Home Extraction of fractional vegetation cover in arid desert area based on Chinese GF-6 satellite
Article Open Access

Extraction of fractional vegetation cover in arid desert area based on Chinese GF-6 satellite

  • Zhengdong Deng , Zhao Lu EMAIL logo , Guangyuan Wang , Daqing Wang , Zhibin Ding , Hongfei Zhao , Haoli Xu , Yue Shi , Zijian Cheng and Xiaoning Zhao
Published/Copyright: April 5, 2021
Download Article (PDF)
Become an author with De Gruyter Brill
Submit Manuscript Author Information
Open Geosciences
From the journal Open Geosciences Volume 13 Issue 1

Abstract

The red edge band is considered as one of the diagnosable characteristics of green plants, but the large-scale remote sensing retrieval of fractional vegetation coverage (FVC) based on the red edge band is still rare. To explore the application of the red edge band in the remote sensing estimation of FVC, this study proposed a new vegetation index (normalized difference red edge index, RENDVI) based on the two red edge bands of Chinese GaoFen-6 satellite (GF-6). The FVC estimated by using three vegetation indices (NDVI, RENDVI1, and RENDVI2) were evaluated based on the field survey FVC obtained in Minqin Basin of Gansu Province. The results showed that there was a good linear correlation between the FVC estimated by GF-6 WFV data and the FVC investigated in the field, and the most reasonable estimation of FVC was obtained based on RENDVI2 model (R 2 = 0.97611 and RMSE = 0.07075). Meanwhile, the impact of three confidence levels (1, 2, and 5%) on FVC was also analyzed in this study. FVC obtained from NDVI and RENDVI2 has the highest accuracy at 2% confidence, while FVC based on RENDVI1 achieved the best accuracy at 5% confidence. It could be concluded that it is feasible and reliable to estimate FVC based on red edge bands, and the GF-6 Wide Field View (WFV) data with high temporal and spatial resolution provide a new data source for remote sensing estimation of FVC.

Keywords: vegetation index; pixel dichotomy model; fractional vegetation cover; red edge band

1 Introduction

As an important part of the ecosystem, the changes of vegetation in its quantity and population proportion will lead to changes in land surface energy, biogeochemical cycle, and hydrology, which is one of the most important links in global change [1,2,3]. To measure the surface vegetation coverage and its changes effectively and quantitatively, the researchers used the concept of fractional vegetation coverage (FVC) [4,5]. FVC is defined as the percentage of the vertical projection area of vegetation (including branches, stems, and leaves) on the ground to the total area of the statistical area [6]. As a comprehensive quantitative index reflecting the surface conditions of vegetation community coverage, FVC is widely used in the ecological environment assessment [7], groundwater enrichment assessment [8], groundwater level monitoring [9], soil degradation, and desertification monitoring [10].

The traditional surface measurement methods for FVC include the photographic method, the sample strip method, the sample point method, the spatial quantitative meter, and so on [11]. Although the accuracy of FVC obtained by these methods is high, due to the characteristics such as small measurement range, time consumption, laborious, and easy to be restricted by natural conditions, these methods cannot measure the FVC of large areas, and the application value is very limited. With the development of remote sensing technology, remote sensing monitoring based on the relationship between vegetation spectral information and vegetation coverage has become the main technical means to obtain FVC in large areas [12]. The current data sources for remote sensing estimation of FVC mainly include Landsat, MODIS (Moderate-resolution Imaging Spectroradiometer), GaoFen (GF), SPOT (Systeme Probatoire d’Observation de la Terre), and so on [13,14,15,16]. The commonly used bands are mainly the blue band (450–520 nm), the green band (520–590 nm), the red band (630–690 nm), and the near infrared band (770–890 nm). The vegetation indices proposed based on the bands include the normalized green-red difference index (NGRDI) [17], the normalized green-blue difference index (NGBDI) [18], the visible-band difference vegetation index (VDVI) [19], and the normalized difference vegetation index (NDVI) [20]. Although the research methods based on these vegetation indices showed good accuracy in many remote sensing estimations of FCV, there were not many applications of FVC estimation in arid and semi-arid regions [21]. Due to the sparse vegetation distribution and special vegetation types in arid and semi-arid areas, a few scholars thought that those general model methods may lose their universal applicability [22].

The red edge band (670–760 nm) is between the red band and the near infrared band. Currently, the researches of remote sensing based on the red edge band mainly includes land classification, chlorophyll content, and biomass at three levels: ground hyperspectral, aviation hyperspectral, and satellite image [23,24,25,26]. Some studies showed that the red edge band can effectively reflect the specific spectral characteristics of crops, and thus, it was considered as one of the diagnosable characteristics of green plants [27,28]. However, the application of the red edge band to the remote sensing estimation of FVC in arid and semi-arid regions is still very rare.

Previously, the main satellite images for the red edge band application study were Rapid Eye [29], EO-1 (Earth Observing-1) Hyperion [30], and Sentinel-2 [31]. Due to the lack of available sensors, the effect of the red edge band on FVC estimation has not attracted much attention [32]. Fortunately, in addition to the common visible and near infrared bands, Chinese GaoFen-6 (GF-6) satellite WFV (Wide Field View) ta also covers two red edge band bands, which provides a new data source for large-scale estimation of vegetation coverage, and there is no research on FVC inversion based on GF-6 data before.

The purpose of this study is to construct a new vegetation index based on the characteristics of GF-6 satellite WFV data and to put forward a new reliable remote sensing estimation method for FVC in an arid environment. Taking sparse vegetation in the arid desert area as the research objective, this article discusses the application of the red edge band in remote sensing estimation of FVC by comparing with the field survey data.

2 Data

2.1 Study area

The study area (38°19′47″–38°44′50″N, 102°48′15″–103°19′1″E) is located on Minqin County, Wuwei City, Gansu Province, China, with an area of about 2,100 km2 (Figure 1). It is adjacent to the Tengger Desert in the east and the Badain Jaran Desert in the west, which have a typical arid desert climate. The geomorphic types are mainly mountains, plains, and sand dunes, with an altitude of 1,400–2,100 m. The natural vegetation in the study includes Nitraria sphaerocarpa, Salsola passerina, Reaumuria soongarica, Ephedra przewalskii, Alhagi sparsifolia, and Rtemisia desertorum, while the artificial vegetation mainly includes Haloxylon ammodendron, Elaeagnus angustifolia, and Hedysarum scoparium [33].

Figure 1 The location of the study area. (a) The red rectangle is the geographic location of the study area. (b) The true color image of GaoFen-6 satellite (GF-6) image after preprocessing.
Figure 1

The location of the study area. (a) The red rectangle is the geographic location of the study area. (b) The true color image of GaoFen-6 satellite (GF-6) image after preprocessing.

2.2 GF-6 WFV data

The Chinese GaoFen-6 satellite (GF-6) was officially put into use on March 21, 2019. It is a low-orbit optical remote sensing satellite, using the CAST 2000 platform. The satellite is equipped with a 2 m panchromatic/8 m multispectral high-resolution camera (PMS) and a 16 m multispectral medium-resolution WFV camera. The observation width of PMS is 90 km and that of WFV is 800 km. The time resolution is 4 days. The GF-6 WFV sensor covers 8 bands. Compared with the GF-1 satellite, in addition to the common four bands (blue, green, red, and near infrared), it also adds two red edge bands, purple band and yellow band. The technical specification for GF-6 WFV data is presented in Table 1.

Table 1

Technical specification of GaoFen-6 Wide Field View (GF-6 WFV) data used in this study

Band number Channel The range of spectrum (nm) The temporal resolution (d) The spatial resolution (m)
1 Blue 450–520 4 16
2 Green 520–590
3 Red 630–690
4 Near infrared 770–890
5 Red edge 1 690–730
6 Red edge 2 730–770
7 Purple 400–450
8 Yellow 590–630

The GF-6 WFV data used in this study were obtained from the Land Survey Satellite Data Service Platform of the China Resources Satellite Application Center. The satellite image was acquired on September 29, 2019, and the weather in the study area was sunny and cloudless. Affected by the adverse factors such as atmosphere, light, and terrain, a remote sensing image may be distorted due to geometric deformation, noise interference, and other reasons; therefore, it is necessary to preprocess the data to get the real surface reflectance. In this study, ENVI 5.5 software was used to complete the radiometric calibration, atmospheric correction, geometric correction, image cropping, and other preprocessing work for the acquired satellite images. Figure 1b shows the true color image of the GF-6 satellite image after preprocessing.

2.3 Field survey FVC

To evaluate the accuracy of FVC estimated by remote sensing, a field survey of FVC was also conducted in the study area. According to the geomorphology and the vegetation coverage type of the study area, 16 sampling points were set up in the study area, and the latitude and longitude coordinates were obtained by UniStrong GPS instrument. The selection of the sampling points followed the rules that the vegetation types and coverage within 2 × 2 pixels (32 m × 32 m) were basically the same. Ten samples were randomly selected at each sampling point, and the size of each sample was 2 m × 2 m. Nikon D7100 digital camera was used to take a photo of the sample at a height of 2 m above the ground. For the convenience of taking pictures, the 2 m × 2 m sample square was divided into four 1 m × 1 m sample squares. After the photo was taken, the four 1 m × 1 m small sample photos were corrected, spliced, and interpreted indoors to obtain the vegetation coverage of the 2 m × 2 m sample. Photoshop software was used to interpret the spliced images. By adjusting hue, saturation and brightness, and other steps, the interpreted images with a gray value were obtained. Through visual interpretation, pixels with a gray value greater than 125 were identified as vegetation pixels, and the proportion of these pixels in the image was calculated, so that the measured vegetation coverage could be obtained.

