Environmental Earth Sciences (2020) 79:347 https://doi.org/10.1007/s12665-020-09066-4

ORIGINAL ARTICLE

Pattern analysis and dating for the Badain Jaran feld, northwestern

Wen‑Xiao Ning1,2 · Zhen‑Ting Wang1

Received: 22 May 2019 / Accepted: 17 June 2020 © Springer-Verlag GmbH Germany, part of Springer Nature 2020

Abstract Pattern analysis can estimate dune formation ages, judge wind conditions and extract environmental signals from the com- mon pattern parameters of active dune feld. For the extraterrestrial dune felds, this method is a great advantage while many in situ and laboratory techniques cannot do anything. An ideal location for the large-scale pattern analysis test is the Badain Jaran where the tallest terrestrial aeolian dune occurs. In this study, the function forms of pattern parameters varying with the dune construction time are established, according to the natural evolution of a dune feld in the duration of over 10 years. It is revealed that the relations between pattern parameters and time display power functions, in which the unknown coefcients in diferent conditions of wind regimes and sand availabilities can be estimated from the absolute ages of local deposits. From the northwest to southeast of the , the dune crest length and mean spacing increase, and ◦ the defect density decreases. The pattern ages change correspondingly. Dune orientation mostly ranges between 10 and ◦ 60 . The crest length ages indicate that the modern active dune feld began to form at about 20–75 ka BP, the in the southeast and northeast are older than those in the western. This work confrms that pattern analysis ofers an efective dating method for hard-to-reach or large-scale dune felds.

Keywords Badain Jaran · Active dune feld · Pattern analysis · Remote sensing · Pattern dating

Introduction defect/bedform repulsion and of-centre collision) efects coexist (Kocurek and Ewing 2005; Ewing and Kocurek Nature provides examples of many kinds of pattern, dune 2010a, b). However, the boundary conditions of the system, feld is one of them. Aeolian dune felds that exist in a simple including wind regime and sediment supply, often restrict form, usually exhibit some complexity, such as the nonlinear the external environment conditions of dune interactions, formation process and self-organized aggregation behavior thus afecting the evolution of dune felds. At the same (Werner 1999). Beginning with small and disordered dunes, time, dune felds also contain some important information dune felds gradually become better organized with longer of paleoenvironment and paleoclimate. Since aeolian dune crests and wider spacing, through interactions between felds are composed of diferent spatial scales of bedform bedforms (Sharp 1963; Kocurek et al. 1992; Werner and pattern (Ewing et al. 2015), the analysis of large-scale pat- Gillespie 1993; Werner 1995; Werner and Kocurek 1997, tern parameters can reveal the response of dunes to long- 1999). In fact, in the process of dune collisions, positive (it term atmospheric changes and sedimentary conditions (Fen- makes the pattern more ordered, e.g. merging and lateral ton et al. 2014; Savage et al. 2014). linking), negative (it produces more disorganized system, In recent years, with the improvement of spectral and e.g. defect creation and bedform splitting) and neutral (e.g. spatial resolution of remote sensing data, the study of aeolian sand dunes has gradually shifted from the map- * Wen‑Xiao Ning ping and classification of dunes in the early stage to [email protected] quantitative analysis of dune morphological dynamics and dune field pattern analysis, see the review of Hugen- 1 Northwest Institute of Eco-Environment and Resources, holtz et al. (2012) and references therein for details. The CAS, Lanzhou 730000, People’s Republic of China common parameters of dune field patterns, i.e. dune spac- 2 University of Chinese Academy of Sciences, Beijing 100049, ing, crest length, defect density and crest direction etc., People’s Republic of China

