Publication n 121 of the International Association of Hydrological Sciences Proceedings of the Anaheim Symposium, December 1976 GROJN'WATER DEPLETION AND LAND SUBSIDENCE IN TAIPEI BASIN ChianMin Wu Water Resources Planning Commission Taipei, Taiwan, Republic of China Abstract In the previous paper presented at Tokyo Symposium on land subsidence (Hwang and Wu, 1969), the plots of subsidence against decline in artesian head and feedback computations of subsidence by Terzaghi's one dimensional consolidation model were used to verify that the decline in artesian pressure was the major cause of subsidence of the land surface in the Taipei Basin. In 1968, a regulatory measure was implemented to limit the groundwater pumpage. As a result of this enforced groundwater use, the pumping center was shifted from the east side to the west side of the basin. Because of this shifting, both the groundwater depression center and the sharp subsidencearea moved correspondingly. This further suggests that the decline in artesian pressure was the major cause of the subsidence. After the Tokyo symposium (1969), the subsidence of Taipei Basin has progressed considerably. This paper describes the subsidence conditions of the basin jp to the year of 1976, and presents the state of knowledge of the land subsidence in the Taipei basin with special stress on the reliability of predicted subsidence which was forecasted in 1969. Through 5 years of subsidence observations it was possible to verify the magnitude of subsidence forecasted in 1969. As a result of this analysis, it was concluded thaï declines in piezometric pressure occur over a very large area such as the Taipei Basin, the simplest one dimensional Terzaghi's consolidation model may offer a reliable forecasting of the future subsidence. Introduction In many areas of the world, land subsidence has been observed to accompany extensive drawdown by excessive pumping from confined aquifers. The cause of the phenomenon is as yet well understood; however, it is believed that reduction of hydrostatic pressure in the aquifer increases the stress on confining clay layers, resulting in their being compressed. Several investigations indicated that the magnitude of land subsidence was controlled by the stratigraphie, and mechanical properties of the soil formation, state of preconsolidation, underground structure of the geological formation, change in the elastic or other physical properties of soils, clay mineral compositions and "ectonic movemen*, etc. In the previous paper presented at Tokyo Symposium on Land Subsidence (Hwang and Wu, 1969), Plots of subsidence against decline in artesian head and feedback computations of subsidence by Terzaghi's onedimensional consolidation model were used to verify that the decline in artesian 389
pressure was the major cause of the subsidence of the land surface in the Taipei Basin, Fig. 1. In 19CJ, a regulatory measure was implemented to limit the groundwater pumpage..as a result of the imposition of strict limitation, the extracted center was shifted from the east part to the west side of the basin. Because of this shifting, both the groundwater depression center and the sharp subsidence area moved correspondingly. KEY MAP FIG.I. LOCATION AND SUBSIDENCE IN TAIPEI BASIN, AREA CONTOURED ON DIFFERENCES IN SUBSIDENCE IN METER, 19571974 390
This again demonstrates that the pressure decline was the major cause of the subsidence. After the Tokyo symposium in 1969, the subsidence of the Taipei Basin has progressed considerably. This paper describes the subsidence of the Taipei Basin up to the year of 1976, and presents the state of knowledge on the land subsidence with special stress on the reliability of predicted subsidence which was forecasted in 1969. BASIN GEOLOGY The basin is described as of tectonic origin associated with a downfaulted structure. Unconsolidated sediments of alternating layers of sand, gravel, clay and silt occupy the basin to depths of 280 m. The basin sediments are classified into four formations, the lowermost being the Tanawan formation (consisting of clay), followed in successive order by the Linkou formation, the Sungshan formation and an upper alluvium of recent age. The Linkou formation is mads up of alluvial deposits, general ly 90 to 130 m thick. It is divided into three units, the upper one is made up of mostly gravel, and the middle and lower units contain clay with intercalated sand and gravel layers. An impervious lateritic cover ranging from 0 to 3 m thick, forms an impervious copping on top of the Linkou formation. The Sungshan formation is 40 to 60 m thick and consists mostly of clay, silt and fine sand with subordinate layers of coarse grained sediments. A recent alluvium overlies