HYDROMETEOROLOGY: CLIMATE and HYDROLOGY of the GREAT LAKES Jan A

Section 4

HYDROMETEOROLOGY: CLIMATE AND HYDROLOGY OF THE GREAT LAKES Jan A. Derecki

4.1 Introduction (5,473 cu. mi.) stored in the lakes, varying from 484 km3 in Lake Erie to 12,234 km3 in Lake Superior, the Great Lakes have a tremendous 4.1.1 Description and Scope heat storage capacity. Through air-water in­ teraction, the lakes influence the climate over The climate of the Great Lakes Region is them and over adjacent land areas. Because of determined by the general westerly atmos­ the lake effect, air tempt!!ratures are moder­ pheric circulation, the latitude, and the local ated, winds and humidities are increased, and modifying influence of the lakes. Due to the precipitation patterns are modified. Although lake effect, the regional climate alternates be­ these phenomena have been recognized for tween continental and semi-marine. The decades, it is only in recent years that inten­ semi-marine climate is more consistent con­ sive programs have been undertaken to de­ tiguous to the lakes, but with favorable termine the more exact nature and mag­ meterological conditions, it may penetrate nitudes of these processes. deeply inland. The Great Lakes climate and The Great Lakes moderate temperatures of hydrology are closely related. Variations in the overlying air masses and surrounding mean lake levels, and consequently lake out­ land areas by acting as heat sinks or sources. flows, are controlled by the imbalance be­ The process of heat exchange between the tween precipitation and evaporation. lakes and atmosphere is both seasonal and The Great Lakes drainage basin is discussed diurnal. During spring and summer, the lakes in other appendixes and a brief summary of are generally colder than air above and have a the climatic and hydrologic elements of the cooling effect on the atmosphere. During fall drainage basin is given in Section 1 of this and winter, the lakes are generally warmer appendix. The major storm tracks affecting than the atmosphere and serve as a heat the Great Lakes Region are indicated in Fig­ source. However, during the winter months, ure 4-13. Distributions ofmean annual values the ice cover reduces the lake effect. for air temperature, precipitation, runoff, and A daily pattern of heat exchange is water losses over the land areas ofthe Basin, superimposed upon the seasonal pattern. This showinglatitudinal and lake-effect variations, daily pattern is produced by land-water tem­ are indicated in Figures 4-15, 17, 19, and 20, perature differences. Because lakes are more respectively. The average monthly means, efficient than land areas in storing heat, lake highs, and lows of overland air temperature temperatures have a tendency to remain sta­ for the individual lake basins and for the total ble, while land temperatures undergo daily Great Lakes Basin are shown in Figure 4-16. variations that are more in line with the air temperatures. When the land is warmer than water, the relatively warmer airover adjacent 4.1.2 Lake Effect land areas tends to rise and is replaced by colder, heavier air from the lakes. When the With a total water volume of 22,813 km3 land is colder, the process is reversed. This

Jan A. Derecki, Lake Survey Center, National Oceanic and Atmospheric Administration, Detroit, Michigan.