Vol.:(0123456789)1 3 347 Page 2 of 11 Environmental Earth Sciences (2020) 79:347 can be easily measured from the remote sensing images. Study area and methods As a convenient method, pattern analysis for active dune fields can obtain some valuable information from com- Study area plex spatial patterns (Ewing et al. 2006; Derickson et al. 2008; Fenton et al. 2014; Savage et al. 2014). are The Badain Jaran desert is located at east of the Heihe alluvial widely distributed in northern China (Wang 2011). As far fan, south of Guaizi lake, west of Zongnai mountain, north as we know, only a few pattern analyses for the Chinese of Heli, Beida, and Yabulai mountains in the northwest of dune fields have been conducted (Wu and Guo 2012; Li China, as shown in Fig. 1. The desert is 442 km long and 4 2 et al. 2016). The linear dunes in the northern Kumtagh 354 km wide, covering an area of 5.2 × 10 km (Zhu et al. desert might have developed since 13 ka BP (Wu and Guo 2010), and has a temperate continental climate with extreme 2012), if the empirical relations of Ewing et al. (2006) drought. It is cold and dry in winter, hot in summer, and the hold almost everywhere. Three possible evolution stages transition between summer and winter is very rapid. Precipita- of the linear dune field on the northern margin of Qarhan tion is scarce and mostly occurs in June, July and August. The Salt Lake were proposed in terms of the statistical char- overall topography of the desert is high in the southeast and acteristics and spatial variation of the pattern parameters low in the northwest. The mega-dunes, mainly concentrated (Li et al. 2016). In the evolution process of an active dune in the southeast of the desert, extend toward northeast–south- field, pattern parameters always change with the passage west, perpendicular to the prevailing northwest wind (Ning of time. But, the expressions of local pattern parameters and Wang 2018). Most of these dunes are between 200 and as functions of time remain unsolved in the aforemen- 300 m in height. The tallest terrestrial dune with the height of tioned works. 460 m is found in the desert (Yang et al. 2011). The types of Since dune fields often develop at the geologic time- mega-dunes include compound transverse dunes and complex scale, it is impossible to record their entire dynamic star dunes (Breed et al. 1979; Walker et al. 1987; Dong et al. process by the conventional field investigations and 2004, 2013). Many aeolian lakes exist between mega-dunes observations. Various chronological techniques have to (Wang et al. 2016). In this study, a few small solidifed or fos- be introduced. Optically stimulated luminescence (OSL) sil dunes occasionally distributed in this desert are not taken proposed by Huntley et al. (1985) is a powerful dating into account. tool for aeolian sediments in practice (Aitken 1998; Win- tle 2008; Zhang et al. 2015a). It is possible to reconstruct Data and methods the developmental stages of individual dunes by combina- tion of grand penetrating radar (GPR) and OSL (Bristow Following Ewing et al. (2006), crest length, dune orientation, et al. 2000, 2007). OSL is more suitable for the dunes mean spacing and defect density are selected as the dominat- with less or no migrations, because this method measures ing pattern parameters. Crest length refers to the boundary the elapsed years since the sediment was last consigned to between the windward and lee slope of dunes, which can darkness by burial. As for actively migrating dunes, OSL be used to measure the continuity of dune (Al-Masrahy and dating produces recent ages, whereas a pattern age rep- Mountney 2013). Dune orientation is defned as the angle resents the total development time of dune fields (Ewing between dune crest and the east direction. Dune spacing is et al. 2006). Moreover, for the extraterrestrial dune fields the horizontal distance between adjacent dune crests, which is (Greeley and Iversen 1985; Lorenz and Zimbelman 2014; closely related to formation time, initial conditions and defect Telfer et al. 2018), the direct dating techniques are cur- behavior of dunes. Given an area, the mean spacing is defned rently not available, pattern analysis is a great advantage. as (Werner and Kocurek 1999) For example, the Titan dunes have grown toward a steady A state for a long period of time (Savage et al. 2014), and Sp = , N L (1) the relation between dune spacing and “interaction” den- c sity on Earth, Mars, and Titan can be well described by where A is the area, and L and Nc are the mean length and a unified inverse quadratic function (Day and Kocurek number of crests in it. Defects are the breaking points or 2018). end points of dunes. The defect density, defned as the num- In the present study, the general relations between pat- ber of pairs of defects per unit of dune crest length, can be tern parameters and dune construction time are established expressed by frst. The local parameters of the Badain Jaran dune feld are then extracted from the remote sensing images and Nd = , (2) DEM, and their statistical characteristics and spatial NcL changes are quantifed. Finally, the pattern ages of the N whole dune feld are computed. where d is the number of defect pairs in the area of A.

1 3 Environmental Earth Sciences (2020) 79:347 Page 3 of 11 347

Fig. 1 Geographic setting of the Badain Jaran desert (BJD). The feld experiment performed by Ping et al. (2014) is sited in the . The sampling grids in the Badain Jaran desert are denoted by small rectangles

The defnitions of (1) and (2) are convenient for dune squares in Fig. 1. Next, dune crestlines are automatically feld-scale surveys and measurements although they could extracted from the DEM via the hydrological analysis model not be optimal. Several algorithms, such as machine learn- in the software of ArcGIS (Tang and Yang 2010). The mis- ing (Azzaoui et al. 2019), fngerprint minutiae extraction takes in the results of dune crests are distinguished and mod- (Scuderi 2019) and manual tracking (Wu and Guo 2012; ifed through visual interpretation and manual tracking, by Li et al. 2016), have been developed to extract the dune comparing multi-spectral remote sensing data (15 m × 15 m ). crest or dune contour in satellite images. Because the dune This method is similar to the extraction process of moun- heights in the Badain Jaran desert are commonly compa- tain ridges, because mega-dunes have the characteristics of rable to these of mountains, here the sophisticated method relatively large height diference between dune crests and of identifying ridgelines from a mountain is adopted. The inter-dune surfaces. Meanwhile, the crest length is computed technical fowchart is given in Fig. 2. The detailed methods and recorded automatically; the numbers of defect pairs and for determining the pattern parameters of L, Sp and are crests are identifed and counted. From their defnitions (1) described as follows. The Landsat 8 satellite images and and (2), the mean spacing and defect density are calculated. ASTER Digital Elevation Model (DEM) of the study area After that, the contour maps for the dune crest length, mean are acquired from the USGS Earth Explorer (USGS 2018). spacing and defect density are drawn up by the pan-Kriging They are used to extract and verify dune crests. First, by interpolation of 61 grids, for the purpose of analyzing how sharpening the true colors multi-spectral bands (4, 3, 2, the pattern parameters changed spatially. Specially, the rea- 30 m × 30 m ) and panchromatic band (8, 15 m × 15 m ) of son for the selection of pan-Kriging interpolation is that 12 landsat 8 remote sensing images from 2016 to 2017 in these parameters are stable in the small grids but unstable ENVI, the multi-spectral images with a spatial resolution throughout the whole study area. Finally, the pattern ages of of 15 m are obtained. Then to estimate the spatial distribu- mega-dunes are computed using the parameter–time rela- tion of dune feld pattern parameters, the study area’s DEM tions which will be described in the next paragraph. The ( 30 m × 30 m ) is embedded by 24 small DEM and divided crest length-weighted average of all dunes in one grid is into 10 km × 10 km grids, with 22 rows and 28 columns in regarded as the mean age of that grid. The whole dune feld total. In this research, 61 grids are selected, denoted by small is further dated by the pan-Kriging interpolation method.