Figure 2 shows the actual photographs and the interpreted renderings of wheat and Haloxylon ammodendron. In the interpreted image, the bright pixels are vegetation, and the dark pixels are other features. Then, the measured FVC of the sample can be obtained by counting the percentage of bright pixels in the entire image. The average FVC of all samples was taken as the FVC of the sampling point.

Figure 2 Interpretation of the measured fractional vegetation cover. Images of (a) wheat and (c) Haloxylon ammodendron taken on-site. Interpretation results of (b) wheat and (d) H. ammodendron.
Figure 2

Interpretation of the measured fractional vegetation cover. Images of (a) wheat and (c) Haloxylon ammodendron taken on-site. Interpretation results of (b) wheat and (d) H. ammodendron.

3 Methodology

3.1 Estimation model of FVC

The remote sensing estimation methods of FVC mainly include the regression model method, the machine learning method, the mixed pixel decomposition model method, and so on. The regression model method estimates FVC by establishing linear or nonlinear regression relationships between certain bands or vegetation indices of remote sensing data and measured FVC. The commonly used bands in the regression model method include the red band, the green band, and the near infrared band [34]. The vegetation indices that have been studied are normalized differential vegetation index (NDVI) [35], enhanced vegetation index (EVI) [36], and modified vegetation index (MVI) [37]. The regression model method is simple and easy to implement, and the estimation accuracy of FVC on a small scale is also high. However, this method is not suitable for large-scale and low-resolution remote sensing data because of the high requirements on the number of measured data.

The common machine learning methods include the neural network [38], decision tree [39], support vector machine [40], and so on. According to the different training samples, machine learning methods can be divided into two categories: one based on remote sensing image classification [41] and the other based on a radiation transfer model [42,43]. Although the accuracy of FVC estimated by the machine learning method is high, the complex surface, sample selection, and model training will all affect the overall accuracy. In areas where there are scarce samples available for model training, the applicability of machine learning methods is limited.

The mixed pixel decomposition model assumes that each component in the pixel contributes to the observation of the remote sensing sensor, and the FVC is estimated by decomposing this mixed pixel. Mixed pixel decomposition models can be divided into linear and nonlinear models. The pixel dichotomy model is a kind of linear mixed pixel model and was also widely used with good effect [44,45,46]. The advantage of this model is that it still can be used without the measured FVC data. Based on the actual conditions of the study, the pixel dichotomy model method was chosen as the estimation model of FVC.

The assumption of the pixel dichotomy model is that the images captured by satellites only contain vegetation and soil. That is, at a certain pixel, there is a single vegetation, a single soil, or both ground objects. At this time, buildings, rivers, and other ground objects are not considered in the model. The pixel dichotomy model assumes that pixels in remote sensing satellite images consist of only vegetation and soil. In other words, the information S captured by the remote sensor can be linearly synthesized by S V contributed by the vegetation and S S contributed by the soil [47]:

(1) S = S V + S S .

Assuming that the proportion of vegetation coverage in a pixel is f c , it can be considered that FVC in that pixel is f c , and then, the proportion of bare soil is 1 − f c . Assuming that the remote sensing information obtained by pure pixels covered by all vegetation is S Veg , then the spectral response contributed by vegetation in the mixed pixels can be expressed as the product of S Veg and f c :

(2) S V = f c × S Veg .

Similarly, assuming the remote sensing information obtained by pure pixels all covered by soil is S Soil , and the information S S contributed by soil in the mixed pixel can be expressed as the product of S Soil and 1 − f c :

(3) S S = ( 1 − f c ) × S Soil .

Based on equations (1)–(3), the spectral response of a mixed pixel can be derived:

(4) S = S V + S S = f c × S Veg + ( 1 − f c ) × S Soil .

Then, the FVC can be obtained by modifying equation (4) as follows:

(5) f c = ( S − S Soil ) / ( S Veg − S Soil ) .

3.2 Vegetation index

The pixel dichotomy model requires that the remote sensing information used must have a good linear relationship with the FVC. Therefore, it is necessary to select the appropriate remote sensing information for the mixed pixel, photosynthetic vegetation end element, and soil end element. It is very limited to extract vegetation information by analyzing and comparing individual or multiple single-band data, but the vegetation index can better reflect the vegetation information. To this end, the reflection characteristics of ground objects in different bands need to be studied.

According to the actual investigation results, four types of features including plant, water body, desert, and bare soil were selected, and the reflectance of the pixels completely covered by these features was extracted from the preprocessed GF-6 image. The reflection characteristics of these typical features in different bands are shown in Figure 3.

Figure 3 Reflectance of typical features in different bands.
Figure 3

Reflectance of typical features in different bands.

Figure 3 shows that, except water body, the reflectance of the other three types of ground objects in the near infrared (B4), red edge 1 (B5), and red edge 2 (B6) bands is higher than that in the other bands. In particular, the reflectance of plant pixels in the near infrared band and the red edge 2 band is much higher than that in the red edge 1 band. According to the spectral characteristics exhibited by plant in the red edge bands, this study attempted to construct a new vegetation index based on red edge bands to invert FVC.

Among the many vegetation indices, NDVI is believed to partially eliminate the effects of changes in radiometric conditions related to the solar altitude angle, satellite observation angle, terrain, clouds, shadows, and atmospheric conditions. NDVI was also a commonly used parameter source in the estimation of vegetation coverage by the pixel dichotomy model [48,49,50]. In this study, NDVI was used to compare with the remote sensing estimation of FVC based on the new vegetation index. NVDI is calculated as follows:

(6) NDVI = NIR − RED NIR + RED .

In equation (6), NIR and RED are the reflectance of pixels in the near infrared band and the red band, respectively.

With reference to the definition of NDVI, a new vegetation index based on the red edge band is proposed: normalized difference red edge index (RENDVI). RENDVI is calculate as follows:

(7) RENDVI = RE − RED RE + RED .

In equation (7), RE is the reflectance of pixels in the red edge band, and RED is the reflectance of pixels in the red band. Since there are two red edge bands in the WFV data, the normalized difference red edge index based on the red edge 1 band and the red edge 2 band is recorded as RENDVI1 and RENDVI2, respectively.

3.3 Remote sensing estimating of FVC

Two parameters S Soil and S Veg in equation (5) were required to estimate FVC when using the pixel dichotomy model. These two parameters are usually determined by the measured method and the confidence method. In this study, the confidence method was used to extract S Soil and S Veg from the vegetation index due to the large area of research and the small number of measured samples. The researchers used different confidence levels in their respective FVC studies [51,52,53]. To compare the influence of confidence on the estimation results, according to the gray distribution of the vegetation index extracted from the entire image, the upper and lower thresholds of the vegetation index were intercepted with confidence levels of 1, 2, and 5%. That is, the vegetation index values with cumulative frequencies of 1, 2, and 5%, respectively, were recorded as S Soil , and the vegetation index values with cumulative frequencies of 99, 98, and 95, respectively, were recorded as S Veg . Then, the FVC thematic map based on a certain vegetation index under different confidence levels can be produced.

3.4 Accuracy evaluation

In this study, the field measured FVC and the FVC estimated based on the pixel dichotomy model were regressed and fitted. The accuracy of the FVC estimated by remote sensing was evaluated by calculating the determination coefficient (R 2) and root-mean-square error (RMSE). For the fitting result, if the R 2 is high and the RMSE is low, it means that the accuracy of remote sensing estimation is high. RMSE measured the overall estimation accuracy and could not check the estimation accuracy of each sample point. Therefore, the relative measurement error (RME) was also calculated to check the estimation accuracy of each sample point. Equations (8), (9) and (10) are the calculation methods of R 2, RMSE and RME, respectively.

(8) R 2 = ∑ i = 1 n P i ^ − P ¯ 2 / ∑ i = 1 n P i − P ¯ 2 ,

(9) RMSE = ∑ i = 1 n P i ^ − P ¯ 2 / n ,

(10) RME = P i ^ − P i / P i .

In equations (8)–(10), P i ^ represents the estimated value of FVC remote sensing of the ith sample point; P i represents the field survey FVC of the ith sample point; P ¯ represents the average FVC of the measured sample points; and n is the number of sample points.

4 Results

4.1 Field survey FVC of sample points

The FVC interpretation results of the sample points in the study area are presented in Table 2. The 16 sites covered four types of vegetation: H. ammodendron, wheat, low bush, and white poplar. The vegetation at sample point 13 was a cluster of woods near Hongyashan Reservoir where the foliage was very lush, and the measured FVC was also the largest, reaching 0.9967. Sample point 11 was located on the edge of the Tengger Desert, with scarce vegetation and the smallest FVC, i.e., only 0.0122.

Table 2

Field survey of fractional vegetation coverage (FVC) of sample points

Sample number FVC Sample number FVC
1 0.2722 9 0.0311
2 0.2756 10 0.0456
3 0.2011 11 0.0122
4 0.2322 12 0.9956
5 0.1267 13 0.9967
6 0.2867 14 0.5011
7 0.4089 15 0.2989
8 0.2333 16 0.2322

4.2 Vegetation indices extraction results

Based on equations (6) and (7), ENVI 5.5 software was used to obtain NDVI, RENDVI1 and RENDVI2, and the results are shown in Figure 4. The statistical results of the three vegetation indices are presented in Table 3.

