A remote sensing extraction method for artificial forests based on the vertical structure complexity of spaceborne laser waveforms

Abstract

The present invention discloses a method for extracting artificial forests remotely sensed based on the vertical structural complexity of satellite-borne laser waveforms. The method comprises the following steps: step 1, obtaining L1B geolocation waveforms, L2A spot-scale global surface height and vegetation height, and L2B spot-scale global crown cover and vertical profile index product data of a satellite-borne full-waveform laser radar (GEDI) in a forest area; step 2, performing a preprocessing operation on the acquired data to obtain high-quality satellite-borne laser radar spots in the area; step 3, performing feature extraction, calculating the information entropy of a single spot echo waveform and the waveform similarity of adjacent spots from the L1B geolocation waveform data, extracting canopy cover and leaf height diversity index from the L2B spot-scale global crown cover and vertical profile index, constructing a linear model of height percentiles in the L2A spot-scale global surface height and vegetation height data, and calculating the slope of the linear model as a feature; step 4, inputting the above five features and using a random forest model to classify and extract artificial forests and natural forests; and step 5, evaluating the extraction accuracy of artificial forests and the importance of features. The present invention obtains multiple vertical structural features of the forest through the waveform information of the satellite-borne laser radar, and combines it with the random forest model to extract the artificial forest stands. It can fully utilize the vertical structural characteristics of the satellite-borne laser waveform in the forest area to distinguish between artificial forests and natural forests.

Description

Technical Field

[0001] The present invention relates to the technical field of artificial forest extraction, and in particular to a remote sensing extraction method for artificial forest based on the vertical structure complexity of satellite-borne laser waveforms. Background Art

[0002] Planted forests and natural forests have different functions and values ​​in terms of carbon storage, biodiversity, and ecosystem services. Accurately identifying the extent of planted forests within forest ecosystems not only helps understand their ecological benefits but also provides a scientific basis for the development of management and conservation measures. The National Forest Resources Inventory is currently the only source of data on the distribution of planted forests nationwide. However, forest inventory data are collected through field surveys and are largely dependent on forest accessibility, remoteness, and ownership. Furthermore, the National Forest Resources Inventory covers a long time span and cannot provide an accurate and up-to-date spatial distribution of China's planted forests. In recent years, advances in remote sensing technology have opened up new possibilities for planted forest extraction. By efficiently extracting and analyzing large-scale forest data, it is possible to timely update the distribution of planted forests over large areas. Current methods for extracting planted forests mainly include extracting spectral and temporal features from satellite imagery for classification, and combining feature recognition with multi-source remote sensing data.

[0003] The paper "Spracklen, B., & Spracklen, DV (2021). Synergistic use of Sentinel-1 and Sentinel-2 to map natural forest and Acacia plantation and stand ages in North-Central Vietnam. Remote Sensing, 13(2), 185. https: / / doi.org / 10.3390 / rs13020185" uses a random forest model to distinguish between plantations and natural forests in North-Central Vietnam based on the texture features of Sentinel-1 and the spectral features of Sentinel-2. In forest areas with a stand age of more than two years, the classification accuracy is significantly reduced. Traditional spectral and texture features are greatly affected by the stand age during the extraction process of plantations.

[0004] The paper "Cheng, K., Su, Y., Guan, H., Tao, S., Ren, Y., Hu, T., Ma, K., Tang, Y., & Guo, Q. (2023). Mapping China's planted forests using high resolution imagery and massive amounts of crowdsourced samples. ISPRS Journal of Photogrammetry and Remote Sensing, 196, 356-371. https: / / doi.org / 10.1016 / j.isprsjprs.2023.01.005" constructed a mapping framework based on Google Earth Engine based on the spectral, temporal, structural, textural and topographic characteristics of Landsat and Sentinel-1 time series images, digital elevation models and China's forest canopy height data, as well as crowdsourced data, and produced the first 30-meter resolution distribution map of China's planted and natural forests. Among them, temporal features are of great importance in extracting artificial forests, but multi-temporal passive optical remote sensing data are difficult to obtain in cloudy and rainy areas.