1 3 347 Page 4 of 11 Environmental Earth Sciences (2020) 79:347

Fig. 2 Flowchart of data processing. It is conveniently performed using the ArcGIS software

An open problem for dune feld pattern analysis is how to time, small transverse dunes appear over several undisturbed quantify the temporal variations of pattern parameters. It was surfaces. We can get some information about the variation of previously revealed that the formation and evolution of aeo- dune feld pattern parameters with time from this case study. lian dune felds pattern in diferent spatial and temporal scales The pattern parameters in 2009, 2013, 2014, 2015, 2016, 2017, are highly similar (Werner and Kocurek 1997; Elbelrhiti et al. 2018 are measured from remote sensing images with the spa- 2005, 2008; Kocurek et al. 2010; Fenton et al. 2014). Owing to tial resolution of 0.3 m × 0.3 m , downloaded from the Google sand bed instability arising from the hydrodynamics (Charru Earth. Due to the short time of dune development, the dune et al. 2013; Parteli et al. 2014), initial small aeolian dunes are crests in this region are identifed and tracked manually. The rapidly formed, and then gradually merge and grow into larger zero time is set to be 2005 when the surface was fatted. The dunes. A feld experiment of this process was carried out by previous wind regime analyses based on the near-surface mete- ◦ Ping et al. (2014). The small experiment site (37 33 20.32 orological data in the Alxa plateau show that drift potential, a ◦ ◦ ◦ –37 33 26.06 N, 105 1 40.81 –105 2 6.84 E) is located widely used variable indicating the sediment transport ability at the southeastern edge of the Tengger desert, as shown in of wind, spatially changes in diferent wind energy environ- Fig. 1. The natural dunes over a large area containing their ments (Zhang et al. 2012; Yang et al. 2014; Zhang et al. 2015b, experiment site was razed to the fat ground for construct- c). One of the most common characteristics is that the prevail- ing factories and houses during 2000 ∼ 2005 . At the present ing wind is from the northwest, both in the deserts of Badain

1 3 Environmental Earth Sciences (2020) 79:347 Page 5 of 11 347

Jaran and Tengger and in the corridors between them. From Sp = t (3b) this point, the obtained function forms of pattern parameters √ from the Tengger desert have a certain representativeness of =∕ t, the Alxa plateau. (3c) √ Results and discussion in which the constant coefcients , , depend on wind regimes and sand availabilities. Expressions for pattern parameters Obviously, with time goes by, the dune crest length and mean spacing increase, while the defect density decreases. The general trends of pattern parameters are consistent Figure 3 gives the computed results of pattern parameter with what Ewing et al. (2006) found in other dune felds, changes during 180 months in the southeastern edge of the but function forms are diferent. Among them, the mean Tengger desert. The outliers at the time of t = 97 months spacing’s variation tendency (Fig. 3b) highly agrees with in Fig. 3a, c are probably caused by the infuence of human the model of Werner and Kocurek (1999). From the view- activities on the dune feld evolution. The data of crest length, point of theory, the physical basis of the expressions of Eqs. mean spacing, defect density can be well ftted by the power (3a–c) is robust. The shapes of dunes belonging to one type functions of time: are similar. For instance, barchan height is proportional to L = t (3a) its width (Wang et al. 2007). Individual barchans may link up to form a transverse dune when sand supply is plenti- √ ful (Pye and Tsoar 2009). For the transverse mega-dunes in the Badain Jaran desert, there are strong linear correlations

Fig. 3 Empirical relations between pattern parameters and dune construction time

1 3 347 Page 6 of 11 Environmental Earth Sciences (2020) 79:347

this empirical expression is not employed to date dunes hereinafter. There are two ways to determine the coefcients in the formulae of (3a)–(3c). One is to measure parameters at dif- ferent time points. Let the crest lengths at t1 and t2 be denoted by L1 and L2 , the frst coefcient can be written as

