• 2000 Volume 55 Issue 3
    Published: 15 May 2000

  • Select all
  • LIU Chang ming, CHENG Li
    2000, 55(3): 257-265.
    Download PDF ( )   Knowledge map   Save
    Viewing the more and more serious phenomena of course drying up of lower Huanghe River, taking 1987 for example, the total duration of drying up reached 226 days. Based on the systematic analysis of the influencing factors, two aspects from both water demand and water supply should be considered. And in the aspect of water supply there exist four layers, that is, precipitation, transpiration, runoff, and groundwater, that affect the drying up of lower Huanghe River. Considering the drying up takes place at the dry season every year, base flow must play an important role. Furthermore, the monthly natural runoff from 18 hydrological stations along the main course of the Huanghe River between the 1950s and the 1990s were divided into four series which are analyzed separately with time series analysis. Theses four series are yearly runoff series, wet season runoff series (from July to Oct), dry season runoff series (from Sep to June next year), and the lowest monthly runoff series as base flow. In this way, the dynamic law about the runoff in the Huanghe River basin was brought to light comprehensively and systematically. Some very meaningful conclusions have been drawn. First, the runoff in lower reaches of the Huanghe River comes mainly from the upstream runoff rather than from local precipitation and transpiration. Second, the drying up is mostly caused by human exploitation. Viewing from these points, a uniform basin exploitation and management should be put into practice. The year of 1999 is the best example. In that year because a uniform basin management concerning water resource utilization started to carry out, the course drying up decreased to 42 days. Third, the base flow of the Huanghe River from the very start of the middle reaches has been keeping on dropping in the last 4 decades, which would lead to a more deep thought about reasons and strategic ideas for the lower Huanghe River drained dry and managerial policy. Because the bed at the lower reaches of the Huanghe River has been raised higher than the surrounding ground, it is impossible for groundwater at the lower reaches to flow into the courses. The decreasing base flow in the middle reaches in the past 40 years caused the decrease of base flow, which might partly lead to the drying up. Last, base flow is not directly related with yearly runoff of the Huanghe River, since they have different periods. Finally, orthogonalization & selection model could be applied satisfactorily into the forecasting of runoff in the Huanghe River. The finial predicting function is just as following. q[DD(-3/4]^(N+t)=φ 0+∑ki=1φ if′ i(N+t) t=1,2,… Here, q[DD(-3/4]^ is the predicting runoff; f′ is mean generation function; and φ 0 and φ i are coefficients. With the model, the similitude precision could be about 90%. For example, the calculated runoff at the monthly lowest in 1987 is only 12 03 hundred million m\+3 per year, which is so low that its anomaly percentage is 14 1%. The result accords with the real drought background, which made the course drying up worst in 1987. In one word, both human and natural factors exert impacts on the courses drying up of the lower Huanghe River to different degreed and from different ways.
  • XU Shi yuan, SUN Yi yi
    2000, 55(3): 266-273.
    Download PDF ( )   Knowledge map   Save
    The concept, theory and research method of geomorphology develops rapidly due to the development of modern science and technique. 1997’s computer simulation and analysis are introduced into geomorphologic research. It makes the research methods move from qualitative description to quantitative expression. This paper introduces a new type of geomorphologic simulation based on the computer graphic theory. It focuses on the geomorphology topic-topography and intends to reduce the parameter definition difficult in the physical models and avoid the non spatial characteristics of statistic models. A new theoretical system is established. Research work included: (1) Design integration system, which contains both digital models and mathematical models. Digital models are a presentation of the real world status, while mathematical model is a presentation of the real world evaluation processes. (2) Use the Digital elevation model as the presentation of 3 dimensional topography. By the 3 dimension concept to 4 dimension expanding we developed the geomorphologic temporal spatial database. Thus make the geomorphologic simulation base on the information system technique. (3) Apply the fractal model. Besides the present continuous process modeling, this model provides a non continues process modeling method. Also, fractal model is the tool to analysis the chaotic set in the dynamic system. The use of fractal model will be important in the further research. (4) By using the new object oriented programming language, 32 bit memory mode and multi thread concept and the interactive GUI of Windows, etc., the 3D visualization tool, profile analysis tool and accumulation-erosion analysis tool are developed. They have high performance efficiency, high analysis functions and high visual effects, and are easy to learn and easy to use. (5) Implement the software in the southern part of Changjiang Estuary and verify the theory and technique of the computer aided geomorphologic simulation. The application of this method has significance in the geomorphology, environment research and decision making support. This study is not the simply application of computer technique or system integration. The result of the work expands the theory and method of multi discipline. The study creates a new research area of geomorphologic simulation. In the real world application, it provides a new viewpoint to the geomorphology research. It will make great contribution to geomorphology modernization process.
