Environ Earth Sci DOI 10.1007/s12665-014-3818-0
ORIGINAL ARTICLE
Assessment of terrain susceptibility to thermokarst lake development along the Qinghai–Tibet engineering corridor, China Fujun Niu • Zhanju Lin • Jiahao Lu Jing Luo • Huini Wang
•
Received: 7 October 2013 / Accepted: 19 October 2014 Ó Springer-Verlag Berlin Heidelberg 2014
Abstract Thermokarst lakes have been developing recently along the Qinghai–Tibet engineering corridor in association with increased human activity and persistent climatic warming. Based on field observations, we assessed the susceptibility of terrain to the development of thermokarst lakes between the Chumaerhe River and Fenghuoshan mountain pass. A susceptibility map of the region was created in a geographic information system by assessing seven controlling factors, ranked using the analytic hierarchy process. The resulting susceptibility values ranged between 0.1 and 0.66. These susceptibility values were divided into four classes (high, moderate, low, and lowest) according to the mutagenesis point method. Areas with values between 0.1 and 0.16 were considered to have the ‘lowest’ susceptibility, while those between 0.26 and 0.66 were considered to have ‘high’ susceptibility. Using SPOT-5 satellite data, we determined that the high-susceptibility areas contained approximately 91 % of the total thermokarst lake area in the study region, and 88 % of the number of lakes. The moderate, low, and lowest level areas accounted for about 52 % of the study region, but only
F. Niu Z. Lin (&) J. Luo State Key Laboratory of Frozen Soil Engineering, Cold and Arid Regions Environmental and Engineering Research Institute, Chinese Academy of Sciences, Lanzhou 730000, China e-mail:
[email protected] F. Niu e-mail:
[email protected] J. Lu Chongqing Institute of Geology and Mineral Resources, Chongqing 400042, China H. Wang Hubei Geomatics Information Center, Wuhan 430074, China
contained 9 % of the total lake area and 12 % of the lakes. Finally, relations between the area of the thermokarst lakes and the main controlling factors, e.g., ground ice content, ground temperature, vegetation type, and altitude were discussed. Keywords Thermokarst lakes Permafrost Susceptibility assessment Qinghai–Tibet engineering corridor
Introduction Permafrost along the Qinghai–Tibet engineering corridor (QTEC) has undergone rapid degradation due to recent human activity and persistent climatic warming on the Qinghai–Tibet plateau (QTP) (Harry and French 1983; French 1996; Wu et al. 2005; Jin et al. 2008a). Ground temperatures and active-layer thicknesses have been increasing, resulting in the thaw of ice-rich permafrost (Wu and Zhang 2008; Lin et al. 2011a). These changes have led to significant ground subsidence and the widespread development of thermokarst features (Niu et al. 2008; Lin et al. 2010). In particular, numerous thermokarst lakes have formed on the QTP (Lin et al. 2010, 2011a; Lin 2011; Niu et al. 2011) and the areal coverage of the water bodies has been increasing (Niu et al. 2008; Lin et al. 2011b). Thermokarst lakes are a major heat source to the surrounding ground (Ling and Zhang 2004). Generally, the mean annual temperature at the lake bottom is greater than 0 °C, except in very shallow lakes (Niu et al. 2011). Therefore, lakes supply heat to the adjacent ground, which warms the surrounding permafrost (Lin et al. 2010, 2011b) and promotes talik formation, increases lake depth, and in some cases completely eradicates underlying permafrost
123
Environ Earth Sci
(b)
Chumaerhe Riverside
Hoh Xil Hill Region
Beiluhe Basin
Fenghuoshan Mountain 5000
Fenghuoshan
Qinghai –Tibet Engineering Corridor (QTEC)
Elevation/m
Chumaerhe
Wudaoliang
Beiluhe
4800
4600
(a)
(c) 4400 0
20
40
60
80
100
120
Distance/km
Fig. 1 The study area on the Qinghai–Tibetan plateau. a The Qinghai–Tibetan engineering corridor (QTEC) and permafrost distribution; b the study area, from the Chumaerhe riverside to the Fenghuoshan mountain. The scene is from SPOT-5 imagery; c the
elevation profile. Permafrost underlies 75 % of the total area of the plateau, ground temperatures are relatively warm, and the permafrost is commonly ice rich. The study area is within the continuous permafrost zone
(Williams and Smith 1989). Talik development may present a significant hazard to infrastructure and change ecosystem dynamics by altering chemical, biological, and physical conditions around thermokarst lakes (Johnston and Brown 1964; Sellmann et al. 1975; Lunardini 1996; Moiseenko et al. 2006). The development of thermokarst features is associated with geo-environmental factors including topography, ground temperature, vegetation cover, surficial deposits, and ground ice content. It is generally believed that changes in local conditions may initiate, delay, or counteract thermokarst activity, but under similar conditions of surface disturbance and climatic warming, regions with high vegetation cover, low ground ice content, and low ground temperature are not as susceptible to the initiation and development of thermokarst lakes as regions with sparse vegetation underlain by warm, ice-rich permafrost. Based on previous studies of thermokarst lakes on the QTP (Lin et al. 2010, 2011a; Lin 2011; Niu et al. 2011), the
purposes of this paper are to: (1) develop a system to assess terrain susceptibility to thermokarst lake development, (2) assess the susceptibility level of terrain to thermokarst lake development along QTEC, (3) map the existing thermokarst lake distribution, and (4) validate the susceptibility assessment by comparing the resulting susceptibility zones to the actual lake distribution.