2 2 L1 − L2 = . (4) t1 − t2

The other is to fnd the absolute ages of individual dunes. Note that the geometrical variations of mega-dunes are not evident in the recent 10 years, we focus on previous chronological studies in the Badain Jaran desert (Yan et al. 2001; Yang 2001; Yang et al. 2003; Mischke 2005; Wang Fig. 4 Crest length–height relation of mega-dunes et al. 2015; Li et al. 2015; Liu et al. 2016; Wang et al. 2017; Chen et al. 2019; Zhao et al. 2019). According to arduous feld surveys and several OSL ages of aeolian and lacustrine between the three spatial scales, i.e. height, crest length, and sediments, Liu et al. (2016) proposed that most mega-dunes spacing. The height–spacing relation has been confrmed growth took place in humid climates, and water acts as a (Dong et al. 2009; Ning and Wang 2018). Figure 4 further cementing agent. Here we use the OSL age of sediments shows that height is a linear function of crest length. Sev- at the toe of windward slope of mega-dunes to represent eral previous numerical simulations of the physical-based the formation age of the adjacent active dune. Two dune ◦ ◦ dune dynamic models imply that the height of a transverse sites are located at ( 39 50 33.18 N, 102 29 1.98 E) and ◦ ◦ dune grows up with the half power of time (Momiji 2001; ( 39 34 9.42 N, 102 25 4.86 E), separately. The absolute Schwämmle and Herrmann 2004; Fischer et al. 2008). Very ages of the lacustrine sediment samples collected in 2014 recently, such an analytical expression for the growth of a are 95.45 ka BP and 116.83 ka BP, respectively (Chen et al. 2-d transverse dune was derived by Ning et al. (2019), under 2019). Substituting these ages, as well as the values of crest the realistic assumptions of shape similarity, the near-surface length, mean spacing, defect density, into Eqs. (3a–c), the −1∕2 airfow independent of height, and the 100% sand trapping mean coefcients are computed to be = 9.79 m yr , −1∕2 −1∕2 efciency of lee face during dune evolution. Consequently, = 8.85 m yr , and = 0.1526 m yr . Detailed infor- the functional forms of Eqs. (3a, b) are reasonable. As for mation are listed in Table 1. Eq.(3c), it is found that defect density is an inverse quadratic function of spacing for linear and crescentic ridge dunes Statistical and spatial characteristics of pattern (Day and Kocurek 2018). Combining this result with Eq. parameters (3b), defect density should be inversely proportion to time. However, the data do not support this conclusion. Thus, The extraction efect of mega-dune crests in the Badain Jaran desert is illustrated in Fig. 5. Aeolian bedform is a

Table 1 Values of the Sand dune ID 1 2 Mean coefcients in Eqs. (3a–c) ◦ � �� ◦ � �� ◦ � �� ◦ � �� Location (39 50 33.18 N, 102 29 1.98 E) (39 34 57.13 N, 102 25 4.86 E) L (m) 3529 2861 Sp (m) 2923 2823 (1/km) 0.466 0.472 t (ka) 116.83 95.45 (m yr−1∕2) 10.32 9.26 9.79 (m yr−1∕2) 8.57 9.14 8.85 (m−1∕2 yr) 0.1592 0.1458 0.1526

The OSL ages are sourced from Chen et al. (2019) L, Sp, , t are crest length, dune spacing, defect density and OSL ages, respectively

1 3 Environmental Earth Sciences (2020) 79:347 Page 7 of 11 347

Fig. 5 Extraction efect of mega-dunes crestlines in the Badain Jaran desert. a The original outputs from the hydrologic analysis model. b The fnal results after fltering small dunes. c The dune crests extracted manually

Fig. 6 Cumulative probability plots for the pattern parameters multi-scale spatiotemporal system. As shown in Fig. 5a, the slope of transverse mega-dunes are often superimposed original outputs from hydrologic analysis model contain with small dunes of other types. Such a classification of many small dunes which can be automatically fltered given dunes from the viewpoint of pattern is consistent with the a critical spatial scale. With the help of visual interpreta- traditional descriptions of desert geomorphology (Breed tion and manual modifcation, the fnal results in Fig. 5b et al. 1979; Walker et al. 1987; Dong et al. 2004). Dune are acceptable, compared with the dune crests in Fig. 5c spacing is regarded as a more convincing parameter than extracted manually. Certainly, the method we used is more crest length in the statistical identification of multiple efcient and suitable for this large dune feld. populations in each separate dune field or sand sea (Der- The logarithmic cumulative probability curves of dune ickson et al. 2008). The data of mean spacing, changing crest length and mean spacing are plotted in Fig 6. The between 1600 m and 5500 m, obey a normal distribution dune morphological information can be identified and (Ning and Wang 2018). Analogously, three populations inferred from these curves. According to the crest length separated by Sp = 2100 m and Sp = 2900 m can be divided. (Fig. 6a) ranging from 1400 to 2400 m, three populations The dune spacing of one population ( Sp ≤ 2100 m ) are of dunes can be roughly classified. The first population intermittent and scattered (Fig. 6b), because small dunes ( L ≤ 1750 m ) is barchans and barchan chains in the west- are more disordered and irregular than large ones. Com- ern and at the margin of the desert. The second population paring the separation or gathering levels of data between (1750 m < L ≤ 2000 m ) represents small dune chains, star Fig. 6a and b, crest length seems better. Different popula- and linear dunes in the northern desert. The third popu- tions can reflect states of pattern development (Derickson lation ( L > 2000 m ), i.e. transverse dunes, accounts for et al. 2008). The evolution of the modern Badain Jaran more than 50% dunes of the whole desert. The upwind dune field has likely gone through three stages. At first,