  • XU Jiong xin, ZHANG Ou yang
    2000, 55(3): 274-280.
    Download PDF ( )   Knowledge map   Save
    Based on the data from 159 flood events at the Tongguan station on the Huanghe River, we have found the rule to describe the complicated relationship between the fill scour features and the event maximum suspended concentration ( C max ). When the C max is low, the depth of scour is large. With an increased C max, the depth of scour decreases and channel fill occurs. However, when the C max further increases and enters the ranges of hyperconcentrated flows, scour occurs again. Controlled by the foregoing complicated fill scour behaviors, the channel adjustment of the Huanghe River is also complicated. When the suspended concentrations are low, channel scour occurs, and thereby the channel shape tends to be narrower and deeper, and the thalweg more sinuous. When the suspended concentration increases, channel fill occurs, leading to an increase in the channel width depth ratio and a decrease in the channel sinuosity. However, when the suspended concentration further increases and exceeds a certain threshold, channel scour occurs again, and the adjustment of channel morphology turns to the opposite direction. The rules outlined in the paper for the channel adjustment of the lower Huanghe River may provide a new thinking for the regulation of the channel morphology of the lower Huanghe River after the closure of the Xiaolangdi Dam, through the regulation of the lower Huanghe River’s water and sediment by choosing some specific operation modes of the reservoir. Either by releasing clear water or by releasing man made hyperconcentrated flows, the downstream channel may tend to be narrower, deeper and more sinuous. If some proper engineering measures are taken afterwards, the channel conditions for a better flood releasing and sediment transport may be greatly improved.
  • WANG Shao wu, GONG Dao yi, YE Jin lin, CHEN Zhen hua
    2000, 55(3): 281-293.
    Download PDF ( )   Knowledge map   Save
    The seasonal and annual precipitation series of 35 stations over eastern China were established using the observations as well as proxy data. All the time series cover the period 1880~1998. During the period 1880~1889, the observed precipitation records are available only for several stations. The huge gap in the data were filled using the historical materials which are used in the past climate reconstruction fruitfully. During the period 1900~1950, the Precipitation Grade Map was applied, and some data gaps were resolved using the documentary evidence too. Since 1951, precipitation observations have been available. During the 3 different periods of time, the observations account for 22 6%, 69 0% and 100% of the total amount of data respectively. All documentary evidence used in this paper is transformed into 5 precipitation categories for every season. The standard is the same as that of used by the China Flood/drought Category Map for the Last Five Centuries, i.e., category 1 is heavy flood, 2 is flood, 3 is normal, 4 is drought and 5 is heavy drought. The frequency of the categories is 1/8 for categories 1 and 5 and 1/4 for the three others. Then, all precipitation categories are transformed into the precipitation amount by the specific functions determined using the recent observed data. Thus, the seasonal precipitation data of all the 35 stations are obtained, and added into the annual amount in the order of spring, summer, autumn and winter. And the possible errors in the transformation are estimated, in averaging less than 28% of the actual rainfall values. The variability of the annual precipitation is preliminarily analyzed. There are no obvious trend, only +0 1%/100a when calculated using the data of whole period 1880~1998. However, there are significant low frequency variations for both the seasonal and annual precipitation series as shown in Figure 2. Power spectral analysis demonstrated that there are evident cycles of 3 3a and 26 7a. The annual time scale period of 3.3a may be related to the ENSO, and the period of 26 7a suggests that there is strong interdecadal variability in the precipitation of China. The significant rainfall variations in the past several decades may be caused by the interdecadal changes, and have no strong association with the global climate trends as expected in some diagnostic and modeling studies.
  • GUO Lai xi, WU Bi hu, LIU Feng, FAN Ye zheng
    2000, 55(3): 294-301.
