Land desertification is a widely concerned ecological environment problem. Studying the evolution trend of desertification types is of great significance to prevent and control land desertification. In this study, we applied the decision tree classification method, to study the land area and temporal and spatial change law of different types of desertification in the North Bank of Qinghai Lake area from 1987 to 2014, based on the current land use situation and TM remote sensing image data of Haiyan County, Qinghai Province, The results show that the area of mild desertification land and moderate desertification land in the study area has decreased, while the area of severe desertification land and extreme desertification land has increased significantly in the past 30 years. The area of desertification land decreased by 4.02 km2, of which the area of mild and moderate desertification land decreased by 39.73 km2 and 36.8 km2 respectively, and the area of severe and extreme desertification land increased by 32.78 km2 and 39.73 km2 respectively. As for the mutual transformation relationship, the transformation from severe desertification land to extreme desertification land is the main, and the junction of severe desertification land and extreme desertification land is the sensitive area of transformation. In the north shore of Qinghai Lake, the sandy land tends to expand eastward. The research provides reference basis for local land desertification monitoring, and has a great guidance for local effective land desertification and soil and water conservation.
Desertification is one of the serious ecological, social, and economic problems currently faced by arid and semi-arid regions worldwide [
Since the 1990s, with the development of “3S” technology, regional desertification research has been further developed [
At present, the research on desertification in Qinghai Lake area focuses on the analysis of the spatial distribution pattern, the interdecadal changes of desertified land, and the evaluation of the stability and sensitivity of the desertification ecosystem [
The remainder of this paper is organized as follows. Section 2 describes the related data sources and processing method. Next, Section 3 demonstrates and discusses the research results in detail, and finally, Section 4 presents the conclusions.
The study area is located in the northeast of Qinghai Tibet Plateau, Haiyan County, Qinghai Province, China (longitude 99°10′E–101°10′E, latitude 36°18′N–37°30′N, and altitude −3194–5174 m above MSL). The study area had a plateau continental climate, with an annual mean temperature of −3.4–6.3°C, annual sunshine hours of 2430–3330 h, and an annual evaporation with 1300–2400 mm. The annual average wind speed is 3.2–4.4 m/s, the gale days are 10.8–13.2 d, and the dust days are 10.8–13.2 d. The annual precipitation in the study area is 200–445 mm. Most of the rainfall is concentrated in May to September, and the rainfall is hot in the same period.
The data used in this paper are mainly remote sensing image data, including Landsat 5 TM image data and Landsat 8 OLI and TIRS image data, with the time span from 1987 to 2014. To better distinguish different types of surface features and monitor the vegetation coverage in the area, the selected image data are mainly concentrated from June to September, because during this period, the snow melts, the vegetation grows well, and the thermal radiation of various surface features significantly differ. The background information and specific parameters of remote sensing image are presented in
Serial number | Type of data | Imaging time | Track number | Number of bands | Spatial resolution (m) |
---|---|---|---|---|---|
1 | Landsat5 TM | 1987.08 | p133r34 | 7 | 30.0 |
2 | Landsat5 TM | 1989.08 | p133r34 | 7 | 30.0 |
3 | Landsat5 TM | 1990.06 | p133r34 | 7 | 30.0 |