Figure 4 Extraction results of (a) normalized differential vegetation index (NDVI), (b) normalized difference red edge index 1 (RENDVI1), and (c) normalized difference red edge index 2 (RENDVI2).
Figure 4

Extraction results of (a) normalized differential vegetation index (NDVI), (b) normalized difference red edge index 1 (RENDVI1), and (c) normalized difference red edge index 2 (RENDVI2).

Table 3

Statistical results of the three vegetation indices

Vegetation index Min value Max value Average value Standard deviation
NDVI −0.340858 0.760981 0.0849 0.1122
RENDVI1 −0.174256 0.350569 0.0309 0.03504
RENDVI2 −0.370259 0.733248 0.05029 0.0991

4.3 FVC estimation based on different vegetation indices

According to the extraction results of vegetation indices, the values of S Soil and S Veg under different confidence levels are presented in Table 4.

Table 4

The values of S Soil and S Veg under different confidence levels

Vegetation index 1% 2% 5%
S Soil S Veg S Soil S Veg S Soil S Veg
NDVI −0.088083 0.482281 −0.016788 0.426108 −0.008164 0.331047
RENDVI1 −0.039448 0.158133 −0.005489 0.135494 0.000685 0.100505
RENDVI2 −0.108627 0.418941 −0.037273 0.358148 −0.028624 0.263279

Substituting the S Soil and S Veg values from Table 4 into equation (5), the FVC can be obtained based on three vegetation indices under different confidence levels. According to the corresponding relationship between soil erosion intensity surface erosion classification and FVC in China’s ‘Classification Standards for Soil Erosion Classification, the FVC in the study area was divided into five levels: the lowest coverage (0–0.3), the lower coverage (0.3–0.45), the medium coverage (0.45–0.6), the higher coverage (0.6–0.75), and the highest coverage (0.75–1). The normal FVC should be between 0 and 1, but the FVC of some ground features (such as water bodies and shadows) may be less than 0. Therefore, it is necessary to remove the outliers from the calculation results. The usual operation method is to assign these pixels to 0.

According to the classification standards, the FVC estimation results in the study area are shown in Figures 5–7. Under the different confidence conditions, the estimation results based on different vegetation indices are significantly different. Figure 5 shows significantly more areas with lower coverage than Figures 6 and 7, which was particularly evident in Figure 5b. The areas covered by different levels of vegetation coverage shown in Figures 6 and 7 are similar, but Figure 7 shows significantly more pixels with vegetation coverage of 0 than Figures 5 and 6.

Figure 5 FVC estimation results based on (a) NDVI, (b) RENDVI1, and (c) RENDVI2 under 1% confidence.
Figure 5

FVC estimation results based on (a) NDVI, (b) RENDVI1, and (c) RENDVI2 under 1% confidence.

Figure 6 FVC estimation results based on (a) NDVI, (b) RENDVI1, and (c) RENDVI2 under 2% confidence.
Figure 6

FVC estimation results based on (a) NDVI, (b) RENDVI1, and (c) RENDVI2 under 2% confidence.

Figure 7 FVC estimation results based on (a) NDVI, (b) RENDVI1, and (c) RENDVI2 under 5% confidence.
Figure 7

FVC estimation results based on (a) NDVI, (b) RENDVI1, and (c) RENDVI2 under 5% confidence.

Tables 5–7 present pixel statistics of FVC grading results estimated by different vegetation indices at 1, 2, and 5% confidence level, respectively. The statistical results presented in Tables 5–7 show the vegetation coverage under different grades more accurately. The statistical results of pixels were consistent with the description of the inversion chart mentioned earlier. Under the 1% confidence level, the proportion of pixels contained in each grade was obviously different based on the inversion results obtained by the three indices, especially the difference between low vegetation coverage and low vegetation coverage was the most obvious. Under the two confidence levels of 2 and 5%, the number of pixels contained in each level in the inversion results obtained by different vegetation indices was different, but the proportion was similar. The proportion of pixels contained in the higher coverage level was very similar in all the inversion results.

Table 5

Pixel statistics of FVC grading results estimated by different vegetation indices at 1% confidence

Grading standards NDVI % RENDVI1 % RENDVI2 %
0 80,664 1.0 78,979 1.0 80,602 1.0
0–0.3 5,286,970 65.3 3,668,751 45.3 5,388,301 66.5
0.3–0.45 1,431,207 17.7 2,963,075 36.6 1,494,900 18.5
0.45–0.6 582,358 7.2 709,201 8.8 503,129 6.2
0.6–0.75 329,782 4.1 332,161 4.1 287,842 3.6
0.75–1 388,671 4.8 347,485 4.3 344,878 4.3
Table 6

Pixel statistics of FVC grading results estimated by different vegetation indices at 2% confidence

Grading standards NDVI % RENDVI1 % RENDVI2 %
0 111,026 1.4 133,255 1.6 116,312 1.4
0–0.3 5,983,771 73.9 5,877,837 72.6 6,176,159 76.3
0.3–0.45 831,920 10.3 978,067 12.1 775,250 9.6
0.45–0.6 429,001 5.3 430,850 5.3 362,254 4.5
0.6–0.75 275,149 3.4 256,180 3.2 236,897 2.9
0.75–1 468,785 5.8 423,463 5.2 432,780 5.3
Table 7

Pixel statistics of FVC grading results estimated by different vegetation indices at 5% confidence

Grading standards NDVI % RENDVI1 % RENDVI2 %
0 241,532 3.0 303,426 3.7 240,653 3.0
0–0.3 5,388,864 66.5 5,142,104 63.5 5,507,465 68.0
0.3–0.45 897,246 11.1 1,097,779 13.6 912,142 11.3
0.45–0.6 502,344 6.2 514,405 6.4 453,322 5.6
0.6–0.75 312,281 3.9 305,558 3.8 269,158 3.31
0.75–1 757,385 9.4 736,380 9.1 716,912 8.9

4.4 Fitting results of correlation between measured FVC and estimated FVC by remote sensing

To assess the accuracy of remote sensing estimation FVC, the linear regression model between the field measured FVC and the FVC estimated by remote sensing is established. R 2 and RMSE were calculated according to equations (8) and (9), and the results are shown in Figure 8.

Figure 8 Linear fitting results between the field survey FVC and the FVC estimated by remote sensing under three confidence levels of (a) 1%, (b) 2%, and (c) 5%.
Figure 8

Linear fitting results between the field survey FVC and the FVC estimated by remote sensing under three confidence levels of (a) 1%, (b) 2%, and (c) 5%.

4.5 Error analysis of remote sensing estimation of FVC in sample points

To investigate the estimation accuracy of FVC of each sample point in the study area, the RME was calculated according to equation (10), and the results are shown in Figures 9–11.

Figure 9 RME of sample points under 1% confidence.
Figure 9

RME of sample points under 1% confidence.

Figure 10 RME of sample points under 2% confidence.
Figure 10

RME of sample points under 2% confidence.

Figure 11 RME of sample points under 5% confidence.
Figure 11

RME of sample points under 5% confidence.

Figure 11 shows that the RME of most sample points is small. However, when the confidence is 1%, the RME of sample points 9, 10, and 11 is high, especially sample point 11. The results show that when the confidence is 1%, the estimation accuracy of FVC in the desert edge with sparse vegetation is relatively poor.

5 Discussion

5.1 Analysis of different vegetation indices extraction results

Table 3 presents statistics on the extraction results of three vegetation indices in the study area. It can be seen from the table that the values of NDVI, RENDVI1, and RENDVI2 are −0.340858 to 0.760981, −0.174256 to 0.350569, and −0.370259 to 0.733248, respectively. In general, the value range of NDVI and RENDVI2 is close, while that of RENDVI1 is relatively narrow. As shown in Figure 3, the reflectance of ground objects in the red edge 1 band is lower than that in the near infrared band and the red edge 2 band, and hence, the value range of RENDVI1 is different from that of NDVI and RENDVI2.

5.2 Analysis of FVC remote sensing estimation results

Figure 8 shows that, in general, R 2 values of the fitting results based on the three vegetation indices are all above 0.9, which indirectly indicates that it is feasible and reliable to use the pixel dichotomy model to estimate FVC. By comparing and analyzing the estimation results based on the three vegetation indices, under the same confidence level, the fitting degree of vegetation coverage estimated based on RENDVI2 is always the highest. Under the 2% confidence, the R 2 of FVC based on RENDVI2 is 0.97635, which is the best of all results. These comparative analysis results show that the vegetation index based on the red edge bands proposed can be effectively used for FVC estimation.

5.3 Influence of confidence levels on FVC remote sensing estimation

In this study, to explore the influence of confidence on FVC estimation, the results of remote sensing estimation of FVC under three different confidence levels were compared. For NDVI and RENDVI2, when the confidence level is 2%, the R 2 values of the estimation results are the largest, which are 0.97115 and 0.97635, respectively; for RENDVI1, when the confidence level is 5%, the R 2 of the estimation result is the largest, which is 0.95981. When the confidence level is 1%, the R 2 values of the estimation results based on the three vegetation indices are the smallest, which are 0.96559, 0.96726, and 0.94723, respectively. For a certain vegetation index, when confidence is used to determine S Soil and S Veg , it does not mean that the greater the confidence, the better the FVC estimation results, or the lower the confidence, the better the FVC estimation results. The results of this article show that it is feasible to use the confidence method to extract S Soil and S Veg when FVC is estimated by the pixel bisection model, even if there is no large amount of the measured data. However, the impact of different confidence levels on the estimated results is clearly visible. Therefore, to obtain better accuracy, it is necessary to select the appropriate confidence level according to the specific conditions of the study area.

