Water content distribution in the surface layer of Maoping slope

Science in China Ser. E Engineering & Materials Science 2005 Vol.48 Supp. 157- -170157Water content distribution in the surface layerof Maoping slopeLIU Yuewu, CHEN Huixin, LIU Qingquan, GONG Xin, ZHANG Dawei& L LianxiangDivision of Engineering Science, Institute of Mechanics, Chinese Academy of Sciences, Beiing 100080,ChinaCorrespondence should be addressed to Liu Yuewu (email: lywu @ imech.ac.cn)Received July 1, 2004; accepted November 9, 2004Abstract The water content distribution in the surface layer of Maoping slope has beenstudied by testing the water content at 31 control sites. The water content profiles at thesesites have also been determined. The water content distributions at different segmentshave been obtained by using the Kriging method of geostatistics. By comparing the watercontent distributions with the landform of the slope, it was shown that the water content isclosely dependent on the landform of the slope. The water content distribution in thesurface layer provided a fundamental basis for landslide predication and treatment.Keywords: landslide, water content, in situ test, Kriging method, geostatistics.DOI: 10.1360/04zze191 IntroductionPre-determination of the water content distribution of a landslide is necessary for cal-culating the landslide volume weight and judging the position of weak intercalary stratain the landslide. It is significant for further investigation of the landslide mechanism, theprediction of occurrence and the control of landslidest- 31. In recent decades, remarkableadvances in water content testing techniques for individual points and local average havebeen achieved along with considerable progress in automatic measurement techni-ques'[4-91Generally, there are three kinds of water content determination methods: (i)Field point measurement methods: these methods include the direct testing method, dry-ing weighing method, time domain reflectometery method, neutron radiation and otherindirect testing methods. The advantage of the field point measurement method is thatthe results possess relatively high precision. The disadvantage is that either the testingsample is small or much tiresome works needs to be done to define the water content fora relatively large region. (ii) Telemetry methods: these mathnds ran he Hivided into ac-tive and passive microwave measuring techniqu中国煤化工; sensing andradar testing methods. Their advantage is that thYHCNMHGover1-1000.Copyright by Science in China Press 2005158Science in China Ser. E Engineering & Materials Science 2005 Vo1.48 Supp. 157- -170km2. However, only the average water content of the whole region can be determined bythese methods. (ii) Numerical simulation methodst0- 13): these methods are based on thegeological and mechanical parameters of the slope layers. The methods partly depend onthe test results of the above two methods, so they are only an evaluation method and donot belong to the range of field tests. There exists a range (about 0.1一10 km) in whichit is difficult to obtain in situ testing results by the first and second methods mentionedabove. It was indeed within this scope where numerous landslides and other geologicalhazards happened. How to determine the water content in this scope is one of the hottopicsl14- 22 in agnicultural engineering, hydraulic engineering and hydrologic geology. Itis also one of the issues far from being resolved, including how to deduce the watercontent at untested points and the space distribution of water content based on the datafrom the tested points.Maoping slope was selected to assess the field test and to evaluate the method of wa-ter content distribution. Through more than two years of geological investigation andfield tests, 31 test sites were selected to test the water content in the surface layers byusing the drying weight and direct measurement methods. The Kriging geostatisticsmethod and other