Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Climatic Change and Water Supply: How Sensitive is the Northeast? 7 INTRODUCTION HARRY E . S CH WARZ Clark University Water supply is an essential service that requires profes- sional water managers to plan continuously and carefully as system expansion is costly and time-consuming. Re- duced to its simplest terms, water-resources planning with its water-supply component is a search for the most cost-effective means of striking a balance between sets of estimated demand with sets of estimated supply. Further, however, water-supply plans must also take into con- sideration constraints imposed by the larger environment within which the water manager operates: limitations imposed by law, custom, institutions, fiscal capabilities, and long-time horizons appropriate to political decision-making. Indeed the time scales for the water- resources planner and the political system are dramati- cally out of synchronization. In most small jurisdictions, for example, the short-range future is tomorrow. Mid- range futures come before the next election, and long- range just beyond that election. Usually, only larger cities, regional jurisdictions, states, or the federal gov- emment can look at planning horizons of 2~50 years. In terms of local decision-making, what might occur in 100 years is almost entirely without significance. In this context, the professional water manager's lot is not a happy one. His customers castigate him in the event of even short-term water shortages, while the academic community, particularly economists and environmen- talists, accuse him of habitually overestimating demand and overbuilding facilities. Further, his operation and construction plans rest on uncertain estimates of demand based, in turn, on equally uncertain population and eco- nomic development projections. Even the range of varia- tion in his supply is based on predictions made from historical data and statistical manipulations. Hence, it is understandable that the introduction of yet another element of uncertaintyclimatic changeleaves the water manager still more vulnerable to accusations that he did not correctly consider the unforeseeable in making his calculations. The water-resources planner needs better information to assess the importance of cli- matic change as a valid planning factor and its acceptabil- ity in the public decision-making process. The probability of climatic change could be considered a goad to action or
112 additional fuel for the "let's wait and see" style of facilities planning. THE REGIONAL SETTING For more than 10 years, uncertainties about the validity of demand projections, fiscal and environmental constraints, and simple procrastination have stalled all major projects that would increase water supplies in the northeastern United States. In 1972, the North Atlantic Regional Water Resources Study (NAR) found that water-quality mainte- nance and water supply for public, industrial, and pow- erplant cooling purposes were the most important water- resource problems of the northeastern United States (NAS, 1972a). Intuitively, water supply is likely to be far more sensitive to climatic change than water quality maintenance, although the latter might also be adversely affected. Future water needs for additional municipal use based on today's experience were estimated to range from a low of 1400 million gallons per day (mad) (61.3 m3/sec) in 1980 to a high of 10,600 mad (464 m3/sec) in 2020 (NAR, 1972a). Even greater quantities have been estimated to be needed for industrial and power cooling. The northeastern United States as defined in the NAB study encompasses all river basins within the United States draining into the Atlantic Ocean north from the Virginia-North Carolina border, the Lake Champlain drainage, and the St. Lawrence River drainage in the United States south of the junction of the St. Lawrence River and the international border. This land area of about 167,000 square miles (432,500 km2) includes all or parts of 13 states and the District of Columbia. The region contains about 5 percent of the nation's land area, but it holds about 25 percent of its population and produces nearly 30 percent of its wealth (NAR, 1972a). On the average, some 1200 water suppliers in the region presently furnish 5.5 billion gallons per day (bad) (241 m3/sec) and serve more than 47 million people or 88 percent of the area's total population. About 4.7 bad (206 m3/sec) is used in domestic supplies. The remainder serves industrial, commercial, and municipal purposes. About 76 percent of the water used in these supplies comes from surface waters, the remainder from ground- water. Depending on which projection of growth is realized, populations are likely to require 9.5 to 10.9 bad (416 to 478m3/sec) by the year 2000 and from 13.1 to 16.0 bad (574 to 701 m3/sec) by 2020. By far the major share of this water will come from surface sources with reservoirs as the major device for collecting supply (NAR, 1972b). Average annual runoff from the region is 163 bad (714 m3/sec) 123 bad (5389 m3/sec) available 90 percent of the time (U.S. Water Resources Council, 1968~. Seasonal var- iability, which is generally moderate, can be high in individual basins. Months of high flow are usually March and April, the lowest flow tending