Agriculture

Landscapes

  • Image Processing (AREA)

Abstract

A vegetation phenological extraction method based on high spatial and temporal resolution remote sensing data comprises the steps of obtaining a sequence remote sensing image of a target area, carrying out cloud detection pretreatment on the remote sensing image, and generating a vegetation NDVI growth curve; designing a noise point removal method to optimize the NDVI growth curve to generate a stable and smooth vegetation NDVI growth curve; extracting three characteristic points of vegetation phenological period based on a maximum change slope method; the design method filters vegetation pseudoperiods. And finally, finishing the extraction of the terrestrial vegetation phenology of the target area. According to the invention, the NDVI curve is used for extracting the phenological period of the surface vegetation, the NDVI is synthesized every 20 days, the synthesized phenological period is subjected to secondary filtration based on noise points of adjacent observation and an optimal synthesis after the phenological period is extracted, all possible phenological feature points in the NDVI curve are automatically extracted, the extraction accuracy is high, the extraction efficiency is high, the speed is high, and technical means and reference can be provided for accurately extracting the phenological period of the surface vegetation in a large area.

Description

Vegetation phenology extraction method based on high spatial and temporal resolution remote sensing data
Technical Field
The invention relates to a phenological extraction method, in particular to a high-spatial-temporal-resolution remote sensing data-based phenological extraction method, which can be used for relieving the technical problem that the prior art cannot accurately extract the phenological of surface vegetation based on an NDVI (normalized vegetation index) time sequence.
Background
The vegetation phenology refers to the phenomenon that the vegetation regularly changes with climatic rainfall, temperature, radiation and the like in the year, and the plants themselves regularly change such as sprouting, twitching, leaf spreading, flowering and fruiting, leaf falling and dormancy. The vegetation climate has higher sensitivity to climate change, the change of the vegetation can not only reveal the growth dynamics of the vegetation, but also visually reflect the response and adaptation process of a vegetation ecosystem to global environmental change, and the vegetation climate is the most sensitive and obvious indicator for representing the climate change.
The vegetation phenological extraction method mainly comprises two methods: traditional ground phenological observation and remote sensing phenological extraction. The traditional phenological extraction method adopts visual observation to visually observe a plurality of phenological stages including germination, leaf growth, flowering, fruiting and the like of individual horizontal species, and can accurately acquire the phenological of individual species. However, the traditional ground phenological extraction is not suitable for extracting phenological data for a long time due to time and labor consumption. The development of the remote sensing technology provides a powerful tool for extracting large-scale long-time sequence phenology, and the real-time observation of various vegetation growth information of pixel scale can be realized. The current methods for extracting the phenological remote sensing commonly used include a threshold value method, a maximum change slope method, a median method, a curve fitting method, a moving average method and the like. Most of the methods need to manually set a threshold, however, the earth surface vegetation phenology presents obvious spatial heterogeneity due to climate environment, geographic position and the like, and it is difficult to find a universal threshold to invert the earth surface vegetation phenology of a large area. The curve fitting method requires that curve shape parameters are fitted, so that the calculation amount is large. In addition, the methods generate vegetation growth curves based on the earth surface vegetation index time sequence, the vegetation index time sequence has more noise points due to the influence of atmospheric conditions when the satellite shoots the earth surface, any unfiltered noise points change the vegetation growth curve shape, pseudo-climate points are extracted by the methods, and the accuracy of extracting earth surface vegetation climate is reduced.
With the rapid development of remote sensing technology, remote sensing data with high space-time resolution, such as Sentinel second (Sentinel-2) satellite, can be used for accurate extraction of earth surface vegetation phenology. However, how to extract the phenological period of the surface vegetation by using remote sensing data with high space-time resolution, such as Sentinel-2 data, so as to accurately extract the phenological period of the surface vegetation becomes a technical problem which needs to be solved in the prior art.
Disclosure of Invention
The invention aims to provide a vegetation phenological extraction method based on high-spatial-temporal-resolution remote sensing data, which is used for solving the technical problem that the prior art cannot accurately extract the surface vegetation phenological.
In order to achieve the purpose, the invention adopts the following technical scheme:
a vegetation phenological extraction method based on high spatial and temporal resolution remote sensing data is characterized by comprising the following steps:
a vegetation NDVI growth curve generation step S110: acquiring a sequence remote sensing image with high space-time resolution of a target area by using, for example, Sentinel-2, and carrying out cloud detection pretreatment on the remote sensing image to generate a vegetation NDVI (normalized vegetation index) growth curve;
NDVI growth curve optimization step S120: designing a noise point removal method to optimize the NDVI growth curve to generate a stable, smooth and reliable vegetation NDVI growth curve;
vegetation phenological period feature point extraction step S130:
using a smooth NDVI curve NDVIsSuper-smooth NDVI curve NDVIssAnd NDVIsCurve rate of change SLP _ NDVIsExtracting three characteristic points of vegetation in a vegetation phenological period based on a maximum change slope method, and performing phenological extraction on earth surface vegetation in a target area, wherein the three characteristic points of the vegetation phenological period are initial SOS (start of growing session), POS (Peak value of growing session) in a growing season and EOS (end of growing session) in a growing season end, and specifically, a time point DOY (day of Yeast) corresponding to the maximum value of an NDVI time sequence in the year is taken as POS;
a target pixel optimal phenology period generation step S140: and removing the pseudo-phenological points, generating the optimal phenological period of the target pixel, and finally finishing the extraction of the phenological of the surface vegetation in the target area.
Optionally, step S110 specifically includes:
detecting and filtering cloud pixels in the sequence remote sensing image by using an FMASK cloud detection algorithm; calculating the NDVI time sequence of each pixel by using the NDVI formula of the filtered pixels, namely formula (1), distinguishing and prolonging two ends of the target year for three months to observe in consideration of poor winter observation quality of the target year to form a long-time sequence NDVI growth curve, wherein the NDVI formula is as follows,
Figure 856915DEST_PATH_IMAGE001
wherein the content of the first and second substances,
Figure 570793DEST_PATH_IMAGE002
is the reflectivity of the near-infrared band,
Figure 271901DEST_PATH_IMAGE003
the reflectivity is in the red light wave band.
Optionally, in step S120, the optimizing the NDVI growth curve by the design noise point removal method includes:
