CN105787457A - Evaluation method for improving vegetation classified remote sensing precision through integration of MODIS satellite and DEM - Google Patents

Abstract

The invention discloses a method for estimating the remote sensing precision of vegetation classification by MODIS satellite integrated DEM, which is characterized by the following steps: step 1, by selecting the corresponding MODIS vegetation index band, establishing a normalized difference vegetation index (NDVI) and an enhanced vegetation index ( EVI); Step 2, using the MODIS-NDVI index and MODIS-EVI index to calculate the required vegetation information respectively; Step 3, using the method of discrete point moving fitting distance weighted average interpolation to calculate the interpolation height of each grid point, and obtain The digital elevation model (DEM) of the experimental area; step 4: extract ground slope information from the digital elevation model (DEM) on the basis of the terrain elevation digital map; step 5: extract ground slope information from the digital elevation model (DEM) Combining the first two bands of MODIS with two time phases and MODIS vegetation index, combined with the basic method of pixel identification and classification based on maximum likelihood, the remote sensing identification of vegetation classification is carried out. The method of the invention has high precision and strong feasibility, and can extract vegetation information in complex terrain areas in a real-time and non-destructive manner.

Description

An Estimation Method for Improving Vegetation Classification Remote Sensing Accuracy by MODIS Satellite Integrated DEM

technical field

The invention relates to an estimation method for improving vegetation classification remote sensing accuracy by integrating DEM with MODIS satellites.

Background technique

In the past ten years, researchers at home and abroad have done a lot of research on how to use a variety of geographic auxiliary data and extract relevant information from images based on geoscience knowledge. The image information of the planting area is extracted first, and then classified, which greatly reduces the classification targets and improves the classification accuracy. Yan Jing et al. used the neural network method to provide multi-source data input and is not limited by data distribution assumptions. They extracted NDVI and diurnal temperature difference values from NOAA images, resampled them, and then added them to the areas that have important effects on rice growth. The ideal soil type, land use type, and elevation distribution information were used to obtain the ideal planting area of double-cropping early rice in Hubei Province; when extracting saline-alkali soil information, it was based on the formation conditions and landform characteristics of saline-alkali soil, as well as the relationship between landform area and salinization degree Correlation, determine the regional parameters for identifying saline-alkali soil, and improve the classification accuracy; apply the method of expert system to the remote sensing survey of grassland resources, use the three bands of MSS4, 5, and 7 for biomass calculation and chromaticity space transformation, and use The combination of geographic auxiliary data and multi-source information can greatly improve the separability of samples (Du Mingyi et al., 2002), and the classification accuracy is higher than that of simply using remote sensing data and maximum likelihood method.

References for the application of the present invention:

[1] Yan Jing, Wang Wen, Li Xiangge. Using Neural Network Method to Extract Rice Planting Area—Taking Hubei Province Double Cropping Early Rice as an Example[J]. Remote Sensing Journal, 2001(03).

[2] Ma Chi. Research on Mechanism and Method of Land Salinization Detection in Songliao Plain [D]. Jilin University. 2011.

[3] Zhang Youshui, Feng Xuezhi. Research on remote sensing image classification based on GIS BP neural network [J]. Journal of Nanjing University. 2003(06).

[4] Jingxia, Wang Jindi, Wang Jihua, Huang Wenjiang. Research on vegetation classification in mountainous areas based on regional and multi-temporal remote sensing data [J]. Remote Sensing Technology and Application. 2008 (04).

[5] Chen Jianyu, Zhan Yuanzeng, Mao Zhihua, Lu Lizhen. High resolution reconstruction of island land cover based on multi-source and multi-temporal remote sensing data [A]. Abstracts of the 17th China Remote Sensing Conference [C]. 2010.

[6] Zhu Changbai. Land Use and Coverage Survey Based on RS and GIS Technology--A Case Study of Chancheng District, Foshan City [J]. Science and Technology Information. 2008(18).

[7] Peng Dailiang. Research on Rice Yield Estimation Method Based on Statistics and MODIS Data [D]. Zhejiang University. 2009.

Contents of the invention

The purpose of the present invention is to provide an estimation method using MODIS satellite integrated DEM to improve the precision of vegetation classification remote sensing, which has high precision and strong feasibility, and can extract vegetation information in complex terrain areas in a real-time and non-destructive manner.

