HYDROLOGY OF THREE SINKHOLE BASINS IN SOUTHWESTERN SEMINOLE COUNTY, FLORIDA

The southwestern part of Seminole County—in east-central Florida—is characterized by sinkholes formed by the subsidence of surficial deposits into solution cavities in the underlying limestone deposits. The area includes three sinkhole basins created by such subsidence: Cranes Roost. Palm Springs, and Grace Lake. Cranes Roost basin (drainage square miles) contains a closed drainage system of lakes and swamps—including and Mobile—that terminates at Cranes Roost sink. It also contains Lake Orienta and two unnamed sinks which do not overflow into Cranes Roost sink. unconnected

Cranes Roost basin (drainage area, 5.02 square miles) contains a closed drainage system of lakes and swamps-including Lakes Adelaide, Florida. and Mobile-that terminates at Cranes Roost sink. It also contains Lake Orienta and two unnamed sinks which do not overflow into Cranes Roost sink.
Palm Springs basin (drainage area, 1.77 square miles) includes Lake Marion, Eleventh Hole Pond, and several small unconnected sinks.
Grace Lake basin (drainage area, 1.64 square miles) includes Island Lake which overflows into Grace Lake.
The recent spread of urban development has tended to encroach on the flood plains of lakes in these sinkhole basins and cause concern over the flood hazard that results. An investigation was made of the area to document the highest known lake levels, to examine possible effects of urbanization with regard to increasing the flood hazard, and to appraise the possibilities of controlling lake levels to reduce or limit the flood hazard.
The highest lake stages of record in the three sinkhole basins occurred in September 1960. Analyses of the hydrologic relations between lake stages, ground-water levels, stream discharges, and rainfall in the area indicated that the lake stages of September 1960 probably were the highest attained since at least 1895.
Cultural development that increases the percentage of a sink basin covered by impervious materials will cause more rapid runoff of a larger part of the rainfall in the basin. Thus, as development progresses, lake stages seldom reached under natural conditions may be reached more frequently unless the lake levels are controlled.

BUREAU OF GEOLOGY
In Cranes Roost basin, the levels of Lakes Mobile, Adelaide, and Florida could be controlled by enlarging their surface outlets and adding control struc.
tures; however, such measures might be ineffective unless the level of Cranes Roost also was controlled. The level of Cranes. Roost could be controlled by removal of water by pumping or by providing a surface outlet; pumping might be ineffective during extreme wet periods because of the sudden surface inflow from upstream lakes and the potential for ground-water inflow from the Floridan aquifer. The level of Lake Orienta could be controlled by removal of water by pumping or by providing a surface outlet.
In the Palm Springs sink basin, the levels of Lake Marion and the several small sinks could be controlled by removal of water by pumping or by pro viding surface outlets. For the several sinks, however, providing a surface outlet would require an extensive excavation and pumping might prove im practical because of seepage induced from the Floridan aquifer.
The level of Island Lake-in Grace Lake sink basin-could be controlled by The scope of the study was such that all pertinent available water-level, streamflow, and rainfall records were evaluated, but field work and collection of new water records were minimal. Periodic water-level data were obtained, however, for five surface-water bodies and one artesian well.