analytical methods were employed to determine the water content dis-tributions in the surface layer of Maoping slope based on the in situ tests. The watercontent distribution in the intermediate range has been obtained by combining the fieldtests and numerical analysis, and it provides the basic data for the stability analysis ofMaoping landslide.2 General geography and landform of Maoping slopeMaoping slope is located on the left bank of the Qingjiang River, 66 km above theGeheyan hydropower station and 25 km below the Shuibuya hydropower station. Ma-oping slope is the largest ancient landslide in the Geheyan Reservoir region, with alength of about 1600 m, a maximum width of 600 m, a thickness of 8- -86 m and a totalvolume of about 2.350x10' m'. Fig. 1 shows the profile and the geologic sketch of Ma-oping landslide in a birds-eye view. If the sliding body goes down at high speed, theQingjiang River would be jammed, and the normal operation of the two hydropowerstations would be affected. Additionally, Maoping slope is an ancient stack on bedrock,and this kind of slope is widely distributed within. Three Gorges Reservoir region. TheGeheyan Reservoir has been operating for more than ten years and its operational modeis similar to that of the Three Gorges reservoir. Hence, further study of the Maopinglandslide will be valuable for the evaluation and control of landslides in the ThreeGorges reservoir area.The early investigations suggested that Maoping slope is relatively stable but that lo-cal collapse or local sliding may occur in the leading edge near the Qingjiang River afterwater storage. The operation of the Geheyan Res中国煤化工j the maximumwater level was reached in the next year. SinceYHC N M H GMaoping slopeCopyright by Science in China Press 2005Water content distribution in the surface layer of Maoping slope159has been monitored by the Changjiang Water Conservancy Committee.241.The moni-toring data indicated that the surface deformation was very small from 1993 to 1995, butthe deformation velocity subsequently increased. Up to 2001, the maximum displace-ment of the sliding body was 2100 mm and the annual avcrage displacement was about25 cm. So the abundant monitoring data is useful and very important to probe themechanism of the initiation and evolution of the landslide.a)Fig. 1. Diagram of Maoping slope in the Qingjiang River. (a) Photo of Maoping slope; (b) geologic ichnograph ofMaoping slope.3 Test approach and instrument3.1Drying weight methodThe drying weight method is widely used to measure soil water content. Its basicprinciple is to dry the collected soil sample in an oven and to calculate the water contentby measuning the weight difference of sample before and after drying. The calculationformula is as follows: .Water ratio (%)=.weight of wet sample - weight of dry sample,x 100%,(1)weight of wet sample - weight of boxwhere the weight of wet sample is equal to the sum of the weight of the box and thesample weight before drying; the weight of dry :中国煤化王of the weightof box and the weight of the sample after drying.TCHCNMHGThe drying machine used in the experiment 1s DHG-9UZ3A electnc constant-tem-www. scichina.comScience in China Ser. E Engineering & Materials Science 2005 Vol.48 Supp. 157--170perature oven (Fig. 2 (a)). The controllable temperature range is 5C- 200°C. An elec-tric scale (made in Germany) with a precision of 0.01 g is also employed (Fig. 2 (b)).Fig. 2. Instruments used in drying weight method. (a) DHG 9023A electic constant-temperature blasting dry box; .