[0005] A Chinese patent document, "A method for spatial pattern recognition of planted forests based on temporal classification and spatial analysis," has been published. Its publication number is CN112380994A. The method involves collecting available Landsat and PALSAR data; classifying land cover types using stochastic gradient boosting based on PALSAR polarization and texture; extracting the cumulative maximum NDVI values ​​for different land cover categories from multi-temporal Landsat imagery to identify phenological features; determining thresholds and criteria for distinguishing forest from non-forest pixels; generating forest and non-forest products based on PALSAR, Landsat, thresholds, and criteria; and finally, combining knowledge-based discrimination with spatial analysis to determine the spatial distribution of planted forests. This method uses a fixed NDVI threshold of 0.72 to identify forest and non-forest pixels. However, regional climate conditions, vegetation phenology, soil background, and other factors can affect the performance of NDVI values. A single threshold may not be adequate for diverse environmental characteristics, especially in areas with significant climate differences between the north and south. A threshold of 0.72 may be too high or too low in some locations. Moreover, only the polarization information and texture features of PALSAR data are used for classification, which fails to fully explore more features in radar images. The surface penetration ability and vertical structure reflection ability of radar images have not been fully utilized.

[0006] A "method for remote sensing identification of evergreen plantations and remote sensing monitoring of their growth" disclosed in Chinese patent literature (CN111310639A) involves determining the evergreen forest cover index at each pixel in the first-type time-series remote sensing images of the work area for each year within a target time range; then, based on the evergreen forest cover index at each pixel, determining the evergreen forest planting area index at each pixel within the target time range; and finally, based on the evergreen forest planting area index at each pixel, determining the historical evergreen forest distribution area within the work area within the target time range. This method is limited in application and cannot meet the needs of mapping different forest types or large-scale areas, resulting in poor versatility. Summary of the Invention

[0007] Purpose of the invention: The present invention provides a remote sensing extraction method for artificial forests based on the vertical structural complexity of satellite-borne laser waveforms. By utilizing the ability of satellite-borne laser full waveform data to reflect vertical structural characteristics, artificial forests and natural forests can be effectively distinguished and artificial forest stands can be extracted, providing a scientific basis for the formulation of management and protection measures.

[0008] Technical Solution: The present invention provides a remote sensing method for extracting plantation forests based on the vertical structure complexity of satellite-borne laser waveforms, comprising the following steps:

[0009] Step 1: Obtain the L1B geolocation waveform, L2A spot-scale global surface height and vegetation height, and L2B spot-scale global canopy cover and vertical profile indicator product data of the spaceborne full-waveform lidar GEDI in the forest area;

[0010] Step 2: Crop the acquired data, screen the quality, and remove the ground point preprocessing operations to obtain high-quality space-borne lidar spots in the area;

[0011] Step 3: Perform feature extraction. Calculate the information entropy of a single spot echo waveform and the waveform similarity of adjacent spots from the L1B geolocation waveform data. Extract canopy cover and leaf height diversity index from the global canopy cover and vertical profile indicators at the L2B spot scale. Construct a linear model of the height percentiles in the global surface height and vegetation height data at the L2A spot scale, and calculate the slope of the linear model as a feature.

[0012] Step 4: Input the above five features and use the random forest model to classify and extract artificial forests and natural forests;

[0013] Step 5: Evaluate the accuracy of plantation extraction and feature importance.

[0014] Furthermore, in step 1, we selected sample areas of artificial and natural forests by combining high-definition images, existing distribution map products of artificial and natural forests, and text records. We then downloaded L1B geolocation waveform data, L2A spot-scale global surface height and vegetation height data, and L2B spot-scale global canopy cover and vertical profile index data from the spaceborne lidar GEDI that passed through the forest area from the NASA Earth Observation Data Center. The data were published in .hdf5 format, and the echo waveforms and other information were stored in gediSDS.

[0015] Furthermore, in step 2, the pre-processing operations such as cropping, quality screening, and removing ground points on the acquired data specifically include the following steps:

[0016] Step 21: Read latitude_bin0 and longitude_bin0 in gediSDS and filter the GEDI light spots within the forest range.

