CN115797501A - Time-series forest age mapping method combining forest disturbance and recovery events - Google Patents
Time-series forest age mapping method combining forest disturbance and recovery events Download PDFInfo
- Publication number
- CN115797501A CN115797501A CN202211624705.1A CN202211624705A CN115797501A CN 115797501 A CN115797501 A CN 115797501A CN 202211624705 A CN202211624705 A CN 202211624705A CN 115797501 A CN115797501 A CN 115797501A
- Authority
- CN
- China
- Prior art keywords
- forest
- age
- recovery
- year
- algorithm
- 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.)
- Pending
Links
Images
Landscapes
- Image Processing (AREA)
Abstract
The invention discloses a time-series forest age mapping method combining forest disturbance and recovery events, which comprises the following steps of: obtaining a remote sensing satellite image data source; (2) constructing a forest age estimation model; (3) Estimating the spatial distribution of forest ages in the reference year by using a random forest algorithm; (4) improving the LandTrendr algorithm; (5) Detecting forest interference and recovery information by using a Landsat interference and trend detection (LandTrendrr) algorithm; (6) integrating results of different change years; and (7) calculating the forest age of the time series. The method adopts LandTrendr algorithm to detect forest change, can identify wide interference intensity, complete the forest age mapping of natural forest and artificial forest, and improve the accuracy of mapping by combining detected forest and recovery event to forest age time sequence estimation.
Description
Technical Field
The invention relates to the field of forest age mapping, in particular to a time-series forest age mapping method combining forest disturbance and recovery events.
Background
The age of forest is a key parameter for determining the carbon sequestration state and potential of the forest ecosystem. The lack of a time series forest age space pattern results in an inability to capture forest disturbance and recovery history in sufficient spatial and temporal domains, thereby also increasing the uncertainty of past and future forest carbon sink estimates. Therefore, the method considers forest disturbance and recovery history, explores forest age dynamic change and has important scientific significance for accurately estimating the carbon sink of the forest ecosystem.
Remote sensing is combined with ground observation records, and an effective and feasible method is provided for obtaining the spatial-temporal distribution of forest ages. At present, the forest age estimation method based on remote sensing tends to establish a regression relationship between forest age and characteristic variables such as spectra and textures of vegetation, and therefore spatial distribution of forest age is predicted. The basic physical mechanism of the method is that different optical characteristics of tree species, branches, leaves, barks, trunks and cones in the forest stand cause different light reflectivity and transmissivity of the crown structure, and in addition, as the forest age increases, the image texture of the forest crown also changes, and changes from a uniform style to a cluster mode, so that the relationship between the crown texture characteristics and the forest age is established. Based on the reasons, researchers at home and abroad invert the spatial distribution of the forest ages by combining spectral and textural features by using regression methods such as multivariate linear regression, artificial neural networks, support vector machines and the like.
However, the above method can only obtain spatial distribution of forest ages for a certain year. How to use the forest ages of a specific year in combination with the interference and recovery history is the key point for estimating the forest ages of the time series. Therefore, the forest age is calculated backwards by taking a specific year as a reference year and combining interference and recovery events, and the spatial distribution chart of the forest age in the time series can be realized.
At present, only one of the most similar prior art inventions is a new method for mapping forest age space of an artificial forest (application number: CN201710592204.2, and an authorization notice number: CN 107247809B). The method comprises the steps of firstly simulating forest age distribution of a certain year by using a random forest algorithm, then detecting a forest interference event by using a Vegetation Change Tracking (VCT) algorithm, and finally calculating time sequence artificial forest age spatial distribution influenced by interference according to the random forest algorithm and the VCT algorithm.
Compared with the VCT algorithm, the Landsat interference and trend detection (LandTrendr) algorithm is widely used on a global scale as an effective method for detecting forest interference and recovery events. The advantage of the LandTrendr algorithm is that the straight line segmentation can detect sharp interference events (all fells), and can detect long-lasting slow events (forest growth, recovery after interference). Meanwhile, the LandTrendr algorithm can set different spectral indexes to obtain different detection results, and different from the VCT algorithm, the LandTrendr algorithm can only be based on the IFZ index. Therefore, the invention tries to more accurately acquire the forest disturbance and the recovery year based on the LandTrendrr algorithm of different spectral indexes by using a Google Earth Engine (GEE) cloud platform, so that a random forest model is coupled to calculate the time series forest age, the process is completely programmed, and the time series forest age mapping of different scales and different time ranges is automatically completed.