To achieve the above object, the present invention can take the following technical solutions:

A MODIS satellite integrated DEM estimation method for improving vegetation classification remote sensing accuracy, comprising the following steps:

Step 1: Collect geographic information and remote sensing data

By selecting the corresponding MODIS vegetation index bands 1, 2 and 3, the normalized difference vegetation index (NDVI) and enhanced vegetation index (EVI) were established;

Step 2: Calculate and obtain vegetation information

Use the MODIS-NDVI index and MODIS-EVI index to calculate the required vegetation information respectively;

Step 3: Create a digital elevation model (DEM)

The interpolation elevation of each grid point is calculated by using the distance-weighted average interpolation method of discrete point moving fitting, and the digital elevation model (DEM) of the experimental area is obtained;

Step 4: Extract ground slope information

On the basis of the terrain elevation digitized map, the ground slope information is extracted from the digital elevation model (DEM);

Step 5: Use the digital elevation model (DEM) to extract ground slope information, then combine the first two bands of MODIS with two time phases and the vegetation index of MODIS, and use the basic method of pixel identification and classification based on maximum likelihood to perform remote sensing of vegetation classification identify.

The range of bands selected for the collection of geographic information and remote sensing data described in step 1 should satisfy Table 1; the MODIS data should be corrected by sun altitude angle, projection transformation and radiation correction, and then strictly registered with the support of GIS. The registration error should be are less than 0.5 pixels.

Table 1 Different vegetation indices and applied bands

The calculation described in step 2 obtains vegetation information, and the specific formula is:

The calculation formula of MODIS-NDVI is as follows:

$\mathrm{<span\; class="notranslate">\; N</span>}\mathrm{<span\; class="notranslate">\; D.</span>}\mathrm{<span\; class="notranslate">\; V</span>}\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; N</span>}\mathrm{<span\; class="notranslate">\; I</span>}\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; R</span>}}{\mathrm{<span\; class="notranslate">\; N</span>}\mathrm{<span\; class="notranslate">\; I</span>}\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; R</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>$

In the formula: NIR and R are near-infrared and red light bands respectively;

The calculation formula of MODIS-EVI is as follows:

$\mathrm{<span\; class="notranslate">\; E.</span>}\mathrm{<span\; class="notranslate">\; V</span>}\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; N</span>}\mathrm{<span\; class="notranslate">\; I</span>}\mathrm{<span\; class="notranslate">\; R</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; R</span>}\mathrm{<span\; class="notranslate">\; E.</span>}\mathrm{<span\; class="notranslate">\; D.</span>}}}{{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; N</span>}\mathrm{<span\; class="notranslate">\; I</span>}\mathrm{<span\; class="notranslate">\; R</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; R</span>}\mathrm{<span\; class="notranslate">\; E.</span>}\mathrm{<span\; class="notranslate">\; D.</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; B</span>}\mathrm{<span\; class="notranslate">\; L</span>}\mathrm{<span\; class="notranslate">\; u</span>}\mathrm{<span\; class="notranslate">\; E.</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; L</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>$

Among them, ρ _NIR , ρ _RED and ρ _BLUE are the spectral reflectance values corresponding to the second near-infrared band, the first red band and the third blue band of the MODIS satellite, L is the background adjustment item, and C ₁ and C ₂ are the simulated Combined coefficients, L=1, C ₁ =6, and C ₂ =7.5.

The establishment of digital elevation model (DEM) described in step 3 adopts the method of discrete point moving fitting distance weighted average interpolation to calculate the interpolation elevation of each grid point, regards DEM as the sum of one or more functions, utilizes this Or these functions derive the terrain factor, the specific formula is:

Set point p(x _p , y _p ) as the grid point to be interpolated, and draw eight direction lines at intervals of 45 azimuths with point p as the center. The distances between these eight direction lines and the closest contour line intersections to point p are d ₁ , d ₂ ,...d ₈ . The elevations of these points are Z ₁ , Z ₂ . . . Z ₈ . If d ₂ =0, (l=1,2,3,...,8), then point p is located on a certain contour line, and the elevation Z _p of this point is the grid elevation sought; Otherwise, point p is not on the contour line, and it is the grid point to be interpolated; when d≠0, set the elevation of the grid to be obtained as Z _i,j , then