[0017] Step 22: Read the stale_return_flag in gediSDS, degrade, filter out the GEDI spots that meet the quality standards, read beamNames, and only keep the GEDI full-power intensity traces of BEAM0101, BEAM0110, BEAM1000, and BEAM1011;

[0018] Step 23: Read the rxwaveform in gediSDS to obtain the DN value of the lidar echo. After normalizing and smoothing the DN value, detect the peak value and delete the spots with only a single peak, that is, the ground point spots. Only the spots with trees in the area are retained.

[0019] Furthermore, in step 3, feature extraction is performed. The information entropy of the single spot echo waveform and the waveform similarity of adjacent spots are calculated from the L1B geolocation waveform data. The canopy cover and leaf height diversity index are extracted from the L2B spot scale global canopy cover and vertical profile index data. A linear model of the height percentiles in the L2A spot scale global surface height and vegetation height data is constructed, and the slope of the linear model is calculated as a feature. The specific steps include the following:

[0020] Step 31: Calculate the information entropy of a single spot echo waveform to measure the complexity of the forest's vertical structure; discretize the DN value into 250 intervals; use a histogram to calculate the probability of each DN value; perform probability distribution calculation and entropy value calculation. The information entropy calculation formula is:

[0021]

[0022] Where i is the index variable, H(X) is the information entropy of the DN value, and x iis the discretized DN value, p(x i ) is the DN value taken to x i probability;

[0023] Step 32: Calculate the similarity between the waveform curves of a single light spot and its spatially adjacent light spots to measure the similarity of the vertical structure between the light spots. Considering the first law of geography, only calculate the similarity of light spots within a distance of 120 meters. Perform linear interpolation on the waveform curves to ensure that the curves have the same number of DN values. After the curve lengths are consistent, calculate the similarity to quantify the correlation in shape between the two curves. The similarity calculation formula is:

[0024]

[0025] where x i and y i are the corresponding point values ​​in the two curves, and is the mean of the two curves, n is the number of DN values;

[0026] Step 33: Extract canopy cover and leaf height diversity index from the L2B spot-scale global canopy cover and vertical profile index data. Canopy cover represents the percentage of the area covered by the vertical projection of the canopy to the spot area. Leaf height diversity represents the density and height distribution of leaves within the canopy.

[0027] Step 34: Construct a linear model of the height percentiles in the L2A spot-scale global land surface height and vegetation height data, and calculate the slope of the linear model as a feature. The slope feature can reflect the density of understory vegetation and tree height. Extract 101 height indicators: rh0-rh100 from the L2A spot-scale global land surface height and vegetation height data, perform a linear regression on the rh value and the square root of the height percentile (0-100), and use the slope of the linear model as a feature. The calculation formula of the linear model is:

[0028]

[0029] Among them, rh x is the height index and x is the height percentile.

[0030] Furthermore, in step 4, five features were input and the random forest model was used to classify and extract artificial forests and natural forests. The dataset was divided into training, validation, and test sets, which accounted for 70%, 20%, and 10% of the total data, respectively. The random forest model was used for classification training, 150 decision trees were set to improve the robustness of the classification, and a random seed was set to ensure the repeatability of the results.

[0031] Furthermore, in step 5, evaluating the plantation extraction accuracy and feature importance specifically includes the following steps:

[0032] Step 51: After the model training is completed, the performance of the model is evaluated using the test set. The confusion matrix, classification accuracy, and classification report are calculated to evaluate the model's accuracy in extracting plantations.

[0033] Step 52: Evaluate feature importance based on Gini Impurity to measure the contribution of each feature in the plantation extraction process.