71 71 Appendi~ 4 processofheatexchange produces lightwinds, thunderstorms. Duringwinter, the conditions which are known as lake breezes. Theoffshore are reversed, and the cold, inland air passing and onshore lake breezes are illustrated in over relatively warm water becomesles8 sta­ Figure 4-14. The direction of lake breezes is ble and picks up moisture, which encourages governed by the land-water temperature dif­ snow flurries. As the winter air masses move ferences and is independent ofthe general at­ over the lakes, the moisture that accumulates mospheric circulation. However, lake breezes in the air produces heavy snowfalls on the lee occuronly duringrelatively calm weather and sides of the lakes, due to orographic effects of affect a limited air mass along the shoreline, the land mass. The fact is well documented rarely extending more than several kilome­ that heavysnowbelt areas resultfrom thelake ters (2-3 miles) inland. The moderation oftem­ effect. These areas include Houghton on Lake peratures by the lakes affects regional ag­ Superior, Owen Sound on Lake Huron,Buffalo riculture by reducing frost hazards in the on Lake Erie, and Oswego on Lake Ontario. early spring and in the fall, thus lengthening the growing season, especially in coastal areas. Examples of this effect are the cherry 4.1.3 Measurement Networks orchards of the Door Peninsula in Wisconsin and theGrandTraverse Bay area in Michigan, Basic meteorological data in the Great and the vineyards of the western Lake Erie Lakes Basin are available from regularobser­ islands in Ohio. vation networks operated by the Nation­ Because lake breezes have a limited range al Weather Service and the Canadian and require special conditions, the lake effect Meteorological Service. The networks consist on winds is minor. A much more important ofa limited number offirst order stationsthat effect is the considerable increase in geo­ provide hourly observations for air tempera­ strophic wind speed over the lakes. This in­ ture, precipitation (total and snow), wind crease is caused by reduced frictional resis­ speed and direction, humidity, and duration of tance to air movement over the relatively sunshine and cloud cover. More numerous co­ smooth water surface, and by the difference in operative stations provide daily observations atmospheric stability created by air-water for air temperature and/or precipitation. Cer­ temperature differences. Recent studies indi­ tain more specialized stations collect addi­ cate thatthe increase in overwaterwind speed tional data on solar radiation, radiosonde in­ varies from approximately 15 percent in mid­ formation, weather radar, and pan evapora­ summer to as much as 100 percent in late fall tion. Other regularly observed data useful in and early winter. The average annual in­ Great Lakes climatology include water tem­ crease is approximately 60 percent. peratures recorded by municipalities at their The Great Lakes also cause an increase in water intake structures and by Federal agen­ overwater humidity by releasing large quan­ cies at ~elected lake perimeter location& tities of moisture through evaporation. On an In addition to the regular networks, more annual basis the humidity over the lakes av­ sophisticateddataarecollected periodicallyor erages 10 percent to 15 percent higher than seasonally for research on lake climatology. that over the land. Seasonal changes in These include special precipitation networks, humidity over the lake compared to that over established and operated on lake islands and land vary from a decrease ofapproximately 10 adjacent shorelines; synoptic surveys con­ percent due to overwater condensation in the ducted by research vessels that take observa­ late spring, to an increase ofapproximately 10 tions for thewhole rangeofhydrometeorologic percent in the summer, 15 percent in the fall, parameters; lake towers that give meas­ and 30 percent in the winter. urements with vertical profiles for selected The Great Lakes also influence the distribu­ parameters for air-water interaction studies; tion ofcloud cover and precipitation. Modifica­ aerial surveys by conventional aircraft for ice tion ofpreeipitation patternsdue to lake effect reconnaissance and water surface tempera­ is caused by thechanges in atmosphericstabil­ tures, using infrared and airborne radiation ity in combination with prevailing wind direc­ thermometer techniques; and weather satel­ tion and topographic effects. During summer, lites that provide useful information for the the air undergoes overland warming before investigation of doud and ice cover on the passing over the lakes. The warm airover rel­ lakes. atively cold water results in the development Hydrologic data on the Great Lakes Basin of stable atmospheric conditions, which dis­ are compiled and pu~lished by several agen­ courage formation of air-mass showers and cies. Records of tributary streamflow to the HrdrofMteorolon 78 lakes are available from the U.S. Geological from appropriate water level gage ratings Survey and the Canada Centre for Inland Wa­ based on periodic current meter flow meas­ ters, Department of the Environment, urements. Canada. The extent of gaged area increased substantially in the late 1930s, giving cover­ age to approximately 50 percent of the Basin. 4.2 Radiation At present approximately 64 percent of the Basin is gaged; gaged areas for Lakes Supe­ rior, Michigan, Huron, Erie, and Ontario ba­ 4.2.1 Total Radiation Spectrum sinsrepresent approximately 53, 71, 66, 67, and 63 percent of their respective basins. In addi­ Total radiation received atthe surface ofthe tion to surface water data, these agencies and earth consists of shortwave radiation coming the Geological Survey of Canada publish ob­ directly from the sun or scattered downward, servation well records providing information and longwave radiation, emitted from the at­ on ground-water conditions. However, the mosphere. Portions of the incoming radiation network of observation wells useful in deter­ in both short and long wavelengths are re­ mining ground-water flow to the lakes is ex­ flected and additional longwave radiation is tremely limited. emitted to the atmosphere. The main radia­ The Great Lakes levels and outflows are tion exchange processes taking place within available from the Lake Survey Center. Lake the terrestrial system (space-atmosphere­ levels are determined from a network ofwater earth) are illustrated in Figure 4-93, present­ level gages maintained by the Lake Survey in ing annual radiation balance, which is based the United States and the Fisheries and largely on information provided by London.503 Marine Service, Department of the Environ­ The net effect of shortwave radiation is the ment, in Canada. Flows in the connecting solar heating of the earth, while longwave channels are determined by the Lake Survey radiation results in cooling.