123
Study area description The study area was 110 km long and lay between the Chumaerhe riverside and Fenghuoshan mountain pass, and extended 5 km in width on either side of the Qinghai–Tibet railway (QTR) (Fig. 1a, b). The region is underlain by continuous permafrost. The study area comprises a major portion of the Hoh Xil nature reserve region, which includes the Chumaerhe high plain, the Hoh Xil hill region, Beiluhe basin, and the Fenghuoshan mountain region.
Environ Earth Sci
Approximately, 90 % of the region lies above 4,500 m in elevation (Fig. 1c). Several infrastructure projects, including the Qinghai–Tibet highway (QTH), the QTR, the electronic transfer project, and the gas pipe line are located within this corridor. The region has a typical high altitude and continental climate. The mean annual air temperature (MAAT) is typically below -4 °C. Air temperatures reach a minimum of -30 °C in winter, and a maximum of approximately 25 °C in summer. The highest mean monthly air temperature recorded was 9.2 °C in July and the lowest was -16.9 °C in January (Niu et al. 2011). The annual precipitation is concentrated between May and August and ranges from 50 mm on the plateau to 400 mm in the mountain ranges. The loose alpine steppe and alpine meadow soils are too dry to support abundant vegetation growth and the development of a protective surface organic layer and root system. As a result, the soils are susceptible to erosion, which results in moisture loss and desertification (Li et al. 1996). No trees grow along the QTR and there is only limited shrub growth. The cold climate, short growing season, and poor soils support simple, uniform, and low-lying (10–15 cm) grassland vegetation communities with low overall biomass. The grasslands and meadows are sometimes sparse and interspersed with bare sandy areas (Li et al. 1996). The sensitive permafrost conditions in the study area are controlled by periglacial processes, regional geography, and microclimate. The depth of the active layer in the study area is between 1.5 and 3 m. The mean annual ground temperature (MAGT) in high plains and valleys is above -1.5 °C, and the permafrost thickness is less than 70 m. In the mountains, the MAGT is less than -1.5 °C, and permafrost thickness exceeds 130 m (Zhou et al. 2000). The warm permafrost in the region commonly contains massive ground ice. Permafrost with volumetric ice contents exceeding 20 % extends 102 km along the corridor, while the section with MAGTs above -1.0 °C (warm permafrost) is 58 km long (Liu et al. 2000; Wu et al. 2002, 2004). The geothermal gradient is 1.5–4 °C/100 m. The ground temperatures are lowest in the middle of February and reach their peak value at the end of August. The lowest ground temperature at 1 m depth was about -11.9 °C at the Fenghuoshan mountain pass, and the average maximum temperature was 3.3 °C at the Beiluhe site.
2009). First, based on our field investigations, factors contributing to the initiation and development of thermokarst lakes were chosen as the assessment indices, and their weights were determined using the analytic hierarchy process (AHP), and our field observations and expertise. Next, a comprehensive assessment system and a model were used to calculate the susceptibility index. Finally, a susceptibility map was produced using the Grid Computing Module of ArcGIS (Li 2008). Assessment model Some models may be used to assess the probability of geological hazards, such as the logistic regression model, comprehensive evaluation model, fuzzy comprehensive evaluation method, and artificial neural network (Liu et al. 1999; Pistocchi et al. 2002; Qiu et al. 2003; Lee et al. 2004; Wang et al. 2009). In this study, a comprehensive evaluation model (CEM) was applied due to the complexity of permafrost conditions. The model is described as follows: n X B¼ bi wi ð1Þ i¼1
where B is the index of the susceptibility to thermokarst lake development, b the assessment factor, w the factor weight, and the subscript i the number of factors. Assessment system
Methods
This assessment of terrain susceptibility to thermokarst lake development mainly focuses on the number of thermokarst lakes that could potentially develop and their areal extent. The factors controlling lake development in the region have been observed to include geological and environmental conditions such as MAGT, ground ice content, surficial deposits, vegetation cover, and slope curvature. According to the distribution of thermokarst lakes and regional conditions in the study region, the geologic and environmental factors considered were the MAGT, the volumetric ice content of permafrost, the quaternary surficial deposits, vegetation cover, and slope curvature; the point density and surface density of existing thermokarst lakes were also included as additional factors. The active-layer thickness is also an important factor for the evaluation of thermokarst lake susceptibility, but associated with the MAGT. Because this factor is not independent, active-layer thickness was excluded in the assessment system.
Assessment process
Weights and assignment of factors
In this paper, the engineering geology analogy (EGA) was adopted to determine the terrain susceptibility to thermokarst lake development in the study region (Yu and Lu
Some methods, such as the eigenvalue method, entropy method, and AHP (Ye 2006), can be used to determine factor weights in assessing geological hazard susceptibility.