[0034] Beneficial effects: Compared with the existing technology, the present invention has the following significant advantages: (1) The present invention uses an active satellite-borne lidar remote sensing method, which is not affected by cloudy and rainy weather and supports the extraction of vertical structural features in any forest area covered by a satellite-borne lidar; (2) The present invention only uses the vertical structural features obtained from the waveform data to extract artificial forests, paying full attention to the differences in vertical structure between artificial forests and natural forests, and achieving better extraction effects; (3) The present invention supports large-scale remote sensing monitoring of artificial forests and can provide policy support for the construction and management of artificial forests by relevant departments. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 Schematic diagram of the method of the present invention.

[0036] Figure 2 Violin plots verifying feature availability for our invention. DETAILED DESCRIPTION

[0037] like Figure 1 As shown in FIG, a remote sensing extraction method for artificial forests based on the vertical structure complexity of spaceborne laser waveforms includes the following steps:

[0038] Step 1: Obtain the L1B geolocation waveform, L2A spot-scale global surface height and vegetation height, and L2B spot-scale global canopy cover and vertical profile indicator product data of the spaceborne full-waveform lidar GEDI in the forest area.

[0039] Combining high-definition images, existing distribution map products of artificial forests and natural forests, and text records, we selected sample areas of artificial forests and natural forests. We downloaded L1B geolocation waveform data, L2A spot-scale global surface height and vegetation height data, and L2B spot-scale global canopy cover and vertical profile index data from the spaceborne lidar GEDI that passed through the forest area from the NASA Earth Observation Data Center. The data were published in .hdf5 format, and the echo waveforms and other information were stored in gediSDS.

[0040] The present invention first selected four northern forest regions, and the regional information is shown in Table 1.

[0041] Table 1 Regional information

[0042]

[0043]

[0044] Step 2: Perform pre-processing operations such as cropping, quality screening, and removing ground points on the acquired data to obtain high-quality space-borne lidar spots in the area. This specifically includes the following steps:

[0045] Step 2.1, read latitude_bin0 and longitude_bin0 in gediSDS, and filter the GEDI light spots within the forest range;

[0046] Step 2.2: Read the stale_return_flag in gediSDS, degrade, filter out the GEDI spots that meet the quality standards, read the beamNames, and only keep the GEDI full-power intensity traces of BEAM0101, BEAM0110, BEAM1000, and BEAM1011.

[0047] Step 2.3: Read the rxwaveform in gediSDS to obtain the DN value of the lidar echo. After normalizing and smoothing the DN value, detect the peak value and delete the spots with only a single peak, that is, the ground point spots. Only the spots with trees in the area are retained.

[0048] Step 3: Perform feature extraction. Calculate the information entropy of a single spot echo waveform and the waveform similarity of adjacent spots from the L1B geolocation waveform data. Extract the canopy cover and leaf height diversity index from the global canopy cover and vertical profile indicators at the L2B spot scale. Construct a linear model of the height percentiles in the global surface height and vegetation height data at the L2A spot scale, and calculate the slope of the linear model as a feature. This specifically includes the following steps:

[0049] Step 3.1: Calculate the information entropy of a single spot echo waveform to measure the complexity of the forest's vertical structure. The specific calculation is: discretize the DN value into 250 intervals; use a histogram to calculate the probability of each DN value; and perform probability distribution and entropy calculations. The information entropy calculation formula is:

[0050]

[0051] Where i is the index variable, H(X) is the information entropy of the DN value, and x i is the discretized DN value, p(x i ) is the DN value taken to x i probability;

[0052] Step 3.2: Calculate the similarity between the waveform curves of a single light spot and its spatially adjacent light spots. This is used to measure the similarity of the vertical structure between the light spots. Considering the first law of geography, only the similarity of light spots within a distance of 120 meters is calculated. The specific calculation is: linearly interpolate the waveform curves to make the curves have the same number of DN values; after the curve lengths are consistent, calculate the similarity to quantify the correlation in shape between the two curves. The similarity calculation formula is:

[0053]

[0054] where x i and y i are the corresponding point values ​​in the two curves, and is the mean of the two curves, n is the number of DN values;

[0055] Step 3.3: Extract canopy cover and leaf height diversity index from the L2B spot-scale global canopy cover and vertical profile index data. Canopy cover represents the percentage of the area covered by the vertical projection of the canopy to the spot area, and leaf height diversity represents the density and height distribution of leaves within the canopy.

