0b

3

4

D1

In headwater areas of the Rio Grande and its principal tributaries, variations in streamflow and in ground water storage and discharge depend upon fluctuations in precipitation, with modifications by geologic factors and by the pattern of water development and use. In downstream areas the surface and ground water resources are replenished not only by local precipitation but also by outflow from the headwaters areas; thus the effects of drought upon those water resources are complex and may be vague and indeterminate.

The Rio Grande basin is one of five areas of the Southwest subdivided for detailed discussion of the effects of drought as distinguished from other water shortages. The Rio Grande (or Ri´o Bravo in Mexico) is an interstate and international stream. It rises in Colorado and flows southward for more than 400 miles across New Mexico, then forms the boundary between Texas and the United States of Mexico for about 1,200 miles to its mouth (fig. 1). The Rio Grande has a total length of more than 1,800 miles and is the second longest river in the United States; its drainage basin encompasses about 182,000 square miles.

Hydrologically the Rio Grande includes an upper river above Fort Quitman, Tex. (about 90 miles downstream from El Paso), which is generated entirely within the United States and almost entirely used up in Colorado and New Mexico, and a lower river that is regenerated chiefly by flow from Mexico.

The upper Rio Grande has a drainage area of about 35,000 square miles, less than a fifth of the water producing area of the Rio Grande basin. According to the report of the Rio Grande Joint Investigation (National Resources Committee, 1938), the total mean annual water production from runoff in the period 1890-1935 was about 3 million acre feet, of which more than half was used consumptively for irrigation in New Mexico and Colorado. The river is generally lowest in average annual flow in the barren reach below Fort Quitman. In the 20 years 1924-43 the average discharge from the upper basin at Fort Quitman was about 200,000 acre feet.

According to records of the International Boundary and Water Commission (1943), the Rio Grande at the gaging station below Presidio, Tex., with drainage area of about 60,000 square miles, had an average annual discharge of 1.4 million acre feet in 1924-43, chiefly from the tributary Ri´o Conchos in Mexico. In the same period the average annual measured runoff of the Rio Grande was 3.5 million acre feet at Laredo, Tex. (drainage basin 133,000 square miles), and 5.1 million at Rio Grande City (drainage basin 174,000 square miles). These figures, as well as those for the upper Rio Grande, are averages based on records for years prior to the beginning of the most recent drought.

The Pecos River, with a drainage basin greater than 38,000 square miles, joins the Rio Grande below the mouth of the Conchos, and is the largest tributary entering the lower Rio Grande from the United States. Like the Rio Grande, the Pecos has headwaters in the Rocky Mountains, its water is used extensively in New Mexico, and the outflow from New Mexico is considerably less than the flow in the river farther up stream. However, the Pecos gains much water as it flows across the Edwards Plateau region in Texas, and as it enters the Rio Grande it has a larger volume than at any other point along its course.

The principal use of water throughout the Rio Grande basin is for irrigation, and the total area irrigated within the basin is estimated (International Boundary and Water Commission, 1950) to have been about 2,700,000 acres in 1950. Of this total about 950,000 acres is in the upper basin above Fort Quitman, "

Water producing area only. The outer rim of the Rio Grande basin encompasses 335,000 square miles, including closed basins between the Pecos and Rio Grande (Thomas and others, 1962b) In the United States, and extensive closed basins bordering the drainage areas of the Rio Conchos, Rio Salado, and Rio San Juan in Mexico.

D2

D3

and about the same acreage is served by major tributaries in the lower basin (Pecos River, 240,000 acres; Ri´o Conchos, 200,000; Ri´o Salado, 125,000; and Ri´o San Juan, 330,000 acres). The remaining 800,000 acres is along the main stem below Fort Quitman, including about 700,000 acres in the lower Rio Grande valley below Rio Grande City.

The headwaters of the Rio Grande are in the rugged San Juan Mountains and Sangre de Cristo Mountains of Southern Colorado. In both ranges there are several peaks that exceed 14,000 feet in altitude, and areas where the mean annual precipitation exceeds 40 inches. In the agricultural valleys along the Rio Grande, however, the average annual precipitation is only 7 to 10 inches, and this is true even in San Luis Valley near the headwaters, where the valley floor ranges from 7,500 to 8,000 feet about sea level. Because of this meager precipitation, irrigation has been prerequisite to successful agriculture along the upper Rio Grande, presumably from the earliest days of habitation. The Spanish explorers in 1540 found that in the many storied towns of the Pueblo Indians of which 17 or 18 are still in existence water was diverted from the river in acequias, or irrigation ditches (Follett, 1898), of which several are still in use. Thus irrigation has been practiced continuously along the upper Rio Grande longer than in any other part of the United States: at least 400 years of recorded history, and probably for several centuries before that.

In comparison to the rest of the Southwest, the upper Rio Grande also has a longer history of water shortages and disputes, and of treaties and decrees and compacts to settle disputes. The town of Albuquerque was founded in 1706, and by 1739 some residents had moved several miles to the south, partly because of shortage of water for the fields at Albuquerque. A water shortage in the early 1890's developed international repercussions because it affected people in Mexico as well as in Texas and southern New Mexico. The shortage was attributed chiefly to the increasing development and use of water for irrigation in San Luis Valley in the preceding decade, but it is to be noted that it coincided with a period of deficiency in precipitation (p. D4). This water shortage was responsible for the "embargo" of 1896 and for the Rio Grande Convention of 1906 between the United States of America and the United States of Mexico. The "embargo" was an order by the Secretary of the Interior which prevented further irrigation development of any magnitude in the Rio Grande basin in Colorado and New Mexico by suspending rights of way across public lands for use of Rio Grande water; the "embargo" was not lifted until 1925. Under the terms of the Treaty of 1906, the United States guaranteed an annual delivery in perpetuity of 60,000 acre feet of water in the Rio Grande at the head of the Mexican Canal near El Paso, Tex.

In 1929 the States of Colorado, New Mexico, and Texas ratified a temporary compact which provided in effect that neither Colorado nor New Mexico would cause or permit the water supply in the Rio Grande to be impaired by new or increased diversions or storage unless and until such depletion was offset by increase of drainage return; this compact was operative until October 1937. In 1938 the same States ratified the Rio Grande Interstate Compact (Witmer, 1956, p. 154-177), which provides for apportionment of the water of the upper Rio Grande basin on the basis of specified indexes of flow at key gaging stations (p. D23-24).

In keeping with the long continued concern over the water of the upper Rio Grande, long records of streamflow are available for many places along the main stem and its principal tributaries; at Embudo, N. Mex., north of Santa Fe, the U.S. Geological Survey initiated its stream gaging program in 1889. However, most of the available records concerning other aspects of hydrology are shorter and less complete, and the data concerning ground water are especially meager.

The report of the comprehensive Rio Grande Joint Investigation (National Resources Committee, 1938), which provided the basic data essential for the negotiation of the Rio Grande Compact of 1938, is of great value in studying the effects of drought upon the basin's water resources. Although that investigation was completed several years before the beginning of the most recent drought, it covered a period (1890- 1935) that began in drought and ended in drought. Because of the intervening wetter years, it was concluded in that report that the median natural streamflow during the period of record was close to the median flow for a much longer period.

In all parts of the basin the natural streamflow has been modified by man to such an extent that there are few places where it can be computed reliably from existing records. Many modifications had been underway long before the beginning of these records, but many others occurred during the period 1890-1935 and have been noted in the report of the Rio Grande Joint Investigation. Such modifications include changes in diversions, reservoir storage, irrigated acreage, and drainage of surface water; changes in groundwater storage; and changes in cover of vegetation that

D4

is dependent upon ground water. In some places the effects of these modifications by man have reached a sort of equilibrium in subsequent years, perhaps different from natural conditions but nevertheless fairly stable. In other places they tend to reflect or even to enhance the natural fluctuations in supply from precipitation (p. D17).

In accordance with natural divisions, the upper Rio Grande basin comprises three principal areas: the San Luis Valley in Colorado, the Middle Valley in New Mexico, and the Elephant Butte Fort Quitman area in New Mexico, Texas, and Mexico. The Rio Grande enters San Luis Valley near Del Norte, Colo., where streamflow records have been collected at a gaging station since 1889. From these and other streamflow records the total annual production of runoff has been calculated (National Resources Committee, 1938, p. 28-37, 175-183), and is shown in figure 2A. This graph indicates natural fluctuations with least modification by man. The total inflow was less than the long term mean in 10 of the 13 years 1892-1904, the first drought recorded in streamflow records, and was greater than the mean in 17 of the 25 years in the following wet period, 1905-29. The drought of 1930-40 resulted in less than average inflow in 7 of the 11 years. After a short wet period, 1941-45, there were several more years of less than average inflow to San Luis Valley, but these were interspersed with years of more abundant streamflow. In the headwater area the 1892-1904 drought was more severe than any subsequent drought, because it included the least annual runoff and the longest succession of consecutive years of low flow (Gatewood and others, 1962). The Rio Grande headwaters are along the margin of the area of 1942-56 drought (Thomas, 1962), and also in a meteorologic region where annual variations in runoff are not as great as in many parts of the Southwest; as shown by Gatewood and others (1963), the standard deviation for Rio Grande at Del Norte is about 240,000 acre feet. In 62 years of record, the runoff was below the mean for not more than 2 consecutive years, except in 2 periods of 4 consecutive years (1899-1902 and 1953- 56).

The runoff of Rio Grande near Lobatos, Colo., represents the residual flow below San Luis Valley. There was a gradual increase in average annual streamflow depletion (ruled pattern in fig. 2) from about 600,000 acre-feet in the 1890's to more than 800,000 acre-feet in the 1920's, during which time the irrigated area increased from 250,000 to 550,000 acres.

Figure 2B shows the flow of Rio Grande into the Middle Valley as measured at Otowi Bridge near San Ildefonso, N. Mex., the total inflow to the Middle Valley as estimated in the Rio Grande Joint Investigation, the measured residual outflow at San Marcial, and the streamflow depletion within the Middle Valley. Here it was found (National Resources Committee, 1938, p. 37-47) that there had been relatively little change in stream depletion from 1890 to 1935, except that due to variation in water supply; there is an apparent downward trend, in both inflow and outflow in the past 50 years. The hydrographs for the Middle Valley show also the effects of long term climatic fluctuations, including the drought of 1892-1904 (interrupted by a wet 1897) and a wetter quarter of a century beginning in 1905. The drought of the 1930's was not so pronounced as in San Luis Valley, but the drought beginning in 1943 was more so.

Figure 2C shows the inflow to Elephant Butte Reservoir (or flow through the reservoir site prior to 1915) and the diversions and other releases from the reservoir; figure 2D shows the yearend reservoir storage. The inflow reflects major climatic fluctuations even though modified by the developments upstream. The outflow has been small and fairly constant since the construction of Elephant Butte Reservoir except in 1942 when there was spill from the reservoir. In all these graphs there is a general downward trend since 1920, interrupted in the wet years 1941-42.