The Vegetation Change Tracking (VCT) algorithm adopted in the prior art detects forest changes, can only identify limited interference intensity, and can only complete the drawing of forest ages of artificial forests; forest recovery events detected by the VCT algorithm are not combined with forest age estimation; or the forest disturbance and recovery detection algorithm can be operated to obtain the corresponding forest change information based on a single spectral index.
Disclosure of Invention
The invention aims to provide a time-series forest age mapping method combining forest interference and recovery events, which aims to solve the problem that the traditional method can only realize medium and small-scale artificial forest age mapping, realize large-scale artificial forest and natural forest age mapping, obtain more accurate forest interference and recovery information, improve the precision of time-series forest age space mapping, and enable a user to set various spectral indexes at will, thereby improving the precision of forest age space mapping by combining the results of all indexes.
The purpose of the invention is realized by the following technical scheme: a time series forest age mapping method combining forest disturbance and recovery events comprises the following steps:
(1) Acquiring a remote sensing satellite image data source:
according to the drawing range, the year and the resolution required by a user, acquiring earth surface reflectivity image remote sensing satellite data of a corresponding position, a corresponding year and a corresponding resolution; acquiring corresponding forest age ground survey data;
(2) Constructing a forest age estimation model:
calculating principal component analysis variables, spectral indexes, tassel-cap transformation and texture characteristic variables of a gray level co-occurrence matrix based on the original spectral band of the earth surface reflectivity data; and combining with forest age ground survey data to construct a training set, and training a random forest regression model by using the training set.
(3) Estimating the spatial distribution of forest ages in the reference year by using a random forest algorithm:
after training of the random forest regression model is completed, masking off non-forest regions in the drawing range, and accordingly simulating and estimating the forest age of the reference year in the drawing range based on the random forest regression model;
(4) Improving the LandTrendr algorithm:
expanding input data of a LandTrenr algorithm, realizing that the LandTrenr algorithm can be used for acquiring surface reflectivity image data of sentinels (Sentinel) and medium resolution imaging spectrometers (MODIS), constructing corresponding time series stack data, namely an image collection (ImageCollection) object in GEE, calculating and segmenting a spectrum track through the object, and segmenting the algorithm according to the track to obtain the change year;
(5) Detecting forest interference and recovery information by using Landsat interference and trend detection (LandTrendr) algorithm:
setting different spectral indexes, and acquiring different forest disturbance and recovery years based on an improved LandTrendrr algorithm;
(6) Integrating different annual changes results:
(6.1) integrating the changing year results of different spectral indexes into an image by a majority voting method;
(6.2) taking the last 5 wave band variables in the results of different spectral indexes as independent variables of the random forest classification model, taking interference and recovery information which is visually interpreted and judged by a user as a label, making a forest interference and recovery training set, training the random forest classification model by using the training set, simulating a forest interference and recovery boundary area, and outputting two binary grid images as a model output result;
(6.3) finally, respectively masking the change year integration result by using the binary raster image so as to respectively obtain forest disturbance and recovery year spatial distribution maps;
(7) Calculating the forest age of the time series:
firstly, calculating the forest age of the next year according to the forest age of the reference year, retrieving a corresponding pixel from the detection result of the LandTrendr forest interference and recovery occurrence year, giving a value of 0 to the pixel subjected to forest interference of the next year, giving a value of 1 to the pixel subjected to forest recovery of the next year, and adding 1 to the age value of the previous year to make the forest age spatial distribution of the next year; then, sequentially and iteratively utilizing the forest age of the previous year to obtain the forest age of the next year; and finally, completing the mapping of the forest ages of the time series required by the user.
Further, the detection result of the spectral index includes 6 bands: year of change, amplitude of change, duration of change, spectral value before change, rate of change of spectrum, signal to noise ratio.
Further, the majority voting method specifically comprises: and when the number of the majority elements is larger than n/2, assigning the value of the majority elements to the pixel. On the contrary, when the majority of elements do not exist or the number of the majority of elements is less than n/2, the value with the highest detection precision in the time sequence is given to the pixel;
further, in step (6.2), the labels are: "forest loss/forest unchanged", "forest recovery/forest unchanged".
The invention has the beneficial effects that:
(1) And the LandTrendr algorithm is adopted to detect forest changes, so that wide interference intensity can be identified, and the forest age mapping of natural forests and artificial forests is completed.
(2) And (3) combining the forest and the recovery event detected by the LandTrendr algorithm to estimate the forest age time sequence, and improving the drawing precision.