\1// ...... - - .... - ....- ...... /8 , SPACE I I\ TOTAl. T014L 1.0NGWAVE SMORTwAVE '''''DIATION 1.055 Al.eEDO UPPER ATMOSPHERE

AeSCReEO I'lIlOM SUNLIGHT

LOWER ATMOSPHERE

SOLAR (SHORTWAVE) RADIATION TERRESTRIAL (LONGWAVE) RADIATION

FIGURE 4-93 Annual Atmospheric Heat Budget. Shows percentage distribution of radiation components for northern hemisphere. After London, 19117 7.. A'PPendi~"

There is some duplication ofnames used for mended an average daily albedo for watersur­ the same radiation components. Shortwave or face of6 percent. The relatively low albedo for global radiation is generally referred to as open water conditions increases drasticaHy solar radiation and both names are used inter­ with ice and snow cover. Bolsenga18 gives al­ changeably in this report. Insolation is inci­ bedo values for various types ofice common on dent or incoming solar radiation. Similarly, the Great Lakes. These values range from 10 terrestrial radiation is used synonymously percent for clear ice to 46 percent for snow ice, with longwave radiation, while longwave both free of snow cover, The presence of par­ radiation from the atmosphere is called at­ tial or complete snow cover on the ice can sig­ mospheric radiation. nificantly increase these values. A regular network for measuring shortwave There is a limited network of regular sta­ radiation has been established but only few of tions that measure incident solar radiation in these are inthe Great Lakes Region. A regular the Great Lakes Region. The average monthly network for total radiation measurements values from these stations are shown in Fig­ (shortwave and longwave) has not been estab­ ure 4-94. Periods ofrecord for the stations vary lished, since there are only a limited numberof from 10 to 50 years. Based on records from the radiometers in operation at various research radiation network, the average monthly in­ installations. Available information indicates coming solar radiation in the Great Lakes that the average monthly all-wave incident Basin varies from a low of approximately 100 radiation in the Great Lakes Region varies ly/day in December (winter solstice) to a high from a winter low of approximately 400 of approximately 550 ly/day in June and/or langleys per day (ly/day) in December to a July (near the summer solstiee), with an aver­ summer high of approximately 1400 ly/day in age annual value of about 320 ly/day. The avo June or July. erage monthly extremes for reflected solar radiation from the lakes representfrom 6 to 33 ly/day (6 percent water surface albedo). 4.2.2 Solar Radiation Beginningin the last decade, direct overwa· ter measurements of solar radiation were in· Solar radiation is reduced by the atmos­ cluded in the synoptic surveys of the Great phere before reaching the earth's surface. At­ Lakes conducted by several research organi· tenuation of the extraterrestrial solar radia­ zations. These measurements are generally tion is caused by scattering, reflection, and limited to the navigation season (April. absorption by gas molecules, water vapor, December), are not continuous, and are some· clouds, and suspended dust particles (Figure what biased towards fair weather conditions, 4-93). As a result of the attenuation, the in­ but nevertheless they represent actual condi­ coming shortwave radiation on a horizontal tions over the lakes and provide a basis for surface arrives partly as direct solarradiation comparison of the overwater and overland and partly as sky radiation (scattered down­ radiation. Richards and Loewen653 conducted ward by atmosphere and diffused through the a preliminary study ofthis type, which shows clouds). Sky radiation is a high percentage of that incident solar radiation over the lakes is the total incident radiation during low decli­ greater than that recorded on adjacent land nation of the sun and on overcast days. stations during summer and smaller during Part of the incoming solar radiation is re­ winter months. This confirms the physical flected from the receiving surface (clouds and concepts ofthe lake effect. Their study is lim­ earth) back to the atmosphere, the amount of ited to four years of data during the April­ reflection depending on the surface albedo or December periods and shows that overwater the ratio ofreflected to incident radiation. Al­ radiation at the beginning and end of the bedo values for a water surface depend on the period amounts to 90 percent of the overland solar altitude (angle ofthe sun above the hori­ radiation. The overwater radiation increases zon), cloud cover, and the roughness of the gradually duringspring and summerto an av­ water surface, but for many practical pur­ erage high ofapproximately 140 percent ofthe poses these factors can be assumed to be con­ overland radiation in the late summer, then it stant for daily or longer periods. During the decreases rapidly in the fall. Lake Hefner study, Anderson16 developed em­ Other recent studies of solar radiation on pirical curves, which interpret water surface the Great Lakes include determination ofthe albedo as a function ofsun altitude for various radiation balance for Lake Ontario(Bruce and cloud cover conditions. Based on results of Rodgers108 and Rodgers and Anderson615). De­ that study, Kohler and Parmele 464 recom- termination of the total atmospheric water HJldrometeorolol1'11 75