Figure 2 shows the effects not only of climatic fluctuations but also of human adaptation to them. Once the irrigated acreage had become stabilized, the consumptive use tended to be more nearly constant from year to year than the natural runoff resulting from precipitation. Although the stream diversions were necessarily less in dry years than in wet years, the proportion of the total runoff diverted in a dry year was generally greater. Thus the fluctuations in the annual outflow from San Luis Valley, if expressed in percentage of the long term average, would be more pronounced than the fluctuations in inflow to that valley. There was even more extreme fluctuation in the outflow from Middle Valley, which constitutes the inflow to Elephant Butte Reservoir. By contrast, the small outflow measured at Fort Quitman results from practically complete consumption of water within the upper Rio Grande basin.

A description of the effects of drought upon the upper Rio Grande is difficult because of the problems of segregating the local effects in specific areas, and integrating those effects upon the resource as it moves downstream; and also the problems of discriminating the effects of recurrent drought from the effects of development and control. For this task we have relatively incomplete basic hydrologic data. The pattern

D5

D6

of the discussion that follows is therefore dictated in part by the availability of those data: first, a summary of the water resource and its development in each of several areas where we have information concerning the ground water as well as the surface water; then a discussion of several regional problems that have been developed or magnified during the recent Southwest drought.

San Luis Valley is a plain about 90 miles long from north to south and 50 miles wide, and thus is nearly as large as the State of Connecticut. Although this plain is 7,500 to 8,000 feet about sea level, it is properly termed a valley because it is bordered on three sides by mountains that are 10,000 to more than 14,000 feet above sea level. The Rio Grande and other streams have had no part in forming the valley, which is a structural trough between two prongs of the Southern Rocky Mountains the San Juan Mountains to the west and the Sangre de Cristo Mountains to the east. The Rio Grande continues southward from San Luis Valley all the way to the Mexican border, in a continuation of this same structural depression. Although the Rio Grande and other streams have not formed the San Luis depression, they have gone far toward filling it, and are primarily responsible for the establishment of the present broad valley floor. The deposition of sediment by the Rio Grande in comparatively recent geologic time has formed a topographic divide in the northeastern part of San Luis Valley and has created a closed basin having a drainage area of 2,940 square miles. Because of this natural condition, several tributaries of San Luis Valley are not tributary to the Rio Grande and make no contribution to the river; indeed, the contribution is in the opposite direction, for the Rio Grande loses some water to the closed basin.

Several major physiographic provinces-the Basin and Range, Colorado Plateaus, Rocky Mountains, and Great Plains provinces, which farther north span the entire breadth of Nevada, Utah, and Colorado all converge in the vicinity of San Luis Valley (National Resources Committee, 1938, p. 198). In such a focal point of geologic activity, it might be expected that San Luis Valley would be like no other in the United States. True, San Luis Valley is quite spectacular in its present setting of rugged mountains, but if the alluvial sediments that underlie the valley floor could be removed, the result would be a gigantic trough almost 3 miles deep, with its bottom below sea level. The total volume of unconsolidated sediments in the valley is unknown in fact, the total thickness is not known at any point; however, in a recent oil test the drill penetrated more than 5,200 feet of sand, gravel, and clay before it struck a thick section of volcanic rocks, beneath which was more sand, gravel, and clay. The well was bottomed in gravel at a depth of 8,023 feet. On the basis of this well log it was estimated that ground water storage in San Luis Valley is of the order of a billion acre feet, which is more than the total estimated water production of the upper Rio Grande in three centuries.

As in many other arid basins of the Southwest, the layers of clay in the fill of San Luis Valley serve as confining beds and create artesian pressure in the underlying beds of sand and gravel. One well drilled to a depth of 1,000 feet crossed more than 50 separate flows of water. As estimated by Powell (1958) there are probably 7,500 flowing wells in the valley; in addition, scores of artesian wells are equipped with large pumps. The flowing and pumped artesian wells have a potential yield of about 500,000 acre feet a year, but many are shut in during part of the year, and the actual yield is not known. Most of the artesian wells are used for irrigation and it is believed that they supply water for the complete or supplemental irrigation of about 150,000 acres. The flow of some wells has diminished over the years, perhaps as a result of deterioration or local interference or overdevelopment. In general the artesian aquifers are regarded as not fully developed.

San Luis Valley also has a shallow unconfined aquifer which is far better than the artesian aquifers as a recorder of the effects of climatic fluctuations and of man's development of the water resources. This unconfined aquifer receives water from the artesian aquifers, both by upward leakage and by downward percolation of water drawn from artesian wells for irrigation. It receives water also by deep percolation of precipitation and especially of surface water applied for irrigation. Water is discharged from the shallow aquifer by irrigation wells, by canals and drains, and by evapotranspiration.

Most of the present shallow aquifer was not saturated until the beginning of large scale diversions of surface water for irrigation. The history of irrigation in the past 80 years includes

• (1) irrigation in the central lowlands, with resulting rise of water table until extensive areas became waterlogged;
• (2) westward migration of farming to the present areas, where, with sufficient surface water and the necessary supply and drain ditches, it is possible to maintain the water table within a few feet of the ground surface for

 

D7

subirrigation ("subbing") of crops; and
• (3) development of pumped wells in the shallow aquifer, especially in years when surface water was insufficient to maintain the "sub."

In a 291square mile area that includes about half the shallow irrigation wells, it was found (Powell, 1958) that the shallow aquifer had an average saturated thickness of 60 feet, a specific yield of 30 percent, and storage of about 3 million acre feet of water. In a 4 year inflow outflow study by Mutz (in Powell, 1958, p. 120-129) the consumptive use of water within the area ranged from 307,000 acre feet in 1949 to 200,000 acre feet in 1951. The quantities of water diverted from streams and pumped from wells are shown by bar graphs in figure 3B. The effects of these additions to and subtractions from the reservoir are shown in the water levels in three wells within the study area, and similar effects may be presumed in years when no data are available as to diversions or pumpage. Thus, after wet 1941, sub irrigation was moderately effective until the extremely dry year 1951. With record precipitation in 1952 the water table rose again to "subbing" level and was high also in 1953. From 1954 through 1956 the water table failed to rise to the "subbing" level and in early 1957 it was at a record low level. With abundant runoff in 1957 and 1958, the water table was again high enough for "subbing" by June 1958.

Increases in ground water storage such as that recorded in 1952 are responsible for part of the difference between stream inflow to and outflow from San Luis Valley (fig. 2). The average annual depletion of Rio Grande within the valley in the 18 years 1936-53 was 800,000 acre feet, the same as was computed for the years 1927-35, but it ranged from about 500,000 acre feet in the dry years 1940 and 1951 to a million acre feet or more in 1941, 1949, and 1952. If 800,000 acre feet represents the quantity of water that must be taken from the Rio Grande in order to sustain San Luis Valley requirements, then surface water could not fulfill the demand in 1940, 1946, 1951 or 1953, because total surface water inflow to the valley was less than 800,000 acre feet. The difference in recent years has been made up by pumping from wells which, however, has reduced the storage in the ground water reservoir. In subsequent wet years this ground water storage was replenished, and the stream depletion in such years exceeded the consumptive use of water by irrigated crops and miscellaneous non beneficial water uses.

The middle section of the upper Rio Grande extends from Lobatos, Colo., near the New Mexico State line, to San Marcial at the head of Elephant Butte Reservoir, a distance of about 270 miles. The northern half of this section is flanked by the southern extensions of the Conejos Range and Sangre de Cristo Mountains, which border San Luis Valley. Throughout most of this reach the Rio Grande flows in canyons or narrow valleys. It is from this part of the drainage area that Rio Grande receives most of its water supply that originates in New Mexico, and most of the water that is used in New Mexico. The principal tributaries are the Rio Chama from the west and several streams rising in the Sangre de Cristos to the east ; winter snows on the higher mountains are an important source of the water in these streams. The annual runoff of the Rio Grande at Otowi Bridge near San Ildefonso, N. Mex., downstream from this producing area, is at least double and in most years is 4 or 5 times as great as the quantity leaving San Luis Valley near Lobatos, Colo.

At a point due west of the city of Santa Fe the Rio Grande emerges from White Rock Canyon and flows southward in a valley that continues all the way to San Marcial, a distance of 150 miles. This is commonly called the Middle Valley; the valley floor is generally 1 to 5 miles wide and is bordered by scarps rising to "mesas" (alluvial fans) that rise hundreds of feet above the valley floor. Most of these lands are within the Middle Rio Grande Conservancy District, which operates El Vado Reservoir (capacity 200,000 acre feet) on the Rio Chama. The mountain ranges bordering the valley are low and receive precipitation chiefly in summer cloudbursts; the tributaries to the Middle Valley therefore are subject to flash floods that produce relatively small total runoff. There is only meager information concerning the tributary inflow and ground water inflow from the sides of the Middle Valley. The report of the Rio Grande Joint Investigation (National Resources Committee, 1938, p. 13) gives the following summary of findings based on available data prior to 1936:

Accurate determination of past stream flow depletion in the Middle Valley is not possible because of the lack of adequate records of tributary inflow and uncertainty with respect to it. An approximation has been derived, based on such data as are available, in order to furnish a reasonable basis for analyses of the effect upon the Elephant Butte Fort Quitman section of present and given future conditions of irrigation development in the San Luis and Middle sections. The mean annual stream flow depletion, 1890-1935, Otowi Bridge to San Marcial, is estimated to have been 586,000 acre feet. The corresponding mean annual tributary inflow derived as a residual

D8

D9

in the method of estimating depletion is 359,000 acre feet. Corrected for present development in San Luis Valley, the derived values for mean annual Middle Valley depletion and San Marcial flow are 580,000 and 1,030,000 acre feet, respectively * * *. Return flow in the Conservancy District, as indicated by the total measured discharge of interior drains in 1936, was 28 percent of the gross diversions. Data were not available on net diversions * * *. The sources of ground water in the Middle Valley are underflow from the mesas on either side and seepage from the river, canals, and irrigated lands. In most areas, seepage from irrigated lands is the principal source, and the water in interior drains is largely derived therefrom. On the other hand, the river is, without doubt, the source of most of the water in the riverside drains. Meager data indicate a total annual underflow from the mesas of between 50,000 and 100,000 acre feet.

Streamflow records since 1935 have provided a basis for extending several of the graphs of figure 2, which for the years before 1935 are based on data from the Rio Grande Joint Investigation. But the available data are insufficient for accurate calculation of the annual inflow to the Middle Valley, which would be desirable for evaluating the effects of drought. One important element in the hydrologic equation is ground water, and concerning it there is very little information. Conover (1956, p. 12) has summarized the ground water situation and need for additional data in the following statement:

In contrast to areas where water essentially is being mined, there are certain areas in the State, particularly along the Rio Grande in places such as the Rincon and Mesilla valleys, where ground water reservoirs are or can be replenished from surface water supplies.