(b) electric scale with the precision of0.01 g.3.2 Direct measurement methodThe direct measurement method directly measures the water content of the test stra-tum with a soil moisture tester. The measuring accuracy of the soil moisture content de-pends on the measuring principle and the accuracy of the tester. The thetaprobe soilmoisture content measuring system made by Eijkelkamp Company of Holland was em-ployed in the experiments. ML2x soil moisture tester and HH2 soil moisture monitor(Fig. 3) are included in this measuring system. The system employed the analytical tech-nique of frequency domain (TDR) to obtain the soil moisture content by measuring thevariation of the electric medium constant and translating it into electric voltage signals.The sensor envelope is waterproof and the four probes penetrate easily into the soil. The中国煤化工Fig. 3. Thetaprobe soil moistureMYHCNMHGCopyright by Science in China Press 2005Water content ditribution in the surface layer of Maoping slope161water content range of this test system is 0% - 55% and the standard calibration preci-sion is 2%. The signal outputis 0- 1 V DC.The soil moisture at the test point was measured segment by segment during the ex-cavation process at the test sites.4 Field test and results4.1 Field testBased on the survey of the geological features and the landform of Maoping slope, 31test sites were selected scattered over the whole sliding body. Fig. 4 shows the distribu-tion of the selected sites on the slope.Fig.5 is the excavated profile chart at one test site. The above-mentioned two meth-ods were adopted to determine the soil water content at different depths throughout theexcavation process. The soil samples were also taken back to the operating room. Afterbeing dried for 24 h at constant temperature (105"C), the sample was weighed and thewater content of the test sample was obtained by using formula (1). Fig. 6 shows the testresults of the water content profile at some test sites.4.2 Comparison of the test resultsTo verify the validity of the testing system and the reliability of the test results, thetest data obtained by the two different methods were compared (Table 1). The maximumrelative error is about 3..343%, which is basically the same as the errors described inmost references. Because the error margin is less than 3.5%, both of the methods cansatisfy the engineering accuracy requirement.8002.●9600-●6●12。19县400100。14 18.21●2413 ]7 20302002316● 22*628029。31~050010001500X/m|中国煤化工Fig. 4. Diagram of test si:MYHCNMHGwww.scichina.com162Science in China Ser. E Engineering & Materials Science 2005 Vol.48 Supp. 157- 170Fig.5. Excavating profile chart in one test site.3:35中3恩30g 2:2g20t点15t0tt 10b)510.152025303505101520253035404550Depth/cm404，3:4(3525820g212010051015202530354045505560010203040506()7()5[50 r++++940I330 t!1(e三1005 10172227 3237 43 48 54 606671021 27 34Fig.6. Testing results of the water content profiles at some test中国煤化工t the test site No.9;(b) water content profile at the test site No. 26; (c) water contd3; (d) water contentprofile at the test site No. 13; (e) water content profile at the tesYHCNMHGrorileattheestsieNo.19.Copyright by Science in China Pess 2005Waler content distribution in the surface layer of Maoping slope163Table 1 Comparison of the test results by two measuring methodsWater content (%)SampleRelativeNo.Drying weightTDRerror (%)Drying weight TDRerTor (%)method123.122.81.2991127.326.72.199226.125.90.7691227.727.21.805319.520.13.0771327.50.72724.425.23.2771420.420.50.49031.630.80.6331S32.931.83.34337.237.10.2691634.333.62.04136.635.91.9261736.51.61732.332.10.6191834.50.58032.431.91.54335.534.91.6901(1.4712036.81.0755 Theoretical basis of the test data analysisIn general, only a few of the water Content values can be obtained from the field test,so it is necessary to use analytical or numerical methods to evaluate the water contentdistribution in the whole slope. Some geo-statistical methods can be used to describe thecharacteristic parameters of the heterogeneous stratum. The Kriging method is one ofthem.5.1 Ordinary Kriging methodOrdinary Kriging is the most widely used Kriging method. It is used to estimate avalue at a point of a region for which the variogram is known, without prior knowledgeof the mean. It can be used to determine the spatially distributed data by using the opti-mized, linear and unbiased interpolative estimation method. The method introduces theweight cofficients ha (c = ,2,.