1 3 347 Page 8 of 11 Environmental Earth Sciences (2020) 79:347

and spacing. The spatial characteristics of parameters, shown in Fig. 8, should be helpful for understanding the dynamic mechanism of dune field. From northwest to southeast, the dune crest length (Fig. 8a) and mean spac- ing (Fig. 8b) gradually increase, and the defect density (Fig. 8c) decreases little by little. The longest crest length is discovered in the southern part of the desert, and the shortest in the western part. However, the correspond- ing extreme values of dune spacing abnormally occur in the eastern and northern of the desert, respectively. The algorithm for indirectly computing spacing, i.e. Eq. (1), results in uncertainty when the dune pattern is not so regular. The manual measurements can improve the pre- cision of the spacing data of mega-dunes really (Dong et al. 2009; Ning and Wang 2018), as recently done by Day and Kocurek (2018) elsewhere. The spatial distribu- tion of defect density is opposed to that of crest length, because the number of defect pairs Nd in Eq. (2) is very close to crest number Nc , and defect density is in inverse proportion to crest length. Fig. 7 Dune orientation frequency’s rose diagram ◦with the zero degree in the east direction and the sector bins of 10 . The direction Pattern ages and length of the rose diagram represent dune orientation◦ ◦ and fre- quency, respectively. Most dunes orient between 10 and 60 The approximate formation ages of the Badain Jaran dune feld can be estimated by crest length in the light of the new simple and compound barchan dunes dominated the con- formula (3a). Mean spacing and defect density are not used struction of dune field. And then, the sand is continuously because they are the derived quantities of crest length in the moving from the northwest to the southeast and piling up, defnitions of Eqs. (1, 2). Other reasons are that the data of due to the abundant supply of sand sources and prevailing mean spacing are not accurate as we expected, and the cal- northwest wind conditions. Coupled with the blocking culated ages are extremely sensitive to the values of defect of the mountains, the mega-dunes gradually formed and density which are much smaller than unit. The interpolated the current dune field pattern appears. Dune orientation contour map of the crest length ages is plotted in Fig. 9. It is an important pattern parameter closely related to local is shown that the dunes in the southeast and northeast are wind conditions. Figure 7 is the rose diagram of dune older, and those in the western are younger. This age con- ◦ orientation frequency in each 10 sector bins. Among the fguration might be explained as follows. The surfcial sedi- ◦ 1259 dunes we calculated, most dunes are between 10 ments in are easily transported by the prevail- ◦ ◦ ◦ and 60 , and more than 50% of the total direct 30 ∼ 50 . ing wind blowing from the northwest in arid climates, and This result agrees with the previous judgment that the the dunes there can be reconstructed by the running water mega-dunes in this desert are commonly perpendicular of the Heihe in humid climates. However, the dunes in the to the prevailing northwest wind (Dong et al. 2009; Ning northeast and southeast restricted by the boundaries of hills and Wang 2018). It is still difficult to get reliable rela- or mountains can continuously develop. Although the initial tions between dune orientation and local sand drift poten- formation of the desert might be 1100 ka BP at least (Wang tials, because the meteorological records in the interior et al. 2015), the crest length ages indicate that the modern of the desert are lacking. The wind regime analyses have active dune feld is likely to begin to form during 20 ∼ 75 ka to depend on 5 ∼ 7 weather stations on the desert mar- BP. As shown in Fig. 10, most active dunes are formed in gins (Yang et al. 2011; Zhang et al. 2015b). Wind energy the period of 10 ∼ 70 ka BP, accounting for more than 75% environment and direction can be distinguished among of the total. Dating deposits in the Badain Jaran desert is an these stations (Yang et al. 2011). But, for a desert of ten ongoing work (Yang et al. 2003; Mischke 2005; Wang et al. thousands of square kilometers in area, the magnitude 2015; Liu et al. 2016; Chen et al. 2019). It was concluded, contours of wind speeds drawn by Zhang et al. (2015b) from the thermoluminescence ages of the aeolian sand sam- is less accurate. ples at the bottom of dunes, that the modern active mega- The dune field evolution process from disorder to order dunes were roughly formed since the last glacial period means the extension and development of the crest length (Yan et al. 2001). The lacustrine sediments in the northwest

1 3 Environmental Earth Sciences (2020) 79:347 Page 9 of 11 347

Fig. 8 Interpolated spatial contour maps of the parameters

Fig. 9 Interpolated spatial contour map of crest length ages Fig. 10 Frequency distribution of crest length ages margin of the desert are dated back to 100 ∼ 190 ka BP at the earliest (Mischke 2005). The upwind with average theoretical time needed to form the current spatial the modern erosion rate of 0.41 ∼ 0.89 mm/yr is considered scale of the desert is estimated to be 80.8 ka (Wang et al. as one mass source of the Badain Jaran desert, and thus the 2019). Compared with these chronological data and theo- retical estimations, the pattern ages we got are acceptable