5.4 Vegetation coverage of study area

Figures 5–7 and Tables 5–7 show that the pixels with minimum vegetation coverage (0–0.3) and low vegetation coverage (0.3–0.45) account for a large proportion in the study area. In addition, except for Hongyashan Reservoir and other water bodies, the proportion of pixels with FVC of 0 is not large, indicating that there are few areas without vegetation coverage. Although the study area is surrounded by the Tengger Desert and the Badain Jaran Desert, the local government and residents have taken measures in recent years to strictly control groundwater exploitation and actively transform the desert. H. ammodendron and other vegetation have gradually appeared in many areas completely covered by desert, which effectively improved the local desertification status [54].

It is worth noting that when the confidence is 1%, in the estimation results of FVC based on RENDVI1, the proportion of pixels with FVC of 0–0.3 is significantly lower than that of other cases under the condition of 1% confidence, while the proportion of pixels with FVC of 0.3–0.45 is significantly higher than that of other eight cases. The possible reason for this phenomenon is that the range of RENDVI1 is narrow. After extracting the relevant parameters of the pixel dichotomy model with 1% confidence, the model would enlarge the FVC of some sparse vegetation coverage areas.

5.5 Analysis of estimation error and uncertainty

The error sources of this study mainly include the selection of sample points for verification, the field survey FVC of sample points, and remote sensing images.

First, it is well known that it is impossible to completely match the position of the measured points on the ground with the corresponding points on the remote sensing image when there is no obvious reference feature. Therefore, it was required that the vegetation difference near the sample points used for verification should not be too large. In the selection of sample points, the principle of little difference in vegetation type and coverage within the range of 2 × 2 pixels (32 m × 32 m) was followed as far as possible, and the field investigation of FVC would not be affected by the vegetation type and the coverage around the sample points consequently.

Second, due to the limitation of test conditions, the field survey FVC was estimated by photographing. Ten samples were randomly selected in each sample point, and the average FVC of all samples was regarded as the FVC of the sample point. The size of each sample is only 2 m × 2 m, while that of the sample point is 32 m × 32 m. The randomness of sample selection will affect the calculation of FVC to a certain extent. However, the size of the sample point is larger than that of the pixel, and the vegetation difference within a sample point is small; therefore, the error caused by sample selection can be ignored.

Finally, in the process of imaging, there will be deformation and dislocation for various reasons in remote sensing images, but these phenomena will be corrected to a large extent after radiation calibration, atmospheric correction, and geometric correction.

6 Conclusion

In this study, GF-6 WFV data were used as the data source, and the FVC estimation results based on NDVI, RENDVI1, and RENDVI2 were compared to explore the application of the red edge band in remote sensing estimation of FVC. Some useful conclusions are obtained.

The vegetation indices (RENDVI1 and RENDVI2) based on the red edge bands proposed in this paper showed good results in the remote sensing estimation of FVC. Whether using NDVI, RENDVI1, or RENDVI2, the estimated FVC based on the GF-6 WFV data had a good linear correlation with the field measured FVC. In terms of accuracy, the accuracy of FVC based on RENDVI1 was worse than that based on RENDVI2 and NDVI. The accuracy of FVC estimated based on RENDVI2 and NDVI was close, but the overall accuracy of RENDVI2 was better. At the 2% confidence, the model based on RENDVI2 generated the best reasonable FVC estimation (R 2 = 0.97611 and RMSE = 0.07075). Studies had shown that the red edge band has a greater advantage in FVC remote sensing estimation in a large-scale background. The vegetation indices defined by the red edge bands of GF-6 WFV data and the FVC estimation method used in this paper can provide a meaningful reference for FVC remote sensing estimation.

According to the principle of the pixel dichotomy model, in theory, S Soil should be close to 0 and S Veg should be close to 1. However, due to various factors such as atmospheric conditions, regions, and vegetation types, these two parameters often change. In this study, three confidence levels (1, 2, and 5%) were used to determine S Soil and S Veg of the model and FVC, respectively. In general, the choice of confidence had an obvious influence on the estimation result of FVC, but there was no general law. At 2% confidence, FVC obtained from NDVI and RENDVI2 has the highest accuracy, while FVC based on RENDVI1 achieved the best accuracy at 5% confidence. When using the pixel dichotomy model to estimate FVC, a reasonable confidence level should be selected based on the land cover and satellite image characteristics of the study area.

In this study, due to various reasons such as equipment and personnel, the photographic method was used to measure the actual vegetation coverage. Due to the limited sample area observed by the camera method, the number and accuracy of the measured samples bring certain difficulties and influences to the verification of FVC. The arid area of Minqin, Gansu Province, was only taken as the research object. The remote sensing estimation method of FVC proposed in the study still needs to be widely verified in different climate and environment areas. At the same time, in order to improve the reliability of verification, it is very important to further enrich the measured FVC date. If conditions permit in the future, high-resolution UAV platform or other more advanced methods can be used to measure FVC, which may be more conducive to the verification of remote sensing estimation of FVC, thereby further promoting the development of this study.

Acknowledgments

The authors would deeply appreciate the anonymous reviewers and the editor for their constructive comments and suggestions, all of which have led to great improvements in the presentation of this article.

  1. Author contributions: Z. D. and Z. L.: conceptualization; Z. L.: methodology; Z. L and G. W.: software; D. W. and Z. D.: validation; H. Z.: formal analysis; G. W. and H. X.: investigation; H. Z. and X. Z.: data curation; Z. L. and G. W.: writing – original draft preparation; Y. S. and Z. C.: writing – review and editing; Z. D.: project administration. All authors have read and agreed to the published version of the manuscript.

  2. Funding information: This research received no external funding.

  3. Conflict of interest: The authors declare no conflict of interest.

References

[1] Parmesan C, Yohe G. A globally coherent fingerprint of climate change impacts across natural systems. Nature. 2003;421(6918):37–42. 10.1038/nature01286.Search in Google Scholar PubMed

[2] Mu XH, Song WJ, Gao Z, McVicar TR, Donhue RJ, Yan GJ. Fractional vegetation cover estimation by using multi-angle vegetation index. Remote Sens Environ. 2018;216:44–56. 10.1016/j.rse.2018.06.022.Search in Google Scholar

[3] Yao YJ, Liang SL, Cheng J, Lin Y, Jia K, Liu M. Impacts of deforestation and climate variability on terrestrial evapotranspiration in subarctic China. Forests. 2014;5(10):2542–60. 10.3390/f5102542.Search in Google Scholar

[4] Gitelson AA, Kaufman YJ, Stark R, Rundquist D. Novel algorithms for remote estimation of vegetation fraction. Remote Sens Environ. 2002;80(1):76–87. 10.1016/S0034-4257(01)00289-9.Search in Google Scholar

[5] Jiapaer G, Chen X, Bao AM. A comparison of methods for estimating fractional vegetation cover in arid regions. Agric For Meteorol. 2011;151(12):1698–710. 10.1016/j.agrformet.2011.07.004.Search in Google Scholar

[6] Zhang WB, Fu SH, Liu BY. Error assessment of visual estimation plant coverage. J Beijing Norm Univ (Nat Sci). 2001;37(3):402–8. 10.3321/j.issn:0476-0301.2001.03.026.Search in Google Scholar

[7] Feng L, Hu WY, Li YX, Zhang EW. Dynamic monitoring of multi-year vegetation coverage in Sichuan province based on google earth engines. For Resour Manage. 2019;4:124–31. 10.13466/j.cnki.lyzygl.2019.04.018.Search in Google Scholar

[8] Taheri F, Jafari H, Rezaei M, Bagheri R. The use of continuous fuzzy and traditional classification models for groundwater potentiality mapping in areas underlain by granitic hard-rock aquifers. Environ Earth Sci. 2020;79(5):1–16. 10.1007/s12665-020-8830-y.Search in Google Scholar

[9] Jin XM, Liu JT, Wang ST, Xia W. Vegetation dynamics and their response to groundwater and climate variables in Qaidam Basin, China. Int J Remote Sens. 2016;37(3):710–28. 10.1080/01431161.2015.1137648.Search in Google Scholar

[10] Gao ZH, Li ZY, Wei HD, Ding F. Quantitative monitoring of vegetation cover change by using remotely sensed data over Minqin Oasis, Gansu. Geogr Res. 2006;25(4):587–93. 10.3321/j.issn:1000-0585.2006.04.004.Search in Google Scholar

[11] Yao Y, Wei X, Gao S, Jiang B, Zhao X. A review on fractional vegetation cover estimation using remote sensing. Adv Earth Sci. 2013;37(3):774–82. 10.11867/j.issn.1001-8166.2013.07.0774.Search in Google Scholar

[12] Zhang ZX, Deng RR, Li H, Chen L, Chen QD, He YQ. Remote sensing monitoring of vegetation coverage in southern China based on pixel unmixing: a case study of Guangzhou city. Remote Sens Land Resour. 2011;3:88–94. CNKI:SUN:GTYG.0.2011-03-018.Search in Google Scholar

[13] Zhou ZM, Yang YM, Chen BQ. Estimating Spartina alterniflora fractional vegetation cover and aboveground biomass in a coastal wetland using SPOT6 satellite and UAV data. Aquat Bot. 2018;144:38–45. 10.1016/j.aquabot.2017.10.004.Search in Google Scholar