-,mn) for each sample to estimate the variation function.By considering the data of each sample Za(a= 1,2,..,n) and the space structure, thegeometric characteristic parameters are obtained.The unbiased estimation condition is(2)立石=The estimation variance is given by暗=E(V,V)-2ExECV,a)+ ZZEgCv."g).(3)a=la=lB=lThat is, to evaluate the weight coefficient●nthe, pstimate variance中国煤化工σ至must reach a minimum value under the unbiaYHCNMHGwww.scichina.com164Science in China Ser. E Engineering & Materials Science 2005 Vol.48 Supp. 157- 170Let F=σ名-2u|(二的-一)one can get the partial derivatives vs ha(Q= 1,2,.,n)and u for F. And then, let the partial derivatives equal 0, the ordinary Kriging equationscan be derived as:|之AypC(va,ya)-μ=C(w,V),β=!(4)Sa=1, (c=,2...n).a=1The estimation variance of the ordinary Kriging equation isσg =C(V,V)-.Z nqC(V,va)+μ. .(5)a=l5.2 Lognormal Kriging methodThe lognormal conservation supposes that when sample values appear to follow alognormal distribution, the sample average values follow a lognormal distribution andtheir combined distribution also remains a lognormal distribution. However, the linearcombination of lognormal distributions disagrees with lognormal conservation.It is suggested that the regional parameter xa defined on the information carrier obeysthe lognormal distribution, where Z is the mathematical expectation, C(vq,va) is thevariance, namely Z=E{x}，D2(xa)=C(va,va). If y=ln(xa) agrees with thenormal distribution, its mathematical expectation is Z = E{ln(xa)}, the variance isC(va,va)= D2 {n(xa)}. The relationship of these two distributions is:(6)2C(va,'a)=Z^[expCE(Vax"a)-1](7)When the regional parameters x0s xp defined on the same information carrierVa,vp(a,β=1,2.,n), obey the combined lognormal distribution with the mathe-matical expectation Z and the covariance C(va,vp)， ln(xa) and ln(xp) agree withthe combined normal distibution with the covariance C(va,vg). Their relationship is中国煤化工C(vanVg)=z2[expC.((8)YHCNMHGCopyright by Science in China Press 2005Watcr content distibution in the surface layer of Maoping slope165With the lognormal Kriging method, it is suggested that if Zv is the average value ofthe sample and Zv indicates the evaluated value, In Zv must be the linear combina-tion of ln(Za)(a= 1,2,.,n):InZiv=C+,之名ln(xa),(9)where C, ha are the undetermined coefficients, xa is defined on the infommation carrierva(a= L,2...,n) based on the observed values of n samples.Similar to the ordinary Kriging method, the lognormal Kriging equations are2qgC(va.Vg)-u=(vw.V),」β=t(10)The matix form is [k,I]=[M,]. .Eq. (10) is linear with n+1 orders. By solving the equations, weight coefficients(a=1,2,,n) and u can be derived. The abbreviated formula of C can be expressedas(11)The unbiased, linear and optimized estimator z; of Zv can be written as2 =exp{(12)where the estimated varianceof Z° isoσ =Z {expC(V,V)+exp- 2exp(13)l _a=1_a=]6 Data analysis and discussionThe data obtained from the field test were analyzed and evaluated with both the ordi-nary Kriging and lognormal Kriging methods to determine the water content distributionin the surface layer of the whole Maoping slope.中国煤化工:nt contours indifferent segments of the surface layer in MaopingYHCNMHGwww. scichina.com166Science in China Ser. E Engineering & Materials Science 2005 Vol.48 Supp. 157- -17080070025 004950029.0811 29.85340029.853 348949,,89493002001000001500X/m31.7931706030.67的Es00 E29.3383)1 30.67]30671” 400E31.79330.062529.3383一上o1000600 .: 50030.824630.5538E37.882527.6333 25.88725.8876Aued(C)031 ,7336600y 967[2≈31.7336): 500E.I、 31.7336543.5805 [25.1283\28 2938g9.5666) 5666200 E中国煤化工X/mrMHCNMHGCopyright by Science in China Press 2005Water content distibution in 1he surface layer of Maoping slope167800300856700民600 23680500269208) 4430.08S87425是4000( 29 44/10.08961257230020030. 