obtaining an NDVI time sequence in every 20 days, sequencing the NDVI values from small to large, taking 80% of the ordered NDVI sequence values as the NDVI of every 20 days, and generating an NDVI time sequence of every 20 days; noise filtering is carried out on the NDVI time sequence every 20 days, NDVI values meeting the following two noise filtering formulas are considered as cloud/cloud shadow pixels omitted by the cloud detection algorithm, the noise filtering formulas comprise formulas (2) and (3),
Figure 339215DEST_PATH_IMAGE005
（2）
Figure 684745DEST_PATH_IMAGE006
（3），
wherein NDVItIs the pixel NDVI value, NDVI of the target pixel at the target time tt-1Is the NDVI value of the t-1 pixel at the previous timet+1Is the pixel NDVI value, NDVI, of the subsequent time t +1maxIs the maximum of the annual NDVI time series.
Optionally, in step S120, the generating a stable, smooth and reliable vegetation NDVI growth curve includes: interpolating default values in the filtered NDVI time sequence by using a linear interpolation algorithm to obtain a day-by-day NDVI time sequence; filtering the interpolated NDVI time sequence by adopting two different time lengths by utilizing a Savitzky-Golay (SG) algorithm to obtain two smooth NDVI curves; calculating the change rate of the NDVI curve smoothed by a short time length, taking the target time as the center, calculating the change slope of the daily NDVI by the following formulas (4) and (5) based on the adjacent 8-day time window,
Figure 428579DEST_PATH_IMAGE008
（4）
Figure 102137DEST_PATH_IMAGE009
wherein
Figure 97775DEST_PATH_IMAGE010
DOY (day of Year) represents the number of days in the year,
Figure 422446DEST_PATH_IMAGE011
for NDVI sequences within a time window, as a dependent variable, COV represents the computation of a covariance matrix of the two variables,
Figure 946968DEST_PATH_IMAGE013
is the covariance of the two variables,
Figure 373401DEST_PATH_IMAGE015
for the variance of the independent variables, SLPtRepresenting the NDVI rate of change at the target time point; cutting redundant time points at two ends, and using the filtered time points with shorter time length as a smooth NDVI curve NDVIsUsing longer time length filtering as the super-smooth NDVI curve NDVIss，SLP_NDVIsIs NDVIsRate of change of curve.
Optionally, specifically, step S130 includes the following steps:
(1) lifting deviceAnd (3) POS of the target pixel phenological period: NDVI (non-uniform color) based on target pixelssThe NDVI is determined by comparing the differences of the NDVI values of three adjacent time pointsssLocal maxima and minima points of the curve, using NDVIssThe purpose of curve determination local extreme points is to roughly determine SOS, POS and EOS positions, and meanwhile, due to the fact that the curve is excessively smooth, more pseudo-object candidate positions can be prevented from being extracted; a plurality of local maximum points are based on NDVIssCarrying out descending order arrangement on the curve values, and then iteratively extracting a plurality of POS, SOS and EOS from a first local maximum value point; respectively extending forward and backward by half a phenological length based on NDVI by taking a local maximum point of target iteration as a center according to a preset phenological period lengthsCurve, search for NDVI in a phenological period centred on a local maximum pointsMaximum NDVI value of the curve, i.e. NDVIiIf NDVI is presentiSatisfying equation (6), the corresponding DOY is used as a valid POS for the waiting period,
Figure 786713DEST_PATH_IMAGE017
wherein NDVIs_minIs NDVIsMinimum of the curve, NDVIs_maxIs NDVIsThe maximum value of the curve;
(2) extracting SOS of the object pixel phenological period: based on the extracted POS position, if the local minimum value point NDVIss(i) Satisfying equation (7), it is considered as a reliable local minimum point before POS,
Figure 841257DEST_PATH_IMAGE019
wherein NDVIss(POS) is NDVI at POSssThe NDVI value corresponding to the curve is extended forward by a preset waiting period length, and the NDVI is obtained between the corresponding time point and the POSsDOY corresponding to the minimum value; between the DOY and the POS, if the formula (8) and the formula (9) are satisfied, thenThis point is considered to be the SOS,
Figure 677626DEST_PATH_IMAGE020
（8）
Figure 450410DEST_PATH_IMAGE021
（9）
wherein SLP _ NDVIs(SOS) represents the rate of change of the curve at SOS as SLP _ NDVIsThe maximum of the curve, and the NDVI ratio between this and the POS is less than 0.95.
(3) Extracting EOS of the target pixel phenological period: based on the extracted POS position, if the local minimum value point NDVIss(i) If equation (7) is satisfied, the local minimum point is considered to be reliable after the POS; extending the local minimum backward by a preset waiting period length, and calculating the NDVI between the POS and the time point corresponding to the extensionsDOY corresponding to the minimum value; between the POS and said DOY, if formula (10) and formula (11) are satisfied, then this point is considered EOS,
Figure 646905DEST_PATH_IMAGE022
wherein SLP _ NDVIs(EOS) for a Rate of Curve change at EOS SLP _ NDVIsAnd the NDVI ratio of that point to the POS is less than 0.95;
(4) and (4) circulating the steps (1) to (3) until traversal is completed until all local maximum values obtained on the NDVIss curve of the target pixel are traversed, and completing extraction of a plurality of phenological periods of the target pixel.
Optionally, in step S140, if only one pixel in the phenological period is extracted, directly storing the optimal phenological period of the target pixel, and recording quality information of the phenological period; if a plurality of phenological periods are extracted, directly taking phenological periods in which three phenological points exist and the POS is close to 180 days as the optimal phenological period, and recording the quality information of the phenological periods; and (3) judging that the influence of noise is caused if two continuous incomplete phenological periods, namely the SOS of the first phenological period and the EOS of the next phenological period are not detected, taking the SOS of the first phenological period as the final SOS, taking the position with the largest NDVI value at the POS of the two phenological periods as the final POS, taking the EOS of the second phenological period as the final EOS, generating the optimal phenological period of the target pixel, and finally finishing the extraction of the terrestrial vegetation phenological of the target area.
A storage medium for storing computer-executable instructions, characterized in that:
the computer executable instructions, when executed by a processor, perform the method of vegetation phenology extraction based on high spatial and temporal resolution remote sensing data described above.
The invention has the following advantages:
(1) utilize the NDVI curve to extract the earth's surface vegetation phenology, the every 20 days earth's surface NDVI that designs in this application is synthetic, and noise point secondary filtration based on adjacent observation after the synthesis to and an optimal synthesis after the phenology extraction, be crucial in the extraction of earth's surface vegetation phenology, can effectively filter phenology curve noise point, improve earth's surface vegetation phenology extraction accuracy degree.
(2) All possible phenological feature points in the NDVI curve can be automatically extracted, the extraction accuracy is high, with the rapid development of current computer hardware, the regional phenological extraction can be completed in a research area of 2000 × 200020 m within 3 hours at present, and the extraction efficiency and the extraction speed are high.
Drawings
FIG. 1 is a flow chart of a method of vegetation phenology extraction based on high spatial and temporal resolution remote sensing data according to the present invention;