123
Environ Earth Sci Table 1 Assessment matrix of all controlling factors
Table 2 The weight coefficients of each factor
Factors
PD
SD
ICP
MAGT
VC
QSD
SC
Factors
PD
SD
ICP
MAGT
VC
QSD
SC
PD
1
2
3
5
7
7
9
0.24
0.19
0.11
0.04
0.04
0.02
1/2
1
2
4
6
6
9
Weight coefficient
0.36
SD ICP
1/3
1/2
1
4
5
5
9
MAGT
1/5
1/4
1/4
1
6
6
8
VC
1/7
1/6
1/5
1/6
1
1
2
QSD
1/7
1/6
1/5
1/6
1
1
3
SC
1/9
1/9
1/9
1/9
1/2
1/3
1
However, due to the complicated interactions between controlling factors in permafrost regions, it is difficult to determine the factors’ weights accurately using a mathematical model. The AHP is useful for simplifying complex problems (Saaty 1980; Li and Xu 1998; Wang and Yi 2009) and is therefore adopted in this study. The assessment of terrain susceptibility to thermokarst lake development in this study was conducted using a decisionmaking process that incorporates multiple controlling factors. The weight of each factor was determined using the AHP, and the specific calculation steps were: 1.
Determine the relevant controlling factors of the system and include in a set function X (Eq. 2). X ¼ fX1 ; X2 ; X3 ; X4 ; X5 ; X6 ; X7 g:
2.
Create a matrix ‘R’ (Eq. 3) for pairwise comparisons of each factor based on their assigned relative importance (see Table 1). In this matrix, the element Rij = 1/ aji, when i = j, Rij = 1. The values in this matrix may vary from 1 to 9, with 1 indicating equal importance between Xi and Xj and 9 indicating that Xi is much more important than Xj. The eigenvector of the largest eigenvalue in the matrix was obtained and the normalized eigenvector was used as the weight of each factor. 8 R11 > > < R21 R¼ ... > > : R71
3.
ð2Þ
R12 R22 ... R72
9 . . . R17 > > = . . . R27 : ... ... > > ; . . . R77
ð3Þ
The largest eigen value e¨max of a reciprocal matrix R is always greater than or equal to m (number of rows or columns). The more consistent the comparisons are, the closer the value of computed e¨max is to m. A consistency index (CI) can be calculated from Eq. (4): CI ¼ ðkmax mÞ ðm 1Þ:
ð4Þ
As the CI is dependent on m, a consistency ratio (CR), which is independent of m, was calculated using Eq. (5):
123
The weight coefficients are the normalized eigenvectors, which correspond to the maximum eigenvalues of the estimated matrix R and are calculated by the root method. These factors are sequenced from high to low according to the weight coefficients, and then distinguished as the main factors (higher), secondary factors (middle), and general factors (lower)
CR ¼ CI RI;
ð5Þ
where RI is the average consistency index of randomly generated comparisons and is obtained according to the size of the matrix. In general, a CR value of 10 % or less is considered to be acceptable. If it is greater, some or all of the pairwise comparisons must be repeated to resolve the inconsistencies. The weight coefficient of each factor is determined in Table 2. Terrain susceptibility to development of thermokarst lakes Gentle slope, high MAGT, exposed mineral soils, ice-rich permafrost, and the presence of scars from past thermokarst processes may all contribute to the development of new thermokarst lakes. Each of these factors is therefore taken into consideration in the assessment of terrain susceptibility to the development of thermokarst lakes. The formula used to calculate the susceptibility index of thermokarst lake occurrence (SIOTL) is expressed as (Olmacher and Davis 2003): SIOTL ¼ Sw Sr þ Iw Ir þ Qw Qr þ Mw Mr þ Vw Vr þ PDw PDr þ SDw SDr ;
ð6Þ
where S is the slope curvature, I the ice content of permafrost, Q the quaternary surface deposit, M the MAGT, V the vegetation cover percentage, PD the point density of thermokarst lakes, and SD the surface density of lakes. The subscript r, used for each factor, corresponds to the rankings given to the different factor ranges, zones, or units, while the subscript w corresponds to the weight of each factor used for calculating the SIOTL index.
Controlling factors and ranking SPOT-5 satellite imagery of the study area, acquired in August 2010, was used in this study. The images have a 2.5 m spatial resolution. The images were geometrically corrected and interpreted to obtain information about the factors that control the terrain susceptibility to thermokarst
Environ Earth Sci Fig. 2 Maps showing a the normalized point density of thermokarst lakes; b normalized lake surface area density
lake development. Quantitative factors such as slope curvature and MAGT were obtained either by spatial analysis of the original data or by interpolation fitting. Qualitative factors, such as the ice content of permafrost, quaternary surface deposit, and vegetation cover, were obtained by creating a grading standard and then quantified based on their contribution to the grading standard. Point density (PD) and surface density (SD) Areas in which thermokarst lakes have already developed are highly susceptible to the development of additional lakes (Cui 2004). Therefore, the point density and surface density of lakes are important factors that influence the terrain susceptibility to thermokarst lake development. PD is the number of lakes per km2 in the study area, while SD is the lake area per km2. Therefore, PD is a quantification of the frequency of lakes, while SD is an indication of the areal coverage. Their formulae are as follows (Liu and Chen 2011): Lakenum Lakearea SD ¼ Unitarea Unitarea Dataunit Datasunit ND ¼ ; Datalunit Datasunit PD ¼
where Lakenum and Lakearea indicate the number and surface area of thermokarst lakes, respectively; Unitarea is the area of the unit grid; ND is normalized data; and the Dataunit, Datas-unit, Datal-unit refers to data of the unit grid, the smallest unit, and the largest unit, respectively. Using these equations, the normalized PD and SD map layers were calculated (Fig. 2). The regions with high PD and SD