085622.742510001500X/m800 FE13.6331 [70060016.3829 005^27.16416 380229.005岳400\ 27.1645.I3 5331y 25.2207/25.2207200 E ,29 00513.633125 22071000E06.19206 ; 26.8797.19206 .30.5622》20:5318/28. 04829.7678|E 40026879 29. 1365|s 29.7678 120531810000.23.6398600 k你不25.8645--25.8645三人313269766500 Y。AR3352808366 25863614258昌400E，292012303135~北4258_29.2012- 32 5381~31.4258- 33.5043253872513650+一~ 33 6504Fig. 7. Water content distribution of surface layer in different d[ content distribu-tion at 5 cm depth; (b) water content distribution at 10 cm depth.中国煤化工is cm deph; (d)water content distibution at 20 cm depth; (e) water content distuYHC N M H G content distribu-tion at 30 cm depth; (g) water content distribution at 35 cm depth; (n) water content alsundulon a 40 cm depth.www. scichina.com168Science in China Ser. E Engineering & Materials Science 2005 Vol.48 Supp. 157- 170By using the Kriging method, several statistical characteristic parameters can also beobtained. These parameters are shown in Table 2. The analytical results indicated thatthe characteristic parameters of the water content distribution at different depths of thesliding body are normally distributed. The following conclusion can be drawn: the watercontent distribution is centralized and mainly at the low end of the value distribution.Table 2 Statistic characteristic paraneters evaluated with Kriging method)LayerMeanVarianceMean squareCEC valueProbabilityNo.deepthderivationvaluedistribution type0028.15625.2075.0210.178-0.1701.979normal distribution0530.51018.1064.2550.1390.3202.487nomnal distribution1029.76439.1516.2570.2100.2442.8921529.48037.0376.0860.2060.2982.6982029.77038.1126.1740.20- -0.3422.1372529.02751.2687.1600.247 -0.7062.329nommal distribution28.78342.0396.4840.225-0.5312.2138528.60137.7396.1430.215-0.1192.037a) CV is the variation cofficient, C3 is the skewness coefficient, EC is the kurtosis coefficient.In addition, the landform map of Maoping slope (Fig. 8) shows that there exist rela-tively large step changes at both the top and the bottom of the landslide. The changes arebasically similar to the water content step changes at the corresponding positions of thelandslide, and the water content distribution changes at different depths also agree withthis change. Therefore, the water content distribution in Maoping slope is in close rela-tion with the slope landfomm.600. 500300 Z2001400”12001001000200800400000 x. 100Fig. 8. Landform of Maoping slope.7 Conclusion中国煤化工Through geologic investigation and field tests,.MH. CNMH G of the Mao-Copyright by Science in China Press 2005Water contcnt distribution in the surface layer of Maoping slope169ping landslide were obtained by using the drying weight method and the direct meas-urement method. On the basis of the test data, the Kri ging geostatistics method was usedto analyze and calculate the water content distribution of Maoping slope. The watercontent distribution at different depths of the slope is in close relation with the practicallandform. The testing and analytical results obtained in this study provided basic data forfurther monitoring and treatment of Maoping landslide.Acknowledgements The authors wish to thank all their colleagues in the lnstitute of Mechanics, CAS, for theirvaluable discussion on various pans of the work descibed in the paper. This work was supported by the SpecialFunds for Major State Basic Research Project (Grant No. 2002CB412703), the National Natural Science Founda-tions of China (Grant No. 10372104), and the Knowledge Innovation Project of Chinese Academy of Sciences(Grant No. KJCX2-SW_L1-4).References1. van den Elsen, E., Xic, Y, Liu, B. et al, Intensive water content and discharge measurement system in a hil]-slope gully in China, Catena, 2003, 54:93- 115.2. Lim, T. T, Rahardjo, H., Chang, M. F. et al, Effect of rainfall on matric suctions in a residual soil slope. Ca-nadian Geotechnical Journal, 1996, 33: 618-628.3. Krahn, J,, Fredlund, D. G, KIassen, M. J.. Effeet of soil suction on slope stability at