(3) The forest change is detected by using various remote sensing characteristic variables (such as original wave bands and spectral indexes) based on a LandTrendr algorithm, and more accurate forest interference and recovery information is obtained by combining a majority voting method.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 is a flow chart of a time series forest age mapping method combining forest disturbance and recovery events according to the present invention.
Detailed Description
The following description will explain embodiments of the present invention in further detail with reference to the accompanying drawings.
As shown in FIG. 1, the time-series forest age mapping method combining forest disturbance and recovery events provided by the invention comprises the following specific steps:
(1) Selecting a remote sensing satellite image data source:
surface reflectance data (https:// developers. Google.com/earth-engine/datasets /) is called by Google Earth Engine (GEE), landsat, MODIS, sentinel, etc. According to the drawing range, the year and the resolution ratio required by a user, selecting corresponding positions, corresponding years and ground surface reflectivity image data with corresponding resolution ratio, and acquiring corresponding forest age ground survey data; so as to realize large, medium and small-scale forest age mapping; the remote sensing satellite image data can be obtained by using a PIE-Engine cloud platform to schedule corresponding data and run the same Javascript (including the inversion of forest age reference years and the LandTrendr algorithm).
(2) Constructing a forest age estimation model:
and calculating characteristic variables such as principal component analysis variables, spectral indexes, tassel-cap transformation, gray level co-occurrence matrix textures and the like based on the original spectral band of the earth surface reflectivity data. And combining with forest age ground survey data collected by a user to construct a sample set, and dividing the sample set into a training set in 70 percent and a testing set in 30 percent. Training a random forest regression model (or multiple linear regression, artificial neural network, support vector machine and the like) by using the training set, and determining the quantitative relation between the forest age and each characteristic variable through modeling.
(3) Estimating the spatial distribution of forest ages in the reference year by using a random forest algorithm:
after the optimization and training of the random forest regression model are completed, the accuracy of the model is verified by using the test set, so that the model is applied to the whole drawing range. Meanwhile, calling a Land cover (Land cover) data set (https:// levelers. Google.com/earth-engine/dates/tags/Land cover) on the GEE, masking off non-forest regions in the drawing range, and estimating the forest age of the reference year in the drawing range based on the random forest regression model simulation.
(4) Improving the LandTrendr algorithm:
input data of the expansion LandTrenr algorithm: in the original LandTrendr algorithm, time sequence stack data are firstly constructed according to Landsat image data provided by GEE, then a spectrum track of an original wave band or a spectrum index is extracted from the time sequence stack data, and then the change year is detected according to a track segmentation algorithm. After the LandTrendr algorithm is improved, the data source of the input end of the LandSat image data is expanded, the LandSat image data called originally is replaced by other earth surface reflectivity image data (such as Sentinel and moderate resolution imaging spectrometer MODIS) through an ee.ImageCollection () function on the GEE to construct corresponding time series stack data (ImageCollection image collection object in the GEE), and then the corresponding segmented spectrum tracks are calculated and obtained through the object according to the LandTrendr spectrum track segmentation algorithm, so that the corresponding change years are obtained. Therefore, the interference and recovery detection of forests with various scales in large, medium and small sizes can be realized.
(5) Detecting forest interference and recovery information by using a Landsat interference and trend detection (LandTrendrr) algorithm:
different spectral indexes are set, and different forest interference and recovery years can be obtained based on an improved LandTrendrr algorithm. The detection result of a certain spectral index comprises 6 wave bands: year of change, amplitude of change, duration of change, spectral value before change, rate of change of spectrum, signal-to-noise ratio.
(6) Integrating results of different change years, and the method comprises the following specific steps:
(6.1) integrating the changing year results of different spectral indexes into an image by a majority voting method, wherein the majority voting method specifically comprises the following steps: and when the number of the majority elements is larger than n/2, assigning the value of the majority elements to the pixel. Otherwise, when the majority elements do not exist or the number of the majority elements is less than n/2, the value with the highest detection precision in the time sequence is given to the pixel;
(6.2) taking other 5 wave band variables except the change year in the results of different spectral indexes as independent variables of the random forest classification model, taking interference and recovery information visually interpreted and judged by a user as a label, making a forest interference and recovery sample set, taking 70% as a training set, and taking 30% as a test set. And training a random forest classification model by using a training set, and verifying the model by using a test set. After the model is established, simulating a forest disturbance and a recovered boundary region, wherein the result is two binary raster images, and labels are respectively as follows: "forest loss/forest unchanged", "forest recovery/forest unchanged";
and (6.3) finally, respectively masking the change year integration result by using the binary grid image, thereby respectively obtaining the forest disturbance and recovery year spatial distribution map. The result can accurately represent which pixels have undergone forest loss or restoration, and more accurately detect the year in which the change occurred. Therefore, the forest disturbance and recovery information obtained through various remote sensing characteristics is more accurate, and the limitation that the prior art can only adopt a single index is solved.