values are generally more susceptible to thermokarst lake development. Rankings from 0 to 1 were given to different regions with normalized PD and SD values from 0 to 1. Ice content of permafrost (ICP) The thaw of ice-rich permafrost is necessary for the development of thermokarst depressions and lakes. Therefore, the ice content of permafrost is a critical factor in their initiation and development. Permafrost in this study was divided into three categories based on the ice content. Ice-poor permafrost had visible ice content less than 20 %, ice-rich permafrost had 20–50 % visible ice content, and massive ground ice had more than 50 % visible ice (Niu et al. 2002). In addition, seasonally frozen ground was also considered as an additional substrate type and contained no ice. This classification has been widely adopted in permafrost engineering and geological investigations along the QTR (China Railway First Survey and Design Institute Group Ltd. 2000; Niu et al. 2002). We refer to ice-poor permafrost as low ice content permafrost, and the ice-rich permafrost and massive ground ice as high ice content permafrost. The ice content of permafrost was determined along the QTR by drilling near the railway. The drilling results were incorporated into a geological map. We extrapolated the ice contents from the geological map to the rest of the study region for different ground surface conditions. When surface conditions or slope angles changed abruptly, the permafrost was sampled directly, either by excavating test pits or drilling boreholes to validate the ice contents. The ice content map for the study area was produced from this
123
Environ Earth Sci
Fig. 3 Normalized map layers of the ice content in permafrost (ICP). ICP was classified into four ranked zones: ice-poor permafrost, icerich permafrost, massive ground ice, and seasonally frozen ground (thawed)
extrapolation and validation process. The normalized ICP is shown in Fig. 3, based on permafrost sampling from deep and shallow drilling, and a range of geophysical and geotechnical surveys. Normalized rankings from 0 to 1 were assigned to areas with low to high ice contents in the first few meters of permafrost, which are most susceptible to thaw following disturbance and changes in climate. Mean annual ground temperature (MAGT) MAGT is another important factor controlling thermokarst lake development in permafrost regions and is closely related to the local environmental conditions. MAGTs from 2000 to 2012 from 29 boreholes along the QTR were used to map the MAGT in the study area. Correlation analysis was conducted between MAGT and several factors, including longitude, latitude, elevation, slope gradient, slope aspect, and vegetation cover. The analysis results showed that the MAGT had a significant negative correlation with latitude, elevation, and normalized slope aspect, with correlation coefficients of -0.45, -0.36, and -0.28, respectively. Therefore, a model was developed using the relations between the MAGT (T in the formula) and the elevation (H), latitude (u), and normalized slope aspects (h) based on multivariate regression analysis (Cheng 1984; Li and Cheng 1999; Wu et al. 2000):
123
Fig. 4 MAGT in the study region. The permafrost on the QTP was also classified into four types: extremely unstable high-temperature permafrost, unstable high-temperature permafrost, relatively stable lowtemperature permafrost, and stable low-temperature permafrost. The first two types ([-1.0 °C) are referred to as high-temperature (warm) permafrost, and the other two (\-1.0 °C) as low-temperature permafrost
T ¼ 65:461 1:222u 0:005H 0:299 cos h: The correlation coefficient (R) of the model was 0.936, which indicates a strong linear relation between MAGT and the three factors. Under the Grid Computing Module of ArcGIS, a simulated map layer of the MAGT was obtained using the regression equation. We conducted error analysis on the simulated data and corrected MAGT using the inverse distance weighted (IDW) method. The resulting MAGT map layer is shown in Fig. 4. Warming permafrost that has MAGT near 0 °C thaws more easily than cold permafrost, and slight disturbances or temperature increases may result in permafrost degradation and thermokarst initiation. Therefore, rankings from 0 to 1 were given to areas with MAGTs from -3.2654 to 0 °C, and a ranking of 0 was given to MAGTs over 0 °C. Vegetation cover (VC) Vegetation cover may be related to permafrost conditions and contribute to thermal stability. Significant vegetation cover helps to protect permafrost from degradation (Jin et al. 2008b). However, if the vegetation is damaged or removed, the permafrost may degrade rapidly. Therefore, the vegetation cover is also an important factor that influences the initiation and development of thermokarst lakes. In the study
Environ Earth Sci
Fig. 5 Normalized NDVI in the study region. It is an important index derived from remotely sensed imagery that indicates the vegetation growth and cover. Areas where the NDVI ranges from 0.2 to 0.8 are considered to be well vegetated. NDVI values were between -0.3786 and 0.3943, highlighting the sparse vegetation cover that characterizes the study region
region, the vegetation is typically sparse, consisting of lowlying species. Based on the field investigation in September 2011 and our interpretation of the remotely sensed images (SPOT-5), we calculated the normalized difference vegetation index (NDVI) shown in Fig. 5. Higher NDVI values indicate greater vegetation cover. In our study area, thermokarst lakes develop more in areas with low vegetation cover. Therefore, rankings from 0 to 1 were given to areas with NDVI values from 0.39 to 0. As NDVI values below 0 generally indicate snow, cloud, or a water body, a ranking of 0 was given to areas with NDVI values from -0.38 to 0. Quaternary surface deposit (QSD) Different surface deposits have different capacities for water absorption, retention, and permeability (Jiang et al. 2007), and different thermal properties. Therefore, they are well related with the initiation and development of thermokarst lakes. Ten different types of deposits were analyzed including the bedrock. A ranking from 0 to 1 was assigned to each deposit type according to the permeability coefficient (Fig. 6). Based on our field investigations, lower rankings (coarse-grained soil) indicate that the substrate is less susceptible to thermokarst processes (e.g., sand or gravel has a ranking of 0). Higher rankings (fine-