1 3 347 Page 10 of 11 Environmental Earth Sciences (2020) 79:347 on the whole, though the more detailed considerations on Bristow CS, Bailey SD, Lancaster N (2000) The sedimentary structure long-term wind regime, sediment composition, and dune of linear sand dunes. Nature 406:56–59 etc. Bristow CS, Duller GAT, Lancaster N (2007) Age and dynamics of internal tectonics are needed to verify them. The spatial linear dunes in the Desert. Geology 35(6):555–558 contour map of Fig. 9 and frequency distribution histogram Charru F, Andreotti B, Claudin P (2013) Sand ripples and dunes. of Fig. 10 will ofer convenience for future studies. Annual Review of Fluid Mechanics 45(1):469–493 Chen T, Lai Z, Liu S, Wang Y, Wang ZT, Miao X, An F, Yu L, Han F (2019) Luminescence chronology and palaeoenvironmental sig- nifcance of limnic relics from the Badain Jaran desert, northern Conclusions China. Journal of Asian Earth Sciences 177:240–249 Day M, Kocurek G (2018) Pattern similarity across planetary dune Pattern analysis provides a convenient method for unknown felds. Geology 46(11):999–1002 Derickson D, Kocurek G, Ewing RC, Bristow C (2008) Origin of a and inaccessible dune felds because pattern parameters complex and spatially diverse dune-feld pattern, Algodones, can quantify dune feld evolution. In the current study, the southeastern California. Geomorphology 99(1–4):186–204 well-known difculty of the temporal functions of pattern Dong Z, Wang T, Wang X (2004) Geomorphology of the megadunes parameters is overcome by a series of remote sensing images in the Badain Jaran Desert. Geomorphology 60(1–2):191–203 Dong Z, Qian G, Luo W, Zhang Z, Xiao S, Zhao A (2009) Geomorpho- of natural dune feld evolution stages. The statistical char- logical hierarchies for complex mega-dunes and their implications acteristics, spatial distribution and ages of pattern param- for mega-dune evolution in the Badain Jaran Desert. Geomorphol- eters of the Badain Jaran dune feld are investigated for the ogy 106(3–4):180–185 frst time. The dunes can be divided into three populations. Dong Z, Qian G, Lv P, Hu G (2013) Investigation of the sand sea with the tallest dunes on Earth: China’s Badain Jaran sand sea. Earth- Dunes with long crest and large spacing belong to transverse Science Reviews 120:20–39 mega-dunes mainly found in the southeast of the desert. Elbelrhiti H, Claudin P, Andreotti B (2005) Field evidence for ◦ ◦ Most dunes orient between 10 and 60 . From the north- surface-wave-induced instability of sand dunes. Nature west to the southeast of the desert, dune crest length and 437(7059):720–723 Elbelrhiti H, Andreotti B, Claudin P (2008) Barchan dune corridors: mean spacing increase and defect density decreases. Cor- Field characterization and investigation of control parameters. respondingly, the modern active dunes in the southeast and Journal of Geophysical Research 113(F2):1–21 northeast are older than those in the western. The initial age Ewing RC, Kocurek G (2010a) Aeolian dune-feld pattern boundary of this dune feld can be estimated or approximated by the conditions. Geomorphology 114(3):175–187 Ewing RC, Kocurek G (2010b) Aeolian dune interactions and dune- pattern ages of crest length. It is concluded that most modern feld pattern formation: White Sands Dune Field, New Mexico. active dunes are formed in the period of 20 ∼ 75 ka BP. The Sedimentology pp 1199–1219 method we developed here can be extended to other ter- Ewing RC, Kocurek G, Lake LW (2006) Pattern analysis of dune- restrial or extraterrestrial dune felds such as these on Mars, field parameters. Earth Surface Processes and Landforms 31(9):1176–1191 Venus, Titan, and Pluto. Ewing RC, McDonald GD, Hayes AG (2015) Multi-spatial analysis of aeolian dune-feld patterns. Geomorphology 240:44–53 Acknowledgements We are grateful to the anonymous reviewers for Fenton LK, Michaels TI, Beyer RA (2014) Inverse maximum gross their insightful comments, to Xu-Yang Liu and Dai-Qiang Lu for their bedform-normal transport 1: How to determine a dune-con- careful proofreading of the original manuscript. This research was structing wind regime using only imagery. Icarus 230:5–14 supported by the Natural Science Foundation of China projects (Nos. Fischer S, Cates ME, Kroy K (2008) Dynamic scaling of desert 41571006 and 41630747). dunes. Physical Review E 77:031302 Greeley R, Iversen JD (1985) Wind as a geological process: On Earth, Mars, Venus and Titan. Cambridge University Press, Cambridge Hugenholtz CH, Levin N, Barchyn TE, Baddock MC (2012) Remote References sensing and spatial analysis of aeolian sand dunes: A review and outlook. Earth-Science Reviews 111(3–4):319–334 Aitken MJ (1998) An introduction to optical dating: the dating of Qua- Huntley DJ, Godfrey-Smith DI, Thewalt MLW (1985) Optical dating ternary sediments by the use of photon-stimulated luminescence. of sediments. Nature 313(5998):105–107 Oxford University Press, Oxford, New York Kocurek G, Ewing RC (2005) Aeolian dune feld self-organization Al-Masrahy MA, Mountney NP (2013) Remote sensing of spatial vari- - implications for the formation of simple versus complex dune- ability in aeolian dune and interdune morphology in the Rub’ Al- feld patterns. Geomorphology 72(1–4):94–105 Khali, Saudi Arabia. Aeolian Research 11:155–170 Kocurek G, Townsley M, Yeh E, Havholm KG, Sweet ML (1992) Azzaoui M, Manare A, Elbelrhiti H, Chaouki I, Masmoudi L (2019) Dune and dune-feld development on Padre Island, Texas, with Detection of crescent sand dunes contours in satellite images implications for interdune deposition and water-table-con- using an active shape model with a cascade classifer. ISPRS - trolled accumulation. SEPM Journal of Sedimentary Research International Archives of the Photogrammetry, Remote Sensing 62:622–635 and Spatial Information Sciences XLII-4/W12:17–24 Kocurek G, Ewing RC, Mohrig D (2010) How do bedform patterns Breed CS, Fryberger S, Andrews S, McCauley C, Lennartz F, Gebel D, arise? New views on the role of bedform interactions within a set Horstman K (1979) Regional studies of sand seas using landsat of boundary conditions. Earth Surface Processes and Landforms (erts) imagery. In: McKee ED (ed) A study of global sand seas, 35(1):51–63 U.S. Government Printing Ofce, Washington, pp 305–397