(7) Calculating the forest age of the time series:
firstly, calculating the forest age of the next year by the reference forest age, retrieving a corresponding pixel from the detection result of LandTrendr forest interference and recovery occurrence years, giving a value of 0 to the pixel subjected to forest interference of the next year, giving a value of 1 to the pixel subjected to forest recovery of the next year, and adding 1 to the age value of the previous year to make the forest age spatial distribution of the next year; then, sequentially and iteratively utilizing the forest age of the previous year to obtain the forest age of the next year; and finally, completing the mapping of the forest ages of the time series required by the user. Therefore, forest disturbance is combined, forest recovery information is integrated, and more accurate estimation of the time series forest age is achieved.
The above-described embodiments are intended to illustrate rather than to limit the invention, and any modifications and variations of the present invention are within the spirit of the invention and the scope of the appended claims.
Claims (4)
1. A time series forest age mapping method combining forest disturbance and recovery events is characterized in that the technical scheme of time series forest age space distribution mapping comprises the following steps:
(1) Acquiring a remote sensing satellite image data source:
acquiring earth surface reflectivity image remote sensing satellite data of corresponding positions, corresponding years and corresponding resolutions according to a drawing range, the years and the resolutions required by a user; acquiring corresponding forest age ground survey data;
(2) Constructing a forest age estimation model:
calculating principal component analysis variables, spectral indexes, tassel-cap transformation and texture characteristic variables of a gray level co-occurrence matrix based on the original spectral band of the earth surface reflectivity data; and combining with forest age ground survey data to construct a training set, and training a random forest regression model by using the training set.
(3) Estimating the spatial distribution of forest ages in the reference year by using a random forest algorithm:
after training of the random forest regression model is completed, masking the non-forest region in the drawing range, and therefore simulating and estimating to obtain the forest age of the reference year in the drawing range based on the random forest regression model;
(4) Improving the LandTrendr algorithm:
expanding input data of a LandTrenr algorithm, realizing that the LandTrenr algorithm can be used for acquiring surface reflectivity image data of sentinels (Sentinel) and medium resolution imaging spectrometers (MODIS), constructing corresponding time series stack data, namely an image collection (ImageCollection) object in GEE, calculating and segmenting a spectrum track through the object, and segmenting the algorithm according to the track to obtain the change year;
(5) Detecting forest interference and recovery information by using a Landsat interference and trend detection (LandTrendrr) algorithm:
setting different spectral indexes, and acquiring different forest disturbance and recovery years based on an improved LandTrendrr algorithm;
(6) Integrating different year-of-change results:
(6.1) integrating the changing year results of different spectral indexes into an image by a majority voting method;
(6.2) taking the last 5 wave band variables in the results of different spectral indexes as independent variables of the random forest classification model, taking interference and recovery information which is visually interpreted and judged by a user as a label, making a forest interference and recovery training set, training the random forest classification model by using the training set, simulating a forest interference and recovery boundary area, and outputting two binary grid images as a model output result;
(6.3) finally, respectively masking the change year integration result by using the binary raster image so as to respectively obtain forest disturbance and recovery year spatial distribution maps;
(7) Calculating the forest age of the time series:
firstly, calculating the forest age of the next year according to the forest age of a reference year, retrieving a corresponding pixel from the detection result of the LandTrendr forest interference and recovery occurrence year, endowing the pixel value with a value of 0 for the pixel subjected to forest interference of the next year, endowing the pixel value with a value of 1 for the pixel subjected to forest recovery of the next year, and adding 1 to the forest pixel not changed in the next year so as to make the forest age spatial distribution of the next year; then, sequentially and iteratively utilizing the forest age of the previous year to obtain the forest age of the next year; and finally, completing the mapping of the forest ages of the time series required by the user.
2. The method for time-series forest age mapping by combining forest disturbance and recovery events as claimed in claim 1, wherein the detection result of the spectral index comprises 6 bands: year of change, amplitude of change, duration of change, spectral value before change, rate of change of spectrum, signal-to-noise ratio.