In such areas efficient utilization of the ground water resource revolves around the long term availability of surface water taking into account the need of downstream users, the capture of water being wasted by native vegetation, and maintenance of soil moisture salinity at a safe level.

In other words, in such stream valleys the total water supply must be considered as a unit, ground water plus surface water. Full integration of the ground water and surface water use in stream valleys apparently could increase measurably the amount of water dependably available for beneficial use.

The ground water reservoir in the Rio Grande valley is very large when compared with present surface reservoirs constructed in the state.

For instance, in the middle Rio Grande valley, it is estimated that nearly half a million acre feet of ground water is stored within 100 feet of the surface under each area of valley floor equivalent to a township (36 square miles). In other words, there is more water stored under five townships than can be stored in Elephant Butte reservoir.

Underground storage generally has the advantage of being relatively immune to direct evaporation losses, a major item in surface reservoirs in this dry country. Because of the large underground storage, utilization of the ground as a regulating reservoir would result in a firmer supply, during droughts, than could be obtained through manmade surface reservoirs alone.

Full utilization of the ground water reservoir in the Rio Grande valley would result in an appreciable lowering of water levels during droughts. This would have a threefold effect :

• (1) Waste of water by water loving plants would be measurably reduced, resulting in an effective increase in water supply;
• (2) the quality of the ground water would deteriorate temporarily, owing to cessation of drain flow ; and
• (3) nearly all water users would of necessity use ground water to secure a dependable water supply.

One problem peculiar to the Middle Valley is the ground water development during the recent drought years in several tributary areas. Some of this development, which doubtlessly was spurred by deficiencies in precipitation, runoff, or spring flow during the drought, provided water for new irrigated areas or industrial uses. Recent studies in the following areas have been concerned chiefly with the water resources in the immediate vicinity of the development, but there has also been some consideration of the effect of pumping from wells upon the flow of the Rio Grande in the Middle Valley.

Southward from the Colorado New Mexico State line the Rio Grande trough (p. D6) is filled partly by basalt lava from volcanoes within the trough, and partly by alluvium derived from the Sangre de Cristo Mountains. The lava forms a plateau occupying the western two thirds of the trough, and is interbedded with the alluvium in most of the eastern third. The alluvium forms a piedmont alluvial plain 7,400 to 7,800 feet above sea level, which extends from the basalt eastward 6 or 8 miles to the base of the Sangre de Cristos. The part of this piedmont that extends from Costilla Creek southward for about 20 miles is called Sunshine Valley. The Rio Grande avoids Sunshine Valley by flowing in a canyon that passes west of two of the extinct volcanoes (Ute and Guadalupe Mountains) but elsewhere the Rio Grande is generally within a mile of the edge of the valley alluvium.

Several irrigation wells were drilled in Sunshine Valley in 1947. These wells were so successful that by 1955 there were 44 wells which pumped about 3,500 acre feet of water for irrigation. Although the Rio Grande does not traverse Sunshine Valley, numerous springs rise along its channel just west of the valley. The possible effect of this ground water development upon inflow to the Rio Grande was therefore an important objective of a recent study (Winograd, 1959). This study included only that part of the piedmont plain in New Mexico, although some irrigation wells are also in the northward extension of the plain in Colorado.

D10

The flow in the Rio Grande and its tributary Red River is known on the basis of records from gaging stations to increase in the reach opposite Sunshine Valley. From these and other data it is estimated that the average ground-water accretion between Ute Mountain and the south end of Sunshine Valley is about 80,000 acre-feet annually, of which about 20,000 comes from the ground-water reservoir underlying Sunshine Valley.

The irrigation wells in Sunshine Valley all obtain water from sand and gravel beds in the alluvium, and yield 600 to 3,000 gpm. Many wells have penetrated more than 400 feet of alluvial materials, but near the western border of the piedmont plain the alluvium thins to less than 50 feet near its contact with the basalt. As much as 100 feet of fine-grained lake beds intervene between the basalt and the alluvium in the west-central part of the valley. The basalt is far more permeable than the average alluvium penetrated by irrigation wells, but to date it has not been tapped by wells.

The natural pattern of ground-water circulation in Sunshine Valley (Winograd, 1959, p. 30-34) involves

• (1) recharge by percolation into the alluvium of water from streams, irrigation ditches, irrigated lands, and direct precipitation, amounting to an estimated 20,000 acre-feet in an average year;
• (2) westward movement within the moderately permeable alluvium under unconfined conditions;
• (3) also downward movement into the basalt, in many places through intervening less permeable lake beds so that the water in the alluvium is semiperched with respect to the basalt;
• (4) westward movement in the exceedingly permeable fractures and interflow zones in the basalt; and
• (5) discharge into the Rio Grande, whose canyon is cut below the level of saturation in the basalt on both sides of the river.

It is concluded that pumping of ground water in Sunshine Valley, on the lava-capped plateau, or in Colorado to the north eventually will reduce the inflow to the Rio Grande by an amount equal to the consumptive use of the water. Owing to the large permeability of the lavas the effect of pumping should rapidly affect accretion to the river unless some wells pump from parts of the alluvium where the underlying lake beds are virtually impermeable-in which case pumping would be from storage for a while and the effect upon the river would be delayed. During the 1955 irrigation season, total pumpage for irrigation in Sunshine Valley (not considering return seepage of the water pumped) is estimated to have been equivalent to 15 or 20 percent of the accretion to the river derived from the ground-water reservoir beneath the valley.

The Santa Fe area as described by Spiegel and Baldwin (1962) is the area upon which the city of Santa Fe and environs depend for water supply, and is also a representative sample of the water-producing area of northern New Mexico. This area is in the eastern part of the Rio Grande trough, and ranges in altitude from 6,000 feet in the southwestern part to 12,400 feet in the Sangre de Cristo Mountains which forms its eastern border. The mountains, which are forested, comprise the eastern fifth of the area, and piedmont plains (grasslands) the remaining four-fifths.

The Santa Fe area has a long history of water use, for the springs and cienagas (areas of spring discharge, locally spelled "cienega") were in use by Indians in 1598 when Spaniards first explored the area. The Spanish had wells for domestic use in 1716, and were using stream water for irrigation in the mid-1700's. Spiegel found a linear relation between precipitation and the logarithm of the altitude- the precipitation increased from an annual average of 13 inches at 6,000 feet to 45 inches at 12,000 feet; this relation is especially observable in winter. The surface runoff from this precipitation averages 0.5 inch (4 percent of the precipitation) in the piedmont plains and 5.8 inches (23 percent of the precipitation) in the mountains. Long-term average yield is therefore 12,000 acre-feet from 24 square miles of mountains and 10,000 acre-feet from 107 square miles of plains. Of this total yield the average contribution to the Rio Grande would be of the order of 10,000 acre-feet under natural conditions. The Santa Fe River, which drains about 60 percent of the Santa Fe area, under natural conditions has an average annual discharge of about 6,800 acre-feet, and the flow is well sustained because its basin includes a comparatively large area above 10,000 feet where snowpack and glacial sediments store large quantities of water.

By 1880 the uncontrolled flow of the Santa Fe River had become inadequate for Santa Fe's needs, and since that time the increasing population has required progressively increasing development and use of storage, as indicated in figure 4. Total surface- reservoir capacity since 1947 has been 4,100 acre-feet, which is only slightly greater than the current annual municipal water requirement.

The unconsolidated sand and gravel of the Santa Fe group of Tertiary age forms a productive ground-water reservoir underneath Santa Fe. Three municipal wells were drilled in 1946 and four others by 1950, but these have been used only in years of inadequate surface supplies. Several irrigation wells

D11

D12

also have been drilled since 1947, and in 1952 there were 10 large-capacity wells in the Santa Fe area, of which 4 were pumped for 6 to 8 months for irrigation and another was used throughout the year. Pumpage from these wells was of the order of 4,600 acre-feet.

According to Spiegel (Spiegel and Baldwin, 1962)- * * * As the runoff of the Santa Fe River in wet years is in excess of that which can be stored in the reservoirs (about 4,000 acre-feet) and used ; therefore, there are occasional periods of spill. The average annual supply available above Santa Fe, under present conditions of storage and use, is about 5,600 acre-feet. For periods as long as 6 years, the average annual discharge of the Santa Fe River has been less than the sum of current use and reservoir losses. The surface discharge of the Santa Fe River must be supplemented by ground-water supplies in many dry years, and if the demand increases as much as 15 percent in the future, supplemental supplies of ground water will be required in most years. Water-level and pumpage data for the city well field suggest appreciable mutual interference by existing wells, amplified by boundary effects. Excessive declines of water level in the easternmost wells in 1 year of pumping are due to boundary effects and to overpumping of individual wells. The surplus surface flow of water of good quality in wet years could be used to replenish the ground-water supplies withdrawn in dry years by (1) regulating the overflow from the reservoirs so as to allow natural infiltration into the bed of the Santa Fe River where the stream crosses relatively permeable sediments, (2) spreading the water on the broad terraces adjacent to the stream in this reach, or (3) direct injection into wells. The installed capacity of existing irrigation wells and city supply wells in and near the city of Santa Fe already exceeds the annual average recharge and ground-water inflow, but actual average withdrawal will probably remain below the inflow and recharge for many years to come.

The Rio Puerco drains about 20 percent of the upper Rio Grande basin but contributes only about 5 percent of the water to the Rio Grande above Elephant Butte. This meager contribution is chiefly stormflow, and the Rio Puerco is ordinarily responsible for more than half the sediment that is carried into Elephant Butte Reservoir. Thus the Rio Puerco is of rather minor and dubious value to the Rio Grande system.

The Grants-Bluewater area is a headwater area in the Rio Puerco basin. Bluewater Creek rises on the northeastern flank of the Zuni Mountains and trends generally northeastward and eastward to join the Rio San Jose north of the village of Bluewater. The Rio San Jose, usually dry above the city of Grants, trends generally southeastward to join the Rio Puerco. Blue-water Lake, with capacity of 46,000 acre-feet, is formed by Bluewater Dam which was constructed in 1927. Uranium deposits of major economic value have been located and are being mined in the region northwest to east of Grants, and the area in recent years has been increasing its use of water for industrial and municipal purposes.