FIG. 2 is a schematic diagram of the logical relationship of the vegetation phenology extraction method based on high spatial and temporal resolution remote sensing data according to the present invention;
FIG. 3 is a schematic diagram of three characteristic points of a phenological period of surface vegetation in accordance with an embodiment of the present invention;
FIG. 4 is a schematic diagram of the NDVI curve processing results during the earth's surface vegetation phenology extraction process according to an embodiment of the present invention;
FIG. 5 is a diagram illustrating a result of phenology extraction of surface vegetation in a study area in accordance with an embodiment of the present invention;
fig. 6 is a diagram showing the details of the terrestrial vegetation phenology extraction results of the study area according to the embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and are not limiting of the invention. It should be further noted that, for the convenience of description, only some of the structures related to the present invention are shown in the drawings, not all of the structures.
The invention is characterized in that: aiming at the problems that the space resolution of the existing earth surface vegetation phenological product is thick and the earth surface phenological space-time difference cannot be accurately reflected, a set of earth surface vegetation phenological extraction method based on the data characteristics is designed by utilizing the high-space-time-resolution remote sensing data, such as satellite sentriel-2 remote sensing data, the space resolution is 20m, the time resolution is 5 days, through noise point processing, the automatic extraction of possible phenological points and the synthetic flow of the extracted optimal phenological, and the technical means and reference can be provided for accurately extracting the phenological period of earth surface vegetation in a large area.
Specifically, referring to fig. 1, a flow chart of a vegetation phenology extraction method based on high spatial and temporal resolution remote sensing data is shown.
The method comprises the following steps:
a vegetation NDVI growth curve generation step S110: the method comprises the steps of obtaining a sequence remote sensing image with high space-time resolution of a target area by utilizing, for example, Sentinel-2, carrying out cloud detection pretreatment on the remote sensing image, and generating a vegetation NDVI (normalized vegetation index) growth curve.
Specifically, step S110 includes:
detecting and filtering cloud pixels in the sequence remote sensing image by using an FMASK cloud detection algorithm; calculating the NDVI time sequence of each pixel by using the NDVI formula of the filtered pixels, namely formula (1), distinguishing and prolonging two ends of the target year for three months to observe in consideration of poor winter observation quality of the target year to form a long-time sequence NDVI growth curve, wherein the NDVI formula is as follows,
Figure 821534DEST_PATH_IMAGE023
wherein the content of the first and second substances,
Figure 563225DEST_PATH_IMAGE002
is the reflectivity of the near-infrared band,
Figure 823305DEST_PATH_IMAGE003
the reflectivity is in the red light wave band.
And an NDVI growth curve optimizing step S120, designing a noise point removing method to optimize the NDVI growth curve and generating a stable, smooth and reliable vegetation NDVI growth curve.
Specifically, in step S120, the optimizing the NDVI growth curve by the design noise point removal method includes:
obtaining an NDVI time sequence in every 20 days, sequencing the NDVI values from small to large, taking 80% of the ordered NDVI sequence values as the NDVI of every 20 days, and generating an NDVI time sequence of every 20 days; noise filtering is carried out on the NDVI time sequence every 20 days, NDVI values meeting the following two noise filtering formulas are considered as cloud/cloud shadow pixels omitted by the cloud detection algorithm, the noise filtering formulas comprise formulas (2) and (3),
Figure 823491DEST_PATH_IMAGE024
（2）
Figure 727993DEST_PATH_IMAGE006
（3），
wherein NDVItIs the pixel NDVI value, NDVI of the target pixel at the target time tt-1Is the NDVI value of the t-1 pixel at the previous timet+1Is the pixel NDVI value, NDVI, of the subsequent time t +1maxIs the maximum of the annual NDVI time series;
in step S120, the generating a stable, smooth, reliable vegetation NDVI growth curve includes: interpolating default values in the filtered NDVI time sequence by using a linear interpolation algorithm to obtain a day-by-day NDVI time sequence; filtering the interpolated NDVI time sequence by adopting two different time lengths by utilizing a Savitzky-Golay (SG) algorithm to obtain two smooth NDVI curves; calculating the change rate of the NDVI curve smoothed by a short time length, taking the target time as the center, calculating the change slope of the daily NDVI by the following formulas (4) and (5) based on the adjacent 8-day time window,
Figure 30799DEST_PATH_IMAGE008
（4）
Figure 637230DEST_PATH_IMAGE025
wherein
Figure 50893DEST_PATH_IMAGE026
DOY (day of Year) represents the number of days in the year,
Figure 75481DEST_PATH_IMAGE027
for NDVI sequences within a time window, as a dependent variable, COV represents the computation of a covariance matrix of the two variables,
Figure 549188DEST_PATH_IMAGE029
is the covariance of the two variables,
Figure 645844DEST_PATH_IMAGE031
for the variance of the independent variables, SLPtRepresenting the NDVI rate of change at the target time point; cutting redundant time points at two ends, and using the filtered time points with shorter time length as a smooth NDVI curve NDVIsUsing longer time length filtering as the super-smooth NDVI curve NDVIss，SLP_NDVIsIs NDVIsRate of change of curve.
Vegetation phenological period feature point extraction step S130:
using a smooth NDVI curve NDVIsSuper-smooth NDVI curve NDVIssAnd NDVIsCurve rate of change SLP _ NDVIsExtracting three characteristic points of a vegetation phenological period based on a maximum change slope method, and performing phenological extraction on the earth surface vegetation of a target area, wherein the three characteristic points of the vegetation phenological period are initial SOS (start of growing session), POS (peak value of growing session) of the vegetation and end of growing session EOS (end of growing session) of the vegetation.
In the present invention, the time point doy (day of Year) corresponding to the maximum value of the NDVI time series in the year is regarded as POS, the time point at which the NDVI value rises most rapidly before the POS is regarded as SOS, and the time point at which the NDVI value falls most rapidly after the POS is regarded as EOS.
Specifically, step S130 includes the following steps:
(1) and (3) POS of the target pixel phenological period is extracted: NDVI (non-uniform color) based on target pixelssThe NDVI is determined by comparing the differences of the NDVI values of three adjacent time pointsssLocal maxima and minima points of the curve, using NDVIssThe purpose of curve determination local extreme points is to roughly determine SOS, POS and EOS positions, and meanwhile, due to the fact that the curve is excessively smooth, more pseudo-object candidate positions can be prevented from being extracted; a plurality of local maximum points are based on NDVIssCarrying out descending order arrangement on the curve values, and then iteratively extracting a plurality of POS, SOS and EOS from a first local maximum value point; respectively extending forward and backward by half a phenological length based on NDVI by taking a local maximum point of target iteration as a center according to a preset phenological period lengthsCurve, search for NDVI in a phenological period centred on a local maximum pointsMaximum NDVI value of the curve, i.e. NDVIiIf NDVI is presentiSatisfying equation (6), the corresponding DOY is used as a valid POS for the waiting period,