to occur in August and September. At the extreme end of the scale, at the height of the drought in the early 1960's, runoff in the affected HARRY E . S C HWARZ area was from between 45 and 60 percent of normal. The total drought period lasted 51 months. Despite the fact that average annual draft on the waters of the region is less than one third of the annual runoff that can be expected to occur 95 percent of the time, severe local shortages have occurred in the past and are likely to occur again in the future. Recurrence of a major drought, coupled with the projected increases in demand, would cause larger and more widespread disruptions than have been experienced in the region so far. This possibility is extremely significant in light of the fact that three metropolitan areas, Washington, New York, and southeastern New England, account for more than 70 percent of the total present average public water supply now furnished in the northeastern United States. From the foregoing it is obvious that water supply is a problem of tremendous magnitude in the northeastern United States and that water supply for these metropolitan re- gions is of particular concern. EVALUATION OF CLIMATIC CHANGE There are two major ways in which climatic change af- fects water supply. One is the effect it may have on demands such as lawn irrigation or the use of showers. Second is the effect on water sources such as streamflows or groundwater recharge. Of these, the second is likely to be far more important to metropolitan water management in the Northeast. Therefore, this paper will deal predom- inantly with changes in water sources and within these restrictions with streamflow, the most common source of metropolitan water supplies here. Four parameters, singly or in combination, could be used as indicators of climatic change. They are the first three moments of the distribution of streamflow and per- sistency and the tendency of dry years to follow dry years and wet to follow wet. Records of streamflow allow us, within the limits imposed by their shortness, to estimate these parameters if we assume some specific distribution function as properly descriptive of streamflow. Climatic change, as expressed in streamflow, can then be de- scribed as a significant change in one or more of these four parameters and maybe even in the distribution func- tion. This last change, however, is unlikely to be detected in practical human time space; and, therefore, changes in the four parameters, mean, standard deviation, skew, and persistency, would suffice to characterize climatic change. To evaluate the effect of climatic change on the water supply systems of the Northeast, the most effective ap- proach would be to project the direction and magnitude of such change, to translate this into streamflow records, and then to use these records to analyze the response of the present and projected future major water-supply systems to forecast change. Unfortunately, this straightforward ap- proach is not feasible. While the specter of major changes in climate has been held before the public for a long time
Climatic Change and Water Supply: How Sensitive is the Northeast? and while serious scientists have long debated this prob- lem, the result of the debate has not been very helpful for the solution of problems faced by water-resources plan- ners and managers. The one conclusion available to prac- ticing professionals is that changes in climate are likely. However, the magnitude and timing and even the sign of the changes are unknown. Under the present state of climatic forecasting and our ability to translate climatic data into streamflow, such an approach is not feasible. Furthermore, the possibility to construct long-term hydrologic records from secondary sources appears limited in the Northeast, although a con- sistent 239-year record of precipitation and temperature has been produced by Landsberg et al. (19681. Unfortu- nately, the errors associated with converting precipitation into streamflow are much higher than rainfall-runoff modelers have led us to believe (Todin and Wallis, 1974~. Thus other ways must be attempted to assess the sensitiv- ity of northeastern water supply systems to climatic change. Three possibilities appear feasible: (1) to review indi- vidual cases and, on the basis of their previous response to existing climatic anomalies and on the basis of ques- tioning some of the water managers, speculate on the response to climatic change; (2) to select certain broad criteria of water systems and speculate on the effect climatic changes might have on each; and (3) to select a method of synthetic streamflow generation, use gener- ated streamflow as a decoupled surrogate for climate, arbitrarily vary parameters, and observe the effect of these variations on the safe yield that could be developed from this streamflow record. Within the space and time limitations of this paper, all three will be attempted. EXAMPLES OF METROPOLITAN WATER- S U PPLY PROB LE M S The Washington Metropolitan Region in Virginia, Mary- land, and the District of Columbia covers about 2800 square miles (7242 Imp. With a population of about 2.4 million, it is the ninth largest metropolitan area in the United States. Population projections postulate a likely growth of this area to 5.2 million people by 2000 and 6.8 