Various hydrologic records for the region, presented graphically in figure 5, indicate the following :

• (1) precipitation was less than the long-term mean in 17 of the 20 years 1937-56-that is, in all years except 1940, 1941, and 1949;
• (2) in the 8 years 1937-44, in spite of this prevailing precipitation deficiency, the quantity of water available from Bluewater Lake for irrigation exceeded 7,000 acre-feet in each year except 1940, doubtless in part because of holdover storage from 1935-36 and 1940-41 when precipitation was greater than normal;
• (3) after 1945, water from the reservoir was used for irrigation only in 1948, 1949, and 1952, whereas in all other years the flow past the gage served chiefly to recharge ground water;
• (4) pumping of ground water began in 1945, and the combined use of ground water plus surface water has exceeded 10,000 acre-feet in each of the 11 years 1947-57;
• (5) the irrigated area fluctuated from 2,000 to 4,300 acres in years before 1945, according to the availability of water from Bluewater Lake, and ranged from 4,500 to 6,000 acres in 1946-55 when wells provided a more stable supply;
• (6) beginning in 1955 there has been increasing use of water for industrial and municipal purposes, but this has been balanced by decrease in irrigation use upon a decreasing area.

• (1) from surface water to ground water as the primary source, at least during drought, which provided not only increased stability of supply but some increase in supply; and, more recently
• (2), an increase in municipal use but no overall increase in use.

The development of ground water has included withdrawal of a considerable quantity from storage. The principal aquifer tapped by irrigation wells is the San Andres limestone of Permian age, which is also a major aquifer in the Pecos River basin (p. D50). In the Grants-Bluewater area the San Andres limestone and the immediately underlying Glorieta sandstone, formerly considered a part of the San Andres, constitute a single aquifer 250 to 300 feet thick having a gentle northeastward dip. Recharge to the aquifer comes from precipitation upon the outcrop area and upon alluvium or basalt where they overlie the aquifer, and from seepage of water from Bluewater Creek, Bluewater Lake, irrigation canals, and irrigated lands. Water is discharged naturally from the aquifer by springs and by evapotranspiration in swampy areas south of Grants. The first successful irrigation well in the area was completed in August 1944, and by 1954 a total of 23 irrigation wells, 3 industrial wells, and 4 municipal wells were in use in the area. Since

D13

D14

1946, the first year of record, ground-water levels have declined 40 to 45 feet under the area north of the town of Bluewater, whereas the decline has been only about 18 to 20 feet under the irrigated area from Bluewater southeastward to Grants. In the upper part of the area where surface water was available for irrigation in 1948, 1949, and 1952, recharge derived from leakage of canals and of the channel of Blue-water Creek temporarily retarded or reversed the downward trend in water levels.

Since the entire history of ground-water use in the Grants-Bluewater area coincides with the Southwest drought, it is difficult to draw conclusions as to the extent to which the present development can be sustained by the average climate.

Pumping is a new discharge imposed upon a previously more or less stable ground-water system, and water levels are expected to decline as long as large-scale pumping continues or until the depletion of storage in the San Andres limestone causes a corresponding reduction in natural discharge by springs and evapotranspiration. It is probable that in some years enough water will be available in Bluewater Lake to provide an adequate amount of surface water for irrigation in the area, in which case the ground-water reservoir will be replenished to some extent. Also, in those years having an adequate supply of surface water, ground-water pumping will be reduced if additional lands are not brought under irrigation. When storage is replenished by surface water, it will be at the expense of flow in one of the tributaries of the Rio Grande. It is doubtful that such depletion will have significant effect upon Rio Grande flow, however, because of the numerous opportunities for evapotranspiration in reaches of the Rio San Jose and Rio Puerco.

The Elephant Butte-Fort Quitman area extends along the Rio Grande from San Marcial, N. Mex., to Fort Quitman, Tex., a distance of about 250 miles. In the upper 65 miles, between San Marcial and Caballo Narrows, the flanking hills are close to the river; Elephant Butte Reservoir occupies the upper 40 miles, and Caballo Dam forms a smaller reservoir in the lower part of this reach. Below Caballo Dam the river enters Rincon Valley, 30 miles long and as much as 2 miles wide, and then traverses a short canyon before entering Mesilla Valley, which extends southward 55 miles and has a maximum width of 5 miles. About 4 miles above El Paso the Rio Grande flows through "The Pass" and enters El Paso Valley, which extends about 95 miles to Fort Quitman. Since 1925 the irrigated area in the Elephant Butte-Fort Quitman section has ranged from about 160,000 to 225,000 acres, of which 130,000 to 180,000 is within the United States. In 1946 the irrigated land in the Rio Grande project of the U.S. Bureau of Reclamation (which does not include the lower 55 miles of the El Paso Valley) was 157,000 acres, of which about 17,000 was in Rincon Valley, 84,000 in Mesilla Valley, and 56,000 in El Paso Valley.

When Elephant Butte Dam was completed in 1915, the reservoir had a usable capacity of 2,635,000 acre-feet, but deposition of sediment had reduced this to 2,185,000 acre-feet by 1951. Additional storage and regulation of the river is provided by Caballo Reservoir, completed in 1938 with a capacity of 346,000 acre-feet. The annual inflow has exceeded reservoir capacity only in 1920 and 1941, and the reservoirs thus afford almost complete regulation of the river. Figure 2D shows the storage in both reservoirs at the end of each calendar year, and thus depicts the quantity of water that is held over in each year for use in the subsequent year. These reservoirs stabilized the supply for downstream users during the alternating drier and wetter years of the droughts of the 1930's and 1940's, when they furnished a fairly consistent supply of 650,000 to 900,000 acre-feet annually. However, beginning in 1951 the reservoirs served only a small part of the needs of the Rio Grande project. A graph of cumulative inflow to Elephant Butte Reservoir (fig. 6) shows the effects of alternating wetter and drier periods, with steeper trend during the wet periods 1919-24 and 1941-42 than in the dry periods 1925-40 and 1943-56, and with a lesser rate of inflow during the most recent drought than in earlier drought periods. The rate of sediment accumulation, as shown by reservoir surveys at intervals of 5 to 10 years since 1916, is not clearly related to the observed fluctuations in climate and inflows of water. Instead the sediment accumulation has been at a progressively reducing rate since the reservoir was placed in operation.

In 1946, when the inflow to Elephant Butte Reservoir was less than in any previous year except 1934, only 11 wells were pumped for irrigation in Mesilla and Rincon Valleys. In 1947, which was almost as dry, the number of irrigation wells increased to more than 50. With continued deficiencies of surface water the ground-water development increased, until by 1955 there were an estimated 1,600 irrigation wells in

D15

A detailed investigation of the ground-water reservoirs of Rincon and Mesilla Valley (Conover, 1954) showed that most wells obtain water from alluvial sand and gravel beds underlying the flood plain to depths of 60 to 100 feet in Rincon Valley, and as much as 200 feet in Mesilla Valley. The ground water in this alluvium is very closely related to the river : it is recharged chiefly from the river, either directly from the channel or from canals and ditches or from river water applied to the land for irrigation; and it is discharged either by evapotranspiration within the valleys or by flow of drains to the river, which in years of normal irrigation supply may include about 250,000 acre-feet discharge through the extensive system of drainage ditches in both valleys. Thus it is concluded (Conover, 1954, p. 2) that "ground water obtained by pumping in the Rincon and Mesilla Valleys does not represent an additional supply or new source of water to the project, but rather a change in method, time, and place of diversion of the supplies already available."

Monthly measurements of water level have been made for a number of years by the Bureau of Reclamation in a network of shallow wells in Mesilla Valley. The hydrograph of average depth to water below land surface since 1946 in 39 of the wells (fig. 7) shows that before 1951 the water table rose to within 6 or 7 feet of the land surface from April to September each year as a result of irrigation from canals, and then dropped a foot or two during the winter. In 1951, however, with a shortage of supplies from Elephant Butte, there was no appreciable rise of the water table during the irrigation season. In 1952 and 1953 the water table was lowered by pumping in the early part of the irrigation season, and then rose

D16

In El Paso Valley (Smith, 1956) about 500 irrigation wells yielded 120,000 acre-feet of water for irrigation of approximately 45,000 acres in 1954; also, 10 large wells were used to produce water for public and industrial supply. This pumping reduced the storage in the ground-water reservoir in El Paso Valley. Water levels were lowered in all 19 observation wells distributed over the valley ; in 10 wells the decline exceeded 3 feet, and in 1 well the water level was lowered 8½ feet during the calendar year 1954. In the lower part of El Paso Valley, below the Rio Grande project lands and where surface water was particularly deficient, water levels are reported to have declined nearly 20 feet in some areas during the period 1950-55.

Throughout the upper Rio Grande basin, each drought has given rise to analysis of water problems, to questions concerning the adequacy of the natural resources to meet the developed needs for water, and to action intended to achieve better balance between supply and demand in the future. The drought of 1892-1904 brought forth the "embargo" of 1896, the treaty of 1906 with Mexico, and plans for the construction of Elephant Butte Reservoir. The drought of 1930-40 spurred the comprehensive Rio Grande Joint Investigation and the subsequent Rio Grande Compact of 1938 between Colorado, New Mexico, and Texas. The most recent drought has been largely responsible for increased utilization of ground-water storage, and has led to further analysis of several problems confronting individual localities and the basin as a whole.

Storage is a major problem because storage of water is a prime requisite for effective use of the highly varying quantities yielded by precipitation. Here we are stressing the storage that can provide fairly constant quantities for use throughout series of wet and dry years, and not concerning ourselves with seasonal storage such as would be required for irrigation in August by water from snow melted in May.

Even if adequate storage capacity is available, there may still be complex problems concerning the conveyance of that water to the reservoirs or other points of distribution. Some water may be "lost" to ground-water reservoirs along the route, and if that loss is determined quantitatively it can perhaps be charged to the users of the recharged reservoirs. There may

D17

also be losses from the river because of water that is consumed by riparian vegetation, which does not benefit anyone and is therefore wasted. This consumptive waste is a natural phenomenon, of course, and it is greater in amount during wet years than drought years; but it receives especial attention in drought because of the deficiency of supplies available to water users.

The quality of the water in the upper Rio Grande basin has generally been poorer in years of drought than in years of more abundant water supply. Inasmuch as there is, at any particular time, a progressive increase in total salt concentration of the river water from the upper to the lower limit of the basin, it is evident that the problem of quality is more critical to the downstream users than to those near the head-waters.

Finally, apportionment of water may be more important and also more difficult during drought when supplies are less than normal. Apportionment is a prime objective of international treaties, interstate compacts, State laws, and court decisions pertaining to water. Current problems of storage development, water salvage, water quality, and apportionment are summarized in the following sections.

It is perhaps trite to remark that history is generally written from a human viewpoint. Histories purportedly concerned primarily with water are characteristically anthropocentric : they tell when man came, when he saw it, and when he conquered it; and generally give for each storage facility details as to when he saw the need, raised the money, constructed it, dedicated it, and began to use it.