$\begin{array}{ccc}{\mathrm{<span\; class="notranslate">\; Z</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; =</span>\frac{\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; 8</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; Z</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}{\mathrm{<span\; class="notranslate">\; d</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}<span\; class="notranslate">\; )</span>}{\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; 8</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; /</span>{\mathrm{<span\; class="notranslate">\; d</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}<span\; class="notranslate">\; )</span>}& {\mathrm{<span\; class="notranslate">\; d</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}<span\; class="notranslate">\; \&NotEqual;</span>\mathrm{<span\; class="notranslate">\; 0</span>}& <span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; .....</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 8</span>}<span\; class="notranslate">\; )</span>\end{array}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; )</span>$

Among them, d _l =|x _l -x _p |(l=1,2); d _l =|y _l -y _p |(l=3,4);

The steps of extracting ground slope information described in step 4 are:

Step 5: Read the 1:2,500,000 scale (format is EOO) contour line data of Zhejiang Province from the National Basic Geographic Information Center into the ARC/INFO software. m*100m grid data;

Step 6: Obtain a DEM with a grid of 250m*250m after resampling, and store the generated raster image in a data set, whose format and location information match the satellite remote sensing data;

Step 7: On the basis of the topographic elevation digital map, obtain the digital elevation model (DEM), and generate the ground slope factor (grid size 250m*250m) from the digital elevation model (DEM).

Compared with the prior art, the beneficial effects of the present invention are: adopting the above-mentioned technical scheme,

1. Using terrain slope information, multi-temporal remote sensing data and a large number of multi-temporal data products provided by MODIS, fully mine geographic information data for research on the research object;

2. Using the slope information generated by DEM and the two time-phase MODIS image data and vegetation index to extract the area, the combination of geographic information and remote sensing data multi-source information can significantly improve the accuracy of area estimation compared with the single-scene image data.

Description of drawings

Figure 1 is the DEM image map of the study area;

Figure 2 is the image map of the ground slope in the study area;

Figure 3 is the diagram of the divergence analysis between samples (the first three bands of MODIS and NDVI);

Figure 4 is the diagram of the divergence analysis between samples (the first three bands in the two phases of MODIS, NDVI, EVI and slope).

detailed description

The present invention is an estimation method for improving vegetation classification remote sensing precision by using MODIS data integration DEM, comprising the following steps:

Steps to collect geographic information and remote sensing data:

By selecting the corresponding MODIS vegetation index bands 1, 2 and 3, the normalized vegetation index (Normalized Vegetation Indices, referred to as NDVI) and enhanced vegetation index (Enhance Vegetation Indices, referred to as EVI) are established;

The different vegetation indexes and bands selected for the collection of geographical information and remote sensing data should meet the requirements of Table 1:

Table 1 Different vegetation indices and applied bands

MODIS data should undergo sun altitude correction, projection transformation and radiation correction, and then be strictly registered with the support of GIS, and the registration error should be less than 0.5 pixels.

Use the NDVI index and MODIS-EVI index to calculate the required vegetation information respectively,

MODIS-NDVI index can be calculated using formula (1),

$\mathrm{<span\; class="notranslate">\; N</span>}\mathrm{<span\; class="notranslate">\; D.</span>}\mathrm{<span\; class="notranslate">\; V</span>}\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; N</span>}\mathrm{<span\; class="notranslate">\; I</span>}\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; R</span>}}{\mathrm{<span\; class="notranslate">\; N</span>}\mathrm{<span\; class="notranslate">\; I</span>}\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; R</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>$

In the formula, NIR and R are near-infrared and red light bands, respectively;

MODIS-EVI index can be calculated with formula (2),

$\mathrm{<span\; class="notranslate">\; E.</span>}\mathrm{<span\; class="notranslate">\; V</span>}\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; N</span>}\mathrm{<span\; class="notranslate">\; I</span>}\mathrm{<span\; class="notranslate">\; R</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; R</span>}\mathrm{<span\; class="notranslate">\; E.</span>}\mathrm{<span\; class="notranslate">\; D.</span>}}}{{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; N</span>}\mathrm{<span\; class="notranslate">\; I</span>}\mathrm{<span\; class="notranslate">\; R</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; R</span>}\mathrm{<span\; class="notranslate">\; E.</span>}\mathrm{<span\; class="notranslate">\; D.</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; B</span>}\mathrm{<span\; class="notranslate">\; L</span>}\mathrm{<span\; class="notranslate">\; u</span>}\mathrm{<span\; class="notranslate">\; E.</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; L</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>$

where ρ _NIR , ρ _RED and ρ _BLUE are the spectral reflectance values corresponding to MODIS near-infrared 2 band, red 1 band and blue 3 band respectively, L is the background adjustment item, C ₁ and C ₂ are the fitting coefficients, L =1, C ₁ =6, and C ₂ =7.5.