million by 2020. This population generates a water de- mand averaging 390 mad (17 m3/sec) on its water supply today and a demand of 720 mad (31.5 m3/sec) and 925 mad (41.5 m3/sec) projected for the years 2000 and 202O, re- spectively (Northeastern United States Water Supply Study, 1975~. The major share of the present water demand, about 60 percent, is supplied by the generally unregulated flow of the Potomac River. The remainder comes from reservoirs on the Patuxent and Occoquan Creek. The two latter rivers are almost fully developed, and the Potomac can, in its present unregulated stage, barely supply present summer peak demands. Plans for the future foresee a ~3 mixture of regulation of the Potomac, local reservoirs, interconnections of water-supply systems, emergency re- strictions, use of water-saving devices in new construc- tion, and reuse of water from the Potomac estuary and advanced waste-treatment plants (Northeastern United States Water Supply Study, 1975~. To date, the existing water-supply sources and the institutions managing them have generally been success- ful. Even during the droughts of the 1930's and 1960's, restrictions did not pass the minor nuisance levels. This situation will not continue, however. Growth to date has been sufficient to outstrip the safe yield of the region's water sources under drought conditions, and even the most optimistic believers in controlled growth cannot show that continued increases in demand, albeit smaller ones than officially estimated, are likely. Planners have predicted this situation for a long time, and expansion plans go back to the 1940's; yet the response has been greater on rhetoric than on action. Only the suburban sys- tems have made many significant improvements in source development, and there has been none at all in the last 10 years. While water planners blame the environmentalists and the environmentalists blame the pigheadedness and lack of imagination of the water engineers, the reasons are more fundamental. One is the diversity of objectives relating to the social and economic development of the region. The other is that the existing institutions are geared to deal with relatively short-term problems in a political climate that demands that decisions be made for clearly visible reasons to achieve immediate results. The water problem of the area is not obviously visible. It is a long-range problem and its possible solutions are based on uncertain assumptions in both the demand and water aspects assumptions that are based on often unpopular political decisions such as growth control or on statistical distributions of natural events that may or may not be valid. In the face of these difficulties in decision making on water-supply development in which steps leading toward stabilization of supply and increases in water reuse are taken extremely slowly and under constant opposition from one side or the other, it would be hard to imagine that the process could include, with some realism, estimates of climatic change that are even more uncertain, less understood, and further in the future than annual and monthly variability now used in water-planning studies. The New York Metropolitan Area covers 9345 square miles (24,200 km2) in the states of New lersey, New York, and Connecticut. It includes not only New York City but a number of municipalities with populations exceeding 150,000 and several hundred smaller communities. The area is not only a population center but contains a large number of manufacturing plants and the nation's largest concentration of financial, trade, professional, and com- munication services. Population is now estimated at about 19 million and expected to reach nearly 24 million by 2000 and nearly 27 million by 2020. Water demands
114 today average 2760 mad (121 m3/sec) and are expected to reach 4050 mad (177 m3/sec) and 5120 mad (224 m3/sec) in 2000 and 2020, respectively. New York City's water- supply system provides more than 65 percent of today's demand, but its share is expected to shrink to less than 60 percent by 2020 (Northeastern United States Water Sup- ply Study, 1975~. The New York metropolitan area is at present supplied by a large number of water-supply systems. The system that serves the city is by far the largest. It draws its supplies from a series of reservoirs located in the Croton, Catskill, and Upper Delaware watersheds. Other systems within the area use the streams of north and central New lersey and western Connecticut. Safe yield of many of these water-supply systems is today barely sufficient to meet demands. The drought of the 1960's required strin- gent emergency restrictions. No major new water sources have been developed since the drought, and a compara- ble period of low supply starting today would likely have even greater effects. Studies of additional sources of water supply show that the most important source for the area would be the Hudson River either by high-flow skimming or reservoir development on one or more of its tributaries. The second regional source, particularly for the New jersey portion, is the Delaware River, through the controversial Tocks Is- land project. Lesser sources are in the smaller streams of New lersey and Connecticut and groundwater in New lersey and on Long Island. Metering of the New York City supply is also