the Sungshan formation and underlies the land surface to a depth of up to 3 m and consists of sand and gravel. The most extensive water bearing zone is found in the upper part of the Linkou formation at depths generally below 50 m from the land surface. Sand, gravel aid cobble beds ranging from about 20 to 30 m thick are commonly associated with the upper part of the Linkou formation, while at lower depths the proportion of finer to coarse tends to increase. Although predominantly fine grained, three water bearing zones have also been identified in the Sungshan formation. The lowermost zone consists of an irregularly distributed sand and gravel, 0 to 13m thick, and is believed to be hydrau llcally interconnected with the upper Linkou formation and with it they form the principal aquifer in the Taipei Basin. The intermediate and upper water bearing zones are found at depths of 30 to 40 m and 10 to 30 m below land surface respectively. They are made up mostly of fine to medium sand, and are widely distributed within the central part of the basin. With the exception of the coarse grained sediments in the Linkou and lower sand and gravel unit of the Sungshan formation, which make up the principal aquifer, the overlying water bearing zones are estimated to have a limited potential of groundwater resources. The top of the principal aquifer occurs near or above sea level in the Hsintien Creek, Tahan Creek and Huang Creek areas. It becomes progresively deeper toward the central and northwestern parts of the basin where it reaches a depth of 70m or more below sea level. A redjction of aquifer thickness from over 50 m to less than 10 m is also evident toward the northwestern part of the basin. Locations having a 391
maximum thickness and higher elevation indicate the source of the aquifer's sediments to be principally from the drainage area of Tahan and Hsintien Creeks and are coincident to the intake areas in which most of the aquifer's recharge originates. The principal aquifer is largely made up of highly permeable sediments which are capable of yielding up to 4.5 m/min to an individual well. The values for the coefficient of permeability range from 30 to 420 m/day, and those for transmissibility from 1,000 to 7,800 m/day. The coefficient of storage is estimated to be 0.2 for unconfined and 1 x 10 for confined aqjifer. The compressibility of Taipei clay was mainly investigated on the undisturbed samples obtained from deep borings of Shungshan formation and Its physical and engineering properties ware given in the previous paper (Hwang and Wu, 1969; Wu 1973). P urn page Wells have been used for domestic water supply in Taipei Basin since about 1895. Uncontrolled large scale development of groundwater started after 1957. In 1957 there were only 240 cased wells, 1765 open pits and 330 bamboo wells with total pumpage of 9 million m^/yr. Since then the groundwater utilization has increased rapidly and reached an estimated maximum pumpage of 435 million nrvyr from 3,846 wells in 1970. As a result of an imposition of strict limitation, the extracted wa'er was reduced to 249 and 212 million m^/yr in 1972 and 1974 respectively. Table 1 shcvs the estimates on pumpage by various use since 1964. The cooling use for air conditioning was once estimated to be 128 million m /yr or 29 percent of the total pumpage in 1970, but the said amount was reduced to 57 million m /yr in 1971 by installation of cooling towers. The basin has been developed into a metropolitan area containing the Taipei city as the center on the east half and the Taipei Hsien as a satellite community on the west half. The pumpages were originally concentrated in the Taipei City. Since 1970 the pumping center has moved from the Taipei City to the west part of the basin. Table 2, shows the area distribution of extracted water and the shifting of groundwater use since 1964. o Table 1. Type of Uses of Groundwater Pumpage (million m /yr) ^ear Cooling Industrial ^HMl _i^pplz_ Others Total 1964 66(20%) 215(66%) 1970 128(29%) 109(25%)) 1971 57(18%) 129(40%) 1972 114(46%) 1973 102(46%) 1974 82(39%) 392 46(14%) 110(25%) 65(20%) 65(26%) 94(42%) 84(40%) 88(21%) 70(22%) 70(28%) 28(12%) 46(21%) 327 435 321 249 224 212
FI S,2 PIEZOMETRIC WATER LEVELS IN METER (MS L), IN TAIPEI BASIN, 1974 Decline of Artesian Head In the Taipei Basin, the water table in the middle water bearing zone is mainly governed by the fluctuations of the stream flow, but has had little change from its initial position. On the contrary, the heavy pumping has lowered the piezometric head in the lower water bearing zone as much as 47 meters since 1957. In 1957, the piezometric head in the whole basin was within the range of +2.0 to +0.0 meters (MSL) It was reported that even the artesian wells could be seen in some parts of 393