GREAT LAKES BASIN

\... eGO .. .::.:...... ~

o J'MAMJJASOND o 0 I 1 J'MAIUJASOND

FIGURE 4-94 Average Daily Solar Radiation (Langleys) in the Great Lakes Basin From Phillips. 1969 vapor over the Great Lakes Basin and deriva­ the net back radiation, a longwave radiation tion of a relationship between atmospheric loss to the atmosphere (Figure 4-83). The net water vapor and surface dew point back radiation is primarily a function of the (Bolsenga 74.77) could contribute to parame­ temperature of the water surface, which con­ terization of the solar radiation term. This trols emitted radiation, and the water vapor might compensate for the lack of recording content ofthe air, which controls atmospheric stations in the lakes. radiation. Other factors that affect the net back radiation include the emissivity ofwater (relative power of a surface to emit heat by 4.2.3 Terrestrial Radiation radiation), which reduces emitted radiation below that of black body (an ideal surface that Terrestrial radiation over a body of water emits maximum radiation); the reflectivity of (or over land) consists of the incident atmos­ water, which controls reflected atmospheric pheric radiation, reflected atmospheric radia­ radiation; and the concentration of carbon tion, radiation emitted by the water body, and dioxide and ozone in the atmosphere, which energy released through the processes of are minor contributors to the atmospheric evaporation, condensation, and precipitation radiation. The reflectivity of a water surface (latent heat), and turbulent heat transfer for atmospheric radiation is 3 percent (Ander­ (sensible heat). The net result ofthe incident, son 18), only about half as much as for the solar reflected, and emitted radiation components is radiation. 76 Appendix"

The earth and the atmosphere can absorb lengths of the lakes to the winds and the lee and emit more than 100 percent of radiation, shores to the maximum lake effect. exceeding the original input from the sun. This is possible because ofthe so-called green­ Because of the lake effect on adjacent land house effect of the atmosphere. By blocking areas, wind data from stations located arOund terrestrial radiation (very small direct loss to the perimeter of the lakes are of particular space), the atmosphere forces the earth sur­ interest to the Great Lakes. Since more repre. face temperature to rise above the value that sentative data were not available, these data would occur in the absence ofthe atmosphere, have often been used in past studies as over­ which in turn produces upward vertical trans­ water winds, frequently without adjustment fer of both latent and sensible heat. for anemometer height or the increase in wind Terrestrial radiation may be determined in­ speed over the lakes. Average monthly directly from the total (all-wave) and solar perimeter wind speeds for the Great Lakes are (shortwave) radiation measurements, but given in Table 4-7. The average annual all-wave measurements are too sparse for this perimeter wind speed generally increases purpose. Atmospheric radiation may also be from north to south, from approximately 4.5 computed utilizing various radiation indices mls (10 mph) to 5.0 mls (11 mph). Average (temperature, percent of sunshine or cloud monthly wind speeds increase from the sum­ mer low of 3.5 mls to 4.0 mJs (~9 mph) to the cover, vapor pressure). Anderson and Baker 15 present a method of computing incident ter­ winter high of 4.5 mls to 5.5 mls (10-12 mph). restrial radiation under all atmospheric con­ A summary of wind direction for selected ditions from observations of surface air tem­ stations around the lakes and the St. Law­ perature, vapor pressure, and incident solar rence River is presented in Figure 4-12. The radiation. Emitted radiation is determined wind roses in this figure show wind direction from water surface temperatures. Based on frequencies for the months of February, May, available information, estimates ofterrestrial August, and November, indicative of the four radiation for the Great Lakes are as follows: seasonal periods. monthly incident atmospheric radiation var­ ies from a winter low of approximately 300 Actual wind conditions on the lakes and fur­ ly/day in December to a summer high of ap­ ther inland vary somewhat from those indi­ proximately 800 ly/day in June or July; re­ cated by shore stations, which are affected to a flected atmospheric radiation for these varying degree by the lakes and lake-land in­ months represents 9 to 24 lylday (3 percent teraction. The perimeter weather stations are reflectivity); monthly emitted radiation from located at some distance inland, and may gen­ the lakes varies from a low of approximately erally be unaffected by lake breezes, but the 400 lylday during winterto a high of900 lylday stations located on the lee sides of the lakes during summer; monthly net back radiation are certainly affected by the lakes during (longwave radiation loss) is roughly 100 lylday winds from the prevailing wind directions. throughout the year (Figure 4-100). The long east-west axis of Lake Superior is divided by the Keweenaw Peninsula, which 4.3 Winds separates the lake into two basins where winds are frequently ofopposite direction. On the western end ofthe lake (Duluth) the winds' 4.3.1 Lake Perimeter Winds are predominantly from the west and north­ westduringcold months and from the east and Winds are a critical factor of lake climate northeast during the warm months. On the because they provide energy for lake waves, easternend ofthe lake(SaultSte. Marie) there constitute a principal force for driving lake are predominantly easterly winds in the cold currents and shifting ofice cover, and through months and westerly winds during warm air movement provide means for the regula­ months. In the middle section of the lake tion ofthermal budget over the lakes and ad­ (Marquette) predominant winds are from the jacent land areas by heat dissipation and northern and southern quadrants. The mean transfer. In the Great Lakes Region the global monthly wind speed at these stations varies atmospheric circulation with prevailing west­ from 3 mls (7 mph) in the summer to 6 mJs (14 erly winds is of particular importance on the mph) in the winter (for average perimeter lower lakes where it coincides with the lon­ wind speeds for the whole lake see Table 4-7). gitudinal axes of the lakes, exposing the full The maximum recorded wind velocity was 41 H"drom.tlteorolotrl 11