1 3 Environmental Earth Sciences (2020) 79:347 Page 11 of 11 347

Li Z, Wang N, Cheng H, Ning K, Zhao L, Li R (2015) Formation Wang F, Li B, Jiang S, Niu D, Wen X, Li Z, Si Y, Guo Y, Yang Z, Liu and environmental signifcance of late Quaternary calcareous E (2017) Climate changes at millennium scale in East Asian mon- root tubes in the deserts of the Alashan Plateau, northwest China. soon during OIS3a in Badain Jaran Desert, China. Environmental Quaternary International 372:167–174 Earth Sciences 76(2):71 Li J, Dong Z, Qian G, Zhang Z, Luo W, Lu J, Wang M (2016) Pat- Wang X, Cai D, Sun J, Lu H, Liu W, Qiang M, Cheng H, Che H, Hua tern analysis of a linear dune feld on the northern margin of T, Zhang C (2019) Contributions of modern Gobi Desert to the Qarhan Salt Lake, northwestern China. Journal of Arid Land Badain Jaran Desert and the Chinese Loess Plateau. Scientifc 8(5):670–680 Reports 9(1):985 Liu S, Lai Z, Wang Y, Fan X, Wang L, Tian M, Jiang Y, Zhao H (2016) Werner BT (1995) Eolian dunes: Computer simulations and attractor Growing pattern of mega-dunes in the Badain Jaran desert in China interpretation. Geology 23(12):1107–1110 revealed by luminescence ages. Quaternary International 410:111–118 Werner BT (1999) Complexity in natural landform patterns. Science Lorenz R, Zimbelman JR (2014) Dune worlds: How windblown sand 284(5411):102–104 shapes planetary landscapes. Springer Praxis Books Geophysical Werner BT, Gillespie DT (1993) Fundamentally discrete stochas- Sciences. Springer, Berlin tic model for wind ripple dynamics. Physical Review Letters Mischke S (2005) New evidence for origin of Badain Jaran Desert of 71(19):3230–3233 from granulometry and thermoluminescence dat- Werner BT, Kocurek G (1997) Bed-form dynamics: Does the tail wag ing. Journal of Palaeogeography 7(1):79–96 the dog? Geology 25(9):771–774 Momiji H (2001) Mathematical modelling of the dynamics and mor- Werner BT, Kocurek G (1999) Bedform spacing from defect dynamics. phology of aeolian dunes and dune felds. PhD thesis, University Geology 27(8):727–730 College London Wintle AG (2008) Fifty years of luminescence dating. Archaeometry Ning WX, Wang ZT (2018) Analysis of morphological parameters of 50(2):276–312 mega-dunes in the southeast of badain jaran desert. Journal of Wu JF, Guo F (2012) Geomorphological patterns in a linear dune feld Fujian Agriculture and Forestry University (natural science edi- and ages of the linear dunes in the northern Kumtagh Desert, tion) 47(6):755–763 (in Chinese) northwest China. Environmental Earth Sciences 66(8):2449–2457 Ning WX, Liu XY, Wang ZT (2019) An analytical model for the Yan MC, Wang GQ, Li BS, Dong GR (2001) Formation and growth of growth and migration of a transverse dune. The European Physi- high megadunes in Badain Jaran Desert. Acta Geographica Sinica cal Journal E 42:42 26:83–91 (in Chinese) Parteli E, Kroy K, Tsoar H, Andrade J, Pöschel T (2014) Morphody- Yang X (2001) Landscape evolution and palaeoclimate in the deserts of namic modeling of aeolian dunes: Review and future plans. The northwestern China, with a special reference to Badain Jaran and European Physical Journal Special Topics 223(11):2269–2283 Taklamakan. Chinese Science Bulletin 46(S1):6–10 Ping L, Narteau C, Dong Z, Zhang Z, Courrech du Pont S (2014) Emer- Yang X, Liu T, Xiao H (2003) Evolution of megadunes and lakes in the gence of oblique dunes in a landscape-scale experiment. Nature Badain Jaran Desert, Inner