[14] Li LL, Wang DW, Ham T. Spatial-temporal dynamics of vegetation coverage and responding to climate change in Shiyang River Basin during 2000–15. J Desert Res. 2018;38(5):212–22. 10.7522/j.issn.1000-694X.2017.00061.Search in Google Scholar

[15] Tao GF, Jia K, Zhao X, Wei XQ, Xie XH, Zhang XW, et al. Generating high spatio-temporal resolution fractional vegetation cover by fusing GF-1 WFV and MODIS data. Remote Sens. 2019;11(19):2324. 10.3390/rs11192324.Search in Google Scholar

[16] Qiao ZM, Wang XB, Yang L, Ma YG. Study on remote sensing inversion model of vegetation coverage using Landsat 8. J Qinghai Norm Univ (Nat Sci). 2019;35(1):59–63. 10.16229/j.cnki.issn1001-7542.2019.01.009.Search in Google Scholar

[17] Tucker CJ. Red and photographic infrared linear combinations for monitoring vegetation. Remote Sens Environ. 1979;8(2):127–50. 10.1016/0034-4257(79)90013-0.Search in Google Scholar

[18] Verrelst J, Schaepman ME, Koetz B, Kneubühler M. Angular sensitivity analysis of vegetation indices derived from CHRIS/PROBA data. Remote Sens Environ. 2008;112(5):2341–53. 10.1016/j.rse.2007.11.001.Search in Google Scholar

[19] Wang XQ, Wang MM, Wang SQ, Wu YD. Extraction of vegetation information from visible unmanned aerial vehicle images. Trans Chin Soc. Agric Eng. 2015;31(5):152–8. 10.3969/j.issn.1002-6819.2015.05.022.Search in Google Scholar

[20] Liu QY, Zhang TL, Li YZ, Li Y, Bu CF, Zhang QF. Comparative analysis of fractional vegetation cover estimation based on multi-sensor data in a semi-arid sandy area. Chin Geogr Sci. 2019;29(1):166–80. 10.1007/s11769-018-1010-2.Search in Google Scholar

[21] Niu BR, Liu JR, Wang ZW. Remote sensing information extraction based on vegetation fraction in drought and half-drought area. Geomatics Inf Sci Wuhan Univ. 2005;30(1):27–80. 10.3321/j.issn:1671-8860.2005.01.007.Search in Google Scholar

[22] Yue J, Mu GJ, Tang ZH, Yang XF, Lin YC, Xu LS. Remote sensing estimation models for vegetation coverage in desert regions of Xinjiang based on NDVI. Arid Land Geogr. 2020;43(1):153–60. 10.12118/j.issn.1000-6060.2020.01.18.Search in Google Scholar

[23] Kanke Y, Tubana B, Dalen M, Harrell D. Evaluation of red and red-edge reflectance-based vegetation indices for rice biomass and grain yield prediction models in paddy fields. Precis Agric. 2016;17(5):507–30. 10.1007/s11119-016-9433-1.Search in Google Scholar

[24] Li D, Cheng T, Zhou K, Zheng HB, Yao X, Tian YC, et al. WREP: a wavelet-based technique for extracting the red edge position from reflectance spectra for estimating leaf and canopy chlorophyll contents of cereal crops. ISPRS J Photogramm. 2017;129:103–17. 10.1016/j.isprsjprs.2017.04.024.Search in Google Scholar

[25] Forkuor G, Dimobe K, Serme I, Tondoh JE. Landsat-8 vs. Sentinel-2: examining the added value of sentinel-2’s red-edge bands to land-use and land-cover mapping in Burkina Faso. Gisci Remote Sens. 2017;55(3):331–54. 10.1080/15481603.2017.1370169.Search in Google Scholar

[26] Thompson CN, Guo W, Sharma B, Ritchie GL. Using normalized difference red edge index to assess maturity in cotton. Crop Sci. 2019;59:2167–77. 10.2135/cropsci2019.04.0227.Search in Google Scholar

[27] Liu J, Wang LM, Teng F, Yang LB, Gao JM, Yao BM, et al. Impact of red-edge waveband of RapidEye satellite on estimation accuracy of crop planting area. Trans Chin Soc Agric Eng (Trans CSAE). 2016;32(13):140–8. 10.11975/j.issn.1002-6819.2016.13.020.Search in Google Scholar

[28] Li XJ, Chen G, Liu JY, Chen WT, Cheng XW, Liao YW. Effects of RapidEye imagery’s red-edge band and vegetation indices on land cover classification in an arid region. Chin Geogr Sci. 2017;27:827–35. 10.1007/s11769-017-0894-6.Search in Google Scholar

[29] Kim HO, Yeom JM. Sensitivity of vegetation indices to spatial degradation of RapidEye imagery for paddy rice detection: a case study of South Korea. Gisci Remote Sens. 2015;5(1):1–17. 0.1080/15481603.2014.1001666.Search in Google Scholar

[30] She B, Huang J, Shi JJ, Wei CW. Extracting oilseed rape growing regions based on variation characteristics of red edge position. Trans Chin Soc Agric Eng (Trans CSAE). 2013;29(15):145–52. 10.3969/j.issn.1002-6819.2013.15.018.Search in Google Scholar

[31] Cai YT, Zhang M, Lin H. Estimating the urban fractional vegetation cover using an object-based mixture analysis method and sentinel-2 MSI imagery. IEEE J-Stars. 2020;13:341–50. 10.1109/JSTARS.2019.2962550.Search in Google Scholar

[32] Wang B, Jia K, Liang SL, Xie XH, Wei X, Zhao X, et al. Assessment of sentinel-2 MSI spectral band reflectances for estimating fractional vegetation cover. Remote Sens. 2018;10(12):1927. 10.3390/rs10121927.Search in Google Scholar

[33] Hua YC, Li ZY, Gao ZH. Variation of vegetation coverage in Minqin city since 2001. Arid Zone Res. 2017;34(2):337–43. 10.13866/j.azr.2017.02.13.Search in Google Scholar

[34] North PRJ. Estimation of fAPAR, LAI, and vegetation fractional cover from ATSR-2 imagery. Remote Sens Environ. 2002;80:114–21. 10.1016/S0034-4257(01)00292-9.Search in Google Scholar

[35] Xiao JF, Moody A. A comparison of methods for estimating fractional green vegetation cover within a desert-to-upland transition zone in central New Mexico, USA. Remote Sens Environ. 2005;98:237–50. 10.1016/j.rse.2005.07.011.Search in Google Scholar

[36] Yang Q, Wang TT, Chen H, Wang YD. Characteristics of vegetation cover change in Xilin Gol League based on MODIS EVI data. Trans Chin Soc Agric Eng (Trans CSAE). 2015;31(22):191–8. 10.11975/j.issn.1002-6819.2015.22.026.Search in Google Scholar

[37] Baloloy AB, Blanco AC, Ana RRCS, Nadaoka K. Development and application of a new mangrove vegetation index (MVI) for rapid and accurate mangrove mapping. ISPRS J Photogramm. 2020;166:95–117. 10.1016/j.isprsjprs.2020.06.001.Search in Google Scholar

[38] Jia K, Liang SL, Liu SH, Li YW, Xiao Z, Yao Y, et al. Global land surface fractional vegetation cover estimation using general regression neural networks from MODIS surface reflectance. IEEE Trans Geosci Remote. 2015;53(9):4787–96. 10.1109/TGRS.2015.2409563.Search in Google Scholar

[39] Hansen MC, DeFries RS, Townshend JRG, Carroll M, Dimiceli C, Sohlberg RA. Global percent tree cover at a spatial resolution of 500 meters: first results of the MODIS vegetation continuous fields algorithm. Earth Interact. 2003;7(1):1–15. 10.3161/150811008X331063.Search in Google Scholar

[40] Su LH. Optimizing support vector machine learning for semi-arid vegetation mapping by using clustering analysis. ISPRS J Photogramm. 2009;64:407–13. 10.1016/j.isprsjprs.2009.02.002.Search in Google Scholar

[41] Han D, Wang HZ, Zheng BY, Wang F. Vegetation type classification and fractional vegetation coverage estimation for an open elm (Ulmus pumila) woodland ecosystem during a growing season based on an unmanned aerial vehicle platform coupled with decision tree algorithms. Acta Ecol Sin. 2018;38(18):6655–63. 10.5846/stxb201803300694.Search in Google Scholar

[42] Liu DY, Yang LQ, Jia K, Liang SL, Xiao ZQ, Wei XQ, et al. Global fractional vegetation cover estimation algorithm for VIIRS reflectance data based on machine learning methods. Remote Sens. 2018;10:1648. 10.3390/rs10101648.Search in Google Scholar

[43] Guo YK, Liu YL, Zhang XJ, Xu M. LAI inversion using radiation transfer model and random forest regression. Eng Surv Map. 2019;28(6):17–21 + 29. 10.19349/j.cnki.issn1006-7949.2019.06.004.Search in Google Scholar

[44] Qi J, Marsett RC, Moran MS, Goodrich DC, Heilman P, Kerr YH, et al. Spatial and temporal dynamics of vegetation in the San Pedro River basin area. Agr For Meteorol. 2000;105:55–68. 10.1016/S0168-1923(00)00195-7.Search in Google Scholar

[45] Li MM, Wu BF, Yan CZ, Zhou WF. Estimation of vegetation fraction in the upper basin of Miyun reservoir by remote sensing. Resour Sci. 2004;26:153–9. 10.1007/BF02973453.Search in Google Scholar