TABLE 4-7 Average Perimeter Wind Speeds fall Cleveland, !uffalo. and London. Ontario: Rochester, Syracuse, Trenton. and Toronto. makes Lake Erie highly susceptible to large­ scale water level motions, especially at the eastern and western extremes ofthe lake. Be­ mls (91 mph) from the south at Marquette in cause of the prevailing wind direction, lake May, 1934. effect on the lee shores is quite pronounced and the monthly wind speeds at Buffalo are Around Lake Michigan the predominant normally somewhat higher than at other sta­ wind direction is from the western quadrant, tions around the lake. Prevailing winds at perpendicular to the long axis ofthe lake. Be­ some locations are from the eastern quadrant, cause ofthe north-south lake orientation, the and in the middle section of the lake (Cleve­ highest seas generally coincide with strong land), prevailing winds during warmer northerly and southerly winds. Prevailing monthsshifttothenorth and south directions. winds from these directions are reported at The maximum velocity recorded was 41 mls (91 some locations, in contrast to the general pre­ mph) from the southwest at Buffalo in dominant westerly direction. The variation in January 1950. prevailing winds is evident in northern Lake The predominant wind direction around Michigan where winds in Traverse City come Lake Ontario is similar to that of Lake Erie, from the south during fall, while Green Bay, with prevailing winds during most months on the opposite (western) shore, is assailed by from the southwest (Rochester, Trenton), westerly winds. Around the southern portion which approaches the direction of the long of the lake prevailing winds in the spring at axis of the lake. During winter months the Milwaukee are from the north, while at Chi­ predominantwind direction shifts to the west. cago they are from the southwest. The mean On the northwestern end ofthe lake (Toronto) monthly wind speed at these stations varies winds frequently prevail from the west, and at from 3m/s to 6 mls (7-14 mph), which is times from the north. The mean monthly wind similar to Lake Superior, but the annual wind speed at these stations varies from 3 mls to 6 speed around Lake Michigan is higher. The mls (7-13 mph). The highest wind velocity re­ highest wind velocity recorded on alltheGreat corded was 33 mls (73 mph) from the west at Lakes, 49 mls (109 mph) from the southwest, Rochester in January 1950. The prevailing occurred at Green Bay in May 1950. winds along the St. Lawrence River are paral­ Winds on Lake Huron may be equally effec­ lel to the river. primarily from the southwest tive on the sea state from all directions due to and secondarily from the northeast. the lake configuration. There is considerable variation in wind direction around the lake, but in general, prevailing winds are from the 4.3.2 Overwater Winds western quadrant. In some locations prevail­ ing winds shift seasonally to the south during Overwater winds differ from overland 18 Appendiz" winds, both daily and seasonally, because of lake-land wind ratios are about two, and for differences in air stability conditions and fric­ stable atmospheric conditions, with air much tional resistance. Daily variation is caused warmer than water, wind speed ratio values mainly by diurnal heating and cooling, which are near one; for the adiabatic or neutral sta­ are more pronounced over land areas than bility conditions values are intermediate. over water and result in larger daily wind var­ Hunt's investigation was conducted mainly iations over land than over water. Seasonal for Lake Erie during navigation season variation is caused by the winter heating and (April-November), with results grouped into summer cooling effects ofthe lakes. The lakes the spring and fall periods. Bruce and Rod­ offer less resistance to wind movement, result­ gersl08 prepared a similar study for Lake On­ ing in considerably higher overwater wind tario. Their investigation was extended by speeds regardless of the season. Lemire 492 who included data from some ofthe The highest wind speeds (one-minute wind other Great Lakes and derived monthly wind gusts) on the Great Lakes, reported from speed ratios for the spring, summer, and fall anemometer-equipped vessels since 1940, are months (March-October). Richards848 ex­ listed for each lake as follows: Lake Superior, tended these ratios for the winter months 42 mls (93 mph) from the northwest in June using partial results determined by Lemire 1950; Lake Michigan, 30 mls (67 mph) from the and extrapolation based on the air-water tem­ west-southwest in November 1955; Lake Hu­ perature difference, along with limited wind ron, 49 mls (109 mph) from the west-northwest observations on Lake Ontario. The variation in August 1965; Lake Erie, 38 mls (85 mph) of wind speed ratios determined in these from the north-northwest in June 1963; Lake studies is shown in Table 4-8. Monthly ratios Ontario, 26 mls (57 mph) from the west­ vary from 1.2 to 2.1, with low values during northwest in November 1964. These velocities summer and high during winter, and an over­ were observed during the navigation season all annual average of about 1.7. and are based largely on observations taken The effects of overwater fetch (length of four times daily during synoptic hours (0100, open water) on lake winds, besides atmos­ 0700, 1300, and 1900 hours, EST). Higher wind pheric stability, were studied by Richards et speeds may have occurred during winter al.,849 who utilized wind data collected during months and at times other than synoptic synoptic surveys on Lakes Erie and Ontario. hours. Most of the shipboard wind directions Their analysis included five stability ranges listed by the National Weather Service verify (from very unstable to very stable), four wind the predominantly westerly wind direction in­ speed classes (3 mls to 8 m/s), and five fetch dicated by the perimeter stations. ranges (10 km to 65 km). They found that the The first intensive effort to determine over­ lake-land wind ratio increases with the at­ water winds utilized various ships-of­ mospheric instability, but the increase is most opportunity programs, which were conducted pronounced in light winds. Under very unsta­ periodically and consisted initially ofcommer­ ble atmospheric conditions (large negative air cial vessels making wind observations four temperature-water temperature difference,