Notch Hill, CanadianGeotechnical Joumal, 1989, 26: 269- 2784. Walker, J. P.. Willgoose, Gi R., Kalma, J. D, In situ measurement of soil moisture: a comparison of tech-niques, Journal of Hydrology.2004, 293(1-4): 8S-99.5. Rothe, A, Weis, W., Kreutzer, K.et al, Changes in soil structure caused by the insallation of time domain re-flectometry probes and their influence on the measurement of soil moisture. Water Resour. Res.1997, 33 (7):1585- 1593.6. Topp, G C. State of the art of measuring soil waler content, Hydrol. Process, 2003, 17: 2993 - 2996.7. Wilson, R. G, Methods of Measuring Soil Mcisture. Technical Manua! Series, otawa, 1971.8. Schmugge, T. J, Jackson, T. J,, McKim, H. L., Survey of methods for soil moisture determination, Water Re-sour. Res, 1980, 16(6): 961 - 979.9. Huisman, J. A., Sperl, C. et al., Soil water content measurements at different scales: accuracy of time domainreflectometry and ground-penetrating radar, Joumal of Hydrology, 2001, 245(1 4): 48- 58.10. Panigrahi, B., Panda, S. N, Field ltest of a soil water balance simwlation model, Agricultural Water Manage-ment, 2003. 58: 223-240.11. Walker, J. P., Willgoose, G R, Kalma, J. D.. One dimensional soil moisture profile retrieval by assimilationof near surface measurements: a simplified soil moisture model and field application, J. Hydromet, 2001, 2:356-373.2. van Genuchten, M. T, A closed-form equation for predicting the hydtaulic conductivity of unsaturated soils,Soil Sci. Soc. Am. J, 1980, 44: 892- -898.3. Knight, J. H., Ferre, P. A., Rudolph, D. L. et al, A Numerical analysis of the effects of coarings and gaps uponrelative dielectric permitivity measurement with time domain reflectometry, Water Resour. Res., 1997, 33 (6):1455 - 1460.4. Westem A.W, Sen-Lin Zhou et al, Spatial corrclation of soil moisture in small catchments and its relationshipto dominant spatial hydrological processes, Joumal of Hydrology, 2004,286(1-4): 113- 13415. Wilson, D. J.. Westem. A. W. Grayson. R. B. et al, Spatial distribution of soil moisture over 6 and 30 cmdepth, Mahurangi river catchment, New Zealand, Journal c中国煤化工4- -274.16. Snepvangers, J. J. J. C, Heuvelink, G B. M., Huisman,_olation using spa-CNMHGtio-temporal kriging with extermal drift, Geoderma, 2003, 1www.scichina.com170Science in China Ser. E Engineering & Materials Science 2005 Vol.48 Supp. 157- 17017. Schume, H., Jost, G, Katzensteiner, K., Spatio temporal analysis of the soil water content in a mixed Norwayspruce (Pica abies (L.) Karst,)-European beech (Fagus sylvatica L,) stand, Geoderma, 2003, 112(3-4): 273-28718. Bruckner, A.. Kandeler, E, Kampichler, C, Plot-scale spatial pattems of soil water content, pH, sub-strate-induced respiration and N mineralizcation in a temperate coniferous forest, Geoderma, 1999, 93(3-4):207-223.19. Western, A.W.. BIoschl, G, On the spatial scaling of soil moisture, Joumal of Hydrology. 1999, 217(3-4): 203-224.20. Grayson, R. B., Western, A. W.. Towards areal estimation of soil water content from point measurements:time and space stability of mean response, Journal of Hydrology, 1998, 207(1-2): 68- 82.21. Westem, A. W., Bloschl, G et al, Geostatistical characterizations of soil moisture patterns in the Tarawarracatchment, Journal of Hydrology, 1998, 205(1 -2): 20- -37.22. Pelletier, J. D., Malamud, B. D., Blodgett, T. D. et al.. Scale-invariance of soil moisture variability and its im-plications for the frequency size distribution of landslides, Engineering Geology, 1997, 48: 255- 268.23. Ma Shuishan, Li Di, The displacement characteristics and prediction of its trend of Maoping landslip mass,Journa! of Yangze River Scientfie Research Instiute (in Chinese), 1994, 11(3):72- -79.24. Wang Shangqing, Yi Qinglin, The synthetic analysis on deformation characteristics and influential factors ofMaoping landslide, Qingjiang, Chinese Journal of Geological Hazard and Control (in Chinese), 1999, 10(2):40-44.中国煤化工MYHCNMHGCopyright by Science in China Press 2005