[46] Cheng DY, Li XD. Vegetation coverage change in a karst area and effects of terrain and population. J Geo-Inf Sci. 2019;21(8):1227–39. 10.12082/dqxxkx.2019.180548.Search in Google Scholar

[47] Zhao GC, Li Z, Han ZX. Study on dynamic change of vegetation coverage in Beipiao city based on pixel bipartite model. Geomatics Spat Inf Technol. 2019;42(7):5–7. 10.3969/j.issn.1672-5867.2019.07.002.Search in Google Scholar

[48] Zhang XF, Liao CH, Li J, Sun Q. Fractional vegetation cover estimation in arid and semi-arid environments using HJ-1 satellite hyperspectral data. Int J Appl Earth Obs. 2013;21:506–12. 10.1016/j.jag.2012.07.003.Search in Google Scholar

[49] Wang J, Yan QW, Tan XL, Zou YQ. Vegetation coverage dynamics and its driving factors in inner Mongolia based on FVC information entropy. For Resour Manage. 2019;4:159–67. 10.3466/j.nki.yzygl.019.4.23.Search in Google Scholar

[50] Gao LM, Zhang LL. Spatiotemporal dynamics of the vegetation coverage in Qinghai Lake basin. J Geo-Inf Sci. 2019;21(9):1318–29. 10.12082/dqxxkx.2019.180696.Search in Google Scholar

[51] Pang GW, Yang QK, Wang CM, Shan LX, Wang BL. Influence of parameter determination methods of the pixel dichotomy model on the estimation accuracy of fractional vegetation cover by GF-1 PMS data. Geogr Geo-Inf Sci. 2019;35(4):27–33. 10.3969/j.issn.1672-0504.2019.04.005.Search in Google Scholar

[52] Shen Q, Zhu C, Zhang X. Fractional vegetation cover estimation in aird and rare vegetation area aided by GF-2 remote sensing data. Bull Surv Map. 2019;7:33–8. 10.13474/j.cnki.11-2246.2019.0214.Search in Google Scholar

[53] Meng CC, Zhao J, Lan YB, Lu LQ, Yang HB, Li ZM, et al. Vegetation coverage extraction model of winter wheat based on pixel dichotomy. J South China Agric Univ. 2020;41(3):126–32. 10.7671/j.issn.1001-411X.201909005.Search in Google Scholar

[54] Lu H, Xie YW, Zhang WP, Wei JJ. Spatiotemporal change of the oasis in Minqin county during the period from 1986 to 2015. Arid Zone Res. 2017;34(6):1410–7. 10.13866/j.azr.2017.06.25.Search in Google Scholar
Received: 2021-01-10
Revised: 2021-01-26
Accepted: 2021-03-10
Published Online: 2021-04-05

© 2021 Zhengdong Deng et al., published by De Gruyter

This work is licensed under the Creative Commons Attribution 4.0 International License.

Articles in the same Issue

  1. Regular Articles
  2. Lithopetrographic and geochemical features of the Saalian tills in the Szczerców outcrop (Poland) in various deformation settings
  3. Spatiotemporal change of land use for deceased in Beijing since the mid-twentieth century
  4. Geomorphological immaturity as a factor conditioning the dynamics of channel processes in Rządza River
  5. Modeling of dense well block point bar architecture based on geological vector information: A case study of the third member of Quantou Formation in Songliao Basin
  6. Predicting the gas resource potential in reservoir C-sand interval of Lower Goru Formation, Middle Indus Basin, Pakistan
  7. Study on the viscoelastic–viscoplastic model of layered siltstone using creep test and RBF neural network
  8. Assessment of Chlorophyll-a concentration from Sentinel-3 satellite images at the Mediterranean Sea using CMEMS open source in situ data
  9. Spatiotemporal evolution of single sandbodies controlled by allocyclicity and autocyclicity in the shallow-water braided river delta front of an open lacustrine basin
  10. Research and application of seismic porosity inversion method for carbonate reservoir based on Gassmann’s equation
  11. Impulse noise treatment in magnetotelluric inversion
  12. Application of multivariate regression on magnetic data to determine further drilling site for iron exploration
  13. Comparative application of photogrammetry, handmapping and android smartphone for geotechnical mapping and slope stability analysis
  14. Geochemistry of the black rock series of lower Cambrian Qiongzhusi Formation, SW Yangtze Block, China: Reconstruction of sedimentary and tectonic environments
  15. The timing of Barleik Formation and its implication for the Devonian tectonic evolution of Western Junggar, NW China
  16. Risk assessment of geological disasters in Nyingchi, Tibet
  17. Effect of microbial combination with organic fertilizer on Elymus dahuricus
  18. An OGC web service geospatial data semantic similarity model for improving geospatial service discovery
  19. Subsurface structure investigation of the United Arab Emirates using gravity data
  20. Shallow geophysical and hydrological investigations to identify groundwater contamination in Wadi Bani Malik dam area Jeddah, Saudi Arabia
  21. Consideration of hyperspectral data in intraspecific variation (spectrotaxonomy) in Prosopis juliflora (Sw.) DC, Saudi Arabia
  22. Characteristics and evaluation of the Upper Paleozoic source rocks in the Southern North China Basin
  23. Geospatial assessment of wetland soils for rice production in Ajibode using geospatial techniques
  24. Input/output inconsistencies of daily evapotranspiration conducted empirically using remote sensing data in arid environments
  25. Geotechnical profiling of a surface mine waste dump using 2D Wenner–Schlumberger configuration
  26. Forest cover assessment using remote-sensing techniques in Crete Island, Greece
  27. Stability of an abandoned siderite mine: A case study in northern Spain
  28. Assessment of the SWAT model in simulating watersheds in arid regions: Case study of the Yarmouk River Basin (Jordan)
  29. The spatial distribution characteristics of Nb–Ta of mafic rocks in subduction zones
  30. Comparison of hydrological model ensemble forecasting based on multiple members and ensemble methods
  31. Extraction of fractional vegetation cover in arid desert area based on Chinese GF-6 satellite
  32. Detection and modeling of soil salinity variations in arid lands using remote sensing data
  33. Monitoring and simulating the distribution of phytoplankton in constructed wetlands based on SPOT 6 images
  34. Is there an equality in the spatial distribution of urban vitality: A case study of Wuhan in China
  35. Considering the geological significance in data preprocessing and improving the prediction accuracy of hot springs by deep learning
  36. Comparing LiDAR and SfM digital surface models for three land cover types
  37. East Asian monsoon during the past 10,000 years recorded by grain size of Yangtze River delta
  38. Influence of diagenetic features on petrophysical properties of fine-grained rocks of Oligocene strata in the Lower Indus Basin, Pakistan
  39. Impact of wall movements on the location of passive Earth thrust
  40. Ecological risk assessment of toxic metal pollution in the industrial zone on the northern slope of the East Tianshan Mountains in Xinjiang, NW China
  41. Seasonal color matching method of ornamental plants in urban landscape construction
  42. Influence of interbedded rock association and fracture characteristics on gas accumulation in the lower Silurian Shiniulan formation, Northern Guizhou Province
  43. Spatiotemporal variation in groundwater level within the Manas River Basin, Northwest China: Relative impacts of natural and human factors
  44. GIS and geographical analysis of the main harbors in the world
  45. Laboratory test and numerical simulation of composite geomembrane leakage in plain reservoir
  46. Structural deformation characteristics of the Lower Yangtze area in South China and its structural physical simulation experiments
  47. Analysis on vegetation cover changes and the driving factors in the mid-lower reaches of Hanjiang River Basin between 2001 and 2015
  48. Extraction of road boundary from MLS data using laser scanner ground trajectory
  49. Research on the improvement of single tree segmentation algorithm based on airborne LiDAR point cloud
  50. Research on the conservation and sustainable development strategies of modern historical heritage in the Dabie Mountains based on GIS
  51. Cenozoic paleostress field of tectonic evolution in Qaidam Basin, northern Tibet
  52. Sedimentary facies, stratigraphy, and depositional environments of the Ecca Group, Karoo Supergroup in the Eastern Cape Province of South Africa