Table 2. Area Distribution of Groundwater Pumpage (million m /yr Year Taipei Wells City Pumpage Satellite Wells Community Pumpage Total for basin Total From Principal Aquifer 1964 1970 1971 1972 1973 1974 1,158 3,408 1,233 966 756 814 187(57%) 320(75%) 118(37%) 91(36%) 51(22%) 41(19%) 1,097 438 1,630 1,384 1,246 1,246 140(43%) 115(25%) 203(63%) 158(64%) 173(78%) 171(71%) 327 435 321 249 224 212 295 392 289 186 188 the basin. However, the increasing withdrawal of groundwater caused a continuous lowering of the artesian head. From 1961 to 1964, the artesian head declined "rom 11 m to 22 m below sea! evel with an average annual rate of 4m. In 1968, the static level was at the elevation of 30 m, forming a cone of depression under the city. From 1970 to 1974, because of shifting groundwater use, the cone of depression expanded and deepened to the depth of 45 m below the mean sea level, Fig. 2. The center of the cone of depression was moved to the western border and established two small closed depression centers with their elevation of 45 m. In recent years, except at the western border area, the rate of artesian head decline has been gradually reduced due to the reduction in pumpage (Table 3.) Subsidence Survey The subsidence of Taipei Basin first was found through the 1955 releveling of a 1950 firstorder level line. It indicated that several centimeters of settlement had Table 3. Lowering of Piezometric Heads of Selected Wells, 19711974 * Piezometric Wells Head: ; in 1974 be low Sea Level(m) Tote il Lowering From 1968 to 1974(m) 1971 Annua 1972 1 Lowering in m. 1973 1974 No. 1 2 3 4 5 6 7 8 9 10 37.94 42.97 42.24 41.11 44.50 46.41 46.99 46.84 40.14 32.52 10.34 13.50 13.24 11.11 12.60 21.41 21.99 26.84 3.63 5.40 5.80 2.30 3.50 4.55 4.63 4.17 0.91 1.03 1.27 1.53 2.22 3.94 3.78 2.00 0.03 10.62 0.10 2.64 2.43 2.35 2.35 2.32 3.81 5.22 3.82 5.43 +0.97 +0.57 +0.26 0.04 0.33 0.95 1.18 1.62 3.39 3.23 * Remark : Wells No. 1 to 3 are located in the Taipei city and the rest in the Taipei Hsien. 394
occurred in the Taipei City area. In order to determine the area of known and suspected subsidence, the firstorder bench mark system established in 1950 was relevelled at least once in a year, releveled 15 times from Nov. 1955 to Oct. 1976. The accuracy of the results of releveling was tested by analyzing the probable error of the disclosures between forward and backward sighting from one benchmark to another. Curve fitting of the frequency data shows that the problable error of recent releveling (1973 and 1975) is equal to + 1.8 mm for a single observation, or the Gaussian constant h2 i s equal to 0.0702. The constant h can be defined by the error function fw = jl e'*'f 2JI An analysis of recorded subsidence in the Taipei Basin shows that the maximum daily subsidence between two adjoining bench marks may reach 0.3 mm. Since the local conditions limit the speed of observations, twentyday interval is required for a series of releveling. The heightdifference may consequently deviate by about 6 mm. Hence the probable error of the disclosure (±1.8 mm.), for a single observation is considered to be adequate for the subsidence survey in the Taipei Basin. The volume of subsidence from 1957 to 1974 was 190 x 10 m, equivalent to nearly 5% of estimated gross pumpage in that period. Thus about 5% of groundwater pumpage in the 17year period was obtained from compaction of the confined aquifer system. This represents reduction in the pore volume of the groundwater reservoir, but the reduction has been principally in the finegrained aquitards, hence it should not affect appreciably the storage capacity or flow characteristics of the sand and gravel aquifers. (i) Subsidence. The general pattern of the land subsidence in the Taipei City could be characterized as a sharp bowl with the bottom at the center part of the basin, Fig. 1. The most rapid rates of 28 cm/yr and 24 cm/yr at the stations of BM 9535 (East side) and No. 009 (West side) were observed in 1968, and their total depths of subsidence reached 1.17m and 0.9 m respectively (19571974). In response to the change of pumpage distribution and enforced groundwater use, the rate of subsidence has gradually reduced and the center of the sharps subsidence area moved westward. In 1974 the largest rate of subsidence was 14 cm/yr located at western border where about 470 hectares of the low farm land has become perennially inundated, Fig. 1. Prediction of Land Subsidence The mechanism of land subsidence is complicated. However, consolidation theory may provide a framework for prediction of probable future subsidence under presumed conditions. The physical phenomenon of land subsidence in the Taipei basin is threedimensional. However, it is believed that declines in piezometric pressure occur over a very large area and the changes of effective stress are essentially onedimensional. Hence the result of the onedimensional analysis is a good approxomation to the problem. 395