    Download PDF ( )   Knowledge map   Save
    Tourist resources are the material basis of sustainable tourism and the potentiality of tourism economic growth. As the working criterion and technical guide for investigating tourist resources, the classification system of tourist resources is a very important premise of the work. Based on the criterion for investigating Chinese tourist resources, the new classification system was put forward. The newly added are three serials of natural, human and service; ten types of geologic landscape, water scene, climate and biology, other natural scene, historic heritage, modern human attractions, abstract human attractions, other human attraction, service resource and other service resources; plus ninety eight patterns. There are three grades to indicate the scale of monocase of tourist resources, which are field, segment and site. The tourist resources also can be divided into three states: developed, to be developed and latent. Based on this new classification system, more consummate characteristic item was revised, and the quantitative evaluation of types tourist resources was carried out by taking the Beihai City as an example to validate this system and methods. In the end, the paper points out that in order to achieve sustainable tourism, China must consummate the criterion for investigating tourist resources, and expand investigations to provide a series of tourist extractive engineering for building China into a country even more attractive to tourists in the world.
  • CHEN Jianyao, YU Jingjie, LIU Changming
    2000, 55(3): 302-308.
    Download PDF ( )   Knowledge map   Save
    Water transfer is a measure to solve water shortage problem, which must be based on efficient utilization of local water resources. Also, water use and water transfer has much to do with population, resources, environment and development (PRED). Only if all these four aspects and their coupled effects have bee considered, would the water transfer scale for different goals such as the basic survival need, the highest economic output, the best eco environmental effect, the most stable social security and sustainable development, and a relatively satisfactory result considering all the above aspects be given. So, PRED synthetic analysis is a very important approach to discuss how the water resources are used and how much water needs to be transferred for a region. PRED synthetic analysis is human oriented. According to the economic development level and the importance of food security in China, the analysis flow of water demand in a region is firstly to consider water requirement for human survival, production and development which is fundamental, then to consider water demand in the human survival environment. By thoroughly diagnosing water demand in every aspects related to population, resources, environment, and development, problems can be found out and the strategies and countermeasures for solving the problems can be given so that water resources management can be best served. The index value for PRED synthetic analysis with different levels such as high, medium and low should be carefully choosen based on local social and natural conditions and be varied with scientific improvment and readjustment of rules and regulations. The analyzing unit should not be too big, in case when a uniform index within an area is used to do comparative ananlysis. This paper takes water resources use in Henan province as a case study and presents the result and use strategy.
  • LI Li juan, ZHENG Hong xing
    2000, 55(3): 309-317.
    Download PDF ( )   Knowledge map   Save
    In this paper, stochastic hydrologic methods are applied to analyze the characteristics and driving forces of annual runoff changes of Chaobaihe River, which is regarded as a typical river in North China. To find out the trend of annual runoff changes, the Kendall method was adopted. The results showed that the amount of annual runoff in the Chaobaihe River tends to decrease exponentially from the 1920s to the 1990s. Moreover, with the application of change point searching technology, it can be concluded that the runoff change history of the river can be divided into five periods, i.e., before 1960, 1960~1970, 1971~1980, 1981~1990 and after 1991. Besides that, the auto correlation function has shown that the runoff series is not an independent one. The auto correlation coefficients of its first, second and third orders are all quite significant. Furthermore, from the annual runoff records, it is found that since the phenomena of flow drying has become more and more serious in the region, the Chaobaihe River tends to be a seasonal river. In order to find out the driving forces of the annual runoff changes, we have analyzed the relationship between precipitation and runoff. Through correlation analysis, it is quite clear that the change in annual runoff was partly dependent on rainfall. However, the rainfall is not the only major driving force of the changes, for changes in annual runoff and rainfall are not entirely the same. Comparing the history of water resources utilization and development with change in annual runoff, we found that the annual runoff responded directly to human activities, especially after 1960. According to the investigation, it was obvious that the annual runoff change is mainly caused by human activitives rather than under the impact of the climate change, though the climate serves as the background of the change. Since the relationship between rainfall and runoff was quite well before 1960, a regressive function is built based on the series of rainfall and runoff from 1920 to 1960. And then, the function is used to extend the series of natural runoff by relying on the records of rainfall after 1960. Thus the estimated rates of water utilization and development of different periods were calculated. The results showed that the rate of water utilization and development has been increasing continuously and has been far beyond the limited rate of 40%. It brooks no delay to control water resources development and utilization in the Chaobaihe River.
  • WANG Zhi yong, WANG Jin feng, YU Jing jie, LIU Chang ming
    2000, 55(3): 318-328.