Topographic Setting
The area is a fairly level plain generally 85 to 100 feet above mean sea level except at the southern edge of the area where one sand hill exceeds 120 feet in altitude. The plain has been extensively altered by the subsidence of the surface materials into cavities in the underlying limestone of the Floridan aquifer. The cavities are caused by solution of limestone by water. The three sink basins studied in this report were formed by such subsidence.
In the three sink basins, circulation of water through the limestone is great because little of the rainfall there escapes as surface outflow. The rainfall which does not return to the atmosphere by evaporation eventually infiltrates to the Floridan aquifer and subsequently is discharged outside the area, mostly at Palm and Sanlando Springs. The voids and cavities, caused by water pass ing through the Floridan aquifer, in time are enlarged to such an extent that the overlying Hawthorn and younger deposits slump or collapse into them. Because of differences in permeability of the limestone and the overlying deposits, infiltration of water and the attendant land subsidence have been greater in the western parts of Cranes Roost and Palm Springs sink basins and in the northern part of Grace Lake sink basin than in other parts of the basins.
The extent and configuration of each of three sink basins was determined by the course that surface outflow would ultimately take if the water level in the individual sink basins were to reach a high enough stage to cause the basin to overflow ( fig. 2). Cranes Roost sink basin, the southernmost of the three, is 5.02 square miles in area and consists of a system of lakes and sinks. Surface drainage begins in several swamps south of Lake Seminole, about halfway between Altamonte Springs and Longwood. This drainage is joined by over flow from Lake Mobile just before it enters Lake Florida. Overflow from Lake 6 BUREAU OF GEOLOGY Florida passes through Lake Adelaide which overflows into Cranes Roost. Lake Orienta and two smaller sinks, though not connected to Cranes Roost on the surface, would drain into Cranes Roost sink if their stages were to rise au-z--z-zz zº ar-zo to about 77 feel above mean sea level. Although this basin does not overflow, Cranes Roost sink would flow into the Little Wekiva River 0.6 mile south of Sanlando Springs were the basin ever to fill. 7 Just north of Cranes Roost sink basin lies the Palm Springs sink basin, which contains Lake Marion and several smaller sinks. The area of this sink basin is 1.77 square miles. None of the sinks are connected on the surface, but if they were, they would constitute a drainage system which upon filling would overflow into the Little Wekiva River at Palm Springs. None of the smaller sinks are named on the map of figure 2. The pond for which stage data were obtained during this investigation is herein referred to as Eleventh Hole Pond because it is near the 11th hole of a golf course.
Grace Lake sink basin, the northernmost, contains Island and Grace Lakes and occupies 1.64 square miles in the northern part of the sinkhole area. Island Lake overflows into Grace Lake; Grace Lake at times overflows into Lake  to the aquifers; some of the lakes are landlocked but others overflow to another lake or sink at a lower altitude. The effects of rainfall and surface flow on lakes are extremely variable and intermittent whereas the effects of ground. water seepage, evaporation, and transpiration are relatively stable and contin uous. The balance between input and output is continually changing; hence.
the level of the lake is nearly always either rising or falling.
During and for a short time after rainfall. surface inflow to the lakes occurs as overland flow and street and sewer drainage; if the lakes are suffici ently filled, surface outflow occurs as overflow to down-gradient lakes or sinks.
Water moves from the lakes to the water-

PEAK STAGES AND PROBABILITY OF OCCURRENCE
The approximate peak stages reached by one or more of the water bodies in the area of investigation in 1954, 1960, and 1970  (d) Stains on posts and wash lines on highway fills.