A history written from the viewpoint of the water- a sort of "hydrocentric" viewpoint-would in most places not differ from the traditional history. But in the upper Rio Grande basin there would be some marked differences. From the viewpoint of the water, the chronologic order of development of major storage facilities was

• (1) San Luis Valley;
• (2) surface reservoirs ; and
• (3) Mesilla, Rincon, and El Paso Valleys.

It is not possible to give proper credit for the development of the largest reservoir in the upper Rio Grande basin, for there is no place to put a cornerstone, and no name to engrave thereon-no man to claim it as a scientific or political achievement, no government agency to embrace it as an essential element in optimum utilization of the water resource. At this late date, almost 80 years after the initial development by man, it appears that the reservoir was created chiefly by unenlightened self-interest.

On the basis of the estimate by Powell (1958, p. 89) that it contains about 3 million acre-feet under 20 percent of the valley area, the unconfined aquifer in San Luis Valley must store water equivalent to several times the capacity of Elephant Butte Reservoir. And, although some of this storage was natural accumulation from precipitation and from leakage of artesian aquifers, most of it is artificial in the sense that it resulted from diversions from the Rio Grande. Depletion of the river flow was inevitable during the filling of this ground-water reservoir, but this did not cause much protest from downstream water users until the effects of artificial depletion were combined with the effects upon natural streamflow of the drought beginning in 1892.

The unconfined aquifer of San Luis Valley is an offstream reservoir : water stored in it moves away from the river and toward a closed basin which is a major area of natural discharge. The closed basin includes the 1,700 square miles of San Luis Valley which is separated by a low divide of alluvial materials from the part of the valley tributary to the Rio Grande. The lowest part of this closed basin is close to the foot of the Sangre de Cristo Mountains on the east side of the valley, and is clearly defined by a succession of alkali flats and by a chain of lakes of which San Luis Lake is the largest. The water table in the closed basin in 1936 had the form of a closed depression, with water moving toward a low point somewhat south of San Luis Lake. The depth to ground water under 90 percent of the valley floor did not exceed 8 feet, and under 70 percent it was less than 5 feet.

The quantities diverted from the river to the closed basin do not reappear in the river at some future date, and thus San Luis Valley has not served as a stabilizer of river flow. Instead, the San Luis Valley's economy, requiring fairly constant amounts of water in wet and dry years, has aggravated the problem of providing a stable supply for use downstream. For example, the inflow to San Luis Valley in 1946 was about half as great as in 1944, but the outflow was less than one-fifth as large. The water supplies for the Middle Valley are far more variable from year to year than are those for San Luis Valley, partly because of this artificial increase in the amplitude of fluctuations in Rio Grande runoff; and also, as shown by Gatewood and others (1963) because there is greater variation in the natural runoff of tributaries entering the river

D18

below San Luis Valley. Typically the natural and artificial fluctuations coincide to make wet years wetter and dry years drier in the Middle Valley than in San Luis Valley. Thus in the wet year 1941 the calculated stream inflow to San Luis Valley was about 3.6 times the inflow in the dry year 1951, but the inflow to the Middle Valley (measured at Otowi Bridge) was 6.7 times as great in 1941 as in 1951.

In years of severe or protracted drought, the streamflow available for diversion into San Luis Valley is insufficient for the requirements of subirrigation of crops. Wells have been pumped to overcome this deficiency, and the pumping has created both local and regional problems. A local problem can be visualized where a man having a primary right to surface water is striving mightily to build up the water table for subirrigation, while his neighbors lacking similar rights are pumping water for irrigation and pulling the water table down in the process; such cross purposes can breed cross neighbors. A regional problem is exemplified by 1952, when the depletion of the Rio Grande in San Luis Valley was 1¼ million acre-feet, greater than in any other year and attributed in part to replenishment of the unconfined aquifer which had been pumped heavily in the preceding 2 years.

Without doubt a reservoir at Elephant Butte was desirable for regulation of the natural flow of the river, but it became essential after the development in San Luis Valley, because that development caused significant river depletion without providing any stabilization of the flow remaining in the river. After it was completed in 1915, Elephant Butte Reservoir released at least 650,000 acre-feet of water each year until 1951, although the annual inflow to the reservoir was less than 500,000 acre-feet in 9 of those years. In 40 years of operation, during which the average inflow was nearly a million acre-feet a year, Elephant Butte proved to be capable of storing for subsequent use all the surplus waters of wet years except during the consecutive years 1941-42. After the filling in 1942, the reservoir was able to provide practically normal supplies for irrigation for 8 consecutive years, during which the inflow was equal to the long-term average in 1944, 1948, and 1949, slightly less than average in 1945, and less than 50 percent of average in the other 4 years. By 1951, however, nearly all the carryover storage had been used, and water requirements for irrigation in the Rio Grande project could be met only where water could be pumped from ground-water reservoirs.

There is little surface-reservoir capacity above Elephant Butte Reservoir. Reservoirs on tributaries to San Luis Valley have aggregate capacity slightly greater than 370,000 acre-feet, but 142,000 of this is in the basins of Costilla, Culebra, and Trinchera Creeks, whose water supplies have long been completely utilized within their basins and contribute practically nothing to the Rio Grande. Even less storage is available for use in the Middle Valley, which is limited essentially to the 200,000-acre-foot capacity of the Middle Rio Grande Conservancy District's El Vado Reservoir.

The development of the ground-water reservoirs in the valleys downstream from Elephant Butte was spurred by the deficiency of surface supplies during the recent Southwest drought, just as the development of Elephant Butte Reservoir was spurred by the deficiency of runoff in an earlier drought. On the basis of ground-water studies, confirmed by the record for the dry years 1951-56, Rincon, Mesilla, and El Paso Valleys have ground-water reservoirs of sufficient capacity to provide supplementary supplies for at least several consecutive years of deficient streamflow.

Pumping in Rincon and Mesilla Valleys may be responsible for increased loss from canals of the Rio Grande project below Caballo Dam. These transmission losses were low before 1950 but had increased to about 65 percent in 1955 and to 75 percent in 1956. The increasing conveyance loss is indicated also by the fact that 544,000 acre-feet released in 1952 was sufficient to provide 2.75 acre-feet per water-right acre, but 247,000 acre-feet released in 1956 provided only 0.3 acre-foot per water-right acre. Reduction in drain outflow has been accompanied by increasing salinity of soils and shallow ground water, because of accumulation of salts that had been dissolved in the water applied for irrigation. Essentially the farmers in Mesilla Valley with their wells have developed a new reservoir having a capacity that may be on a par with that of Elephant Butte Reservoir, and have assured themselves of supplies throughout a drought that overtaxed the regulatory ability of Elephant Butte. In such a development there is also a possibility of creating problems of salt accumulation and of reservoir depletion that may continue to be troublesome even after increased supplies again become available from the river-both in Mesilla Valley and in deliveries of water to users farther downstream. However, the rapid refilling of the Mesilla Valley ground-water reservoir in 1958 (fig. 7) is an encouraging sign that the present development does not exceed the capabilities of the system for perennial yield.

D19

Conover (1954, p. 3) concluded in advance of the recent ground-water development that- in a hypothetical year having only 50 percent of a normal supply of surface water available for diversions, the project lands would require an additional acre-foot per acre of water from wells to assure successful irrigation of the crops. However, because of the reduction in flow of the drains caused by pumping and because of losses in distribution, the use of water from wells to supply this deficit would require pumping 2.42 acre-feet per acre, or 213,000 acre-feet a year for the 88,000 acres of water-right land in New Mexico. Of the amount pumped, it is calculated that all but 63,000 acre-feet would be diverted from surface-water flow. If supplemental pumping were resorted to for 5 successive dry years, continued pumping would be necessary for 3 to 4 years after a return to normal surface supply so as to permit bypassing of the required share of water to the El Paso district, awaiting the restoration of ground-water storage by recharge from surface water.

The 5 dry years hypothesized by Conover occurred in 1951-56.

Obviously not all the seepage from streams into ground-water reservoirs can be attributed to pumping from wells, and the evaluation of the proportions of river losses due respectively to natural causes and to man's activities requires a large amount of hydrologic data and a good working knowledge of the interrelations of surface and ground water. Partial information concerning these relations commonly raises more questions than it answers.

In spite of the complex problems in water regulation generated by the recent utilization of ground-water reservoirs, more complete and more flexible utilization of the water resources is possible because of ground-water development. The combined capacities of surface and subsurface reservoirs are sufficient to overcome the effects of long and intense droughts. On the other hand, with the total facilities for storage now available, it is unlikely that Fort Quitman will again see annual runoff as great as a million acre-feet from the upper basin.

Consumptive waste, or "nonbeneficial consumptive use," of water has been a major problem throughout the upper Rio Grande basin for a long time. It was concluded during the Rio Grande Joint Investigation (National Resources Committee, 1938, p. 92, 121) that in 1936 the streamflow depletion by irrigated acreage -the "directly beneficial consumptive use"-was slightly less than half the total streamflow depletion in the upper Rio Grande basin; the other half was due to evaporation from water surfaces and bare lands, and especially to transpiration by native vegetation, which was responsible for the great bulk of the consumptive waste. The San Luis section, Middle section, and Elephant Butte-Fort Quitman section were each responsible for about one-third of the total consumptive waste from the upper basin. The computations for San Luis Valley alone in 1936 were 476,000 acre-feet of water used by irrigated lands and 636,000 acre-feet consumed nonbeneficially; similar data are not available for subsequent years.

In each of the principal valleys of the upper Rio Grande there are extensive areas where the water table is less than 10 feet below the land surface, and where conditions are therefore favorable for consumptive waste of large quantities of water by evaporation and by transpiration of native vegetation. In several areas the shallow ground water is part of the natural environment, maintained in large part by the river even before the advent of white man, and forming swampy areas or "bosques" having a dense cover of vegetation. A sample of this environment is preserved for future generations in the Bosque del Apache National Wildlife Refuge north of San Marcial, N. Mex.

In most areas where surface water has been used for irrigation the water table has risen to within a few feet of the surface, some irrigated lands have become waterlogged, and drainage systems have been constructed to lower the water table. In San Luis Valley the construction of drains began in 1910, and their general effect has been to lower the water table a few feet without changing its form, and to maintain it at a more or less uniform depth under the drained area; in July 1936 (National Resources Committee, 1938, p. 13, 226) the depth to water was less than 5 feet in most of the irrigated area of San Luis Valley. In the irrigated lands of the Middle Valley Conservancy District the water table became high during years of abundant water supply in the 1920's and necessitated construction of an extensive drainage system, which was completed between 1930 and 1935. In 1936 the water table over the entire district was 3 feet lower, on the average, than in 1927 before the drainage construction. Even so, the depth to water in October 1936 was less than 4 feet in 15 percent of the total valley area, less than 6 feet in 61 percent, and less than 8 feet in 89 percent of the Middle Valley. In Mesilla and Rincon Valleys the average depth to water prior to 1950 was 9 to 10 feet in the winter, and 1 or 2 feet less during the irrigation season; this annual cycle, with minor variations, has been observed since water levels in wells were first measured in 1925.