considered as a method to reduce demand (Northeastern United States Water Supply Study, 1975). Here again, acrimonious debates have taken the place of action. Arguments over the merits of the Tocks Island area as a park or as a reservoir have been decided in favor of park development, removing the Delaware River as an expansion source for the New York-New Jersey met- ropolitan area. Equally, reservoir development in the Adirondacks had been practically outlawed by action of New York's electorate. Water-supply interconnections in New Jersey, installed during the drought, have been disconnected. As a result, the region is more susceptible to shortages now, and its choices for the future have been significantly narrowed. Institutional arguments between the state of New York and New York City, the emotional issue of New York City metering, and the city's monetary problems further complicate the situation. A Hudson River high-flow skimming project coupled with metering in the city appears to have some chance to be im- plemented and to solve New York City and state of New York water problems 10 years hence. In New Jersey, there is as yet no plan. Even with a Hudson source development, the region's water system would be sensitive to climatic changes that change the runoff regimen of the source streams. Three points make it so. First, the Delaware's delicate balance between upstream diversions to New York City and the HARRY E. SCHWARZ downstream needs for municipal and industrial water supply and estuarine salinity control could be easily upset by a relatively small reduction in long-term average runoff or the occurrence of long droughts. This balance is codified by existing legal decisions and interstate agreements; climatic change would bring changes here too, as discussed in Chapter 4. Second, the amounts avail- able from highflow skimming of the Hudson, with con- sideration of both salinity control and fish spawning and migration, depend on the time-flow distribution. Changes in that distribution through climatic change could upset yield estimates. Third, groundwater yield on Long Island is directly related to local rainfall. Consistent short falls, for example, would rapidly reduce aquifer yield or in- crease the danger of saltwater intrusion. The southeastern New England Metropolitan area in Massachusetts and Rhode Island encompasses 357 municipalities with an estimated population of 6.5 mil- lion today. Population projections foresee a growth to 8.5 and 9.7 million for the years 2000 and 2020, respectively. Water demands in this area were 749 mad (33 m3/sec) in 1965, and these demands are projected to increase to 1519 mad (66 m3/sec) in 2000 and to 1893 mad (83 m3/sec) in the year 2020 (Northeastern United States Water Supply Study, 1975). This area is served today by 369 public water systems with a combined safe yield of 970 mad (42 m3/sec). The Metropolitan District Commission (MDC) system serving Boston and 41 nearby communities is by far the largest. Its supplies are drawn from tributaries of the Connecticut and the Merrimack through the Quabbin, Wachusetts, and Sudbury reservoirs. Other supplies come from the basins of the Ipswich, North, Taunton, Pawtuxet, Blackstone, Thames, and Powcatuck Rivers. Impound- ments in these basins are the usual source of the water supplies. Present demands are approximately equal to safe yield, although the demand in the MDC system ex- ceeds its safe yield by almost 1 percent now. During the drought of the 1960's, reservoir levels in the MDC system were very low, and drought of a longer duration would have caused significant shortages. Other communities in the area resorted to emergency restrictions, some of quite stringent nature. Ongoing planning includes additional diversion of water from the Connecticut Basin and the Merrimack Basin, either through high-flow skimming or reservoir development, as well as through new reservoirs in the other aforementioned streams in the area and groundwa- ter development (Northeastern United States Water Sup- ply Study, 1975~. The major problem here is that there is a different con- stituency for the source area western Massachusetts- and for the user area Boston. These constituencies are economically and socially divergent and see little need to accommodate each other. To this must be added the perception of the supply. The reservoirs are full right now, and the drought is forgotten. Climatic
Climatic Change and Water Supply: How Sensitive is the Northeast? changes that would produce longer dry and wet periods would tend to inhibit development further, as little can be completed even during a 5- or 6-year drought, and 10 years or longer of normal or above normal supply would allow people to forget the lessons of the last drought completely. In a discussion of the problems of water supply and climatic change with the Chief of the Water Divisions of the Metropolitan District Commission, that official re- marked that long-term changes in climate reflected in the availability of water would increase or decrease the vul- nerability of this supply system but would not change his basic problem. "We look ahead, but not necessarily act on that look" is the way he put it (Matera, 1976~. A decrease in the mean annual yield would be in his opinion most detrimental to Boston's water supply. Increased