    Download PDF ( )   Knowledge map   Save
    This paper gives the method for computing the marginal revenue of water in industrial sector and agricultural sector and illustrated the process of data treatment taking Baoding district as an example. Before calculating the marginal revenue of water, a lot of work has to be done to make the production conditions comparable. For the sake of data restriction, we take the fixed capital as the primary consideration while calculating the marginal revenue of water in industrial sector. The fixed capital always coordinates with a certain amount of floating capital. Thus, to some extent, change in fixed capital stands for the total capital amount in work. We also consider the elasticities of output to capital and to labor respectively, which we calculate as the weight in computation. We assume the elasticity of output to capital equals to 0 91, while 0 1 as the elasticity of output to labor. Therefore, the weight of each calculation is around 1 09, which equals to the reciprocal of elasticity of output to labor. While in calculation, we use the following formula: MR=Y i+t ∏tk=1g i+k ·w t-Y iX i+t -X i where Y is the total output in industrial sector, X is the water amount used in industrial sector, g is the growth rate of fixed capital investment and the w means the weight got from elasticity output to capital and to labor. We assume that there is no delay in capital investment, which means all the investment will be in force right after the plan is fulfilled. Actually, if data is available, we should consider the time lag of investment. As referring to the calculation of marginal revenue of water in agriculture, we use similar means to compute. Taking the agricultural sector as a whole, we consider the total mechanic force as the fixed capital in agriculture, which measures the primary agricultural devices. Besides, we consider the natural disasters that affect the output in agriculture. There also exist the elasticities of output to capital and to labor. Thus we have the weight to reflect the elasticities and natural disasters. Using the similar way, we can also calculate the aggregated marginal revenue of water both in industrial and agricultural sector. The conclusion of our computation and analysis is that the marginal revenue of water is diminishing, regardless the sectors that use water. In general, industrial sector has the higher value of marginal revenue of water than agricultural sector. Different districts have different average marginal revenues of water due to differences in capital amount.
  • XIA Zheng kai, DENG Hui, WU Hong ling
    2000, 55(3): 329-336.
    Download PDF ( )   Knowledge map   Save
    The Xar Moron River Basin is located in the interlock area of farming and pasturing, Northern China, where the present climate is semi arid, and the ecosystem is very sensitive to climatic changes and human activities. Based on geomorphologic observations and archeological excavations, the authors analyzed the relationship between geomorphic evolution and prehistoric cultural development during the Megathermal in Xar Moron River Basin.
    The main geomorphologic processes were loess sedimentation, gully aggravation, degradation, deserts spreading and retreating during the Megathermal in this area. This had exerted important influences on the changes of prehistoric cultures in the Xar Moron River Basin.
    The Xinglongwa Culture (8000~6500aBP), which is characterized by hunting and collecting, and the Xiajiadian Upper Layer Culture (3300~2800aBP), which is characterized by Animal husbandry, can be discovered in the whole Xar Moron River Basin. Since loess is beneficial to agriculture, the Zhaobaogou Culture (6500~6000aBP), Hongshan Culture (6000~5000 aBP) and Xiajiadian Lower Layer Culture (4000~3300aBP), which have characteristics mostly in agricultural cultures, are found only in the Loess region.
    There is a close relationship between alluvial landforms and localities of culture sites. Human changed habitats along with gully degradations. The Xinglongwa Zhaobaogou sites are located on the loess tablelands. The Hongshan sites are distributed on the loess tablelands and are discovered in the sediments of the second terraces of the Xar Moron River. The Xiajiadian sites are on the surface of the loess tablelands and the second terraces, the relics of the sites also can be discovered in the sediments of the first terraces. The Liao Dynasty sites are on the surface of the first terrace.The climate changes during the Holocene resulted in the spreading and retreating of deserts in this area, causing the rise and fall of prehistoric cultures. The first expansions of desert (5000~4000aBP) resulted in the fall of Hongshan culture. The second desert expansion (3300~2800aBP) caused the transition from agricultural culture of Xiajiadian Lower Layer to nomadic culture of Xiajiadian Upper Layer.
  • XU Shang ping, TAO Shu, XU Fu liu, CAO Jun
    2000, 55(3): 337-345.