During the period of the Southwest drought, the water table has been lowered by pumping in many valley areas that had previously been irrigated exclusively

D20

by surface water, as in San Luis Valley (fig. 4), Mesilla Valley (fig. 7), El Paso Valley, and doubtless several localities in the Middle Valley. It is assumed that consumptive waste in these areas also has been reduced, but quantitative data are not available.

Drought was one factor that led to the construction of a low-flow channel and cleared floodway for the Rio Grande in a reach of 71 miles above the Elephant Butte Reservoir for the purpose of reducing consumptive waste and thus salvaging water for the Rio Grande project. The low-flow channel, of which 31 miles was completed by 1954, has a capacity of 2,000 cfs and traverses an area of shallow water table and high consumptive waste by phreatophytes. In 1955 the inflow to Elephant Butte Reservoir was about 45,000 acre-feet greater than would be expected on the basis of stream-depletion data before the channelization. Part of this water was salvaged from consumptive waste and part was a nonrecurring benefit that resulted from dewatering sediments along the channel.

As pointed out by Robinson (1957, p. 1), "phreatophytes are plants that depend for their water supply upon ground water that is within reach of their roots. Although not confined to the arid regions of the western United States, their occurrence there is commoner, more spectacular, and, because of their effect on water supply, more important than it is in humid or subhumid regions." In the upper Rio Grande baisn, phreatophytes are responsible for most of the consumptive waste of ground water and thus for a large part of the total streamflow depletion. Phreatophytes include many types of plants not genetically related and whose only common bond is their penchant for using ground water. Included in the group are the cottonwood and willow trees that are well-known markers for watercourses throughout the West. Alfalfa is probably the phreatophyte of greatest economic value, and its high consumption of water is always rated as "beneficial consumptive use" rather than the "consumptive waste" charged to most phreatophytes.

In the Rio Grande basin, by far the most abundant, prolific, and aggressive phreatophyte is the saltcedar (Tamarix gallica and T. pentandra). Saltcedar is in direct competition with man for the limited water supplies of the Rio Grande, and so far has been winning. Saltcedar was first observed in New Mexico in 1910 in Mesilla Valley. By the time of the Rio Grande Joint Investigation in 1936 saltcedar covered about 5,500 acres in the Middle Valley; in 1947 a survey by the U.S. Bureau of Reclamation showed an infested area of about 26,000 acres, and by 1960 the area of infestation had increased to about 60,000 acres in the Middle Valley and tributaries. The significance of saltcedar in this report on drought in the Southwest is that the increasing area and density of infestation, and the accompanying increase in consumptive waste of water, have been going on throughout the drought period. Some of the shortage of water for beneficial use during these dry years was thus caused by saltcedar rather than by climatic fluctuations.

According to analyses made throughout the year 1936, the water in the Rio Grande at Del Norte was of excellent quality-a natural calcium bicarbonate water containing less than 100 ppm of dissolved solids. At Fort Quitman at the other end of the upper basin, in that same year, the water had more than 2,000 ppm of dissolved solids, chiefly sodium chloride and sulfate. Thus the problems of water quality are concentrated in the lower part of the basin. Broadly speaking, the deterioration in quality is another aspect of aridity. Evapotranspiration within the upper basin-consumptive use plus consumptive waste-accounts for more than 90 percent of the runoff produced in the basin; all the materials dissolved by that water remain behind, in surface water or ground water, or in soil or rock material.

The progressive increase in total salt concentration of the river water from the upper to the lower limit of the basin is shown in figure 8. Weighted annual averages of dissolved solids of the Rio Grande since 1946 show that the water leaving San Luis Valley near Lobatos and that entering the Middle Valley at Otowi Bridge are calcium bicarbonate waters having dissolved-solids contents less than 300 ppm and changing little from year to year. At San Acacia in the Middle Valley, and at San Marcial at the head of Elephant Butte Reservoir, the concentration of dissolved solids is ordinarily at least twice as great as at the Otowi Bridge and there is a marked increase in sulfate, largely reflecting the occurrence of gypsum in the drainage basin tributary to the Middle Valley. The average concentration at both stations varies appreciably from year to year; it was 600 ppm or more in 1934, 1940, 1951, and 1954, all drought years of low runoff, and it was less than 400 ppm in 1942 when runoff was high.

The graph for El Paso (fig. 8A) represents the water available for irrigation on the lower part of the Rio Grande project and in the vicinity of Juarez, Mexico. The water is more mineralized than that coming into Elephant Butte at San Marcial, chiefly because of increase in sodium and chloride. The

D21

D22

records for El Paso show the stabilizing and mixing effect of Elephant Butte Reservoir, for in 20 years the water has contained an annual average of 1.0 to 1.2 tons of dissolved solids per acre-foot, except in wet 1942 when the content dropped to 0.8 ton per acre-foot, and in the dry years 1934, 1935, 1940, 1951, and 1954 when it exceeded 1.2 tons per acre-foot. The water in the river at Fort Quitman has a high concentration of sodium chloride, but has a far lower concentration of the less soluble calcium carbonate and sulfate than at San Marcial.

Figure 8B shows the fluctuations in total load of dissolved solids at various gaging stations along the river. The graphs, representing the product of total runoff (fig. 2) times concentration, show that the greatest load of dissolved matter is carried in years of high runoff, and the least in years of drought. The total load carried into Middle Valley at Otowi during drought years (1950-52) is almost as great as the total carried into Elephant Butte Reservoir; even in years of average runoff (1948-49) three-fourths of the total entering the reservoir originated above Otowi.

Comparison of the graphs of total load at San Marcial and Fort Quitman indicates the relation of salt inflow to and outflow from the Rio Grande project. Differences between these amounts (fig. 8C) indicate an increasing residue of salt left in the project area prior to 1941, a decrease from 1942 to 1946, and essential balance from 1947 through 1951. During the drought years since 1951 there has been a return to unfavorable salt balance, similar to that in later years of the 1930-40 drought. Wilcox (1957) shows that the salt balance in Rincon Valley was favorable (that is, output exceeded input) in 17 of the years 1934-53, but Mesilla Valley had a favorable balance in only 11 of the 20 years. Input exceeded output, and salts accumulated in Mesilla Valley in the drought years 1934-36, and in 5 of the 7 drought years 1947-53. In confirmation, Chang (1957) has tabulated the limited evidence that there has been a general increase in salinity of solids of Mesilla Valley in the period 1949-54.

The Hudspeth County Conservation and Reclamation District, southeast of El Paso Valley, is most widely affected by salinity; Chang reports that 75 to 80 percent of the 13,000 acres has saline or saline-alkali soils. The use for irrigation of drainage water from the Rio Grande project lands has been a contributing factor in the past accumulation of salts here. However the supply of waste water dwindled to 3,800-feet in 1955 and practically none in 1956, so that two-thirds of the district land was abandoned. The remaining 5,500 acres was irrigated by some 57 wells, and these have created another problem in quality. Lyerly (1957) reports that the average salt content of wells in lower El Paso Valley in 1954 was 3.2 tons per acre-foot of water, with a range from 1½ to 7½ tons per acre-foot. In Hudspeth County in 1955 the salinity of well water ranged from 2½ to 10 tons per acre-foot, and the average was about 5 tons; this average increased to 5½ in 1956. Such high-salinity waters are classified as unsuitable for irrigation in standards published by the U.S. Salinity Laboratory (1954). Lyerly (1957) states that if water contains 3 tons of salt per acre-foot, at least as much water is needed for carrying salt residues below the root zone as is needed for irrigation of a crop; and if the salt content is 4 tons per acre-foot, the water requirement for leaching is 3 times the irrigation requirement. With water of still higher salinity, maintenance of yield even of moderately salt-tolerant crops may be impossible.

The effects of drought upon the lower El Paso Valley include

• (1) reduction in surface water available for irrigation;
• (2) reduced outflow of soluble salts, as shown by measurements at Fort Quitman, and consequent accumulation within the valley ;
• (3) increasing use of ground water, which has been derived in part from water applied for irrigation and is therefore more saline than the surface water originally used; and
• (4) excess application of water to leach the salts from the soil, thus further increasing the content of salt in the ground-water reservoir.

To summarize the preceding paragraphs, a major asset of the upper Rio Grande is the facilities for water storage, partly in surface reservoirs but dominantly in ground-water reservoirs. These were tapped during the drought and served to offset the deficiencies in rainfall and runoff. The efforts to utilize all available water, including that in surface and subsurface reservoirs, were so successful that outflow from the upper basin was reduced to negligible quantities in the years 1951-57.

A major liability of the upper Rio Grande is the mineral matter dissolved by the water, which prior to 1951 was carried in the outflow at an average rate exceeding half a million tons a year. With practical cessation of outflow in 1951-57 this elimination of

D23

wastes also stopped. And although this cessation has so far occurred only in drought, it could become chronic if storage and use of water in the upper basin were such as to inhibit outflow even in periods of abundant rainfall and runoff. Of course, even under the best of conditions the outflow carries only a part of the total solids dissolved by water within the upper basin; the remainder accumulates wherever water is consumed and returned to the atmosphere in irrigated soils, and if leached by application of additional water, then in underlying ground-water reservoirs; and in areas of shallow water table, including notably the closed basin in San Luis Valley.

Consumptively wasted water must be listed as a liability, in that it promotes deterioration of soils and of ground water because of accumulation of saline residues; but it can be converted to an asset wherever means can be found for salvaging the water for beneficial use. Carrying of saline residues to places where they can do no harm could well be included among the beneficial uses of water.

Many of the problems that have loomed large during the drought have been studied extensively, and some have been found to be complex, but physical solutions are possible for most of them. Physical solutions, however, are not enough: a practical solution must conform to the systems that have been established and accepted by the people of the basin for apportionment of the water and for regulation of water use -or alternatively those systems must be modified to embrace management of the water resource for optimum use. In summarizing the problems of the upper Rio Grande, Duisberg (1957, p. 68) states:

The water problem is more than a matter of drought and it is apparent that shortages to existing works will continue even under conditions of normal precipitation. It is also obvious that central problems such as establishing a practical relationship between ground- and surface-water use, developing a policy for the retirement of marginal land and land about to be permanently ruined, creation of means for encouraging and applying the results of research, and informing the people throughout the watershed of their mutuality of interest, must be faced.

To date, intrariver relationships between projects have been characterized by self-righteous attitudes, dependence on legal force, and a search for legal loopholes. The confusion created by the Supreme Court ruling of 1957 in the Texas-New Mexico suit may encourage those who think in terms of loophopes, and technicalities, to the point of destroying any future basis for trust and cooperation between the projects. On the other hand, each state has a moral obligation as a signatory of the Compact regardless of technicalities. It has been suggested that certain changes in the Rio Grande Compact would be both realistic and fair in the light of experience of the past 16 years. This may be an opportune time for these to be considered. Eventually, however, the people of the Rio Grande Watershed must recognize the necessity of closer working relationships between the major sections than is envisioned by the Compact. In working out these relationships they may well have to initiate concepts and ideas never tried before anywhere else.