Mongolia, China during the last 31,000 Geoscience 7(2):99–103 years. Quaternary International 104(1):99–112 Pye K, Tsoar H (2009) Aeolian sand and sand dunes. Springer, Berlin Yang X, Scuderi L, Liu T, Paillou P, Li H, Dong J, Zhu B, Jiang W, Savage CJ, Radebaugh J, Christiansen EH, Lorenz RD (2014) Implica- Jochems A, Weissmann G (2011) Formation of the highest sand tions of dune pattern analysis for Titan’s surface history. Icarus dunes on Earth. Geomorphology 135(1–2):108–116 230:180–190 Yang Y, Qu Z, Shi P, Liu L, Zhang G, Tang Y, Hu X, Lv Y, Xiong Y, Schwämmle V, Herrmann H (2004) Modelling transverse dunes. Earth Wang J, Shen L, Lv L, Sun S (2014) Wind regime and sand trans- Surface Processes and Landforms 29(6):769–784 port in the corridor between the Badain Jaran and Tengger deserts, Scuderi L (2019) The fngerprint of linear dunes. Aeolian Research central Alxa Plateau, China. Aeolian Research 12:143–156 39:1–12 Zhang K, Qu J, An Z (2012) Characteristics of wind-blown sand and Sharp RP (1963) Wind ripples. The Journal of Geology 71(5):617–636 near-surface wind regime in the Tengger Desert, China. Aeolian Tang G, Yang X (2010) ArcGIS geographic information system spatial Research 6:83–88 analysis experiment tutorial, 2nd edn. Science Press, Beijing Zhang KQ, Wu ZH, Lü TY, Feng H (2015a) Review and progress Telfer MW, Parteli EJR, Radebaugh J, Beyer RA, Bertrand T, Forget of OSL dating. Geological Bulletin of China 34(1):183–202 (in F, Nimmo F, Grundy WM, Moore JM, Stern SA, Spencer J, Lauer Chinese) TR, Earle AM, Binzel RP, Weaver HA, Olkin CB, Young LA, Zhang Z, Dong Z, Li C (2015b) Wind regime and sand transport in Ennico K, Runyon K, The New Horizons Geology, Geophysics China’s Badain Jaran Desert. Aeolian Research 17:1–13 and Imaging Science Theme Team (2018) Dunes on Pluto. Sci- Zhang ZC, Dong ZB, Wen Q, Jiang CW (2015c) Wind regimes and ence 360(6392):992–997 aeolian geomorphology in the western and southwestern Tengger USGS (2018) Earth explorer. http://earth​explo​rer.usgs.gov/, accessed Desert, NW China. Geological Journal 50(6):707–719 October 1, 2018 Zhao T, Liu W, Xu Z, Liu T, Xu S, Cui L, Shi C (2019) Cosmogenic Walker AS, Olsen JW, Bagen (1987) The Badain Jaran desert: Remote nuclides ($^{10}$Be and $^{26}$Al) erosion rate constraints in sensing investigations. Geographical Journal 153(2):205–210 the Badain Jaran Desert, northwest China: implications for surface Wang T (2011) Deserts and aeolian desertifcation in China. Science erosion mechanisms and landform evolution. Geosciences Journal Press, Beijing 23(1):59–68 Wang ZT, Tao SC, Xie YW, Dong GH (2007) Barchans of Minqin: Zhu JF, Wang NA, Chen HB, Yu DC, Zhang HA (2010) Study on the Morphometry. Geomorphology 89(3–4):405–411 boundary and the area of badain jaran desert based on remote Wang F, Sun D, Chen F, Bloemendal J, Guo F, Li Z, Zhang Y, Li sensing imagery. Progress in Geography 29(9):1087–1094 (in B, Wang X (2015) Formation and evolution of the Badain Jaran Chinese) Desert, North China, as revealed by a drill core from the desert centre and by geological survey. Palaeogeography, Palaeoclima- Publisher’s Note Springer Nature remains neutral with regard to tology, Palaeoecology 426:139–158 jurisdictional claims in published maps and institutional afliations. Wang ZT, Chen TY, Liu SW, Lai ZP (2016) Aeolian origin of inter- dune lakes in the badain jaran desert, china. Arabian Journal of Geosciences 9(3):190

1 3