persis- tency, longer periods of less flow, would be the next most damaging manifestation of climatic change. EVALUATION OF METROPOLITAN WATER SUPPLIE S There is a set of attributes that appears common to all the examples of major water-supply problem areas that can be used to evaluate the effect of climate change. While these attributes are common to all, the importance of each varies from area to area. Nine such attributes are considered here. First, there are four that are related to the resource itself: yield from unregulated streamflow, yield from res- ervoirs, yield from groundwater, and quality of raw wa- ter. The next two related to the effectiveness of water- supply works: overall systems reliability and effective- ness of regional interconnections. The last three involve the management of water supply: magnitude and control of demand, cost of operation, and pressure on the water- supply system to expand and its ability to respond to change. These attributes of metropolitan water-supply systems can then be evaluated against the four parameters of climatic change mentioned in the previous section. An additional characteristic of climatic change should also be considered. This is the speed with which these changes become manifest. Reactions to such changes are likely to be considerably different if change manifests itself over the span of a few years or comes gradually over periods as long as 20, 50, or even 100 years. A speculative effect matrix could be constructed show- ing the effects of the enumerated manifestations of cli- matic change on the nine attributes of water systems. Table 7.1 presents such a matrix. A SIMULATION APPROACH A third way to evaluate the possible significance of cli- matic changes would be through the use of synthetic ~5 hydrology. Using streamflow as a surrogate for climate, alternative hydrologic scenarios could be constructed to cover a wide range of possibilities, permitting an evalua- tion of system sensitivity to change. There are two major problems to this approach. First, the relationship between climatic variations, the way climatologists see them, and streamflow is largely unde- fined. This is particularly true when a specific stream and location on that stream must be considered. The second problem is the selection of a generating algorithm. Sev- eral streamflow generation models exist, each having its partisans and detractors within the hydrologic commu- nity. For the initial analysis of this paper, it was assumed that monthly streamflow adequately described the water re- source and that its variations could be represented by a skewed logarithmic lag-one Markovian model. Such a model can be characterized by four parameters. These are mean, standard deviation, skew, and serial correlation coefficients. These parameters can be used to generate streamflow traces representing alternative futures. By varying these four parameters, we can simulate some of the effects of climatic change on systems response. A computer program developed by Leo R. Beard and the Hydrologic Engineering Center is available that allows generation of such alternative synthetic streamflow rec- ords using observed or modified statistics (U.S. Army Corps of Engineers, 1971~. Using the aforementioned mathematical model and computer program, alternative streamflow records were generated for the Potomac River at the Point of Rocks gauge, where the drainage area is 9651 square miles (24,996 amp. Statistics were computed from the existing record. Records from 1895 to 1970 were used to compute the logarithmic mean, standard deviation, skew, and se- rial correlation coefficient for each month. These statistics were then used to generate a 1000-year record and eight alternative "climatically changed" 1000-year records. For each of the alternative records, one set of statistics was varied. Table 7.2 shows the assumptions used for each data generation and the highest and lowest 1-month, 6-month, and 54-month flows, as well as the long-term average. As no estimates of likely climatic change were avail- able, the variations in statistics were arbitrarily chosen. As can be seen from Table 7.1, variations in the standard deviation not only greatly increase the range of monthly flows, they also increased the mean flow. Changes in the mean generally shift the entire population of data, and increases in the serial correlation coefficients have rela- tively little influence. A change of the skew coefficient had a far greater effect on the one-month extremes than on the longer and more critical high- or low-flow periods, a result to be expected since monthly correlations of the generating algorithm were rather small. The 1000-year synthetic records generated were pro- cessed through a computer program developed by the
by A . - 4- o x ce ;~ is .~ 'e - of En be ct c) hi .- . - 4- o hi em pa 116 ,c) 3 ~ o to cn ~ ._ ~ ~ o ~ C ._ _ a; Cd ~ c) cn o o 3 U. o ~ c a, _1 Cal C) ._ 3 .= ~ Us Ct Id ~ cn o . ~ .- 3 o C en C, ~ A; U. 4 - U: s-4 U) Cd ._ ;> rid ~ ~ o Cal _ _ _ ~ ~ C) Cal Cal ._ ._ ._ ._ ~ ~ z ~ ;z; zo .,= ~ V o ~ ss C .fj, E ~ ~ ~ ~0 ~ ° c~ ._ u' O .= ~ ~ g .N ~ ~ ~ ~ ~ ' ~ ~ ~ ° ~ ~ df ~ ' ~ · 8$ Z G ~ 2, e ~ V) ^_ ~ ~ ~ ~ Q^ ~'= j C 0 =5 ~ ~ C E ' ~ ~ ~ S 3 G 0 _ 0 E C ~ ~ ~ G ;,, ~ - ~ ~ -.? .= G _ e ;> C,) ._ ._ ~ C~ . - ~o 2 C, 3 3 £ ~ _ E ,;. E E ~ c;i c~ .