Figure 597620DEST_PATH_IMAGE017
wherein NDVIs_minIs NDVIsMinimum of the curve, NDVIs_maxIs NDVIsThe maximum value of the curve;
(2) extracting SOS of the object pixel phenological period: based on the extracted POS position, if the local minimum value point NDVIss(i) Satisfying equation (7), it is considered as a reliable local minimum point before POS,
Figure 742293DEST_PATH_IMAGE032
wherein NDVIss(POS) is NDVI at POSssThe NDVI value corresponding to the curve is extended forward by a preset waiting period length, and the NDVI is obtained between the corresponding time point and the POSsDOY corresponding to the minimum value; between the DOY and the POS, if the formula (8) and the formula (9) are satisfied, the point is regarded as the SOS,
Figure 777114DEST_PATH_IMAGE020
（8）
Figure 233503DEST_PATH_IMAGE021
（9）
wherein SLP _ NDVIs(SOS) represents the rate of change of the curve at SOS as SLP _ NDVIsThe maximum of the curve, and the NDVI ratio between this and the POS is less than 0.95.
(3) Extracting EOS of the target pixel phenological period: based on the extracted POS position, if the local minimum value point NDVIss(i) If equation (7) is satisfied, the local minimum point is considered to be reliable after the POS; extending the local minimum backward by a preset waiting period length, and calculating the NDVI between the POS and the time point corresponding to the extensionsDOY corresponding to the minimum value; between the POS and said DOY, if formula (10) and formula (11) are satisfied, then this point is considered EOS,
Figure 129915DEST_PATH_IMAGE033
wherein SLP _ NDVIs(EOS) for a Rate of Curve change at EOS SLP _ NDVIsAnd the NDVI ratio of that point to the POS is less than 0.95.
(4) And (4) circulating the steps (1) to (3) until traversal is completed until all local maximum values obtained on the NDVIss curve of the target pixel are traversed, and completing extraction of a plurality of phenological periods of the target pixel.
A target pixel optimal phenology period generation step S140: and removing the pseudo-phenological points, generating the optimal phenological period of the target pixel, and finally finishing the extraction of the phenological of the surface vegetation in the target area.
Through the iteration of each sub-step in step S130, a plurality of phenological periods may be generated, and therefore, it is necessary to filter the pseudo-phenological points to generate the optimal phenological period for the target pixel.
Specifically, if only one pixel of the phenological period is extracted, the optimal phenological period of the target pixel is directly stored, and the quality information of the phenological period is recorded; if a plurality of phenological periods are extracted, directly taking phenological periods in which three phenological points exist and the POS is close to 180 days as the optimal phenological period, and recording the quality information of the phenological periods; and (2) as the influence of noise is not identified, two continuous incomplete phenological periods may be generated, for two continuous incomplete phenological periods, namely the SOS of the first phenological period and the EOS of the next phenological period are not detected, judging that the influence of noise is generated, taking the SOS of the first phenological period as a final SOS, taking the POS of the two phenological periods with the largest NDVI value as a final POS, taking the EOS of the second phenological period as a final EOS, generating an optimal phenological period of the target pixel, and finally finishing the extraction of the phenological of the earth surface vegetation in the target area.
Example (b):
the invention provides a logic relation schematic diagram of a vegetation phenology extraction method based on high-space-time resolution remote sensing data, wherein the high-space-time resolution remote sensing data is based on a Sentinel-2 NDVI time sequence, referring to fig. 2, a sequence remote sensing image of a target area is obtained, cloud detection pretreatment is carried out on the remote sensing image, and a vegetation NDVI growth curve is generated; designing a noise point removal method to optimize the NDVI growth curve to generate a stable and smooth vegetation NDVI growth curve; as shown in fig. 3, three feature points of the vegetation phenological period are extracted based on the maximum variation slope method, and the pseudophenological points are finally removed to generate the optimal phenological period of the vegetation.
As an implementation, the vegetation phenological extraction method comprises the following steps:
step 1: all available Sentinel-22A/B atmospheric top reflectivity data from 9/2019 to 3/2021 were screened, the data had a spatial resolution of 20m and a temporal resolution of up to a 5-day revisit period due to combined 2A and 2B observations over the earth. The area of the research area is on the Sentinel-218 SUJ picture frame and is 2000 x 200020 m pixels in size.
Step 2: based on the selected Sentinel-22A/B atmosphere top layer reflectivity data, an FMASK open source packet is adopted to extract cloud/cloud shadow from the Sentinel-2 time sequence data, and pixels polluted by cloud, cloud shadow, water and ice and snow are filtered. The Sentinel-2 cloud detection result information is specifically as follows:
1 = land;
2 = water;
3 = cloud/cloud shadow;
4 = snow and ice;
and step 3: based on the selected Sentinel-22A/B atmosphere top layer reflectivity data, an FMASK open source packet is adopted to extract cloud/cloud shadow from the Sentinel-2 time sequence data, and pixels polluted by cloud, cloud shadow, water and ice and snow are filtered. The Sentinel-2 cloud detection result information is specifically as follows:
and 4, step 4: based on the selected Sentinel-22A/B atmospheric top reflectivity data, the NDVI vegetation index is calculated as shown in FIG. 4 (a), and the calculation formula is as follows:
Figure 847204DEST_PATH_IMAGE034
in the above formula, NDVI is the normalized vegetation index,
Figure 662714DEST_PATH_IMAGE002
is the reflectivity of the near-infrared band,
Figure 747344DEST_PATH_IMAGE003
the reflectivity is in the red light wave band.
And 5: taking the target pixel as an example, based on the cloud detection result, after removing cloud/cloud shadow, water body, and ice and snow, as shown in fig. 4 (b), the NDVI data of every 20 days are arranged in ascending order, and only the NDVI at 80% of the ordered NDVI is retained as the NDVI of every 20 days.
Step 6: taking the target pixel as an example, based on a cloud detection result, after cloud/cloud shadows, water bodies and ice and snow are removed, the NDVI data of every 20 days are arranged in an ascending order, and only 80% of the NDVI in the ordered NDVI is reserved as the NDVI of every 20 days.
And 7: noise filtering is performed again on the NDVI time series every 20 days, and if the NDVI at the time t satisfies the following two formulas, it is considered as a cloud/cloud shadow pixel missed by the cloud detection algorithm and needs to be filtered again, as shown in fig. 4 (c).
Figure 696715DEST_PATH_IMAGE036
In the above formula, NDVItIs the pixel NDVI value, NDVI of the target pixel at the target time tt-1Is the NDVI value of the t-1 pixel at the previous timet+1Is the pixel NDVI value, NDVI, of the subsequent time t +1maxIs the maximum of the annual NDVI time series;
and 8: considering that many earth cover vegetation changes NDVI faster during rapid growth periods, it is more meaningful to provide the phenological feature points in days. Taking the target pixel as an example, the filtered NDVI time sequence every 20 days has an default, and the default in the NDVI time sequence in the step 7 is interpolated by using a time sequence linear interpolation method to obtain a complete NDVI time sequence day by day.