Fig. 6 Normalized map layer of quaternary surface deposits. Quaternary units were assigned ranks (1–10) based on their influence on thermokarst processes
grained soil) indicate substrates that are associated with thermokarst development (e.g., clay has a ranking of 1). Surface curvature (SC) The majority of thermokarst scars in the study region are in flat areas with slopes that are near 0°. Thermokarst lakes were rarely observed on slopes greater than 10° during our field investigations along the QTEC. This is because plains or basins are favorable to the accumulation of melt water, precipitation, and runoff, while the steeper slopes do not allow water to accumulate. As a result, there is little risk of thermokarst lake development on slopes. In this study, slope percentage was calculated from the 2010 digital elevation model (DEM), and the resulting surface curvature map is shown in Fig. 7. Surface curvature was divided into three classes. A ranking of 0 was assigned to areas with surface curvature over 0 (convex), and for areas with surface curvature B0 (concave) rankings from 0 to 1 were assigned based on the magnitude of the curvature. Results and discussion Zoning and assessment The integrated SIOTL values obtained from the weighted sum of the different factors ranged from 0.1 to 0.66. The
123
Environ Earth Sci
curve of susceptibility values shows three distinct mutations at 0.16, 0.19, and 0.26 (Fig. 8). As a result, these mutation values are viewed as boundaries of probability levels (Liu et al. 2004). Based on the mutation point method (Liu et al. 1999, 2004), the results were divided into four levels of susceptibility to thermokarst lake development: high, moderate, low, and lowest. The integrated map of the terrain susceptibility to thermokarst lake development is shown in Fig. 9. Values between 0.1 and 0.16 indicate that there is little to no chance of thermokarst lake development (lowest level). Values between 0.16 and
0.19 indicate a low level of susceptibility, while those between 0.19 and 0.26 indicate a moderate level. Finally, values between 0.26 and 0.66 indicate a high susceptibility to lake development. The results indicate that highly susceptible areas are located predominantly in the Chumaerhe high plateau, Wudaoliang basin and Beiluhe beach. These areas are flat, have high MAGTs, and are underlain by ice-rich permafrost or massive ground ice. The surface in these regions is sparsely vegetated or bare, and near-surface sediments are fine-grained to gravelly. Summaries of environmental conditions in each region are shown in Table 3. In general, areas with moderate susceptibility to thermokarst lake development are areas where permafrost poses problems when it is thawed, however, it is impossible that the massive ground ice exists underlying the surface. Lastly, the low and lowest susceptibility areas are found on rocky or steep slopes lacking ground ice. Analyzing relations between controlling factors and probability
Fig. 8 The curve of SIOTL values in the study area. Three well-defined mutations are considered as division values for different levels of susceptibility to thermokarst lake development
Ration of area
Fig. 7 Map of surface curvature (SC) in the study region. SC was classified into three types: concave (SC \ 0), flat (SC = 0), and convex (SC [ 0)
Areas classified as highly susceptible commonly had high MAGTs, ranging between -1.5 and -0.5 °C (Table 4). High MAGTs facilitate the initiation and development of thermokarst lakes, because increases in ground temperature lead directly to permafrost degradation, whereas in regions underlain by cold permafrost, temperature perturbations result in the warming of permafrost rather than immediate degradation. Therefore, areas with MAGTs near 0 °C are particularly susceptible to environmental changes including disturbances from human activities. In all of the highly susceptible areas, permafrost is ice rich or contains massive ice. Therefore, the thaw of permafrost in these areas would cause ground subsidence and facilitate the accumulation of meltwater in thaw depressions. In contrast, terrains underlain by ice-poor sediments are more stable upon thawing and are only subject to thaw consolidation.
Limit value
123
Environ Earth Sci
Quaternary surface deposits are mainly composed of alluvial or diluvial fine sands, clay, or silty clay, with thicknesses of approximately 5 m. Fine-grained soils facilitate the accumulation of segregated ice lenses near the permafrost table (e.g., O’Neill and Burn 2012), and their low permeability may also help meltwater, precipitation, and runoff accumulation in thaw depressions. There are no trees along the QTEC and very few shrubs. A robust vegetation cover serves to protect permafrost. However, if
it is damaged or removed, the insulating effect is lost and soil erosion may easily expose ground ice to thaw. In the majority of highly susceptible areas, vegetation cover is sparse and the resistance to erosion is low. These areas are generally basins, plains, or slight slopes on terraces, and the flat ground, in conjunction with high MAGTs, frost-susceptible soils, and sparse vegetation, promotes thermokarst lake development. Distribution of thermokarst lakes based on SPOT-5 imagery
Fig. 9 Integrated map of the terrain susceptibility to thermokarst lake development in the study region. Susceptibility was divided into four categories, including lowest (0.1–0.16), low (0.16–0.19), moderate (0.19–0.26), and high (0.26–0.66)
Table 3 Ice content of permafrost, MAGT, quaternary surface deposit, vegetation, and terrain in high-susceptibility regions along QTEC