BUREAU OF GEOLOGY
The available lake-stage data are insufficient for evaluating the frequency of occurrence of specific lake stages. However, lakes, streams, and aquifers are part of a hydrologic system, and rainfall provides the water than keeps the system viable. As a consequence, fluctuations of lake levels are tied to the fluctuations of ground-water levels and stream discharges. Therefore, some insight as to the frequency of occurrence of selected lake stages can be gained from a study of the available records of ground-water levels, stream discharges, and rainfall, which together span a large number of years and a wide range of hydrologic conditions. From November 1953 to December 1956 the trend of the water level of Lake Orienta was generally the same as the trend of the level of the potentio metric surface of the Floridan aquifer, which is represented by the level of water in well 841-121-1 ( fig. 4). The potentiometric surface of the Floridan aquifer was substantially higher in September 1960 than in January 1954, or in the last few months of 1953. This is consistent with the fact that the stage of Lake Orienta was substantially higher in September 1960 than in January 1954 as the table shows.  4). Sharp upswings of the discharge and water-level graphs generally correspond with large monthly rainfalls at Orlando and sustained downswings of the graphs gen erally corresponds with a series of small monthly rainfalls; the exceptions are an indication that rainfall over the drainage basin of concern was not uni formly the same as rainfall at Orlando. The average annual discharge of Econlockhatchee River near Chuluota ( fig. 7) also was greater in 1960 than in any year since at least 1936. In relation to Wekiva River, Econlockhatchee River has a larger surface-water component and a smaller ground-water component. The annual discharge of Econlockhatchee River tends to reflect more directly the variations in the yearly rainfall.  Thus, from the general relation apparent between lake levels, ground water levels, and the measured discharges of different types of streams of different sizes, it reasonably can be concluded that the stages of Lake Orienta and other lakes in the area of investigation, especially lakes in closed basins.
were higher in September 1960 than they were at any time since at least 1936.
On the assumption that the discharge of Wekiva River near Sanford pro vides a valid index of the general level of nearby lakes, the average annual i 12 BUREAU OF GEOLOGY discharge of Wekiva River was estimated for years before 1936 by regression methods (Riggs, 1968) using the following equation as a model: (1) Q, = a + b,P, + b,P, + b.P.
Where Qo = average annual discharge of Wekiva River (   According to Riggs (1968, p. 19, 20), as variables are added to or de leted from a regression, a change in sign of the coefficient of one of the variables indicates that the effect of the variable is small in relation to the sampling error. Seemingly, the same conclusion would apply if the use of different but comparable data in a regression produces a change in sign of the coefficient of one of the variables. The effect of antecedent rainfall on the current year's discharge doubtless decreases with each preceding year. Thus, for the purpose at hand, there is no need to consider the effect of rainfalls past the second preceding year.
On basis of the foregoing analysis of the average annual discharge of Wekiva River near Sanford, and other hydrologic data and relations pre viously described, it is concluded that the peak stages attained by lakes in  HYDROLOGIC SETTING ON SEPTEMBER 30, 1960 In the area investigated the potentiometric surface of the Floridan aquifer in 1960 was recorded in only one well (number 841-121-1 at Island Lake) which has served as a moniter well since 1951. Consequently, the maximum altitude of the potentiometric surface elsewhere in the area can only be estimated.
The estimated altitude and configuration of the potentiometric surface on September 30, 1960 (fig. 2) are based on the altitudes of water levels in wells observed by Barraclough (1962b) in January 1954 and June 1956. The 1960 altitudes of water levels in these wells were estimated on the assumption that the ratio of the difference between their 1951 and 1956 water levels to the difference between their 1956 and 1960 water levels was the same as the ratio of these differences for well 841-121-1. This assumption is reasonable because experience has indicated that correlation from well to well is fairly tight for wells tapping the Floridan aquifer. not only in this area but else.
where in Florida as well. Along the western and southern boundaries of the area the potentiometric contours as shown in figure 2 conform with the po tentiometric contours presented by Lichtler (1968, p.   The high water table east of Lake Orienta (fig. 10) is typical of the con dition existing east of a line running in an arc from just east of Lake Orienta. northeast to the area between Lakes Florida and Mobile, and thence northwest around the west side of Island Lake where it curves east. The Hawthorn de posits in the area of investigation lying east of this line must be relatively impermeable in order to permit head differences as much as 30 feet between the water table and Floridan aquifers. West of the line, smaller head differ. ences (5 to 8 feet) between the aquifers indicate that the Hawthorn deposits are much more permeable than those to the east.
Ordinarily, evapotranspiration is sufficient to prevent the water table from rising to land surface in places where the terrane slopes to the lakes and sinks west of the area of high water table previously described. During wet periods, however, ground water comes to the surface along these slopes, as evidenced by the stench of septic-tank effluent in the subdivision east of Lake Orienta in the spring of 1970 (William Bush, Jr., oral commun., 1971). Surface outflow -Overflow to down-gradient lakes.
Ground-water inflow -Lateral seepage from surrounding water-table aquifer.
Ground-water outflow -Minor vertical seepage through the bottom to the aquifers beneath the lakes.
Note.-Lake levels were higher than the potentiometric surface of the Floridan aquifer but probably lower than the surrounding water table.