And step 9: after interpolation and filling, cloud/cloud shadows and the like which are not identified by the FMASK algorithm still appearThe affected pixels are fitted based on two time sequence windows (41 days and 91 days) and a quadratic polynomial, and two smooth NDVI curves are formed by adopting a Savitzky-Golay filtering algorithm and are respectively NDVI41E.g., (e) and NDVI in FIG. 491As shown in fig. 4 (d).
Step 10: based on NDVI41Taking the time t as an example, calculating the change slope of the NDVI at the time t based on the time windows of the adjacent 8 days before and after, and finally generating the NDVI41Rate of Curve Change (SLP _ NDVI)41). The calculation formula is as follows:
Figure 409456DEST_PATH_IMAGE008
Figure 5653DEST_PATH_IMAGE037
in the above formula, the first and second carbon atoms are,
Figure 436635DEST_PATH_IMAGE038
the DOY sequences in adjacent time windows at time t, as arguments,
Figure 921187DEST_PATH_IMAGE039
for NDVI sequences within the time window of phase t, as a dependent variable, COV represents the computation of a covariance matrix of the two variables,
Figure 754014DEST_PATH_IMAGE040
is the covariance of the two variables,
Figure 521113DEST_PATH_IMAGE041
for the independent variable variance, the calculation is completed directly by adopting a Python numpy open source packet.
Step 11: NDVI of the year 9/9 to the year 2021/341、NDVI91、SLP_NDVI41Cutting the materials in front and back to generate 2020 NDVI41、NDVI91、SLP_NDVI41。
Step 12: base ofIn NDVI91The curve, through the difference to the NDVI value of adjacent 3 days, produces NDVI91And (4) carrying out descending sorting on the maximum value points according to the NDVI values and traversing and iterating to extract the candidate points.
Step 13: respectively extending forwards and backwards for 90 days by taking the local maximum value point as a center according to a preset phenological period length (120 days), and calculating corresponding NDVI (normalized difference of variance) in a time window of 180 days adjacent to the local maximum value41Maximum NDVI value of the curve (NDVI)i) And NDVIiIf the following calculation formula is satisfied, the corresponding DOY is used as the valid POS of a waiting period, the POS quality value is set to 1, no POS is extracted, and the POS quality value is set to 0.
Figure 829603DEST_PATH_IMAGE043
In the above formula, NDVI41_minIs NDVI41Minimum of the curve, NDVI41_maxIs NDVI41The maximum value of the curve.
Step 14: based on the extracted POS, NDVI if located before the POS91The NDVI of the local minima point to the NDVI at the POS is less than 0.9, and the local minima point is considered reliable. The DOY corresponding to the local minimum point is extended forward for 180 days, and NDVI is searched between the corresponding DOY and the POS41And finding a DOY corresponding to the minimum value in the DOY and POS time window, wherein the DOY is considered as a valid SOS if the following two formulas are adopted, the SOS quality value is set to be 1, no SOS is extracted, and the SOS quality value is set to be 0.
Figure 730563DEST_PATH_IMAGE044
In the above formula, SLP _ NDVI41(SOS) represents the rate of change of the curve at SOS as SLP _ NDVI41The maximum of the curve, and the NDVI ratio between this and the POS is less than 0.95.
Step 15: similar to step 14, only after the POSReliable local minima points. But the DOY corresponding to the local minimum value point is backwards prolonged by 180 days, and NDVI is searched between the POS and the DOY corresponding to the POS41And finding a DOY corresponding to the minimum value in the POS and DOY time windows, wherein the DOY is considered to be a valid EOS if the following two formulas are adopted, the EOS quality value is set to be 1, no EOS is extracted, and the EOS quality value is set to be 0. And finishing the extraction of one phenological period of the target pixel.
Figure DEST_PATH_IMAGE045
In the above formula, SLP _ NDVI41(EOS) for a Rate of Curve change at EOS SLP _ NDVI41The minimum value of the curve, and the NDVI ratio between this and the POS is less than 0.95.
Step 16: and (5) circulating the steps 12 to 15 until the target pixel NDVI is traversed91And (4) extracting a plurality of phenological periods of the target pixel by using all local maximum values obtained on the curve.
And step 17: for a target pixel, if only one pixel of a phenological period is extracted, directly storing the optimal phenological period of the target pixel, and recording the quality information of the phenological period; if a plurality of phenological periods are extracted, directly taking phenological periods, in which three phenological points exist and the POS is closer to the 180 th day, as optimal phenological periods, and recording quality information of the phenological periods; for two consecutive incomplete phenological periods, if the SOS of the first phenological period and the EOS of the next phenological period are not extracted, it is determined as noise influence, the SOS of the first phenological period is taken as the final SOS, the POS with the largest NDVI value at the two phenological periods POS is taken as the final POS, and the EOS of the second phenological period is taken as the final EOS, so as to generate the optimal phenological period of the target pixel, as shown in (f) of fig. 4. The quality information record of each pixel phenological period is recorded by using 0-2 bit formats, as follows:
1,1,1: the pixel has effective SOS, POS and EOS at the same time;
0,1,1: the pixel only has POS and EOS and no SOS;
1,1,0: the pixel only has SOS and POS and does not have EOS;
0,1,0: the pixel only has POS and does not have SOS and EOS;
step 18: and traversing 2000 pixels in the research area, repeating the steps 4 to 17, and completing the extraction of the earth surface vegetation phenological data of the whole research area, wherein the extraction result is shown in fig. 5 and 6.
The invention further discloses a storage medium for storing computer-executable instructions which, when executed by a processor, perform the vegetation phenology extraction method based on high spatial and temporal resolution remote sensing data.
Aiming at the problems that the space resolution of the existing earth surface vegetation phenological products is thick and the earth surface phenological space-time difference cannot be accurately reflected, the earth surface vegetation phenological extraction method based on the data characteristics is designed based on the latest high-space-time-resolution remote sensing data, such as Sentinel-2 remote sensing data, the space resolution is 20m, the time resolution is 5 days, through noise point processing, automatic extraction of possible phenological points and the synthetic flow of the extracted optimal phenological, and the technical means and reference can be provided for accurately extracting the phenological period of earth surface vegetation in a large area.
The invention has the following advantages:
(1) utilize the NDVI curve to extract the earth's surface vegetation phenology, the every 20 days earth's surface NDVI that designs in this application is synthetic, and noise point secondary filtration based on adjacent observation after the synthesis to and an optimal synthesis after the phenology extraction, be crucial in the extraction of earth's surface vegetation phenology, can effectively filter phenology curve noise point, improve earth's surface vegetation phenology extraction accuracy degree.
(2) All possible phenological feature points in the NDVI curve can be automatically extracted, the extraction accuracy is high, with the rapid development of current computer hardware, the regional phenological extraction can be completed in a research area of 2000 × 200020 m within 3 hours at present, and the extraction efficiency and the extraction speed are high.
While the invention has been described in further detail with reference to specific preferred embodiments thereof, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.