~ - - o ~ ~ ~ ~ o ~~ a, z Go z _ , ,,S . . ~ , ,, ~ 5 s a-~~ c) c - - ~ ~ ~ o -~ ~ o ~ c) ~ ~ ~ ~ ~ c : o o ~ ~ -= ,~ ~ ~ o z ~ ~ 3 j== ~ it, , ~ , ;, ~ ~ _ ~ ~ ~ ~ . .2 s ~ so a, ~ s ~ ~ ~ ~ _ E ,: ~ ~ ~ ~ o ~ ~ ~ ~ ° ~ ~ D ~ O o ~ ~ z 0 o ~ ~ ~ cc ~ ~ 3 ~ ~ 3,~ - o ~ o ~ ~ E ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ · - Gil 2 ~ ~ ~ ~ =~ do ~ , t ~ car' en ~ c;, 117
118 TABLE 7.2 Alternative Streamflow Records for Potomac River at Point of Rocks Average Flow in Cubic Feet per Second HARRY E. SCHWARZ 1-Month 6-Month 54-Month - Total Basis for Record Generation Highest Lowest Highest Lowest Highest Lowest Record 76-year historical record Actual record 1895-1970 68,359 706 26,225 943 12,171 6205 9061 1000-year simulated records Observed statistics Standard deviation increased 10~o Standard deviation increased 20% Log mean decreased logo Log mean decreased 20~o Serial correlation coefficient increased logo Serial correlation coefficient increased 20`Yo Skew coefficient increased +0.5 Skew coefficient increased -0.5 80,529 112,808 128,932 70,647 43,104 75,762 75,786 167,855 57,011 439 273 192 329 284 358 358 736 159 34,437 1132 36,728 866 42,514 761 28,464 893 25,306 789 32,136 936 31,901 882 51,001 1223 25,720 823 13,528 5455 9076 15,871 5353 9428 17,214 5270 9813 12,846 4896 8152 11,412 4344 7339 14,449 5376 9073 14,636 5315 9073 16,996 5356 9233 12,748 5579 8938 author for the NAB study and adapted for use here by Bruce Morris, a student at Clark University. This program subjects the flow of the river under study to varying levels of drafts while varying the amount of usable storage in the system from O to that amount needed to meet all drafts without a month of failure. The program tabulates for each combination of draft and storage the number of months in which the target draft could not be furnished, the number of shortage periods of various length, the highest and the average shortage, and other information. Relationships between draft, storage, percent chance of meeting demand, and the different assumptions of statis- tical variations were then analyzed. The result of this analysis was generally disappointing to those who be- lieve that climatic change should radically alter the water-supply planning process. No clear picture of the susceptibility of large water-supply shortages attributable to a specific statistical parameter developed. For instance, if 95 percent assurance of supply is required, the Potomac in its present state can supply about 1400 cubic feet/see (cis) (938 mad or 130 m3/sec). With the most unfavorable alternative tested, the yield was 1100 cfs (737 mad or 122 m3/sec) or a reduction of about 20 percent. With the same unfavorable trace, the full 1400 cts were assured 91 per- cent of the time. The trace that produced this change assumed a reduction of the log-mean by 20 percent. The other alternative traces have less influence on the yield of the free-flowing river. Figure 7.1 shows the plot between draft and percentage of time the draft requirements can be fulfilled. Adding storage into the system complicates the analysis and makes it even less satisfactory. The number of points now available for analysis of the apparently most impor- tant relationship, that between storage, draft, and shortages, is small; and storage and shortages vary from run to run even with the same statistics because of the random component in the data generation. Therefore, plots can only be sketched by smoothing and with a resulting reduction in accuracy. There are strong indica- tions, however, that major increases in standard deviation are as important as reductions in log-mean. There are furler indications that sensitivity to changes in the streamflow trace increases with the degree of develop- ment and with a decrease in the level of acceptable risk. In other words, the increase in storage needed to com- pensate for alternative scenarios and to maintain a spe- cific draft increases with the size of the draft and with a decrease in the acceptable shortage index. The results of this analysis can be accepted only as very preliminary. Various phenomena observed, such as the changes in mean flows accompanying changes in stan- dard deviations, need further analyses. Also, other generating algorithms, including non-Markovian models, should be studied. Work toward this end has been ini- tiated by the author in cooperation with lames Wallis and others. CONCLUSION What does all this mean to those who must plan and manage the water supply of the northeastern United States? Apparently it has no importance to them now. There is not yet sufficient knowledge of climatic change or its effect on water supply to suggest any rational change in public policy. If, in addition, we add in the uncertainty of the timing of possible climatic changes, then it becomes even more certain that current planning does not have to be concerned with climatic change. On balance, it is the more easily perceived uncertainties at