  53. Water deep mapping from HJ-1B satellite data by a deep network model in the sea area of Pearl River Estuary, China
  54. Identifying the density of grassland fire points with kernel density estimation based on spatial distribution characteristics
  55. A machine learning-driven stochastic simulation of underground sulfide distribution with multiple constraints
  56. Origin of the low-medium temperature hot springs around Nanjing, China
  57. LCBRG: A lane-level road cluster mining algorithm with bidirectional region growing
  58. Constructing 3D geological models based on large-scale geological maps
  59. Crops planting structure and karst rocky desertification analysis by Sentinel-1 data
  60. Physical, geochemical, and clay mineralogical properties of unstable soil slopes in the Cameron Highlands
  61. Estimation of total groundwater reserves and delineation of weathered/fault zones for aquifer potential: A case study from the Federal District of Brazil
  62. Characteristic and paleoenvironment significance of microbially induced sedimentary structures (MISS) in terrestrial facies across P-T boundary in Western Henan Province, North China
  63. Experimental study on the behavior of MSE wall having full-height rigid facing and segmental panel-type wall facing
  64. Prediction of total landslide volume in watershed scale under rainfall events using a probability model
  65. Toward rainfall prediction by machine learning in Perfume River Basin, Thua Thien Hue Province, Vietnam
  66. A PLSR model to predict soil salinity using Sentinel-2 MSI data
  67. Compressive strength and thermal properties of sand–bentonite mixture
  68. Age of the lower Cambrian Vanadium deposit, East Guizhou, South China: Evidences from age of tuff and carbon isotope analysis along the Bagong section
  69. Identification and logging evaluation of poor reservoirs in X Oilfield
  70. Geothermal resource potential assessment of Erdaobaihe, Changbaishan volcanic field: Constraints from geophysics
  71. Geochemical and petrographic characteristics of sediments along the transboundary (Kenya–Tanzania) Umba River as indicators of provenance and weathering
  72. Production of a homogeneous seismic catalog based on machine learning for northeast Egypt
  73. Analysis of transport path and source distribution of winter air pollution in Shenyang
  74. Triaxial creep tests of glacitectonically disturbed stiff clay – structural, strength, and slope stability aspects
  75. Effect of groundwater fluctuation, construction, and retaining system on slope stability of Avas Hill in Hungary
  76. Spatial modeling of ground subsidence susceptibility along Al-Shamal train pathway in Saudi Arabia
  77. Pore throat characteristics of tight reservoirs by a combined mercury method: A case study of the member 2 of Xujiahe Formation in Yingshan gasfield, North Sichuan Basin
  78. Geochemistry of the mudrocks and sandstones from the Bredasdorp Basin, offshore South Africa: Implications for tectonic provenance and paleoweathering
  79. Apriori association rule and K-means clustering algorithms for interpretation of pre-event landslide areas and landslide inventory mapping
  80. Lithology classification of volcanic rocks based on conventional logging data of machine learning: A case study of the eastern depression of Liaohe oil field
  81. Sequence stratigraphy and coal accumulation model of the Taiyuan Formation in the Tashan Mine, Datong Basin, China
  82. Influence of thick soft superficial layers of seabed on ground motion and its treatment suggestions for site response analysis
  83. Monitoring the spatiotemporal dynamics of surface water body of the Xiaolangdi Reservoir using Landsat-5/7/8 imagery and Google Earth Engine
  84. Research on the traditional zoning, evolution, and integrated conservation of village cultural landscapes based on “production-living-ecology spaces” – A case study of villages in Meicheng, Guangdong, China
  85. A prediction method for water enrichment in aquifer based on GIS and coupled AHP–entropy model
  86. Earthflow reactivation assessment by multichannel analysis of surface waves and electrical resistivity tomography: A case study
  87. Geologic structures associated with gold mineralization in the Kirk Range area in Southern Malawi
  88. Research on the impact of expressway on its peripheral land use in Hunan Province, China
  89. Concentrations of heavy metals in PM2.5 and health risk assessment around Chinese New Year in Dalian, China
  90. Origin of carbonate cements in deep sandstone reservoirs and its significance for hydrocarbon indication: A case of Shahejie Formation in Dongying Sag
  91. Coupling the K-nearest neighbors and locally weighted linear regression with ensemble Kalman filter for data-driven data assimilation
  92. Multihazard susceptibility assessment: A case study – Municipality of Štrpce (Southern Serbia)
  93. A full-view scenario model for urban waterlogging response in a big data environment
  94. Elemental geochemistry of the Middle Jurassic shales in the northern Qaidam Basin, northwestern China: Constraints for tectonics and paleoclimate
  95. Geometric similarity of the twin collapsed glaciers in the west Tibet
  96. Improved gas sand facies classification and enhanced reservoir description based on calibrated rock physics modelling: A case study
  97. Utilization of dolerite waste powder for improving geotechnical parameters of compacted clay soil
  98. Geochemical characterization of the source rock intervals, Beni-Suef Basin, West Nile Valley, Egypt
  99. Satellite-based evaluation of temporal change in cultivated land in Southern Punjab (Multan region) through dynamics of vegetation and land surface temperature
  100. Ground motion of the Ms7.0 Jiuzhaigou earthquake
  101. Shale types and sedimentary environments of the Upper Ordovician Wufeng Formation-Member 1 of the Lower Silurian Longmaxi Formation in western Hubei Province, China
  102. An era of Sentinels in flood management: Potential of Sentinel-1, -2, and -3 satellites for effective flood management
  103. Water quality assessment and spatial–temporal variation analysis in Erhai lake, southwest China
  104. Dynamic analysis of particulate pollution in haze in Harbin city, Northeast China
  105. Comparison of statistical and analytical hierarchy process methods on flood susceptibility mapping: In a case study of the Lake Tana sub-basin in northwestern Ethiopia
  106. Performance comparison of the wavenumber and spatial domain techniques for mapping basement reliefs from gravity data
  107. Spatiotemporal evolution of ecological environment quality in arid areas based on the remote sensing ecological distance index: A case study of Yuyang district in Yulin city, China
  108. Petrogenesis and tectonic significance of the Mengjiaping beschtauite in the southern Taihang mountains
  109. Review Articles
  110. The significance of scanning electron microscopy (SEM) analysis on the microstructure of improved clay: An overview
  111. A review of some nonexplosive alternative methods to conventional rock blasting
  112. Retrieval of digital elevation models from Sentinel-1 radar data – open applications, techniques, and limitations
  113. A review of genetic classification and characteristics of soil cracks
  114. Potential CO2 forcing and Asian summer monsoon precipitation trends during the last 2,000 years
  115. Erratum
  116. Erratum to “Calibration of the depth invariant algorithm to monitor the tidal action of Rabigh City at the Red Sea Coast, Saudi Arabia”
  117. Rapid Communication
  118. Individual tree detection using UAV-lidar and UAV-SfM data: A tutorial for beginners
  119. Technical Note
  120. Construction and application of the 3D geo-hazard monitoring and early warning platform
  121. Enhancing the success of new dams implantation under semi-arid climate, based on a multicriteria analysis approach: Case of Marrakech region (Central Morocco)
  122. TRANSFORMATION OF TRADITIONAL CULTURAL LANDSCAPES - Koper 2019
  123. The “changing actor” and the transformation of landscapes
https://doi.org/10.1515/geo-2020-0241
Keywords for this article
vegetation index; pixel dichotomy model; fractional vegetation cover; red edge band
Creative Commons
BY 4.0