[0056] Step 3.4: Construct a linear model of the height percentiles in the L2A spot-scale global surface height and vegetation height data, and calculate the slope of the linear model as a feature. The slope feature can reflect the density of understory vegetation and tree height. The specific calculation is: extract 101 height indicators: rh0-rh100 from the L2A spot-scale global surface height and vegetation height data, perform linear regression on the rh value and the square root of the height percentile (0-100), and use the slope of the linear model as a feature. The calculation formula of the linear model is:

[0057]

[0058] Among them, rh x is the height index and x is the height percentile.

[0059] In order to verify that the extracted features are different between artificial forests and natural forests and can be used for classification, the present invention uses violin plots to compare the differences of the five features in the two forest stands. The results are as follows: Figure 2 shown.

[0060] Step 4. Input the above five features and use the random forest model to classify and extract artificial forests and natural forests; divide the dataset into training set, validation set, and test set, which account for 70%, 20%, and 10% of the total data, respectively. Use the random forest model for classification training, set 150 decision trees to improve the robustness of the classification, and set a random seed to ensure the repeatability of the results.

[0061] Step 5: Evaluate the extraction accuracy and feature importance of plantations; specifically, the following steps are included:

[0062] Step 5.1: After model training is completed, use the test set to evaluate the performance of the model. Calculate the confusion matrix, classification accuracy, and classification report to evaluate the model's accuracy in extracting plantations.

[0063] Step 5.2: Evaluate feature importance based on Gini impurity to measure the contribution of each feature in the plantation extraction process.

[0064] To evaluate the accuracy of plantation extraction, the present invention calculated the accuracy, producer precision, user precision, and F1 score using the confusion matrix. The results are shown in Table 2. The results show that the proposed method achieves good results in plantation extraction.

[0065] Table 2 Accuracy, producer precision, user precision and F1 score

[0066]

[0067] In order to evaluate the contribution of each feature to the extraction of artificial forests in random forests, the present invention evaluates the importance of features based on Gini impurity, and the results are shown in Table 3.

[0068] Table 3 Evaluation of feature importance based on Gini impurity

[0069]

Claims

1. A remote sensing extraction method for artificial forests based on the vertical structure complexity of satellite-borne laser waveforms, characterized in that: The steps include: Step 1: Obtain the L1B geolocation waveform, L2A spot-scale global surface height and vegetation height, and L2B spot-scale global canopy cover and vertical profile indicator product data of the spaceborne full-waveform lidar GEDI in the forest area; Step 2: Preprocess the acquired data to obtain high-quality space-borne lidar spots in the area; Step 3: Perform feature extraction. Calculate the information entropy of a single spot echo waveform and the waveform similarity of adjacent spots from the L1B geolocation waveform data. Extract the canopy cover and leaf height diversity index from the global canopy cover and vertical profile indicators at the L2B spot scale. Construct a linear model of the height percentiles in the global surface height and vegetation height data at the L2A spot scale, and calculate the slope of the linear model as a feature. This specifically includes the following steps: Step 31: Calculate the information entropy of a single spot echo waveform to measure the complexity of the forest's vertical structure; Step 32: Calculate the similarity between the waveform curves of a single light spot and its spatially adjacent light spots to measure the similarity of the vertical structure between the light spots. Considering the first law of geography, only calculate the similarity of the light spots within a distance of 120 meters. Step 33: Extract canopy cover and leaf height diversity index from the L2B spot-scale global canopy cover and vertical profile index data. Canopy cover represents the percentage of the area covered by the vertical projection of the canopy to the spot area. Leaf height diversity represents the density and height distribution of leaves within the canopy. Step 34: construct a linear model of the height percentiles in the L2A spot-scale global land surface height and vegetation height data, and calculate the slope of the linear model as a feature. The slope feature can reflect the density of understory vegetation and tree height. Extract 101 height indices (rh0-rh100) from the L2A spot-scale global land surface height and vegetation height data, perform a linear regression on the rh value and the square root of the height percentiles 0-100, and use the slope of the linear model as a feature. Step 4: Input the above five features and use the random forest model to classify and extract artificial forests and natural forests; Step 5: Evaluate the accuracy of plantation extraction and feature importance.