Steps to create a digital elevation model (DEM):

The interpolation elevation of each grid point is calculated by using the distance-weighted average interpolation method of discrete point moving fitting, and the digital elevation model (DEM) of the experimental area is obtained. The specific formula is:

Set point p(x _p , y _p ) as the grid point to be interpolated, and draw eight direction lines at intervals of 45 azimuths with point p as the center. The distances between these eight direction lines and the closest contour line intersections to point p are d ₁ , d ₂ ,...d ₈ . The elevations of these points are Z ₁ , Z ₂ . . . Z ₈ . If d ₂ =0, (l=1,2,3,...,8), then point p is located on a certain contour line, and the elevation Z _p of this point is the grid elevation sought; Otherwise, point p is not on the contour line and is the grid point to be interpolated. When d≠0, set the elevation of the grid to be Z _i,j , then

$\begin{array}{ccc}{\mathrm{<span\; class="notranslate">\; Z</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; =</span>\frac{\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; 8</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; Z</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}{\mathrm{<span\; class="notranslate">\; d</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}<span\; class="notranslate">\; )</span>}{\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; 8</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; /</span>{\mathrm{<span\; class="notranslate">\; d</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}<span\; class="notranslate">\; )</span>}& {\mathrm{<span\; class="notranslate">\; d</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}<span\; class="notranslate">\; \&NotEqual;</span>\mathrm{<span\; class="notranslate">\; 0</span>}& <span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; .....</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 8</span>}<span\; class="notranslate">\; )</span>\end{array}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; )</span>$

Among them, d _l =|x _l -x _p |(l=1,2); d _l =|y _l -y _p |(l=3,4);

The elevation of each grid point is calculated by the above formula, and finally the DEM of the entire test area is obtained.

Steps to extract ground slope information:

On the basis of the terrain elevation digital map, the ground slope information is extracted from the digital elevation model (DEM), and the steps are as follows:

Read the 1:2,500,000-scale (format EOO) contour line data of Zhejiang Province from the National Basic Geographic Information Center into the ARC/INFO software, the contour line spacing is 50 meters, and use the above-mentioned method to establish a digital elevation model (DEM) Generate 100m*100m grid data;

After resampling, the DEM of 250m*250m grid is obtained, and the generated raster image is stored as a data set, and its format and location information match the satellite remote sensing data;

On the basis of the topographic elevation digital map, a digital elevation model (DEM) is obtained, and the ground slope factor (grid size 250m*250m) is generated from the digital elevation model (DEM), as shown in Figures 1 and 2.

Steps to classify and extract vegetation information:

Using digital elevation model (DEM) to extract ground slope information, combined with the first two bands of MODIS in two time phases and MODIS vegetation index composite, using the basic method of pixel identification and classification based on maximum likelihood to carry out remote sensing identification of vegetation classification; The comparison does not consider the DEM data, but only single-view MODIS images for classification.

Under the same conditions, when there is only one MODIS image classification, the classification accuracy is 53.3% lower; using the digital elevation model (DEM) to extract ground slope information, and then combining the first two bands of MODIS with two time phases and the composite classification accuracy of MODIS vegetation index increased to 79.8%. This shows that the introduction of ground slope and multi-temporal image data has a significant effect on improving the accuracy of area information extraction and classification quality.

Claims (5)

1. the integrated DEM of MODIS satellite improves an evaluation method for vegetation classification remote sensing precision, and its feature includes Following steps:

Step one: gather geography information and remotely-sensed data

By choosing corresponding MODIS vegetation index wave band 1,2 and 3, set up normalized differential vegetation index (NDVI) and Strengthen vegetation index (EVI)；

Step 2: calculate and obtain vegetation information

It is utilized respectively MODIS-NDVI index and MODIS-EVI index is calculated required vegetation information；

Step 3: set up digital elevation model (DEM)

The method using discrete point to move the distance weighted average interpolation of matching calculates each mesh point interpolation elevation, obtains Obtain the digital elevation model (DEM) of test block；

Step 4: extract ground line gradient information

On the basis of landform altitude digitalized maps, from digital elevation model (DEM), extract ground line gradient information；

Step 5: utilize digital elevation model (DEM) to extract ground line gradient information, in conjunction with two phases MODIS The first two wave band and MODIS vegetation index, the basic skills utilizing maximum likelihood to be main pixel identification classification,

Carry out vegetation classification remote sensing recognition.