Articles in the same Issue

  1. Regular Articles
  2. Lithopetrographic and geochemical features of the Saalian tills in the Szczerców outcrop (Poland) in various deformation settings
  3. Spatiotemporal change of land use for deceased in Beijing since the mid-twentieth century
  4. Geomorphological immaturity as a factor conditioning the dynamics of channel processes in Rządza River
  5. Modeling of dense well block point bar architecture based on geological vector information: A case study of the third member of Quantou Formation in Songliao Basin
  6. Predicting the gas resource potential in reservoir C-sand interval of Lower Goru Formation, Middle Indus Basin, Pakistan
  7. Study on the viscoelastic–viscoplastic model of layered siltstone using creep test and RBF neural network
  8. Assessment of Chlorophyll-a concentration from Sentinel-3 satellite images at the Mediterranean Sea using CMEMS open source in situ data
  9. Spatiotemporal evolution of single sandbodies controlled by allocyclicity and autocyclicity in the shallow-water braided river delta front of an open lacustrine basin
  10. Research and application of seismic porosity inversion method for carbonate reservoir based on Gassmann’s equation
  11. Impulse noise treatment in magnetotelluric inversion
  12. Application of multivariate regression on magnetic data to determine further drilling site for iron exploration
  13. Comparative application of photogrammetry, handmapping and android smartphone for geotechnical mapping and slope stability analysis
  14. Geochemistry of the black rock series of lower Cambrian Qiongzhusi Formation, SW Yangtze Block, China: Reconstruction of sedimentary and tectonic environments
  15. The timing of Barleik Formation and its implication for the Devonian tectonic evolution of Western Junggar, NW China
  16. Risk assessment of geological disasters in Nyingchi, Tibet
  17. Effect of microbial combination with organic fertilizer on Elymus dahuricus
  18. An OGC web service geospatial data semantic similarity model for improving geospatial service discovery
  19. Subsurface structure investigation of the United Arab Emirates using gravity data
  20. Shallow geophysical and hydrological investigations to identify groundwater contamination in Wadi Bani Malik dam area Jeddah, Saudi Arabia
  21. Consideration of hyperspectral data in intraspecific variation (spectrotaxonomy) in Prosopis juliflora (Sw.) DC, Saudi Arabia
  22. Characteristics and evaluation of the Upper Paleozoic source rocks in the Southern North China Basin
  23. Geospatial assessment of wetland soils for rice production in Ajibode using geospatial techniques
  24. Input/output inconsistencies of daily evapotranspiration conducted empirically using remote sensing data in arid environments
  25. Geotechnical profiling of a surface mine waste dump using 2D Wenner–Schlumberger configuration
  26. Forest cover assessment using remote-sensing techniques in Crete Island, Greece
  27. Stability of an abandoned siderite mine: A case study in northern Spain
  28. Assessment of the SWAT model in simulating watersheds in arid regions: Case study of the Yarmouk River Basin (Jordan)
  29. The spatial distribution characteristics of Nb–Ta of mafic rocks in subduction zones
  30. Comparison of hydrological model ensemble forecasting based on multiple members and ensemble methods
  31. Extraction of fractional vegetation cover in arid desert area based on Chinese GF-6 satellite
  32. Detection and modeling of soil salinity variations in arid lands using remote sensing data
  33. Monitoring and simulating the distribution of phytoplankton in constructed wetlands based on SPOT 6 images
  34. Is there an equality in the spatial distribution of urban vitality: A case study of Wuhan in China
  35. Considering the geological significance in data preprocessing and improving the prediction accuracy of hot springs by deep learning
  36. Comparing LiDAR and SfM digital surface models for three land cover types
  37. East Asian monsoon during the past 10,000 years recorded by grain size of Yangtze River delta
  38. Influence of diagenetic features on petrophysical properties of fine-grained rocks of Oligocene strata in the Lower Indus Basin, Pakistan
  39. Impact of wall movements on the location of passive Earth thrust
  40. Ecological risk assessment of toxic metal pollution in the industrial zone on the northern slope of the East Tianshan Mountains in Xinjiang, NW China
  41. Seasonal color matching method of ornamental plants in urban landscape construction
  42. Influence of interbedded rock association and fracture characteristics on gas accumulation in the lower Silurian Shiniulan formation, Northern Guizhou Province
  43. Spatiotemporal variation in groundwater level within the Manas River Basin, Northwest China: Relative impacts of natural and human factors
  44. GIS and geographical analysis of the main harbors in the world
  45. Laboratory test and numerical simulation of composite geomembrane leakage in plain reservoir
  46. Structural deformation characteristics of the Lower Yangtze area in South China and its structural physical simulation experiments
  47. Analysis on vegetation cover changes and the driving factors in the mid-lower reaches of Hanjiang River Basin between 2001 and 2015
  48. Extraction of road boundary from MLS data using laser scanner ground trajectory
  49. Research on the improvement of single tree segmentation algorithm based on airborne LiDAR point cloud
  50. Research on the conservation and sustainable development strategies of modern historical heritage in the Dabie Mountains based on GIS
  51. Cenozoic paleostress field of tectonic evolution in Qaidam Basin, northern Tibet
  52. Sedimentary facies, stratigraphy, and depositional environments of the Ecca Group, Karoo Supergroup in the Eastern Cape Province of South Africa
  53. Water deep mapping from HJ-1B satellite data by a deep network model in the sea area of Pearl River Estuary, China
  54. Identifying the density of grassland fire points with kernel density estimation based on spatial distribution characteristics
  55. A machine learning-driven stochastic simulation of underground sulfide distribution with multiple constraints
  56. Origin of the low-medium temperature hot springs around Nanjing, China
  57. LCBRG: A lane-level road cluster mining algorithm with bidirectional region growing
  58. Constructing 3D geological models based on large-scale geological maps
  59. Crops planting structure and karst rocky desertification analysis by Sentinel-1 data
  60. Physical, geochemical, and clay mineralogical properties of unstable soil slopes in the Cameron Highlands
  61. Estimation of total groundwater reserves and delineation of weathered/fault zones for aquifer potential: A case study from the Federal District of Brazil
  62. Characteristic and paleoenvironment significance of microbially induced sedimentary structures (MISS) in terrestrial facies across P-T boundary in Western Henan Province, North China
  63. Experimental study on the behavior of MSE wall having full-height rigid facing and segmental panel-type wall facing
  64. Prediction of total landslide volume in watershed scale under rainfall events using a probability model
  65. Toward rainfall prediction by machine learning in Perfume River Basin, Thua Thien Hue Province, Vietnam
  66. A PLSR model to predict soil salinity using Sentinel-2 MSI data
  67. Compressive strength and thermal properties of sand–bentonite mixture
  68. Age of the lower Cambrian Vanadium deposit, East Guizhou, South China: Evidences from age of tuff and carbon isotope analysis along the Bagong section
  69. Identification and logging evaluation of poor reservoirs in X Oilfield
  70. Geothermal resource potential assessment of Erdaobaihe, Changbaishan volcanic field: Constraints from geophysics
  71. Geochemical and petrographic characteristics of sediments along the transboundary (Kenya–Tanzania) Umba River as indicators of provenance and weathering
  72. Production of a homogeneous seismic catalog based on machine learning for northeast Egypt
  73. Analysis of transport path and source distribution of winter air pollution in Shenyang
  74. Triaxial creep tests of glacitectonically disturbed stiff clay – structural, strength, and slope stability aspects
  75. Effect of groundwater fluctuation, construction, and retaining system on slope stability of Avas Hill in Hungary
  76. Spatial modeling of ground subsidence susceptibility along Al-Shamal train pathway in Saudi Arabia
  77. Pore throat characteristics of tight reservoirs by a combined mercury method: A case study of the member 2 of Xujiahe Formation in Yingshan gasfield, North Sichuan Basin
  78. Geochemistry of the mudrocks and sandstones from the Bredasdorp Basin, offshore South Africa: Implications for tectonic provenance and paleoweathering
  79. Apriori association rule and K-means clustering algorithms for interpretation of pre-event landslide areas and landslide inventory mapping
  80. Lithology classification of volcanic rocks based on conventional logging data of machine learning: A case study of the eastern depression of Liaohe oil field
  81. Sequence stratigraphy and coal accumulation model of the Taiyuan Formation in the Tashan Mine, Datong Basin, China
  82. Influence of thick soft superficial layers of seabed on ground motion and its treatment suggestions for site response analysis
  83. Monitoring the spatiotemporal dynamics of surface water body of the Xiaolangdi Reservoir using Landsat-5/7/8 imagery and Google Earth Engine
  84. Research on the traditional zoning, evolution, and integrated conservation of village cultural landscapes based on “production-living-ecology spaces” – A case study of villages in Meicheng, Guangdong, China
  85. A prediction method for water enrichment in aquifer based on GIS and coupled AHP–entropy model
  86. Earthflow reactivation assessment by multichannel analysis of surface waves and electrical resistivity tomography: A case study
  87. Geologic structures associated with gold mineralization in the Kirk Range area in Southern Malawi
  88. Research on the impact of expressway on its peripheral land use in Hunan Province, China
  89. Concentrations of heavy metals in PM2.5 and health risk assessment around Chinese New Year in Dalian, China
  90. Origin of carbonate cements in deep sandstone reservoirs and its significance for hydrocarbon indication: A case of Shahejie Formation in Dongying Sag
  91. Coupling the K-nearest neighbors and locally weighted linear regression with ensemble Kalman filter for data-driven data assimilation
  92. Multihazard susceptibility assessment: A case study – Municipality of Štrpce (Southern Serbia)
  93. A full-view scenario model for urban waterlogging response in a big data environment
  94. Elemental geochemistry of the Middle Jurassic shales in the northern Qaidam Basin, northwestern China: Constraints for tectonics and paleoclimate
  95. Geometric similarity of the twin collapsed glaciers in the west Tibet
  96. Improved gas sand facies classification and enhanced reservoir description based on calibrated rock physics modelling: A case study
  97. Utilization of dolerite waste powder for improving geotechnical parameters of compacted clay soil
  98. Geochemical characterization of the source rock intervals, Beni-Suef Basin, West Nile Valley, Egypt
  99. Satellite-based evaluation of temporal change in cultivated land in Southern Punjab (Multan region) through dynamics of vegetation and land surface temperature
  100. Ground motion of the Ms7.0 Jiuzhaigou earthquake
  101. Shale types and sedimentary environments of the Upper Ordovician Wufeng Formation-Member 1 of the Lower Silurian Longmaxi Formation in western Hubei Province, China
  102. An era of Sentinels in flood management: Potential of Sentinel-1, -2, and -3 satellites for effective flood management
  103. Water quality assessment and spatial–temporal variation analysis in Erhai lake, southwest China
  104. Dynamic analysis of particulate pollution in haze in Harbin city, Northeast China
  105. Comparison of statistical and analytical hierarchy process methods on flood susceptibility mapping: In a case study of the Lake Tana sub-basin in northwestern Ethiopia
  106. Performance comparison of the wavenumber and spatial domain techniques for mapping basement reliefs from gravity data
  107. Spatiotemporal evolution of ecological environment quality in arid areas based on the remote sensing ecological distance index: A case study of Yuyang district in Yulin city, China
  108. Petrogenesis and tectonic significance of the Mengjiaping beschtauite in the southern Taihang mountains
  109. Review Articles
  110. The significance of scanning electron microscopy (SEM) analysis on the microstructure of improved clay: An overview
  111. A review of some nonexplosive alternative methods to conventional rock blasting
  112. Retrieval of digital elevation models from Sentinel-1 radar data – open applications, techniques, and limitations
  113. A review of genetic classification and characteristics of soil cracks
  114. Potential CO2 forcing and Asian summer monsoon precipitation trends during the last 2,000 years
  115. Erratum
  116. Erratum to “Calibration of the depth invariant algorithm to monitor the tidal action of Rabigh City at the Red Sea Coast, Saudi Arabia”
  117. Rapid Communication
  118. Individual tree detection using UAV-lidar and UAV-SfM data: A tutorial for beginners
  119. Technical Note
  120. Construction and application of the 3D geo-hazard monitoring and early warning platform
  121. Enhancing the success of new dams implantation under semi-arid climate, based on a multicriteria analysis approach: Case of Marrakech region (Central Morocco)
  122. TRANSFORMATION OF TRADITIONAL CULTURAL LANDSCAPES - Koper 2019
  123. The “changing actor” and the transformation of landscapes
Sign up now to receive a 20% welcome discount
Institutional Access
How does access work?
Have an idea on how to improve our website?
Please write us.
© 2025 De Gruyter Brill
Downloaded on 8.9.2025 from https://www.degruyterbrill.com/document/doi/10.1515/geo-2020-0241/html
Scroll to top button