    Download PDF ( )   Knowledge map   Save
    Soil samples representing various soil horizons (A, B, and C) were collected from 349 sites in Inner Mongolia Autonomous Region and the total contents of 10 trace elements, namely copper, lead, zinc, cadmium, nickel, chromium, mercury, cobalt, vanadium and manganese were measured with atomic adsorption spectrophotometry. Preliminary data analysis showed that the concentration of each element conforms to logarithmic normal distribution, therefore, logarithmic transformation was performed in order to reduce the possible scale effect. Factorial Kriging Analysis (FKA) was then applied to analyze the spatial correlation pattern among various trace element contents. As Principal Component Analysis (PCA) showed that the primary two components account for the main source of variance, a series of rotation methods have been used and spatial correlation patterns of these two primary factors under different rotation schemes were compared with respect to spatial orthogonality. Since rotations increased significantly the spatial dependency between the two components in this case, the principal components without any rotation were deployed as the basis of FKA. It is further demonstrated by computing and comparing partial β values that parent materials and soil class are the primary factors controlling the concentrations and spatial distribution patterns of the trace elements in the soils of the studied area. The variogram of the primary factor, which mainly represents the effect of parent materials, consists of nugget effect and two spherical models with ranges of 200 and 400 km, respectively. A linear model with a 1 000 odd km range and nugget effect, on the other hand, constitutes the variogram model of the second component, which can be referred to as the representative of soil class. Based on the variogram model of the principal components, the variogram of the content of each element is therefore modeled by manual coefficient adjustment under the constraint of Schwarz inequality. Although the influence of the principal factors varies among the spatial distribution patterns of different elements, viewed from a comprehensive perspective and compared with the effect of parent materials, the role of soil class is much weaker. The variograms of the contents of one element in different soil horizons share the same primary pattern, while there exists such a trend that the influence of parent materials is increasing from top to bottom soil. Under the assumption that the theoretical covariance between experimental errors of different elements is zero, by checking carefully the coefficient matrix, it was concluded that the nugget effect in the variogram models of most of the trace elements was mainly composed of small range variation except for lead, cadmium and mercury. The effect of experimental errors of the latter three elements seems to be much more noticeable. Spatial correlation analysis was preformed to examine the relative correlation among the trace elements at various scales defined by the variogram models of principal factors. The result demonstrates that the concentrations of copper, zinc, nickel, chromium, cobalt, vanadium and manganese showed significant positive correlation, while lead, cadmium and mercury are more or less independent. This spatial correlation did not show any significant variations from top to bottom soil, or from small to large scale. Considering the fact that the parent material is the primary influencing factor of soil trace element contents for our case, the ion radii of each trace element were examined and compared with each other. As the radii of mercury, lead and cadmium are remarkably larger than those of the other elements, isomorphism during the endogenic action, which is controlled by the radius of ion of the trace elements, provides the most possible interpretation.
  • WANG Ying, WANG Zheng
    2000, 55(3): 346-353.
    Download PDF ( )   Knowledge map   Save
    Location is a common appearance in geography. In 1826, Von Thünen put forward the Agricultural Location. In 1909, Weber studied the Industrial Location. After World War II, the study of the location of the third industrial, the Facility Location, became the focus of locational theory. But in the Facility Location theory, the facilities are some public service facilities, such as the hospital, the store, and the library. They are different from the scenic spots. The tourist attractions are unchangeable. So the Tourism Location is different with the Facility Location. The time of visiting and travelling is the key elements that influence the decision of the tourists. The marginal utility of the scenic spots, f, is defined, f= Δ T y/ Δ T x, where T y is the time of visiting in the scenic spot, and T x is the time of travelling from the tourism distributing center to the scenic spots. Based on the study of distribution and development of the scenic spots of Yunnan province, the Tourism Location is got. From the distributing center to the outer region, four rings appear: ① historic sites, ② fancy scenic spots, ③ tourists stagnation, and ④ folkways of minority nationalities & beauty spots. The location is similar to the Thünen ring, but it is decided by the marginal utility of the scenic spots, f, not by the product profit. Because different kinds of scenic spots provide tourists different visiting time, T y, and the tourist attractions are unchangeable, only T x of the scenic spots in their relevant location can satisfy the consumptive psychology of the common tourism. The tourist location also existed in other part of China, but its scale is larger. Because of the Tourists Stagnation, tourism is not suitable to be developed in some regions.
  • CHEN Hao
    2000, 55(3): 354-363.