times daily. The data collection program has T A - T w) the wind ratio increases gradually now been expanded to off-shore towers, buoys, from 1.4 for strongwinds to 3.0 for light winds, research vessels, and research stations lo­ with a 2.2 value for all winds. Under very sta­ cated on small islands in the Great Lakes. Be­ ble atmospheric conditions(large positive T A­ cause of practical limitations imposed on Tw difference) the wind ratio increases gradu­ measurement of wind data for prolonged ally from 0.8 for strong winds to 1.4 for light periods oftime overthe lakes, the primary aim winds, with a value of 0.9 for all winds. Thus, ofwind measurement programs was to deter­ under very stable conditions the lakes may mine the relationship between overland and reduce the wind speed, especially in strong overwater winds. Several studies ofthis type winds. The effect ofoverwater fetch was not as have been conducted, relating shore data with pronounced and somewhat erratic. Under un­ observations from ships and islands located on stable atmospheric conditions the wind speed the Great Lakes (Hunt,397.398 Lemire,492 Bruce ratio increases with the overwater fetch, but 649 and Rodgers,I08 and Richards et aI. ). only for lengths smaller than 50 km (25 nauti­ The ratios of wind speed over water to wind cal miles). Under stable atmospheric condi­ speed over land vary diurnally and seasonally, tions the relationshipbetween wind ratios and and are a function mainly ofthe stabilityofthe overwater fetch was highly erratic. Sum­ air. For unstable atmospheric conditions, with marized results of this study are listed to­ water temperature much higher than air, gether with other wind studies in Table 4-8. Hydrometeorology 79