4 | Landsat5 TM | 1993.08 | p133r34 | 7 | 28.5 |
5 | Landsat5 TM | 1995.08 | p133r34 | 7 | 30.0 |
6 | Landsat5 TM | 1996.06 | p133r34 | 7 | 28.5 |
7 | Landsat5 TM | 1997.08 | p133r34 | 7 | 28.5 |
8 | Landsat5 TM | 1998.07 | p133r34 | 7 | 28.5 |
9 | Landsat5 TM | 1999.07 | p133r34 | 7 | 28.5 |
10 | Landsat5 TM | 2000.08 | p133r34 | 7 | 30.0 |
11 | Landsat5 TM | 2001.07 | p133r34 | 7 | 30.0 |
12 | Landsat5 TM | 2002.07 | p133r34 | 7 | 30.0 |
13 | Landsat5 TM | 2003.09 | p133r34 | 7 | 30.0 |
14 | Landsat5 TM | 2004.09 | p133r34 | 7 | 30.0 |
15 | Landsat5 TM | 2005.09 | p133r34 | 7 | 30.0 |
16 | Landsat5 TM | 2006.08 | p133r34 | 7 | 30.0 |
17 | Landsat5 TM | 2007.08 | p133r34 | 7 | 30.0 |
18 | Landsat5 TM | 2008.07 | p133r34 | 7 | 28.5 |
19 | Landsat5 TM | 2009.08 | p133r34 | 7 | 30.0 |
20 | Landsat5 TM | 2010.07 | p133r34 | 7 | 30.0 |
21 | Landsat5 TM | 2011.06 | p133r34 | 7 | 30.0 |
22 | Landsat8 OLI | 2013.09 | p133r34 | 11 | 30.0 and 100.0 |
23 | Landsat8 OLI | 2014.07 | p133r34 | 11 | 30.0 and 100.0 |
During remote sensing imaging, various systematic and random errors will occur. Before the analysis and information extraction of remote sensing image, the remote sensing image must be preprocessed, including radiometric calibration, geometric correction, and atmospheric correction. In this study, the gain and bias parameters given by USGS are used for image radiometric correction, which was calculated by the following equation:
In geometric correction, taking TM L4 products as the benchmark, quadratic polynomial and nearest neighbor resampling methods are used for geometric correction of the image. In this study, 55 geometric correction points are selected, all of which are evenly distributed on the image, and the error is controlled within one pixel.
In this study, the environment for visualizing images software with fast line-of-sight atmospheric analysis of spectral hypercubes module for atmospheric correction is used. The relevant parameters can be obtained from the MTL file attached to the image.
NDVI is a comprehensive reflection of vegetation type, growth status, and cover morphology in a unit pixel. When the pixel is completely covered by vegetation, the vegetation index is 1; when no vegetation cover exists, the NDVI value is between −1 and 0, such as water, desert, and so on; If the vegetation fails to cover the whole pixel completely, the NDVI value is between 0 and 1, and this kind of pixel is called mixed pixel. NDVI was given as:
In this study, the land cover types are divided into the following seven categories: water body, severe desertification land, severe desertification land, moderate desertification land, mild desertification land, non-desertification land, and others (including cloud and snow). NDVI can well distinguish vegetation area and non-vegetation area. Vegetation coverage can well distinguish non desertification land, mild desertification land, moderate desertification land, and severe desertification land, whereas surface temperature can well distinguish water body and severe desertification land, albedo can separate cloud and snow from non-vegetation area, soil moisture and modified soil adjusted vegetation index can further distinguish the desert, water, cloud, and snow in the areas with no vegetation or less vegetation. On the basis of the above characteristics, the decision tree is constructed, and the specific process is shown in
On the basis of the interpretation of the desertification results, the desertification vector diagrams of 1990 and 2000, 2000 and 2009, and 2009 and 2014 were superimposed and calculated to obtain the years 1990 to 2000, 2000 to 2009, and the transfer matrix of various types of desertified land in the three periods from 2009 to 2014 analyzes the evolution mechanism of different types of desertified land in two different time dimensions: interdecadal and interannual.