Through the above assessment, the terrain susceptibility to thermokarst lake development was mapped. However, validation of the classification results with the actual thermokarst lake distribution in the study area is needed to determine whether the assessment system was reliable. We determined the number and area of thermokarst lakes in the study area using the SPOT-5 satellite imagery. Lakes that were supplied by rivers or streams were excluded from the analysis. In the August 2010 imagery, we counted 2,610 thermokarst lakes, which covered approximately 1.54 9 107 m2. We compared this result to our field investigations from October 2011. From a random sample of 80 thermokarst lakes, 76 were correctly interpreted in the remotely sensed imagery, which corresponds to an accuracy of 95 %. This indicates that the lake inventory derived from the SPOT-5 image interpretation was reasonable. The ‘high’ susceptibility areas occupied about 48 % of the total study area contained approximately 88 % of the thermokarst lakes, and over 91 % of the total lake area. The moderate susceptibility areas accounted for about 27 % of the study area, 6.6 % of the lakes, and 8 % of the lake area. The lower and lowest susceptibility areas covered approximately 25 % of the study area, but contained only
Sections
Ice content of permafrost
MAGT (°C)
Quaternary surface deposit
Vegetation
Terrain
Chumaerhe high plateau
Massive ground ice or icesaturated soil
-0.6 to -1.2
Clay, fine or gravelly sands, or mixture of sand and clay
Sparse vegetation and sandy surface or dune in most sections
High plateau, slight slope
Wudaoliang basin
Massive ground ice and ice-rich permafrost
-0.6 to -1.5
Sparse vegetation and frozen soil wetland
Basin or hills
Beiluhe beach
Massive ground ice or icesaturated soil
-0.5 to -1.0
Sparse vegetation in most sections, but good in minor sections
Basin, but undulating terrain, low hills and depression
Thickness: 5–8 m Alluvial or diluvial fine sands, clay, or silty clay Thickness: 3–5 m Alluvial or diluvial fine sands, silty clay with gravel Thickness: 3–5 m
123
Environ Earth Sci Table 4 The distribution of thermokarst lakes in four susceptibility categories. The majority of thermokarst lakes are located in basins or plains, which facilitate the accumulation of meltwater, precipitation, and runoff
Probable level
Surface area (km2)
Number and surface of lakes
Distribution regions
Number
Number rate (%)
Surface (km2)
Surface rate (%)
High level
485.1
2,297
87.9
17.5
91.2
Moderate level
278.4
171
6.6
1.5
8.0
Qingshuihe terraces, Chumaerhe south, Hoh Xil hill region and Xiushuihe terraces
Low level
129.1
118
4.5
0.1
0.6
Hoh Xil hill region and Fenghuoshan hills
Lowest level
118.7
26
1.0
0.1
0.3
Rock outcrop or steep mountain region, e.g., Fenghuoshan mountain, Xiushuihe hill region, etc.
Majority regions of Chumaerhe high plateau, Wudaoliang basin and Beiluhe beach
Fig. 10 The integrated map of the terrain susceptibility combined with the actual distribution of thermokarst lakes in the study region
5.4 % of the lakes and \1 % of the lake area (Table 4). The areas designated as highly susceptible to thermokarst lake development also, in reality, are areas where lakes were widespread, while low-susceptibility areas contained very few lakes (Fig. 10). This suggests that the assessment method is reliable. Frequency distribution of lake area and thematic factors This study has highlighted the important relations between the five main assessment factors and the development of thermokarst lakes. The following section analyzes relations
123
between total lake area and lake frequency and four of these factors, based on the field observations made in October 2011. We excluded quaternary sediment as a factor in this analysis due to the lack of accurate geological data. The statistical data showed that approximately 43 % of the total lake area occurred in regions underlain by ice-rich permafrost, while 30 % occurred in regions with icy soil. Areas with ice-poor permafrost contained 27 % of the total lake area. Therefore, about 73 % of the total thermokarst lake area occurred in high ice content permafrost. Just over half of the total lake area occurred in regions with high MAGT (C-1.0 °C), while 45 % occurred in
Environ Earth Sci
areas with low-temperature permafrost (\-1.0 °C). This suggests that the occurrence of the thermokarst lakes is more closely related to ground ice content than to MAGT. For vegetation cover, about 51 % of the lake area occurred where there was an exposed surface (NDVI \ 0), 38 % occurred in grassland (0 \ NDVI \ 0.15), and 11 % in meadow (NDVI [ 0.15). This result suggests that high vegetation cover serves to protect permafrost from thermokarst development. Altitude is associated with the surface curvature and lake distribution in our study area. Thermokarst lakes mainly occurred in plains or basins with altitudes between 4,500 and 4,600 m. These areas contained about 56 % of the total lake area. Regions below 4,500 m contained less than 30 % of the total lake area. This may be because MAGT is higher at lower elevation, and permafrost has already degraded to the point where taliks exist and allow infiltration of surface water. High elevation areas ([4,600 m) contained only 16 % of the total lake area, likely because steep slopes in these areas limit water accumulation.
Conclusions 1.
2.
3.
(4)
In accordance with the principles of engineering geological analogy (EGA), combined with expert estimation (EE) and field observations, terrain susceptibility to thermokarst lake development in a region along the QTEC was comprehensively assessed and mapped. Integrated susceptibility values ranged from 0.1 to 0.66 and the study area was divided into four categories. Values between 0.1 and 0.16 indicated that there was a low susceptibility to thermokarst lake development and those between 0.26 and 0.66 indicated a high level. Interpretation of SPOT-5 satellite data showed that the actual distribution of thermokarst lakes in the study area was consistent with the results of the susceptibility assessment. The ‘high’ susceptibility areas occupied about 48 % of the total study area and contained over 88 % of the lakes in the region, which corresponded to approximately 91 % of the total lake surface area. The moderate, low, and lowest level areas accounted for about 52 % of the total study area, but contained only 12 % of the lakes in the region and 9 % of the total lake surface area. Correlation analysis between lake area and the controlling factors indicated that thermokarst lakes in the study region were associated with icy soil, warm permafrost, low vegetation cover, and low altitudes.
