LARGE-SCALE FOREST HEIGHT REMOTE SENSING RETRIEVAL METHOD CONSIDERING ECOLOGICAL ZONING

Abstract

A large-scale forest height remote sensing retrieval method includes: acquiring Ice, Cloud and land Elevation Satellite (ICESAT-2) tree height data, Landsat data, Shuttle Radar Topography Mission (SRTM) data, Worldclim data, forest type data and ecological zoning data within a target zone, and preprocessing the data; carrying out georeferencing on the processed data to generate a first data set; calculating spectral features, terrain features and climatic factor features of an image, and combining the calculated features with the ecological zoning data and the forest type data to obtain a second data set; extracting eigenvalues of a same geographical location from the second data set, and combining the extracted eigenvalues with the tree height data to generate training data; constructing a random forest model covering a large zone as an ecological zoning tree height retrieval model, and dividing the obtained training data into a training sample and a verification sample.

Claims

1. A large-scale forest height remote sensing retrieval method, the method comprising: step 1: acquiring Ice, Cloud and land Elevation Satellite (ICESAT-2) tree height data, Landsat data, Shuttle Radar Topography Mission (SRTM) data, Worldclim data, forest type data and ecological zoning data within a target zone, and preprocessing the data; step 2: carrying out georeferencing on the processed Landsat data, SRTM data, Worldclim data, forest type data and ecological zoning image data to generate a first data set; step 3: calculating spectral features, terrain features and climatic factor features of an image according to the first data set, and combining the calculated features with the ecological zoning data and the forest type data to obtain a second data set; wherein, in step 3, the spectral features of a Landsat image comprise six original spectral wavebands B2, B3, B4, B5, B6 and B7, an normalized differential vegetation index (NDVI), a difference vegetation index (DVI), a ratio vegetation index (RVI), a soil-adjusted vegetation index (SAVI), an enhanced vegetation index (EVI), a leaf area index (LAI), a tasseled cap brightness TCB, a tasseled cap greenness TCG, a tasseled cap wetness TCW, a Contrast texture, a Dvar texture and an Inertia texture; the terrain features of SRTM comprise a terrain altitude DEM, a slope, an aspect, and hill shades under the solar azimuth angles of 0°, 60°, 120°, 180°, 240° and 300°; the climatic factor features of Worldclim comprise 19 biologically-related climatic factors bio1-bio19; thus, the finally-obtained second data set comprises a total of 48 features consisting of the ecological zoning data, 18 spectral features, 9 terrain features, 19 climatic factors and the forest type data; step 4: extracting eigenvalues of a same geographical location from the second data set by using a latitude and longitude coordinate of a spot center corresponding to the ICESAT-2 tree height data, and combining the extracted eigenvalues with the tree height data to generate training data; step 5: constructing a random forest model covering a large zone as an ecological zoning tree height retrieval model, and dividing the obtained training data into a training sample and a verification sample, wherein the training sample is used to train the model, and the verification sample is used to verify the model; and step 6: estimating a spatially-continuous forest height of an entire research zone by using the ecological zoning tree height retrieval model trained in step 5 to obtain a tree height spatial distribution map.

2. The method of claim 1, wherein step 1 comprises the following steps: step 1.1: collecting the Landsat data, the SRTM data, the Worldclim data and the forest type data within the target zone; step 1.2: employing a data quality layer in a cloud masking method CFmask to remove cloud and cloud shade pixels in the Landsat image to obtain high-quality Landsat data. step 1.3: resampling the SRTM data and the Worldclim data to be consistent with Landsat resolution; step 1.4: carrying out category data re-encoding on each category of the forest type data to obtain a forest type 1, a forest type 2, a forest type 3 . . . a forest type M with corresponding codes 1, 2, 3 . . . , M respectively; step 1.5: acquiring the ICESAT-2 tree height data within the target zone, and employing terrain filtering, canopy height filtering and photon number filtering to remove low-quality laser spot data to obtain high-precision tree height data Hcanopy and a longitude and latitude coordinate of a corresponding spot center; and step 1.6: collecting the ecological zoning data within the target zone to obtain a boundary range of each ecological zone, wherein N sub-ecological zones are comprised in total, carrying out category data re-encoding on each sub-ecological zone to obtain an ecological zone 1, an ecological zone 2, an ecological zone 3 . . . an ecological zone N with corresponding codes 1, 2, 3 . . . , N respectively.