2. The artificial forest remote sensing extraction method based on vertical structure complexity of satellite-borne laser waveform according to claim 1, characterized in that: In step 1, we selected sample areas of planted and natural forests by combining high-definition imagery, existing planted and natural forest distribution map products, and textual records. We then downloaded L1B geolocation waveform data, L2A spot-scale global surface elevation and vegetation height data, and L2B spot-scale global canopy cover and vertical profile indicator data from the NASA Earth Observation Data Center, which was obtained from the spaceborne LiDAR (GEDI) that passed through the forested area. The data were published in .hdf5 format, and the echo waveforms and other information were stored in gediSDS.

3. The artificial forest remote sensing extraction method based on vertical structure complexity of satellite-borne laser waveform according to claim 1, characterized in that: In step 2, the acquired data is cropped, quality screened, and ground point preprocessing operations are performed.

4. The artificial forest remote sensing extraction method based on vertical structure complexity of satellite-borne laser waveform according to claim 3, characterized in that: The pre-processing steps for clipping, quality screening and removing ground points of the acquired data include the following: Step 21: Read latitude_bin0 and longitude_bin0 in gediSDS and filter the GEDI light spots within the forest range. Step 22: Read the stale_return_flag in gediSDS, degrade, filter out the GEDI spots that meet the quality standards, read beamNames, and only keep the GEDI full-power intensity traces of BEAM0101, BEAM0110, BEAM1000, and BEAM1011; Step 23: Read the rxwaveform in gediSDS to obtain the DN value of the lidar echo. After normalizing and smoothing the DN value, detect the peak value and delete the spots with only a single peak, that is, the ground point spots. Only the spots with trees in the area are retained.

5. The artificial forest remote sensing extraction method based on vertical structure complexity of satellite-borne laser waveform according to claim 1, characterized in that: In step 31, the DN value is discretized into 250 intervals; the probability of each DN value occurring is counted using a histogram; Probability distribution calculation and entropy value calculation are performed. The information entropy calculation formula is: Where i is the index variable, H(X) is the information entropy of the DN value, and x i is the discretized DN value, p(x i ) is the DN value taken to x i probability.

6. The method for remote sensing extraction of artificial forests based on vertical structure complexity of satellite-borne laser waveforms according to claim 1, characterized in that: In step 32, linear interpolation is performed on the waveform curves so that the curves have the same number of DN values. After the curve lengths are consistent, similarity is calculated to quantify the correlation between the two curves in shape. The similarity calculation formula is: where x i and y i are the corresponding point values ​​in the two curves, and is the mean of the two curves, and n is the number of DN values.

7. The method for remote sensing extraction of artificial forests based on vertical structure complexity of satellite-borne laser waveforms according to claim 1, characterized in that: In step 34, the calculation formula of the linear model is: Among them, rh x is the height index and x is the height percentile.

8. The method for remote sensing extraction of artificial forests based on vertical structure complexity of satellite-borne laser waveforms according to claim 1, characterized in that: In step 4, five features were input and the random forest model was used to classify and extract plantations and natural forests. The dataset was divided into training, validation, and test sets, which accounted for 70%, 20%, and 10% of the total data, respectively. The random forest model was used for classification training, 150 decision trees were set to improve the robustness of the classification, and a random seed was set to ensure the repeatability of the results.

9. The method for remote sensing extraction of artificial forests based on vertical structure complexity of satellite-borne laser waveforms according to claim 1, characterized in that: In step 5, the evaluation of plantation extraction accuracy and feature importance includes the following steps: Step 51: After the model training is completed, the performance of the model is evaluated using the test set. The confusion matrix, classification accuracy, and classification report are calculated to evaluate the model's accuracy in extracting plantations. Step 52: Evaluate feature importance based on Gini impurity to measure the contribution of each feature in the plantation extraction process.

Images