Acknowledgments This work was supported by the Major State Basic Research Development Program of China (973 Plan, 2012CB026101), the Western Project Program of Chinese Academy of Sciences (KZCX2-XB3-19), and the Open Foundation of Key Laboratory of Highway Construction and Maintenance Technology in Permafrost Region, CCCC First Highway Consultants Co. Ltd. The authors would like to express their gratitude to the editors and the two anonymous reviewers who provided insightful suggestions, which significantly benefited the authors during the revision process. We are also grateful to Brendan O’Neill, Ph.D. candidate at Carleton University, who helped improve the English in the manuscript.
References Cheng GD (1984) Problems on zonation of high-altitude permafrost. Acta Geogr Sin 39:185–193 (in Chinese) China Railway First Survey and Design Institute Group Ltd (2000) The report on temporary regulations for reconnaissance of constructions at the long term permafrost area in the Qinghai– Tibet plateau (in Chinese) Cui AP (2004) Geological hazard zoning in Lueyang by information system spatial analysis and harassing coefficient method. J Catastrophol 19(2):51–55 French HM (1996) The periglacial environment, 2nd edn. Longman Group, London Harry DG, French HM (1983) The orientation and evolution of thaw lakes, southwest Banks Island, Canadian Arctic. In: Proceedings of the 4th international conference on permafrost. National Academy Press, Washington, pp 456–461 Jiang L, Wang LJ, Zhang XF (2007) Classification of frost heave for subgrade sand soil of highways in seasonal frost regions. J Eng Geol 15(5):639–645. doi:10.3969/j.issn.1004-9665.2007.05.010 (in Chinese) Jin HJ, Yu QH, Wang SL, Lu¨ LZ (2008a) Changes in permafrost environments along the Qinghai–Tibet engineering corridor induced by anthropogenic activities and climate warming. Cold Reg Sci Technol 53(3):317–333. doi:10.1016/j.coldregions. 2007.07.005 Jin HJ, Sun LP, Wang SL, He RX, Lv LZ, Yu SP (2008b) Dual influences of local environmental variables on ground temperatures on the interior-eastern Qinghai–Tibet plateau (I): vegetation and snow cover. J Glaciol Geocryol 30(4):535–545 (in Chinese) Johnston GH, Brown RJE (1964) Some observations on permafrost distribution at a lake in the Mackenzie delta NWT, Canada. Arctic 17:162–175. doi:10.14430/arctic3501 Lee S, Ryu JH, Won JS, Park HJ (2004) Determination and application of the weights for landslide susceptibility mapping using an artificial neural network. Eng Geol 71(3/4):289–302. doi:10.1016/S0013-7952(03)00142-X Li YH (2008) Susceptible degree zoning of mountain torrent disasters in Shaanxi based on ArcGIS. J Catastrophol 23(1):37–42. doi:10. 3969/j.issn.1000-811X.2008.01.009 (in Chinese) Li X, Cheng GD (1999) A GIS-aided response model of high altitude permafrost to global change. Sci China D42(1):72–79. doi:10. 1007/BF02878500 Li EK, Xu GH (1998) Comprehensive evaluation of information systems using the analytic hierarchy process. J Chin Soc Sci Technol Inform 17(6):431–438 (in Chinese) Li BY, Gu GA, Li SD (1996) The series of the comprehensive scientific expedition to the Hoh Xil region-physical environment of Hoh Xil region, Qinghai. Science Press, Beijing (in Chinese) Lin ZJ (2011) Study of thermokarst lake and its influence on the permafrost environment and engineering. Dissertation, Cold and
123
Environ Earth Sci Arid Regions Environmental and Engineering Research Institute (in Chinese) Lin ZJ, Niu FJ, Xu ZY, Xu J, Wang P (2010) Thermal regime of a thermokarst lake and its influence on permafrost, Beiluhe basin, Qinghai–Tibet plateau. Permafr Periglac Proc 21(4):315–324. doi:10.1002/ppp.692 Lin ZJ, Niu FJ, Liu H, Lu JH (2011a) Disturbance-related thawing of a ditch and its influence on roadbeds on permafrost. Cold Reg Sci Technol 66:105–114. doi:10.1016/j.coldregions.2011.01.006 Lin ZJ, Niu FJ, Liu H, Lu JH (2011b) Hydrothermal processes of Alpine Tundra lakes, Beiluhe basin, Qinghai–Tibet plateau. Cold Reg Sci Technol 65:446–455. doi:10.1016/j.coldregions.2010. 10.013 Ling F, Zhang TJ (2004) Modeling study of talik freeze-up and permafrost response under drained thaw lakes on the Alaskan Arctic coastal plain. J Geophys Res 109(01111):57–65. doi:10. 1029/2003JD003886 Liu XL, Chen YJ (2011) Influence of debris flow spot and area densities on regional hazardousness of debris flow: a comparative study. J Nat Disasters 20(2):36–43 (in Chinese) Liu DL, Tham LG, Lee PKK, Tsui Y, Wang YH (1999) The stability of rock slope and fuzzy comprehensive evaluation method. Chin J Rock Mech Eng 18(2):170–175. doi:10.3321/j.issn:1000-6915. 