Lakes Adelaide and Florida
Surface inflow -Street and storm-sewer drainage, overflow from up gradient lakes, possibly minor overland inflow.
Surface outflow . Overflow to Cranes Roost.
Ground-water inflow -Lateral seepage from the surrounding water-table aquifer; vertical seepage through the bottom from the aquifers be neath the lake.
Note.-Lake levels probably were below both surrounding water table and potentiometric surface.

Cranes Roost
Surface inflow . Street and storm-sewer drainage, inflow from Lake Adelaide, possibly minor overland inflow.
Ground-water inflow -Probably none.
Ground-water outflow -Lateral seepage to the surrounding water- Ground-water inflow -Lateral seepage from the water-table aquifer.
Ground-water outflow -Vertical seepage through bottom to the aquifers beneath the lake.
Note.-The adjacent water table probably was higher than the lake. The lake level was higher than the potentiometric surface beneath the lake.

Lake Marion
Surface inflow -Street and storm-sewer drainage, possibly minor over land flow.
Ground-water inflow -Lateral seepage from the adjacent water-table aquifer on east.
Ground-water outflow -Lateral seepage to adjacent water-table aquifer on west and vertical seepage through the bottom to the aquifers beneath the lake.
Note.-Lake level probably lower than adjacent water table at east end but higher than water table at west end. Lake level higher than potentiometric surface.

Palm Springs sinks (including Eleventh Hole Pond)
Surface inflow -Street and storm-sewer drainage, possibly some over land runoff from lawns and golf course fairways.
Ground-water inflow -Lateral seepage from the surrounding water-table aquifer and vertical seepage through bottom from the aquifers beneath the sinks if the sink levels were below the potentiometric surface.
Ground-water outflow -Vertical seepage through bottom to the aquifers beneath the sinks if the sink levels were above the potentiometric surface.
Note.-Sink levels probably were slightly above potentiometric surface if the stage was falling and slightly below if the stage was rising.

REPORT OF INVESTIGATION NO. 81 21
Grace Lake Sink Basin Island Lake Surface inflow -Street and storm-sewer drainage, possibly overland inflow.
Surface outflow -Overflow to down-gradient lakes.
Ground-water inflow -Lateral seepage from surrounding water-table aquifer.
Ground-water outflow -Minor vertical seepage through the bottom to the aquifers beneath the lakes.
Note.-Lake levels were higher than the potentiometric surface but probably lower than the surrounding water table.
Grace Lake Surface inflow -Overflow from Island Lake, highway drainage, possibly minor overland inflow.
Surface outflow -Possibly overflowing to down-gradient sinks and lakes.
Ground-water inflow -Probably none.
Ground-water outflow -Vertical seepage through the bottom to the aquifers beneath the lake.
Note:-Lake level probably higher than both surrounding water table and potentiometric surface.