Among the very early titles to land in the Southwest are those granted by the King of Spain before 1821 and by the Government of Mexico before 1848; and of these the "pueblo colonization" grants are of especial interest here because they have been interpreted in some jurisdictions to include a right to the water needed for the future growth and expansion of the pueblo. Such rights would doubtless encourage colonization of an arid region, but as the use of water increases toward the limit that can be sustained by the natural resource the right may become difficult to maintain. In New Mexico the significance of the "open end" of pueblo rights was not realized until a recent decision by the New Mexico Supreme Court (Cartwright et al v. Public Service Co. of New Mexico) which construed the pueblo rights of Nuestra Senora de Las Dolores de Las Vegas, predecessor of the Town and City of Las Vegas, N. Mex.

Generally in New Mexico as well as in Colorado, water rights are based upon the appropriation doctrine, and are thus on the basis of priority of beneficial use. The systems of apportionment of water are separate for each State, and apply only to the water users within the respective States. The Rio Grande Convention of 1906 (Witmer, 1956, p. 408-412) and the Rio Grande Compact of 1938 (Witmer, 1956, p. 154-177) together constitute the basis of apportionment of the water among the States of Colorado, New Mexico, and Texas, and the United States of Mexico.

The Rio Grande Compact of 1938 apportions the water of the river among the three natural divisions of the upper basin described on pages D4-D16. Under "

As pointed out in the decision on El Paso County Water Improvement District no. 1 v. El Paso, 133 F. Supp. 894 (D.C.W.D. Tex., 1955, p. 909) : This Compact has a number of peculiar provisions. For example, the water New Mexico must pass to Texas is delivered not where the two States meet, but at San Marcial, New Mexico, more than 100 miles above the point where the Rio Grande leaves New Mexico. This delivery is made into the reservoir of the Elephant Butte Dam, the principal structure of the Rio Grande Project. Some of this water eventually goes to Mexico. The Compact, instead of leaving the Texas share of the water open for disposition under the general water statutes of Texas, plainly directs same for irrigation in the Project. A large part of the Project lands are in New Mexico and, consequently, this water delivered to Texas goes to irrigate not only Texas lands, but also New Mexico lands in the Project. The apparent reason for all this is that when the Compact was negotiated, the Rio Grande Project, in all of its far flung works and physical properties was, and for some time had been, superimposed on the Rio Grande and its adjoining valleys all the way from the Elephant Butte Reservoir in New Mexico to a point below Fabens in Texas and that fait accompli colored the whole Compact as between New Mexico and Texas. Perhaps the problem was handled in the only practicable way.

D24

its terms, Colorado is obligated to deliver to New Mexico water in the Rio Grande as measured near Lobatos, Colo., in each calendar year, on the basis of the flow as measured at specified index gaging stations in the headwaters and a relation that had been established between the flow at those stations and the flow near Lobatos, representing the outflow from San Luis Valley. Similarly New Mexico (that is, the Middle Valley) is obligated to deliver water to Elephant Butte Reservoir in accordance with a relation between the measured flow in the Rio Grande at Otowi Bridge (adjusted for storage as specified) and inflow to the reservoir as established by past records. The compact includes provisions for annual credits and debits, and for accrual of these credits and debits up to specified limits. Basically the compact undertakes to maintain the status quo by allocating to the three divisions of the upper Rio Grande basin the same proportionate flows at all stages of the river that had been received by those divisions over a period of years (1928-37) before the negotiation of the compact. Its provisions are not dependent upon storage, because they do not guarantee specific quantities to anyone (except indirectly by recognizing the obligations of the United States to Mexico and to Indian tribes). It recognizes existing surface storage, however, and includes some provisions pertaining to it. It also permits the development of additional storage, so long as that development does not reduce the proportional supplies available to the division downstream.

The terms of the compact have not been met during the recent years of drought. As of early 1957 the total unofficial debit of Colorado to New Mexico was about 350,000 acre-feet, and that of New Mexico to Texas about 530,000 acre-feet, although the compact states that Colorado's accrued debit must not exceed 100,000 acre-feet, and New Mexico's must not exceed 200,000 acre-feet, at any time. Texas in 1951 brought suit to force New Mexico to deliver debit water, but this suit was dismissed in February 1957 by a ruling of the U.S. Supreme Court, which did not touch on the legality of the compact or the validity of Texas' interpretation, but held that the United States Government should have been an indispensable party to the suit by virtue of its administration of 8,000 acres of irrigated Indian lands and its ownership of various structures in the Rio Grande Project. Thus the enforceability of the compact is in doubt.

Because of the shortage in deliveries of water from Colorado and from New Mexico, Mexico has in some years received less than the 60,000 acre-feet annual allotment under the terms of the Treaty of 1906. The treaty provides that in case of extraordinary drought Mexico's quota may be diminished in the same proportion as the water delivered to lands on the American side, and accordingly Mexico received about 8,200 acre-feet in 1955 and 7,800 in 1956. In recent years more than 25,000 acres of formerly irrigated land in Juarez Valley has been abandoned or retired from production; about 27,000 acres has been irrigated from wells and only 2,500 from the river.

Although perhaps 90 percent of the water users in the upper basin-including those in New Mexico, in Texas, and in Mexico-received during the drought less than the share of water that was apportioned to them by interstate and international agreements, relatively few were forced to abandon their enterprises that depended upon water. Many were able to continue through the drought years with no diminution of water supply, because of development of ground water. This development and use of ground water, and the close physical relation of ground water to surface water, were responsible at least in part for the inability to apportion water in accordance with the provisions of compact and treaty. These instruments specify the apportionment of surface water among States which, at least in the first several years of drought, did not undertake to regulate the use of ground water in the upper basin. This situation is now rectified in part by the New Mexico State Engineer, who has declared the entire Rio Grande valley in New Mexico to be subject to regulation of all water development and use, both surface water and ground water (Reynolds, 1958, p. 15-28).

At all levels of administration of the water-rights system governing apportionment in the basin, there are opportunities for rigidity of operation that tend to paralyze the upper Rio Grande basin's water economy, but there may be opportunities also for modifying the present pattern so as to achieve more effective use of the water resource. For example, it is fundamental in the appropriation doctrine to recognize and protect the earliest rights in perpetuity, even though they require use of water that seems wasteful or damaging in the light of the present economy ; a less rigid concept would be to recognize such rights as analogous to other property rights, not necessarily sacrosanct but capable of exchange, adjustment, or purchase on the basis of their actual economic value. Some readjustment of water-use pattern may be desirable, for instance, in San Luis Valley where contiguous surface-water "subbers" and ground-water pumpers work at cross-purposes, yet where there are doubtless some areas in which the physical characteristics of soil and subsoil are better suited to irrigation by pumping, and others to subirrigation, in years of less than average inflow to the valley.

D25

As a problem of larger dimension, it would be desirable to incorporate the unconfined aquifer of San Luis Valley into the Rio Grande reservoir system, so that it could discharge water for the benefit of downstream users, rather than for consumptive waste in the closed basin. This problem has been studied extensively and a projected network of drains, analyzed by the U.S. Bureau of Reclamation, would salvage a quantity of water estimated at 19,000 to 33,000 acre-feet annually (Powell, 1958, p. 112-117). But evapotranspiration from the closed basin has continued for so long that there is now a considerable accumulation of salt in the soil and water. The findings of the Rio Grande Joint Investigation (National Resources Committee, 1938, p. 123-126) were that the water initially salvaged by a gravity drain might carry 1½ tons of salt per acre-foot, "with a remarkably unfavorable preponderance of sodium combinations in its constituent parts," and according to Howard (in Powell, 1958, p. 110) this is the quality than can be expected in the specific drain system proposed. This is approximately equivalent to the average quality of water entering Elephant Butte reservoir since 1953, and is considerably better than the water leaving the upper basin at Fort Quitman in most years. However, the Rio Grande Compact of 1938 includes in its Article III a proviso concerning any water salvaged from the closed basin:

In event any works are constructed after 1937 for the purpose of delivering water into the Rio Grande from the Closed Basin, Colorado shall not be credited with the amount of such water delivered, unless the proportion of sodium ions shall be less than forty-five percent of the total positive ions in that water when the total dissolved solids in such water exceeds three hundred fifty parts per million.

Studies in recent years have therefore led to proposals that would leave the salt and the sump as they are, but intercept water by means of wells where it is still of good quality, and divert that water to places of use (U.S. Bureau of Reclamation, 1956).

The effect of the recent drought upon volume of streamflow in the lower Rio Grande is indicated by figure 9: the runoff in the Rio Grande at Rio Grande City, Tex., was lower in 1950, 1951, and 1952 than in any year since records began, and in the minimum year 1952 was only about 20 percent of the 1924-53 mean. But even in 1952, according to the International Boundary and Water Commission, 100,000 acre-feet of unused streamflow passed the lowest diversion works at Brownsville, Tex. In the 30 years 1924-53 the average unused streamflow passing that point was about 2½ million acre-feet. As in other basins, the fluctuations in supply in the lower Rio Grande basin create difficulties in providing adequate flows for developed requirements, but these difficulties are less than in basins where the total requirements are a larger proportion of the mean river flow. Many difficulties of the past are now alleviated by storage and regulation in surface reservoirs in Mexico with total capacity (at spillway level) of 4,665,000 acre-feet, and in the international Falcon Reservoir 85 miles downstream from Laredo, which was completed in December 1952 and has a capacity of 3,350,000 acre-feet.

The average annual runoff from the upper basin at Fort Quitman in the 32 years 1924-55 was about 204,000 acre-feet, compared to an average for the same years of 195,000 as measured at the Upper Presidio, Tex., gaging station, 205 miles downstream from Fort Quitman. The long record from the Upper Presidio station is the basis of the lowest graph of figure 9. In some flood years, as for example in 1905-07 and 1911-12, the discharge from the upper basin has been of the order of a million acre-feet or more, but that flood discharge represented only 10 to 20 percent of the total discharge of the river near its mouth. Since Elephant Butte Reservoir began operation in 1916, the annual runoff at the Upper Presidio station has ranged from 439,000 to 2,500 acre-feet except in the flood years 1941 and 1942.

By the time the Rio Grande reaches Langtry, Tex., it is carrying the drainage from a basin more than twice the size of the upper basin alone. The flow is far more than twice the outflow from the upper basin, chiefly because of inflow of the Rio Conchos from Mexico. Before the Elephant Butte Reservoir began operation, the years of greatest runoff from the Conchos coincided rather closely with those of greatest runoff from the upper basin.