Claims (2)

1. A vegetation phenological extraction method based on high spatial and temporal resolution remote sensing data is characterized by comprising the following steps:
a vegetation NDVI growth curve generation step S110: acquiring a sequence remote sensing image with high space-time resolution of a target area, and carrying out cloud detection pretreatment on the remote sensing image to generate a vegetation NDVI growth curve;
NDVI growth curve optimization step S120: designing a noise point removal method to optimize the NDVI growth curve to generate a stable, smooth and reliable vegetation NDVI growth curve;
vegetation phenological period feature point extraction step S130:
using a smooth NDVI curve NDVIsSuper-smooth NDVI curve NDVIssAnd NDVIsCurve rate of change SLP _ NDVIsExtracting three characteristic points of a vegetation phenological period based on a maximum variation slope method, and performing phenological extraction on the earth surface vegetation of a target area, wherein the three characteristic points of the vegetation phenological period are initial SOS (sequence of events) of a vegetation growing season, POS (Point of season) of the vegetation growing season and final EOS (sequence of events) of the vegetation growing season, specifically, a time point DOY (day of Yeast) corresponding to the maximum value of an NDVI time sequence in a year is taken as POS, the time point at which the NDVI value rises fastest before the POS is taken as the SOS, and the time point at which the NDVI value falls fastest after the POS is taken as the EOS;
a target pixel optimal phenology period generation step S140: removing pseudo-phenological points, generating an optimal phenological period of the target pixel, and finally completing the phenological extraction of the surface vegetation in the target area;
step S110 specifically includes:
detecting and filtering cloud pixels in the sequence remote sensing image by using an FMASK cloud detection algorithm; calculating the NDVI time sequence of each pixel by using the NDVI formula of the filtered pixels, namely formula (1), distinguishing and prolonging two ends of the target year for three months to observe in consideration of poor winter observation quality of the target year to form a long-time sequence NDVI growth curve, wherein the NDVI formula is as follows,
Figure FDA0003328145600000021
where ρ isnirIs the reflectivity of the near infrared band, rhoredIs the reflectivity of red light wave band;
in step S120, the optimizing the NDVI growth curve by the design noise point removing method includes:
obtaining an NDVI time sequence in every 20 days, sequencing the NDVI values from small to large, taking 80% of the ordered NDVI sequence values as the NDVI of every 20 days, and generating an NDVI time sequence of every 20 days; noise filtering is carried out on the NDVI time sequence every 20 days, NDVI values meeting the following two noise filtering formulas are considered as cloud/cloud shadow pixels omitted by the cloud detection algorithm, the noise filtering formulas comprise formulas (2) and (3),
Min(NDVIt-1-NDVIt,NDVIt+1-NDVIt)/Max(NDVIt-1-NDVIt,NDVIt+1-NDVIt)＞0.45(2)
Max(NDVIt-1-NDVIt,NDVIt+1-NDVIt)＞NDVImax*0.3 (3)，
wherein NDVItIs the pixel NDVI value, NDVI of the target pixel at the target time tt-1Is the NDVI value of the t-1 pixel at the previous timet+1Is the pixel NDVI value, NDVI, of the subsequent time t +1maxIs the maximum of the annual NDVI time series;
in step S120, the generating a stable, smooth, reliable vegetation NDVI growth curve includes: interpolating default values in the filtered NDVI time sequence by using a linear interpolation algorithm to obtain a day-by-day NDVI time sequence; filtering the interpolated NDVI time sequence by using a Savitzky-Golay algorithm with two different time lengths to obtain two smooth NDVI curves; calculating the change rate of the NDVI curve smoothed by a short time length, taking the target time t as the center, calculating the change slope of the NDVI curve per day by the following formulas (4) and (5) based on the adjacent 8-day time window,
ncovt＝COV((DOYt-8,...DOYt-1,DOYt,DOYt+1,..DOYt+8),(NDVIt-8,...NDVIt-1,NDVIt,NDVIt+1,..NDVIt+8))(4)
SLPt＝ncovt[0,1]/ncovt[0,0] (5)
wherein (DOY)t-8,...DOYt-1,DOYt,DOYt+1,..DOYt+8) Is a DOY sequence in a time window as an independent variable, the DOY represents the number of days in the year, (NDVI)t-8,...NDVIt-1,NDVIt,NDVIt+1,..NDVIt+8) For NDVI sequences within a time window, as a dependent variable, COV represents the calculation of the covariance matrix of the two variables, ncovt[0,1]Is the covariance of two variables, ncovt[0,0]For the variance of the independent variables, SLPtRepresenting the NDVI rate of change at the target time point; cutting redundant time points at two ends, and using the filtered time points with shorter time length as a smooth NDVI curve NDVIsUsing longer time length filtering as the super-smooth NDVI curve NDVIss，SLP_NDVIsIs NDVIsRate of change of curve;