Climatic Change and Water Supply: How Sensitive is the Northeast? 100 90 FIGURE 7.1 Potomac River at Point of Rocks. Draft-duration relationship without storage. - o O_ Q 80 . _ a: en . _ ~ 70 . _ o a' cn a' cat a) cat 60 50 40 -' hand that are the appropriate guides to policy-making. If this study has any relevance to ongoing water-supply planning, it might reinforce the need for flexibility in plans and argue for a more adequate safety margin in the development of water supplies. The conclusions of this chapter are as follows: 1. A full understanding of the possible relationships between climatic changes and water supply, its planning, and management is a desirable goal. 2. At minimum, a range of likely climatic-change scenarios should be available to the water-resources planner. 3. Much more detailed and complete analyses are re- quired to assess the sensitivity of existing or projected water supplied to climatic variations. Such work should: (a) Consider several streams throughout the region. (b) Test different models for streamflow generation including those that preserve skewness directly without a logarithmic transformation and those,that allow for / 119 all other futures tested log M -1 C)% I\ I\\ 109 M - 20% \\ \ 1 1 1 1 1 1 1 1 1 1 1 0 1 2 3 4 5 6 7 8 9 10 Draft in 1000 cfs greater long-term persistence than is possible with the Markov model used in this study. (c) Postulate smaller increments between alterna- tives and make a far greater number of iterations for each alternative so that observed differences cannot be at- tributable to noise in the data. (d) Use more detailed models of urban water-supply systems, including economic loss functions. Beyond these preliminary results, it is appropriate to speculate on the changes in water-supply planning and decision-making occasioned by findings that significant climatic change is at hand or that such changes are of no consequence. In the latter case, unwarranted con- siderations should not be allowed to complicate and delay water supply plans and their implementation. If, how- ever, significant climatic changes are likely, present planning methods must take them into account. The most straightforward incorporation of information on climatic change would be the development of a series
120 of water-resource projections analogous to Bureau of the Census population projections. These could be prepared on a national basis and then adapted to specific regional characteristics and uses. In this fashion, consistency could be maintained, while variations dictated by specific local circumstances could still be considered. Such alter- native projections would complicate the hydrologic studies needed for water-supply planning and manage- ment, but the institutional decision process would not be significantly changed for near-term and midterm con- siderations. Should really large climatic changes appear likely for Me more distant future, political decision- makers must be sensitized to deal with long-range plan- ning fixtures and their associated increase in uncertain- ties. REFERENCES Landsberg, H. E., C. H. Yu, and L. Huang (1968~. Preliminary Reconstruction of a Long Time Series of Climatic Data for the Eastern United States, Tech. Note BN-571, Sept. HARRY E. SCHWARZ Matera, J. (1976~. Director, Division of Water Supply, Metropoli- tan District Commission, Boston, Mass., personal communica- tion. NAR (1972a). North Atlantic Regional Water Resources Study, Coordinating Committee, North Atlantic Regional Water Re- so~4rces Study Report, Fords Atlantic Div.? U.S. Arrny Corps of Engineers, New York, June. NAR (1972b). North Atlantic Regional Water Resources Study, Coordinating Committee, North Atlantic Regional Water Re- sources Study,Append~x R. Water Supply' North Atiantic Div., U.S. Army Corps of Engineers, New York, June. Northeastern United States Water Supply Study (1975~. Interim Report~ritical Choices for Critical Years, Norm Atlantic Div., U.S. Army Corps of Engineers, New York, Nov. Todin, E., and J. R. Wallis (1974~. Using cars for duly or longer period rainfall-runoR modeling, paper presented at Workshop on Mathematical Models in Hydrology, Pisa, Italy, Dec. U.S. Army Corps of Engineers (1971~. Hydrologic Engineenng Center, HECK Monthly Streamflow Simulation, Davis, Celia, Feb. U.S. Water Resources Council (1968~. The Nation's Water Re- sources, U.S. Government Pnnfing Office' Washington, D.C.