STAGE FLUCTUATIONS AND SEEPAGE RATES
To provide some insight to the nature of water-level fluctuations of the surface-water bodies and their relation to the Floridan aquifer, stage data were collected during the investigation for five water bodies and well 841 121-1 which taps the Floridan aquifer near Island Lake. These data are shown in figures 11-15; the hydrograph for the well is repeated on the hydrograph for each of the water bodies for comparison. For three of the water bodies-Lake Orienta, Cranes Roost, and Grace Lake-the approximate altitude is shown for the peak stages occurring about mid- March, 1970. A similarity in the patttern of fluctuation in the water levels of Island Lake and well 811-121-1 is evident in figure 11, although the range of fluctu ation is much less for the lake than for the well. The lake level being about 40 feet higher than the well level suggests a poor connection between the lake and the Floridan aquifer. This would preclude an appreciable amount of seepage from the lake to the aquifer.
On the assumption that seepage from Island Lake to the Floridan aquifer is insignificant, the hydrologic conditions at Island Lake between January 1  and May 31,1971 were favorable for making an estimate of the difference between rainfall and lake evaporation for the same 5-month period. The lake level was low enough that there was no surface-water outflow. Surface-water inflow probably was minimal because the highly permeable surficial sand in most parts of the contributing drainage area is conducive to the infiltration of rainfall. Rainfall during this period-10.7 inches at Orlando; 15.8 inches at Sanford; mostly in February and May-was not great enough to cause large quantities of runoff but was sufficient to account entirely for the few rises in lake level indicated by the hydrograph. Small quantities of runoff would REPORT OF INVESTIGATION NO. 8] 23 not contribute appreciably to a rise in lake level because the runoff would be distributed over the entire lake area which takes up a large percentage of the total drainage area. Ground-water inflow to the lake from the water-table aquifer probably was small because rainfall was over ll inches below normal in the 6 months preceding January 1971. Thus, the 0.6 foot decline in the stage of Island Lake between January and May 1971 is considered to be a good measure of the excess of evaporation over rainfall for that period.
The estimate of 0.6 foot for the excess of evaporation over rainfall between January and May 1971 probably is applicable to other water bodies in the area for the same period. For example, the hydrograph for Lake Orienta (fig. ure 12) shows that during this period the lake level declined about 1.1 feet. Therefore, the net of seepage into the lake from the water-table aquifer and out of the lake to the water-table and Floridan aquifers plus any surface-water inflow that occurred was 0.5 foot. (Although figure 10 shows that the ground water level in the adjacent water-table aquifer is above the level of Lake Ori enta on both the east and west sides of the lake, as it probably was at the end of the unusual wet period ending September 1960, during dry periods, such as January-May 1971, the water table probably slopes continuously downward from Lake Orienta toward the nearby sink to the west; if so, during dry periods some water would move from Lake Orienta through the water-table aquiſer to the sink.) Inasmuch as the rainfall on the lake was sufficient to account for the few rises in lake level indicated by the hydrograph, surface water inflow probably was minimal during this period. The magnitude of the individual components of seepage moving into and out of the lake cannot be determined from the data at hand.
Water probably moves always downward from Lake Orienta to the Flori dan aquifer but the rate of movement varies appreciably between wet and dry periods. The level of Lake Orienta is several feet above the level of the potentiometric surface of the Floridan aquifer at Island Lake from the com parison of the lake-level graph with the water-level graph for well 841-121-1 (figs. 4 and 12); however, the level of the potentiometric surface changes appreciably over a distance of a few miles. Water-level data for -56 (Bar raclough, 1962b indicate that the level of the potentiometric surface at Lake Orienta-as represented locally by the water level in well 839.   (fig. 2)-on January 11, 1954, was 5.2 feet higher than the water level in well 841-121-1, and on January 5, 1956, was 4.2 feet higher. Thus, it is evident from figure 4 that the level of Lake Orienta locally was about 4 feet higher than the potentiometric surface in January 1954 and about 8 feet higher in January 1956. This indicates that from January 1954 to January 1956 the head differential between the two water bodies-and the resultant rate of water movement-increased by a factor of about 2. Just as the level of the potentiometric surface declined more than the level of Lake Orienta in 1954-56, so it would have risen more than the lake level as they both rose to their peak stages of September 1960. Consequently, the head differential between the two water bodies-and the resultant rate of water movement-in September 1960 would have been substantially less than in 1956 or at other times when water levels were generally lower. The fluctuation in the water level of the water table aquifer adjacent to the lake between wet and dry periods also probably is greater than the fluctu ation in level of Lake Orienta. Consequently, the inflow to the lake increases during wet periods and decreases during dry periods. Thus at a time when ground-water outflow from Lake Orienta to the Floridan aquifer is at its least, ground-water inflow to Lake Orienta from the water-table aquifer is at its greatest.