Utilization MODIS data the most according to claim 1 improve the evaluation method of vegetation classification remote sensing precision, It is characterized in that, the collection geography information described in step one should meet table 1 with the wavelength band selected by remotely-sensed data； MODIS data should be corrected through sun altitude, projective transformation and radiant correction, then carries out strict under GIS supports Registration, registration error should be respectively less than 0.5 pixel.

The different vegetation index of table 1 and the wave band applied

The utilization integrated DEM of MODIS satellite the most according to claim 1 improves vegetation classification remote sensing precision Evaluation method, it is characterised in that the calculating described in step 2 obtains vegetation information, and concrete formula is:

MODIS-NDVI computing formula is as follows:

$NDVI=\frac{NIR-R}{NIR+R}---\left(1\right)$

In formula: NIR and R is respectively near-infrared and red spectral band；

MODIS-EVI computing formula is as follows:

$EVI=\frac{{\mathrm{\&rho;}}_{NIR}-{\mathrm{\&rho;}}_{RED}}{{\mathrm{\&rho;}}_{NIR}+C_1{\mathrm{\&rho;}}_{RED}-C_2{\mathrm{\&rho;}}_{BLUE}+L}(1+L)---\left(2\right)$

Wherein, ρ_NIR、ρ_REDAnd ρ_BLUEIt is corresponding MODIS satellite near-infrared second band, ruddiness first respectively Wave band and the spectral reflectance values of blue light the 3rd wave band, L is that background adjusts item, C₁And C₂It is fitting coefficient, L= 1,C₁=6, and C₂=7.5.

Utilization MODIS data the most according to claim 1 improve the evaluation method of vegetation classification remote sensing precision, It is characterized in that, described in step 3, set up digital elevation model (DEM), use discrete point to move matching distance and add The method of weight average interpolation calculates the interpolation elevation of each mesh point, and DEM regards as the sum of one or more function, Utilizing this or these function to derive terrain factor, concrete formula is:

Set up an office p (x_p,y_p) it is mesh point to be interpolated, centered by p point, draw eight direction lines by 45 azimuthal spacings, These eight direction lines are respectively d with the distance of p point immediate contour intersection point₁,d₂,......d₈, the elevation of these points For Z₁,Z₂.....Z₈.If d₂=0, (l=1,2,3 ... .., 8), then p point is positioned on a certain contour, the elevation Z of this point_p It is required gridded elevation；Otherwise p point is not on contour, for mesh point to be interpolated.When d ≠ 0, if The elevation of required grid is Z_i,j, then

$\begin{array}{cc}{Z}_{i,j}=\frac{\underset{i=1}{\overset{8}{\&Sigma;}}\left(Z_l{d}_l\right)}{\underset{i=1}{\overset{8}{\&Sigma;}}(1/d_l)}& d_l\&NotEqual;0\end{array},(l=1,2,3,...\mathrm{..},8)---\left(3\right)$

Wherein, d_l=| x_l-x_p| (l=1,2)；d_l=| y_l-y_p| (l=3,4)；

Utilization MODIS data the most according to claim 1 improve the evaluation method of vegetation classification remote sensing precision, It is characterized in that, the extraction ground line gradient information described in step 4 the steps include:

Step 5: by 1:250000 ten thousand engineer's scale (form is EOO) of National Foundation Geography Information Center Zhejiang Province's contour line data is read into ARC/INFO software, and contour interval is 50 meters, uses the side of step 3 Method generates 100 meters of * 100 meters of Grid squares；

Step 6: obtain the DEM of 250 meters of * 250 meters of grids through resampling, the grating image of generation is deposited with data set Putting, its form and positional information match with Value of Remote Sensing Data；

Step 7: on the basis of landform altitude digitalized maps, obtains digital elevation model (DEM), and from numeral Elevation model (DEM) produces the ground line gradient factor (grid size 250 meters * 250 meters).