The records of runoff at Laredo, Tex., and particularly at Rio Grande City, Tex., suggest that the Rio Grande basin below Langtry is in a climatic region distinct and appreciably different from that of the basin above Langtry. At one or both of these stations the runoff was above the 1924-55 average in several years (1930, 1933, 1935, 1936, 1944) when the runoff at Langtry was less than average. The contrast in climatic regions is indicated also by the graphs showing the trends in precipitation in the region below Rio Grande City and in the region between Fort Quitman and Presidio (fig. 9). The periods of greatest precipitation deficiency in the lower Rio Grande valley (below Rio Grande City) were in 1895-1902, 1907-11, and 1917-21, of which only the years 1907-11 were markedly dry in the upper part of the lower basin.

D26

D27

Judging by these graphs, the Southwest drought did not encompass the lower Rio Grande basin until 1951, although precipitation was below average in 1947 and 1948 in the upper part of that basin. Beginning in 1951 the entire basin was affected by drought.

Under the terms of the Rio Grande, Colorado, and Tijuana Treaty of 1944 between the United States of America and the United States of Mexico, the United States is allotted one-third of the flow reaching the main channel of the Rio Grande (Rio Bravo) from the Conchos, San Diego, San Rodrigo, Escondido and Salado Rivers and the Las Vecas Arroyo, provided that this third shall not be less, as an average amount in cycles of five consecutive years, than 350,000 acre-feet annually.

* * * In the event of extraordinary drought or serious accident to the hydraulic systems on the measured Mexican tributaries, making it difficult for Mexico to make available the run-off of 350,000 acre-feet annually allotted * * * to the United States as the maximum contribution from the afore-said Mexican tributaries, any deficiencies existing at the end of the aforesaid five-year cycle shall be made up in the following five-year cycle with water from the said measured tributaries.

Whenever the conservation capacities assigned to the United States in at least two of the major international reservoirs, including the highest major reservoir, are filled with waters belonging to the United States, a cycle of five years shall be considered as terminated and all debits fully paid, whereupon a new five-year cycle shall commence.

Prior to consummation of the treaty, one-third of the combined outflow of the named Mexican tributaries exceeded 350,000 acre-feet in all years of record except 1934, 1937, and 1940; and this third did not drop below an annual average of 350,000 acre-feet in any period of 5 consecutive years. The Southwest drought began soon after the effective date of the treaty, however; one-third of the combined flow of the named tributaries was less than 350,000 acre-feet in 1945. 1948 and in each of the 8 years 1950-57. One-third of the annual average flow in the 5-year period 1948-52 was 275,000 acre-feet, and in the following period (1953-57) the comparable average was 210,000 acre-feet.

Most of the contribution from the United States to the lower Rio Grande enters the river between Presidio and Eagle Pass, Tex. Under the terms of the Treaty of 1944, the United States is allotted all the water contributed to the Rio Grande by the following measured tributaries in this reach : Pecos and Devils Rivers, Goodenough Spring, and Alamito, Terlingua, San Felipe and Pinto Creeks. The combined flow of these sources is shown in figure 9 by the shaded, dotted, and black patterns. A substantial part of this total comes from springs issuing from the Edwards Plateau.

The Rio Grande traverses the Edwards Plateau from the time it completes its semicircuit around the Big Bend country until after it passes Del Rio. The river gains substantially in this reach by inflow of measured tributaries and by unmeasured inflow, and most of this gain comes from ground water discharged from aquifers underlying the Edwards Plateau.

According to reconnaissance geologic maps (Darton, 1933; U.S. Geological Survey, 1932), the rocks of the Edwards Plateau are chiefly in the Comanche series of Early Cretaceous age, and this series includes several limestone formations, of which the Edwards and associated limestones are important aquifers farther east, in the San Antonio region (Thomas and others, 1963a). As described by Roberts and Nash (1918) some of the limestone beds are exceedingly cavernous and honeycombed, and yield abundant supplies of water to wells in the vicinity of Del Rio and Pumpville. However, extensive areas of the plateau, particularly near the Rio Grande, are capped by the less permeable Eagle Ford formation or Austin chalk of Late Cretaceous age. The Lower Cretaceaus rocks also include less permeable sediments, of which the Walnut clay (Del Rio clay of former usage) is an example. Thus, from the meager data now available, the Edwards Plateau apperas to have a major ground-water reservoir which in many places can be readily recharged by precipitation upon permeable limestone outcrops, but which in other places has a relatively impermeable surface from which there is overland runoff during intense rainstorms.

The Rio Grande has cut into the plateau to form bluffs more than 200 feet high both upstream and downstream from Langtry. The Pecos River, Devils River, and other tributaries flow through limestone-walled canyons for many miles before joining the Rio Grande. There are many springs, both large and small, in these canyons, generally at altitudes not far above that of Rio Grande (Roberts, 1918, pl. 1). The concentration of springs within Val Verde County may mean only that here are the topographically lowest outlets available to the ground-water reservoir. However, the geologic structure may also be a factor as shown by King (1942, fig. 23), the limestones under most of the plateau have a very gentle southeastward dip, but there is a marked structural depression just south of the Rio Grande opposite the mouths of the Pecos and Devils Rivers.

Although there are some wells of high yield in the part of the Edwards Plateau within the Rio Grande

D28

basin, the total production from wells is very small, either in comparison with pumping from the ground-water reservoir with Edwards limestone farther east (Thomas and others, 1963a) or in comparison with spring discharge from the Edwards within the Rio Grande basin. Judging by records from a few observation wells, chiefly near Del Rio, there have been fluctuations in storage in the ground-water reservoir but no progressive depletion during the recent drought. In well XV-3 (fig. 10) the water level was as high in 1949 and again in 1954 as the maximum stage reached in previous years, notably in 1938 and 1943. Thus there was full recovery from intervening drier years such as 1946, when the level was 7 feet lower, and 1953, when it was 10 feet lower than this maximum stage.

Goodenough Springs are among the largest of the springs that discharge water from the Edwards Plateau, and their discharge has been gaged since 1929. The springs are so close to the Rio Grande that the gaging station is affected by backwater when the river flow exceeds 35,000 cfs, but otherwise the record shows natural ground-water discharge. Evidently that discharge is not derived solely from local sources, for the fluctuations do not correspond with those of water levels in wells or of stream discharge in the adjacent Devils River basin. Thus the peak of discharge in 1941, and especially the high discharge of 1946-47, are not replicated by high water levels in the well at Del Rio, or by high discharge in the Devils River (fig. 10). However, water levels and spring discharge were high in both periods in the limestone ground-water reservoir farther east (Thomas and others, 1963a).

The hydrograph showing discharge of Devils River near Juno has many features in common with the graphs of discharge of Goodenough Springs and of water level in the well near Del Rio. The gaging station near Juno, operated from 1926 to 1949, measured the runoff from the upper two-thirds (2,730 square miles) of the Devils River basin. Although there was storm runoff at least once in nearly every year, the discharge is believed to be primarily of ground water from the plateau. In general form the hydrograph for the river is similar to that for the Del Rio well: a gradual decline from 1938 to mid-1942, then a sharp rise by the end of that year, followed by a decline until 1945, little change through 1946 and 1947, and then a rise in the following two years. The chief similarity between the hydrographs of the river near Juno and of Goodenough Springs is in the sharp increases in discharge (generally during periods of storm runoff in the river) and the gradual decline thereafter, following a trend that approaches a straight line on a semilog plot. The lines in successive years are approximately parallel, and those for the river and for Goodenough Springs are also virtually parallel. Troxell (1953) found similar depletion curves in hydrographs for streams in southern California, and attributed them to "perennial ground-water runoff." The depletion curves for Goodenough Springs and for Devils River near Juno indicate that if there were no replenishment to the reservoirs from which these flows are derived, the flows would be decreased by about 50 percent every 2 years.

The gaging station on Devils River at Del Rio (4½ miles above the confluence with the Rio Grande) measures the runoff from 4,185 square miles. Although this drainage basin is only 50 percent larger than that above the Juno gaging station, the discharge is at least 3 times as great, and storm runoff may be more than 30 times as large. Thus the lower third of the Devils River basin is the principal runoff-producing area in the basin.

Devils River is noted for its "constant flow of clear sparkling water that rises in great springs and flows down the Rio Grande in a stream which for many miles keeps separate from the muddy water of the main river" (Darton, 1933). The importance of the ground-water contribution to the river was especially noteworthy in 1933, when rainfall over the drainage basin was only about 50 percent of the long-term mean, and there was no indication of storm runoff at any time, yet the total runoff in 1933 was 50 percent greater than in the wet year 1941, when rainfall was 25 percent above average.

On the other hand, Devils River is capable of tremendous flash floods. The maximum discharge of 597,000 cfs on September 1, 1932, is equivalent to 143 cfs per square mile, a record for drainage basins of its size. In frequency of runoff-producing storms, 1949 was the best year, with eight separate flood peaks between February and October-one in every month except March. Thus, in spite of the large ground-water contribution, the runoff of Devils River ranges widely from year to year, chiefly because of stormflow. Statistics for the period 1924-53 indicate that the mean annual runoff was 406,000 acre-feet, the mean deviation 194,000 acre-feet, and the standard deviation about 164,000 acre-feet (Gatewood and others, 1963).

Casual inspection of the record of runoff of Devils River indicates that the effects of the recent drought have been very marked. In 47 years of record (1901-

D29

D30

13, 1924-57) the 5 years of least runoff were 1956, 1952, 1953, 1951, and 1947; 1951 and 1952 were also the years of least precipitation on the drainage basin in the 47 years. However, there is considerable variation in the precipitation-runoff relation from year to year : in 1910 and in 1933, for instance, precipitation was less than in 1953, but runoff was almost three times as great. There is a similar variation in the relation in wet years : 1932 and 1935 rate first and second in both annual precipitation and runoff, but the third ranking year in runoff (1948) was a year of only average precipitation over the basin. Figure 11 shows the variation in precipitation from year to year, the variation in runoff, and the variable relation of precipitation to runoff.

Doubtless some of the irregularity in the relation of precipitation to runoff can be traced to inadequate data concerning precipitation, which may occur in intense but localized storms. For example, the maximum discharge during 1942-an instantaneous peak of 25,100 cfs and mean daily of 11,900 cfs-occurred on November 6, following 10 days when discharge had been less than 500 cfs. The storm responsible for this flood discharge apparently slipped through the net of precipitation stations, for they recorded an average of only 0.58 inch of rain during the entire month of November.

The principal variable in the precipitation-runoff relation, however, is undoubtedly the lag between the time of precipitation and the time of runoff. Some water runs off concurrently-within hours or days, or at least within the same month-with precipitation, but other water is stored in the ground-water reservoir for months or years before it reaches the stream. To analyze the effect of drought upon