3. The method of claim 1, wherein, in step 4, a total of 50 eigenvalues are extracted from a same geographical location, and comprise longitude, latitude, re-encoded forest type number, reencoded ecological zoning number, B2, B3, B4, B5, B6, B7, NDVI, DVI, RVI, SAVI, EVI, LAI, TCB, TCG, TCW, Contrast, Dvar, Ierita, DEM, slope, aspect, hillshade 0°, hillshade 60°, hillshade 120°, hillshade 180°, hillshade 240°, hillshade 300° and bio1-bio19.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0025] FIG. 1 is a flow chart of an embodiment of the disclosure; and

[0026] FIG. 2 is an example of ecological zoning in the eastern region of China according to an embodiment of the disclosure; and

[0027] FIG. 3 is an example of the mapped forest height in the eastern monsoon ecozone of China according to an embodiment of the disclosure.

DETAILED DESCRIPTION OF THE EMBODIMENTS

[0028] The disclosure provides a large-scale forest height remote sensing retrieval method considering ecological zoning. The following further describes the technical solution of the disclosure with reference to Landsat8 image data of the eastern region of China in 2019 and a random forest model constructed by the Google Earth Engine, and in combination with accompanying drawings and embodiments.

[0029] As shown in FIG. 1, a flow of an embodiment of the disclosure comprises.

[0030] At step 1: ICESAT-2 tree height data, Landsat data, SRTM data, Worldclim data, forest type data and ecological zoning data within a target zone are acquired, and then these data are preprocessed, where step 1 specifically comprises:

[0031] at step 1.1: the Landsat data, SRTM data, Worldclim data and forest type data of the eastern region of China in 2019 are collected by using the Google Earth Engine;

[0032] at step 1.2: a data quality layer in a cloud masking method CFmask is employed to remove cloud and cloud shade pixels in a Landsat image to obtain high-quality Landsat data;

[0033] at step 1.3: the SRTM data and the Worldclim data are resampled to a 30 m resolution;

[0034] at step 1.4: each category of the forest type data is subjected to category data re-encoding to obtain codes 1, 2, 3 respectively corresponding to a coniferous forest, a broad-leaved forest and a mixed broadleaf-conifer forest;

[0035] at step 1.5: the ICESAT-2 tree height data of the year 2019 is acquired, and low-quality laser spot data is removed by using terrain filtering, canopy height filtering and photon number filtering, so as to obtain high-precision tree height data Hcanopy and a longitude and latitude coordinate of a corresponding spot center; and

[0036] at step 1.6: the ecological zoning data of the eastern region of China is collected to obtain a boundary range of each ecological zone, where 33 sub-ecological zones are included in total, and uploaded to the Google Earth Engine. As shown in FIG. 2, each sub-ecological zone is subjected to category data re-encoding to obtain an ecological zone 1, an ecological zone 2, an ecological zone 3 . . . an ecological zone 33 with corresponding codes 1, 2, 3 . . . , 33 respectively.

[0037] At step 2: the Landsat data, SRTM data, Worldclim data and forest type data of the corresponding ecological zones are respectively extracted from the Google Earth Engine to generate a first data set. The data format of the first data set is shown in Table 1.

TABLE-US-00001 TABLE 1 Data Format of the First Data Set Number of Forest Ecological Landsat SRTM Worldclim Type Zones Ecological Zones type Data Data Data Data 1 Mountain deciduous raw raw raw raw coniferous forest ecoregion in raster raster raster raster northern Greater Khingan image image image image Mountains 2 Ecoregion of coniferous raw raw raw raw broad-leaved mixed forest in raster raster raster raster Lesser Khingan Mountains image image image image 3 Agriculture and wetland raw raw raw raw ecological area in Sanjiang raster raster raster raster Plain image image image image . . . . . . . . . . . . . . . . . . 33  Hainan central mountain raw raw raw raw tropical rain forest and raster raster raster raster monsoon rain forest ecological image image image image zone

[0038] At step 3, spectral features, terrain features and climatic factor features of an image are calculated according to the first data set, and then combined with the ecological zoning data and the forest type data to obtain a second data set.