    Download PDF ( )   Knowledge map   Save
    The relationships between erosion and sediment yield, and the effect of rainfall runoff and silt laden runoff on sediment delivery ratio of drainage systems in the loess gully hilly areas are discussed based on data of Dali He drainage basin system. The characteristics of space time distributoin of delivery ratio in the basins of various scales and the power mechanism of single rainstorm event with flood due to increase and decrease of flood peak or runoff depth are illustrated by analyzing more annual single rainfall events. The results are as follows: (1) that the delivery ratio is close to 1 for long term (average value of more annual rainfall) while that estimated to be >1 or <1 for short term (single enent or annual rainfall) for various events in Dali He drainage basin. This phenomenon will frequently occur as a result of sediment retaining as well as re erosion and delivery of the retained sediment in a short period of time. (2) Sediment delivery ratio is closely related to the spatial distribution of rainfall magnitude of rise and fall of peak flood and that of runoff depth in the basin system. Changes of mechanism of delivery ratio ( D r ) can be explained by runoff depth ratio ( H b ) or rise of flow shear force ( τ ) with falling of runoff depth in the basin system. The flow shear force can be described as: τ=γ mhJ as a unit drainage basin τ=γ m(H b)hJ or as accretion of basin areas (3) Delivery ratio ( D r ) is closely related with runoff depth ratio ( H b ), single event rainfall distributing ratio ( P b ) and change amplitude of peak flood ratio ( H f ) in the basin system. The interrelation can be expressed as: D r=0 403P b 0 37 H b 1 066 H f 0 191 r=0 996 n=125 in Chabegou (4) Delivery ratio is close to 1 as H b is 1 due to even distribution of rainfall runoff; D r is >1 as H b >1 for flow kinetic energy increase and D r is <1 as H b <1 for flow kinetic energy decay.
  • SHEN Xiao ping
    2000, 55(3): 365-374.
    Download PDF ( )   Knowledge map   Save
    Regional geography is one important part of geographic education and research. In the Era of globalization, the demands for regional geographic courses have increased significantly. The importance of regional geographic education has drawn much more attention, especially. Those regions of emerging economies such as Asia. This paper exams the transition of regional Educationin American universities during the last three decades (1965~1998) using Asian geography courses as a case study. Transition and development of regional education in the United States are influenced by two major factors: position of geographic education in the over all higher education, and the importance of the region in the world and the relationship with the United States. Asia is one of the most dynamic regions in the world and draws increasing attention in the last 30 years. Asia has become the largest trade partner of the United States since the late 1980s. These changes have brought a rapid increase in enrollment of Asian geography courses during the past ten years. Compared with other regional courses, such as Europe and South America, this research studies the transition and spatial distribution of Asian geography courses in American universities.
  • XUE Xian, ZHANG Wei min, WANG Tao
    2000, 55(3): 375-383.
    Download PDF ( )   Knowledge map   Save
    For resolving the wind erosion problem that has menaced the Dunhuang Mogao Grottos for a very long time, this paper studies the inhibiting effects of gravel mulch on wind erosion. Wind tunnel experiments and field surveys have been used to support these findings. At first, we divided the Gobi bed of Mogao Grottos districts into three zones according to the gravel cover and wind states, then took photos and some samples for examination and testing in different zones. According to analytic results of the size distribution and gravel cover density, we defined three kinds of beds as stable, semi stable and unstable. The experimental work for this paper was conducted in a straight line puff field wind tunnel in Shapotou. The tunnel was 37 m long and a 21 m long working section had a cross sectional area of 1 44 m 2 (width 1 2 m, height 1 2 m). The test beds were located 16~20 m downwind of tunnel inlet of the working section. The Pitot tube measuring profiles of wind velocity was installed in the middle of the test bed area at the 19 8 m mark. The Pitot tube was moved along a slide to the following positions of 4, 10, 30, 50, 100, 200, 250, 300, 450 and 500 mm above the bed surface. The wind tunnel experiment consisted of two parts: (1) the relation of gravel shape, height and roughness length; (2) the characters of the particle size distribution of the stable bed surface. In the experiment, there were two kinds of wind states (with and without sand) and five kinds of wind velocity (6, 8, 10, 12 and 15 m/s). We came to these conclusions: (1) In the same gravel cover states with low wind velocity, gravel shape affects roughness length more than gravel size, and vice versa. (2) The two basic characters of stable bed surface: >65% gravel cover and specific particle size distribution. In addition, a strong stable bed surface can inhibit erosion by the low velocity wind containing sand to some extent, but when high velocity wind containing sand goes across a bed surface, the bed surface will change from stable state to unstable state quickly. Finally, according to the findings in the Mogao Grottos Gobi and wind tunnel experiment, we suggest that: (1) for keeping the bed surface stable, it is important to cut off the source of sand and form wind without sand; (2) in the district Ⅱ, crushed gravel (>10 mm) should be paved in order to prevent strong wind erosion; due to the low wind velocity in district Ⅲ, 5~10 mm small gravel should be in higher ratio than >10 mm gravel; (3) to prevent gravel from being driven by sand salutation it is also imperative to control the particle distribution according to the appro priate standards; and (4) human activity must be control in order to keep the bed stable for a long time.