specifically, step S130 includes the following steps:
(1) and (3) POS of the target pixel phenological period is extracted: NDVI (non-uniform color) based on target pixelssThe NDVI is determined by comparing the differences of the NDVI values of three adjacent time pointsssLocal maxima and minima points of the curve, using NDVIssThe purpose of curve determination local extreme points is to roughly determine SOS, POS and EOS positions, and meanwhile, due to the fact that the curve is excessively smooth, more pseudo-object candidate positions can be prevented from being extracted; a plurality of local maximum points are based on NDVIssCarrying out descending order arrangement on the curve values, and then iteratively extracting a plurality of POS, SOS and EOS from a first local maximum value point; respectively extending forward and backward by half a phenological length based on NDVI by taking a local maximum point of target iteration as a center according to a preset phenological period lengthsCurve in parts ofSearching for NDVI in a phenological period centered on a maximum pointsMaximum NDVI value of the curve, i.e. NDVIiIf NDVI is presentiSatisfying equation (6), the corresponding DOY is used as a valid POS for the waiting period,
NDVIi＞(NDVIs_min+(NDVIs_max-NDVIs_min)/2)*NDVIi/NDVIs_max， (6)
wherein NDVIs_minIs NDVIsMinimum of the curve, NDVIs_maxIs NDVIsThe maximum value of the curve;
(2) extracting SOS of the object pixel phenological period: based on the extracted POS position, if the local minimum value point NDVIss(i) Satisfying equation (7), it is considered as a reliable local minimum point before POS,
NDVIss(i)/NDVIss(POS)＜0.9 (7)，
wherein NDVIss(POS) is NDVI at POSssThe NDVI value corresponding to the curve is extended forward by a preset waiting period length, and the NDVI is obtained between the corresponding time point and the POSsDOY corresponding to the minimum value; between the DOY and the POS, if the formula (8) and the formula (9) are satisfied, the point is regarded as the SOS,
SLP_NDVIs(SOS)＝＝Max(SLP_NDVIs) (8)
NDVIs(SOS)/NDVIs(POS)＜0.95 (9)
wherein SLP _ NDVIs(SOS) represents the rate of change of the curve at SOS as SLP _ NDVIsThe maximum value of the curve, and the NDVI ratio between this point and the POS point is less than 0.95;
(3) extracting EOS of the target pixel phenological period: based on the extracted POS position, if the local minimum value point NDVIss(i) Satisfying equation (7), it is considered as a reliable local minimum point after POS,
extending the local minimum backward by a preset waiting period length, and calculating the NDVI between the POS and the time point corresponding to the extensionsDOY corresponding to the minimum value; between POS and said DOY, if satisfiedEquation (10) and equation (11), then the point is considered as EOS,
SLP_NDVIs(EOS)＝＝Min(SLP_NDVIs) (10)
NDVIs(EOS)/NDVIs(POS)＜0.95 (11)，
wherein SLP _ NDVIs(EOS) for a Rate of Curve change at EOS SLP _ NDVIsAnd the NDVI ratio of that point to the POS is less than 0.95;
(4) and (4) the steps (1) to (3) are circulated until the target pixel NDVI is traversedssExtracting a plurality of phenological periods of the target pixel by using all local maximum values obtained on the curve;
step S140 specifically, if only one pixel of the phenological period is extracted, directly storing the optimal phenological period of the target pixel, and recording the quality information of the phenological period; if a plurality of phenological periods are extracted, directly taking phenological periods in which three phenological points exist and the POS is close to 180 days as the optimal phenological period, and recording the quality information of the phenological periods; and (3) judging that the influence of noise is caused if two continuous incomplete phenological periods, namely the SOS of the first phenological period and the EOS of the next phenological period are not detected, taking the SOS of the first phenological period as the final SOS, taking the position with the largest NDVI value at the POS of the two phenological periods as the final POS, taking the EOS of the second phenological period as the final EOS, generating the optimal phenological period of the target pixel, and finally finishing the extraction of the terrestrial vegetation phenological of the target area.
2. A storage medium for storing computer-executable instructions, characterized in that:
the computer executable instructions, when executed by a processor, perform the method of high spatial and temporal resolution remote sensing data-based vegetation phenology extraction of claim 1.
CN202111018642.0A 2021-09-01 2021-09-01 Vegetation phenology extraction method based on high spatial and temporal resolution remote sensing data Active CN113469145B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111018642.0A CN113469145B (en) 2021-09-01 2021-09-01 Vegetation phenology extraction method based on high spatial and temporal resolution remote sensing data