[0039] Where, the spectral features of the Landsat image comprise six original spectral wavebands (B2, B3, B4, B5, B6 and B7), a normalized differential vegetation index (NDVI), a difference vegetation index (DVI), a ratio vegetation index (RVI), a soil-adjusted vegetation index (SAVI), an enhanced vegetation index (EVI), a leaf area index (LAI), a tasseled cap brightness TCB, a tasseled cap greenness TCG, a tasseled cap wetness TCW, a Contrast texture, a Dvar texture and an Inertia texture.

[0040] The terrain features of SRTM comprise a terrain altitude DEM, a slope, an aspect, and hill shades under the solar azimuth angles of 0°, 60°, 120°, 180°, 240° and 300°.

[0041] The climatic factor features of Worldclim comprise 19 biologically-related climatic factors bio1-bio19.

[0042] Thus, the finally-obtained second data set comprise a total of 48 features consisting of the ecological zoning data, 18 spectral features, 9 terrain features, 19 climatic factors and the forest type data.

TABLE-US-00002 TABLE 2 Data format of the second data set Number of ecological Forest type zones Landsat data SRTM data Worldcli data data 1 18 spectral 9 terrain 19 climatic raw raster features features factors image 2 18 spectral 9 terrain 19 climatic raw raster features features factors image 3 18 spectral 9 terrain 19 climatic raw raster features features factors image . . . . . . . . . . . . . . . 33  18 spectral 9 terrain 19 climatic raw raster features features factors image

[0043] At step 4, eigenvalues of a same geographical location are extracted from the second data set by using the latitude and longitude coordinate of a spot center corresponding to the high precision tree height data Hcanopy extracted using ICESAT-2, and then combined with Hcanopy to generate training data.

[0044] A total of 50 eigenvalues are extracted from a same geographical location, comprising longitude, latitude, re-encoded ecological zoning number, B2, B3, B4, B5, B6, B7, NDVI, DVI, RVI, SAVI, EVI, LAI, TCB, TCG, TCW, Contrast, Dvar, Ierita, DEM, slope, aspect, hillshade 0°, hillshade 60°, hillshade 120°, hillshade 180°, hillshade 240°, hillshade 300°, bio1-bio19 and forest type.

[0045] At step 5: a random forest model covering a large area is constructed as an ecological zoning tree height retrieval model, the obtained training data is divided into a training sample and a verification sample, where the training sample is configured to train the model, and the verification sample is configured to verify the model.

[0046] By taking the features extracted in step 4 as prediction variables and taking the Hcanopy extracted by ICESat-2 as response variables, the training data is divided into a training sample and a test sample at a ratio of 7:3, the training sample is used to train the random forest model, and the test sample is used to test the random forest model, so as to obtain an ecological zoning tree height retrieval model.

[0047] At step 6: a spatially-continuous forest height of an entire research zone is estimated by using the ecological zoning tree height retrieval model trained in step 5. Each pixel value of the second data set (including latitude, re-encoded ecological zoning number, B2, B3, B4, B5, B6, B7, NDVI, DVI, RVI, SAVI, EVI, LAI, TCB, TCG, TCW, Contrast, Dvar, Ierita, DEM, slope, aspect, hillshade 0°, hillshade 60°, hillshade 120°, hillshade 180°, hillshade 240°, hillshade 300°, bio1-bio19 and forest type) in step 3 is substituted into the ecological zoning tree height retrieval model, and the ecological zoning tree height retrieval model is run in Google Earth Engine, so as to obtain a prediction value of each pixel of forest height, finally obtaining a continuous forest height distribution map of the eastern region of China in 2019, as shown in FIG. 3.

[0048] During specific implementation, the above flow may automatically run by using a computer software technology.

[0049] The specific embodiments described herein are only examples to illustrate the spirit of the disclosure. Those skilled in the art may make various amendments or supplements to the specific embodiments described or replace them in similar ways without deviating from the spirit of the disclosure or going beyond the scope defined in the appended Claims.