TABLE 4-8 Lake-Land Wind Speed Ratios for the Great Lakes Hunt Lamire Richards, Dragert, McIntyre (1958) (1961) (1966) Stability Range TA-T Period Ratio Period Ratio w (OC) Ratio January 1.961 February 1.941 March 1.88 April 1.81 • Spring 1. 35 May 1.71 >-12.6 2.24 June 1.31 -12.5 to - 4.1 1.88 July 1.16 - 4.0 to 4.0 1.44 August 1. 39 4.1 to 12.5 1.06 Fall 1.82 September 1.78 => 12.6 0.92 October 1.99 November 2.091 December 1.981 Navigation Season 1.58 1.63 Annual 1.75 1.51

1Va1ues for winter months were extended by Richards (1964) through extrapolation.

4.4 Air Temperature Because of the lake effect and lack of direct overwater measurements for any longer period oftime, various investigators used data 4.4.1 Lake Perimeter Temperature from perimeter stations to estimate air tem­ perature over the lakes. The average monthly Temperature is one of the principal indi­ and annual air temperatures for the individual cators of climate and exerts a large influence lakes, based on perimeter data for the 1931-69 on other climatic elements, such as precipita­ period, are listed in Table 4-9. Average annual tion and evaporation. The vast water ex­ temperatures vary from 4°C (39°F) on Lake panses of the Great Lakes moderate air tem­ Superior to 9°C (48"F) on Lake Erie, with in­ perature over the lakes, which in turn has a termediate values on other lakes. Average moderating effect on adjacent land areas. An monthly temperature extremes also occur on indication of the lake effect on shoreline tem­ Lakes Superior and Erie, and vary from peratures is given in Figure 4-95 prepared by monthly lows of approximately -HOC and Pond,822 which compares mean hourly air -4°C (13 and 26°F) in January to monthly temperatures for March,June, and September highs of about 18"C and 22"C (65 and 71°F) in at Douglas Point, located on the eastern July, on the two lakes, respectively. shores of Lake Huron, to those at Paisley The air temperature decreases as latitude climatological station, some 20 km (12 miles) increases, being lowest for Lake Superior and inland. During late winter (March), the lake­ highest for Lake Erie. Disregarding minor shore station is consistently warmer by 2°C to local variations,the average annual tempera­ 4°C (3-T'F) than the inland station. This effect ture on Lake Superior varies from 1°C (34°F) changes gradually during spring to the sum­ along the extreme northern shore to 5°C (41°F) mer effect, providing daytime cooling of the along the southern shoreline(see Figure 4-15). lakeshore station by as much as 4°C (T'F) and Distribution of average annual temperature nighttime warmingby 2°C (3°F) inJune. Inthe on Lake Michigan varies from approximately early fall the effect reverses again and the ~C (43°F) in the north to lOoC (50°F) in the lakeshore station is consistently warmer south. On Lake Huron, the average annual throughout the day. temperature increases southward from 5°C to 80 Appendix"