As was stated in the previous paper ( Huang and Wu, 1969), a stud/ was made to predict the future subsidence of the Taipei Basin due to groundwater depletion. In computing the ultimate subsidence, the consolidation equation ç Cc H. PO+AP,. S = HoLog g (2) 1 + e 0 Po was used, where S is the amount of the ultimate subsidence, Ho is the initial thickness of soil layer, e o is the initial void ratio, Po is the initial overbarden pressure, AP is the change in effective pressure and Cc is the compression index. Using the graphical method proposed by Terzaghi and Frohlich, the computed ultimate subsidence was further distributed to determine the relationship between the elapsed time and the degree of consolidation. Table 4 showsa typical result of the subsidence computations using well log data of Ambassador Hotel, about 1 km north of BM9536, and the recorded rate of piezometric head change. As was stated in the previous report, the graphs of computed subsidence is about 75% less than the actual. This was expected because subsidence in the cohesiveless strata was not taken into consideration. Again, other causes such as loading at land surface, vibrations at or near land surface, compaction due to seepage water and tectonic movement might contribute appreciably to the overall landsubsidence. As subsidence due to these factors can not be determined quaitatively in 1969prediction, the ratio of the computed subsidence to the observed subsidence has been used to adjust the magnitude of the computed future subsidence. This overall calibration, though is not backed up by any theory, is considered to be justified. Following surveys in 1971 to 1976 showed that the forecasted values were fairly close to the observed values, the differences were is the range of + 2 cm, (see Table 4) When the prediction was made in 1969, the piezometric head at the center of the basin was only 10+5 meters above the top of the confined aquifer. It was assumed "hat the drawdown would continue until the piezometric head fell below the top of the aquifer. Since then, continued drawdown would interfere with pumping operations, and the peizometric head in the confined aquifer would stablize itself Table 4. Comparison of forecasted and observed subsidence Year 1961 1970 1971 1972 1973 1974 1975 1976 1981 Ultimate forecasted subsidence m 0.00 1.32 1.48 1.59 1.64 1.73 1.77 1.83 2.20 2.30 observed subsidence m Differ m 1.50 0.02 1.60 0.01 1.65 0.01 1.72 0.01 1.78 0.01 396
DATUM PLANEORIGINAL BENCH MARKS, 1955 EAST Scale FIG.3. PROFILE ACROSS BENCH MARKS IN TAIPEI BASIN 397
automatically. Atomospheric pressure would be prevailed between the lower face of the aquitard and the water table in the aquifer. With these assumptions, it was estimated that the additional amount of subsidence due to groundwater drawdown may reach 50 to 80 cm over a period of 10 years for the downtown Taipei and 80 to 200 cm for the surrounding area respectively. Even if the drawdown of the confined aquifer were to cease, the land subsidence would continue to settle for several years. In the paper entitled "Groundwater Depletion and Land Subsidence in Taipei Basin" in 1969, the subsidence due to cohesive less materials was neglected. Though this difference might be covered by the model calibration, however, comparison of the volume of subsidence and the total pumpage during the period from 1957 to 1974 showed that only 5% of ground water pumpage was obtained from the compaction of the confined aquifer. Hence the reduction in the pore volume could be neglected. This again shows that the settlement has been principally in the finegrained compressible aquitards. In the Taipei basin, though further predictions of future subsidence by two dimensional and threedimensional models were performed 3y various authors, however, historical data were not long enough for verifcarion. Hence on the basis of presently available data it may be concluded that the simplified onedimensional model may be used to estimate the location, rate, and magnitude of future subsidence of the land surface with the applications of the laboratory physical properties of well logs and field drawdown data. References J.M. hwang and C. M. Wu, "Land Subsidence Problems in Taipei Basin", Proc. Intnl. Symposium on Land Subsidence, Tokyo, 1969. C M. Wu "Groundwater Depletion and Subsidence Problems in Taipei Basin", Proc. Intnl. Symposium on Development of Ground Water Resources, Vol. 1. 1973 pp IIb 1728 398