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111018642.0A CN113469145B (en) 2021-09-01 2021-09-01 Vegetation phenology extraction method based on high spatial and temporal resolution remote sensing data

Publications (2)

Publication Number Publication Date
CN113469145A CN113469145A (en) 2021-10-01
CN113469145B true CN113469145B (en) 2021-12-21

Family

ID=77867101

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111018642.0A Active CN113469145B (en) 2021-09-01 2021-09-01 Vegetation phenology extraction method based on high spatial and temporal resolution remote sensing data

Country Status (1)

Country Link
CN (1) CN113469145B (en)

Families Citing this family (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114092831B (en) * 2021-12-02 2023-03-24 中国科学院东北地理与农业生态研究所 Method for extracting vegetation phenological information of herbaceous swamp
CN114092835B (en) * 2022-01-24 2022-04-19 中国测绘科学研究院 Normalized vegetation index data space-time fusion method based on different space-time resolutions
CN114544515B (en) * 2022-02-23 2024-05-14 中国矿业大学 Grassland object remote sensing monitoring method and system
CN116721361B (en) * 2023-06-09 2024-01-02 中国测绘科学研究院 Wetland remote sensing extraction method compatible with space-time discontinuous images
CN116542403B (en) * 2023-07-06 2023-10-20 航天宏图信息技术股份有限公司 Crop yield prediction method, device, electronic equipment and readable storage medium
CN116597157B (en) * 2023-07-11 2023-09-26 中国科学院地理科学与资源研究所 Plant climate extraction method and device based on characteristic spectrum change
CN117216444B (en) * 2023-09-06 2024-04-19 北京林业大学 Vegetation weather parameter extraction method and device based on deep learning
CN117315505B (en) * 2023-11-28 2024-03-15 成都理工大学 Method for quantifying influence of arid and semiarid region mining activities on vegetation

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104933699A (en) * 2015-04-22 2015-09-23 吉林大学 Method for automatically extracting phenology information of earth surface vegetation based on fitting variance of Gaussian function
CN112304902A (en) * 2020-11-02 2021-02-02 航天宏图信息技术股份有限公司 Real-time monitoring method and device for crop phenology
CN112800973A (en) * 2021-01-29 2021-05-14 宁波大学 Spartina alterniflora extraction method based on vegetation phenological feature decision
CN113221765A (en) * 2021-05-18 2021-08-06 河海大学 Vegetation phenological period extraction method based on digital camera image effective pixels

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102592181A (en) * 2011-12-30 2012-07-18 中国农业科学院农业资源与农业区划研究所 Method for optimizing spatial distribution of statistical data about crop planting area
CN102609726B (en) * 2012-02-24 2014-04-16 中国科学院东北地理与农业生态研究所 Method for classifying remote sensing images blended with high-space high-temporal-resolution data by object oriented technology
CN104915674B (en) * 2014-10-24 2018-12-14 北京师范大学 The method that Landsat8 and MODIS constructs high-spatial and temporal resolution data identification autumn grain crop
CN108020511B (en) * 2016-11-01 2020-06-30 中国科学院遥感与数字地球研究所 Remote sensing monitoring method and device for water quality parameters of shallow grass type lake
CN106897707B (en) * 2017-03-02 2020-08-28 苏州中科天启遥感科技有限公司 Characteristic image time sequence synthesis method and device based on multi-source midsplit
CN109919083B (en) * 2019-03-06 2020-11-17 武汉大学 Early automatic winter wheat mapping method based on Sentinel-2 image data
CN110544357B (en) * 2019-08-28 2021-03-02 中科海慧（天津）科技有限公司 Agricultural, forest and grass composite district fire prediction and early warning method and device and electronic equipment
CN110766299B (en) * 2019-10-11 2022-11-29 清华大学 Watershed vegetation change analysis method based on remote sensing data

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104933699A (en) * 2015-04-22 2015-09-23 吉林大学 Method for automatically extracting phenology information of earth surface vegetation based on fitting variance of Gaussian function
CN112304902A (en) * 2020-11-02 2021-02-02 航天宏图信息技术股份有限公司 Real-time monitoring method and device for crop phenology
CN112800973A (en) * 2021-01-29 2021-05-14 宁波大学 Spartina alterniflora extraction method based on vegetation phenological feature decision
CN113221765A (en) * 2021-05-18 2021-08-06 河海大学 Vegetation phenological period extraction method based on digital camera image effective pixels

Also Published As

Publication number Publication date
CN113469145A (en) 2021-10-01

Similar Documents

Publication Publication Date Title
CN113469145B (en) Vegetation phenology extraction method based on high spatial and temporal resolution remote sensing data
CN106897707B (en) Characteristic image time sequence synthesis method and device based on multi-source midsplit
CN107730527B (en) Remote sensing satellite image-based plateau region ice lake extraction method
Tan et al. An enhanced TIMESAT algorithm for estimating vegetation phenology metrics from MODIS data
Pan et al. Winter wheat area estimation from MODIS-EVI time series data using the Crop Proportion Phenology Index
CN111340826B (en) Aerial image single tree crown segmentation algorithm based on super pixels and topological features
CN106599844A (en) Method for automatically extracting paddy rice growing region based on MODIS
CN113221765B (en) Vegetation phenological period extraction method based on digital camera image effective pixels
CN112348812B (en) Forest stand age information measurement method and device
Zhang et al. Quantifying latitudinal variation in land surface phenology of Spartina alterniflora saltmarshes across coastal wetlands in China by Landsat 7/8 and Sentinel-2 images
CN111832518A (en) Space-time fusion-based TSA remote sensing image land utilization method
CN112215169B (en) Crop plant height and biomass self-adaptive resolving method based on unmanned aerial vehicle passive remote sensing
CN111222539B (en) Method for optimizing and expanding supervision classification samples based on multi-source multi-temporal remote sensing image
Rodrigues et al. Monitoring vegetation dynamics inferred by satellite data using the PhenoSat tool
CN116188465B (en) Crop growth state detection method based on image processing technology
Hashim et al. Determination of seagrass biomass at Merambong Shoal in Straits of Johor using satellite remote sensing technique
CN109359617B (en) Method for identifying rainstorm based on grid rainfall data
CN114387516A (en) Single-season rice SAR (synthetic aperture radar) identification method for small and medium-sized fields in complex terrain environment
CN115062863A (en) Apple flowering phase prediction method based on crop reference curve and accumulated temperature correction
CN116091938A (en) Multisource remote sensing monitoring method for single-cropping rice planting area
CN115810155B (en) Tidal wetland classification method
CN112580595A (en) Double-cropping rice field remote sensing identification method based on Shapelet
CN111985433A (en) Rice remote sensing information extraction method and system
CN117274798A (en) Remote sensing rice identification method based on regularized time sequence variation model
CN114332492A (en) Vegetation feature extraction and selection method based on long-time sequence

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