1990–2000 | No desertification (km2) | Light desertification (km2) | Moderate desertification (km2) | Severe desertification (km2) | Extreme desertification (km2) | Total (km2) |
---|---|---|---|---|---|---|
Non-desertification | 302.90 | 55.37 | 4.62 | 0.19 | 0.02 | 363.10 |
Light |
0.53 | 5.76 | 65.55 | 24.04 | 0.57 | 96.45 |
Moderate desertification | 43.11 | 166.07 | 46.92 | 0.56 | 0.05 | 256.71 |
Severe desertification | 2.71 | 68.98 | 222.16 | 7.61 | 0.21 | 301.67 |
Extreme desertification | 0.00 | 0.23 | 5.85 | 33.87 | 169.07 | 209.02 |
Total | 349.26 | 296.41 | 345.10 | 66.27 | 169.91 | 1226.95 |
2000–2009 | No desertification (km2) | Light desertification (km2) | Moderate desertification (km2) | Severe desertification (km2) | Extreme desertification (km2) | Total (km2) |
---|---|---|---|---|---|---|
Non-desertification | 333.02 | 68.63 | 4.15 | 1.06 | 0.02 | 406.90 |
Light |
24.82 | 143.27 | 44.61 | 3.03 | 0.44 | 216.16 |
Moderate desertification | 3.16 | 42.32 | 210.54 | 30.47 | 1.40 | 287.88 |
Severe desertification | 0.89 | 1.47 | 37.98 | 46.09 | 6.47 | 92.89 |
Extreme desertification | 0.02 | 0.12 | 2.52 | 13.75 | 187.78 | 204.19 |
Total | 361.91 | 255.80 | 299.80 | 94.39 | 196.11 | 1208.02 |
2009–2014 | No desertification (km2) | Light desertification (km2) | Moderate desertification (km2) | Severe desertification (km2) | Extreme desertification (km2) | Total (km2) |
---|---|---|---|---|---|---|
Non-desertification | 362.86 | 33.75 | 2.54 | 0.90 | 0.01 | 400.06 |
Light |
32.20 | 140.79 | 46.52 | 1.16 | 0.11 | 220.78 |
Moderate desertification | 5.93 | 34.20 | 180.46 | 23.29 | 1.22 | 245.09 |
Severe desertification | 2.38 | 4.80 | 51.50 | 49.30 | 7.65 | 115.62 |
Extreme desertification | 0.04 | 0.22 | 3.67 | 17.56 | 184.17 | 205.66 |
Total | 403.41 | 213.75 | 284.69 | 92.20 | 193.16 | 1187.21 |
On the basis of the interpretation results, remote sensing software was used to analyze the spatial changes of different types of desertified land from 1987 to 2014. The results are shown in
Taking the north shore of Qinghai Lake District as the research object, this paper interprets the TM remote sensing images from 1987 to 2014 by using remote sensing technology and referring to the land use status map of Haiyan County in Qinghai Province. On the basis of the decision tree classification method and the interpretation of desertification information, the land area and temporal and spatial transformation rules of the different types of desertification in the study area in recent 30 years were obtained. Based on the research findings, in the past 30 years, the area of desertified land in the study area has decreased by 4.02 km2, of which the area of lightly desertified land has decreased by 39.73 km2, the area of moderately desertified land has decreased by 36.8 km2, and the area of severely desertified land has increased by 32.78 km2. The land area has increased by 39.73 km2, and the overall characteristic is that lightly and moderately desertified lands have a reversal trend, whereas severely desertified land and severely desertified land have a significant increasing trend. In addition, from the perspective of mutual transformation relationship, the desertified land is mainly transformed into severely desertified land. The junction of the severely desertified land and severely desertified land is a sensitive area for transformation. Third, the desertified land in the study area has an eastward development and expansion trend. Finally, this paper only analyzed the spatial-temporal evolution mechanism of desertification from 1987 to 2014 using TM images. The spatial-temporal evolution mechanism of desertification from 2015 to 2020 and the driving mechanism of the spatial-temporal evolution of desertified land in recent 35 years in this region will be further studied.
This research was supported by the National Nature & Science Foundation of China “Study on the dynamic mechanism of grassland ecosystem response to climate change in Qinghai Plateau”(No. U20A2098), the second Tibetan plateau scientific expedition and research program (STEP, No. 2019QZKK0804) and China Huaneng Group Co. Science and Technology Program “The research development and implement on the evaluation technology of wind resource” (No. HNKJ18-H31).