1999.02.012 (in Chinese) Liu YZ, Wu QB, Zhang JM, Tong CJ, Sheng ZY (2000) Study on ground temperature field in permafrost regions of Qinghai–Tibet plateau. Highway 2:4–8 (in Chinese) Liu CZ, Li TF, Cheng LP, Wen MS, Wang XP (2004) A method by to analyse four parameters for assessment and early warning on the regional geo-hazards. Hydrogeol Eng Geol 4:1–8. doi:10.3969/j. issn.1000-3665.2004.04.00.1 (in Chinese) Lunardini VJ (1996) Climatic warming and the degradation of warm permafrost. Permafr Periglac Proc 7:311–320. doi:10.1002/ (SICI)1099-1530(199610)7:4\311:AID-PPP234[3.0.CO;2-H Moiseenko TI, Voinov AA, Megorsky VV, Gashkina NA, Kudriavtseva LP, Vandish OI, Sharov AN, Sharova Yu, Koroleva IN (2006) Ecosystem and human health assessment to define environmental management strategies: the case of long-term human impacts on an Arctic lake. Sci Total Environ 369:1–20. doi:10.1016/j.scitotenv.2006.06.009 Niu FJ, Zhang JM, Zhang Z (2002) Engineering geological characteristics and evaluations of permafrost in Beiluhe testing field of Qinghai–Tibetan railway. J Glaciol Geocryol 24(3):264–269. doi:10.3969/j.issn.1000-0240.2002.03.007 (in Chinese) Niu FJ, Xu J, Lin ZJ, Wang P (2008) Engineering activity induced environmental hazards in permafrost regions of Qinghai–Tibet plateau. In: Kane DL, Hinkel KM (eds) Proceedings of the 9th international conference on permafrost, Faribanks, pp 1287–1292 Niu FJ, Lin ZJ, Liu H, Lu JH (2011) Characteristics of thermokarst lakes and their influence on permafrost in Qinghai–Tibet plateau. Geomorphology 132:222–233. doi:10.1016/j.geomorph.2011.05. 011 O’Neill HB, Burn CR (2012) Physical and temporal factors controlling the development of near-surface ground ice at Illisarvik, western Arctic coast, Canada. Can J Earth Sci 49(9):1096–1110. doi:10.1139/e2012-043
123
Olmacher GC, Davis JC (2003) Using multiple logistic regression and GIS technology to predict landslide hazard in northeast Kansas, USA. Eng Geol 69:331–343. doi:10.1016/S0013-7952(03) 00069-3 Pistocchi A, LuziL Napolitano P (2002) The use of predictive modelling techniques for optimal exploitation of spatial databases: a case study in landslide hazard mapping with expert system-like methods. Environ Geol 41(7):765–775. doi:10.1007/ s002540100440 Qiu XR, Yuan RM, Xu WW (2003) Fuzzy mathematics method for evaluation of slope stability. Res Soil Water Conserv 10(3):26–36. doi:10.3969/j.issn.1005-3409.2003.03.007 (in Chinese) Saaty TL (1980) The analytic hierarchy process. McGraw-Hill, New York Sellmann PV, Week WF, Campbell WJ (1975) Use of sidelooking airborne radar to determine lake depth on the Alaska North Slope. Cold Regions Research and Engineering Laboratory (CRREL) Special Report 230, Hanover, USA Wang WD, Chen YP, Zhong S (2009) Landslides susceptibility mapped with CF and logistic regression model. J Cent South Univ Sci Technol 40(4):1127–1133 (in Chinese) Wang Z, Yi FC (2009) AHP-based evaluation of occurrance easiness of geological disasters in Mianyang City. J Nat Disaster 18(1):14–23. doi:10.3969/j.issn.1004-4574.2009.01.003 (in Chinese) Williams PJ, Smith MW (1989) The frozen earth: fundamentals of geocryology. Cambridge University Press, Cambridge Wu QB, Zhang TJ (2008) Recent permafrost warming on the Qinghai–Tibetan plateau. J Geophys Res 113(D13108):1–22. doi:10.1029/2007JD009539 Wu QB, Li X, Li WJ (2000) Computer simulation and mapping of the regional distribution of permafrost along the Qinghai-Xizang highway. J Glaciol Geocryol 22(4):323–326. doi:10.3969/j.issn. 1000-0240.2000.04.005 (in Chinese) Wu QB, Liu YZ, Tong CJ (2002) Interactions between the permafrost and engineering environments in the cold regions. J Eng Geol 8(3):281–287 (in Chinese) Wu QB, Cheng GD, Ma W (2004) The impact of climate warming on Qinghai–Tibetan railroad. Sci China D47(Supp I):122–130 (in Chinese) Wu QB, Dong XF, Liu YZ (2005) Response of permafrost to climate change and engineering activity along the Qinghai–Tibet highway. J Glaciol Geocryol 27(1):50–54. doi:10.3969/j.issn.10000240.2005.01.007 (in Chinese) Ye YC (2006) The method of comprehensive assessment and its application. Metallurgy Industry Press, Beijing (in Chinese) Yu HM, Lu HQ (2009) The genetic analysis and occurrence probability of the debris flow around a watershed in Xian-ju county, Zhejiang province. China J Geol Hazard Control 20(1):21–26. doi:10.3969/j.issn.1003-8035.2009.01.005 (in Chinese) Zhou YW, Guo DX, Qiu GQ, Cheng GD, Li SD (2000) Geocryology in China. Science Press, Beijing