3. The time-series forest age mapping method combining forest disturbance and recovery events according to claim 1, wherein the majority voting method specifically comprises the following steps: and when the number of the majority elements is larger than n/2, assigning the value of the majority elements to the pixel. Otherwise, when the majority elements do not exist or the number of the majority elements is less than n/2, the value with the highest detection precision in the time sequence is given to the pixel;
4. a method for time series forest age mapping in combination with forest disturbance and recovery events as claimed in claim 1, wherein in step (6.2), the labels are: "forest loss/forest unchanged", "forest recovery/forest unchanged".
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211624705.1A CN115797501A (en) | 2022-12-16 | 2022-12-16 | Time-series forest age mapping method combining forest disturbance and recovery events |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211624705.1A CN115797501A (en) | 2022-12-16 | 2022-12-16 | Time-series forest age mapping method combining forest disturbance and recovery events |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115797501A true CN115797501A (en) | 2023-03-14 |
Family
ID=85425460
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211624705.1A Pending CN115797501A (en) | 2022-12-16 | 2022-12-16 | Time-series forest age mapping method combining forest disturbance and recovery events |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115797501A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116297223A (en) * | 2023-03-24 | 2023-06-23 | 南京大学 | Forest deforestation recovery remote sensing monitoring method and system |
-
2022
- 2022-12-16 CN CN202211624705.1A patent/CN115797501A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116297223A (en) * | 2023-03-24 | 2023-06-23 | 南京大学 | Forest deforestation recovery remote sensing monitoring method and system |
| CN116297223B (en) * | 2023-03-24 | 2023-11-21 | 南京大学 | Forest deforestation recovery remote sensing monitoring method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112836610B (en) | Land use change and carbon reserve quantitative estimation method based on remote sensing data | |
| Li et al. | A 30-year (1984–2013) record of annual urban dynamics of Beijing City derived from Landsat data | |
| Münzinger et al. | Mapping the urban forest in detail: From LiDAR point clouds to 3D tree models | |
| CN113449594A (en) | Multilayer network combined remote sensing image ground semantic segmentation and area calculation method | |
| CN116580318A (en) | Soil nutrient inversion method, device, equipment and medium for crop coverage area | |
| CN109063660B (en) | Crop identification method based on multispectral satellite image | |
| US11379742B2 (en) | Method for predictive soil mapping based on solar radiation in large flat area | |
| Șerban et al. | Mapping thermokarst lakes and ponds across permafrost landscapes in the Headwater Area of Yellow River on northeastern Qinghai-Tibet Plateau | |
| Khare et al. | A 15-year spatio-temporal analysis of plant β-diversity using Landsat time series derived Rao’s Q index | |
| Oehmcke et al. | Deep point cloud regression for above-ground forest biomass estimation from airborne LiDAR | |
| CN117475313A (en) | Large-scale winter wheat early-stage identification method based on automatic extraction of training samples | |
| CN115797501A (en) | Time-series forest age mapping method combining forest disturbance and recovery events | |
| CN114612315B (en) | High-resolution image missing region reconstruction method based on multitask learning | |
| CN115953683A (en) | Method for detecting hyperspectral change through learning of small samples across heterogeneous domains based on bidirectional generation | |
| Tamber et al. | Accounting for label errors when training a convolutional neural network to estimate sea ice concentration using operational ice charts | |
| CN114266932B (en) | Self-learning-based remote sensing multispectral data semi-supervised labeling method | |
| Zhang | A time-series approach to detect urbanized areas using biophysical indicators and landsat satellite imagery | |
| CN117075138A (en) | Remote sensing measurement and calculation method, system and medium for canopy height of 30-meter forest in area | |
| CN115345069A (en) | Lake water volume estimation method based on maximum water depth record and machine learning | |
| CN116863341B (en) | Crop classification and identification method and system based on time sequence satellite remote sensing image | |
| CN117115671A (en) | Soil quality analysis method and device based on remote sensing and electronic equipment | |
| CN115147726B (en) | City form map generation method and device, electronic equipment and readable storage medium | |
| CN115984689A (en) | Multi-scale earth surface complexity feature extraction and land utilization segmentation method | |
| Julien et al. | TISSBERT: A benchmark for the validation and comparison of NDVI time series reconstruction methods | |
| Jemy et al. | Quadruple stacked-based concept: a novel approach for change detection |
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 |