Whereas ground-water outflow from Lake Orienta was greater than ground-water inflow during the relative dry period of  fer is about 3 to 4 feet lower at Grace Lake than at Island Lake. In January May 1971, therefore, the level of Grace Lake was probably about 20 feet above the potentiometric surface locally. The rate of outseepage from the lake to the Floridan aquifer is proportional to the head differential between the two water bodies. Thus, if the area of the hydraulic connection between the lake and the Floridan aquifer remains the same irrespective of lake stage, outseepage from the lake to the Floridan aquifer was about one-third less in September 1960 than it was in January-May 1971. In terms of effect on the lake level, the difference in outseepage would be even greater because the lake area was considerably greater in September 1960 than in January-May 1971.
Grace Lake apparently overflows to sinks north of the lake during rel 'atively wet periods. In August 1974 one of the authors observed water flowing at a rate of about 10 cfs through a normally dry culvert under the road north of the lake. Antecedent rainfall in the general area was ample but not out standingly great. The occasional overflowing of Grace Lake accounts for the peak stage of 1960 being only slightly greater than the peak stage attained in 1970; in contrast, the 1960 stages of other water bodies, such as Lake Orienta and Cranes Roost, were much above those attained in 1970.
The fluctuation in the water level of Cranes Roost is greater than that of any of the other water bodies investigated and is controlled primarily by sur face inflow from Lake Adelaide and outseepage to the Floridan aquifer. From January 1 to May 31, 1971, the water level in Cranes Roost declined about 5.3 feet as shown in figure   14.
Both the surface-water inflow from Lake Adel.
aide and inseepage from the water- At the end of a dry period the potentiometric surface of the Floridan aqui fer in response to rainfall probably rises faster than the lake level until such time as rainfall is sufficient to cause surface inflow from Lake Adelaide. Under such conditions the potentiometric surface may briefly rise above the water level in Cranes Roost, thereby causing water to move upward from the Flori dan aquifer into Cranes Roost. Had there been no surface inflow from Lake Adelaide, the water level in Cranes Roost still would have reached an altitude greater than 50 feet because the water level in the sink cannot naturally decline below the level of the potentiometric surface.
Conditions in the Palm Springs sink area are such that the pond levels, including Eleventh Hole Pond, probably are always close to the level of the potentiometric surface of the Floridan aquifer. The many sink holes and the extent of land subsidence in this area indicates a good hydraulic connection between the surficial material and the Floridan aquifer. The pond levels are about the same in altitude which indicates that movement of water through the Floridan aquifer is relatively unrestricted in this area.
In February-March 1971 the level of Eleventh Hole Pond (figure 15) declined slightly more rapidly than the water level in well 141-121-1 at Island Lake; this indicates that the pond level was above the local potentiometric sur. face at that time. However, the rise in the pond level in July-August 1971 was much greater than that of any of the other surface-water bodies investigated, and it follows closely the trend of the water level in well 841-121-1; this sug.
gests that the potentiometric surface in the Palm Springs sink area probably was above the pond level during this period. The water level in the Palm Springs sinks probably are controlled almost exclusively by the potentiomeric surface of the Floridan aquifer; water seeps out of the pond into the Floridan aquifer when the potentiometric surface is falling and seeps into the pond from the Floridan aquifer when it is rising.

EFFECTS OF CULTURAL DEVELOPMENT IN THE SINK BASINS
The preceding remarks concerning water-level fluctuations and peak stages of different water bodies are applicable under the physiographic conditions that existed in the area at the time of the investigation and in some instances before 1960. The area has undergone considerable urbanization since 1960 and doubtless will be further urbanized; this development will affect the hydrologic relations in the basin.

BUREAU OF GEOLOGY
Cultural development that increases the percentage of a basin covered by impervious materials, such as roofs and pavement, causes more of the rainfall to run off. Replacement of the natural soil cover with grass sod also may cause more rapid and voluminous runoff. The higher velocities associated with greater and more rapid runoff can cause greater erosion, especially where the soil is disturbed during construction. Eroded organic matter and clay accumu lating in the bottoms of sinks or lakes may reduce the rate of seepage out of the sinks and lakes. The reduction in seepage coupled with the increased rate and volume of inflowing water can cause higher than normal water levels.
Conceivably, lake stages that are seldom reached or exceeded under natural conditions could be reached or exceeded frequently under conditions of full cultural development. The hydrologic effects of future developments could be