-F. "C. values for the average conditions on the lakes 1 (Hunt,·11 Powers et al.,1S4 Snyder,151 Rodgers 0 MARCH and Anderson,·'711 Derecki,114 and Richards848). -1 ,,---, -2 "." , -3 " ...., // '-, 4.4.2 Overwater Temperature -4 / / ' .... -5 .... _---_/ 10£'" CAllE TE_~ATUIII 13'" -6 The difference between air temperature IA 2A over lake and land areas in the Great Lakes 00 02 01 oa 01 lO 12 .. • zo zz Basin has been studied by several inves­ 22 ___UPOINT tigators. In describing the climate of South 21 ,-.".-...... " JUNE 848 20 --...... lY ,,/ , Bass Island in western Lake Erie, Verber / \ 19 / \ compared island records (Put-in-Bay) with 18 I \ perimeter and inland stations and concluded I 17 / that the mean mid-summer temperature 16 I (July) decreases gradually from Put-in-Bay to 15 perimeter stations (Sandusky and Toledo) and 14 to inland stations (Tiffin and Bucyrus) located 13 _AN LAKE TEMPERATURE: ...., 12 within 80 km (50 miles) south of the lake. He 11 attributes the higher overlake temperatures 00 02 01 oa oa lO 12 IA .. I' zo zz M to the increased solar radiation due to less precipitation and cloud cover over the lake. 16 During mid-Winter (January) the reverse is 15 SEPTEMBER /'--...... true, because the portion of Lake Erie around ~ ,/ \ 13 /' \ the island region is shallow and normally 12 / , freezes over and is largely ice covered, thus / "- 11 I ...... reducing the lake effect. Although Put-in-Bay 10 I _AN CAlCE TEIIP£RllTUII£ ..., ...... has the highest mean July temperature and 9 ....,1 the lowest mean January temperature of the 800 02 01 01 oa 10 12 IA .. • zo 22 M five stations, its average monthly range dur­ TIME IESTI ing the year is the smallest, because thermal FIGURE 4-95 Mean Hourly Temperatures for stability over the lake acts as a damper Douglas Point and Paisley, Lake Huron, for the against sudden heating or cooling. The aver- Months of March, June, and September, 1962 From Pond. 1964 TABLE 4-9 Average Perimeter Air Tempera- 8"C (41 to 4'T'F). The annual temperature on ture for the Great Lakes, 1931-1969 (Degrees Lake Erie increases from a low of goe (47"F) Centigrade) along the northeastern shore to a high of11°C Lake (52"F) along the southwestern shoreline. On Pet"iod Superior Michigan Huron Erie Ontario Lake Ontario, the annual air temperature in­ January -11.2 - 6.4 - 7.4 - 3.8 - 5.2 creases from 7°C to goe (45 to 48°F) between the February -10.4 - 5.6 - 8.0 - 3.7 - 5.2 Karch - 4.7 - 0.4 - 3.5 0.9 - 0.1 northern and southern shores. April 3.1 6.8 4.1 7.4 6.8 It should be noted that the air temperatures May 9.2 12.7 10.2 13.6 13.1 June 14.6 18.4 15.8 19.2 18.7 discussed above are based on lake perimeter July 18.1 21.3 18.9 21.8 21.4 stations and may be different from those rep­ August 17 .4 20.4 18.7 20.9 20.3 resenting mid-lake conditions. Some differ­ September 13.0 16.4 14.2 17.2 16.4 October 7.2 10.3 8.7 11.1 10.2 ence in these temperatures is introduced by November - 0.7 2.8 2.0 4.3 3.8 the land effect, which takes place not only at December - 7. 7 - 3.6 - 4.2 - 1. 7 - 2.8 shore stations but also in the shallow coastal Annual 4.0 7.8 5.8 8.9 8.1 waters. Furthermore, most first order perime­ ter stations used to derive temperature esti­ Values are based on data for the following stations: mates are located some distance inland, where Superior: Sault Ste. Marie, Marquette, Duluth, and the land effect is more pronounced. Neverthe­ Thunder Bay. less, because of data limitations, air tempera­ Michigan: Milwaukee. Muskegon, and Green Bay. Huron: Alpena, Gore Bay, and Wiarton. tures from the perimeter stations around the Erie: Toledo, Cleveland, Buffalo, and London. lakes are generally used as representative Ontario: Rochester, Syracuse, Trenton, and Toronto. Hydrometeorology 81

age monthly maximum-minimum tempera­ higher than water surface temperature, while ture range increases gradually inland from air temperature at Toronto displays a differ­ approximately 8°C (14°F) at Put-in-Bay to ap­ ent pattern and is on the average approxi­ proximately 12°C (2~F) at Bucyrus, and the mately 1rc (30°F) higher than water temper­ frost-free season decreases gradually inland ature. These figures are based on only three from more than 200 days to approximately 150 days of data from a single cruise, and their days. Also, the hottest days in July show magnitudes may not be valid for longer higher temperatures on mainland stations periods. Rodgers and Anderson875 state that than at Put-in-Bay. During the winter an ice there are insufficientdatato provide a reliable cover around the island region, usually form­ conversion ofland station temperatures to the ing in January and lasting through February, overwater air temperatures. At present, with acts as an insulator between the warm water approximately a decade ofdata available, this and cold air, producing enough change in the difficulty has been overcome but the conver­ normal temperature pattern to make Feb­ sion factors have not been developed. There ruary colder than January. On exceptional oc­ are no published reports presenting overwa­ casions when the lake is free of ice during ter temperatureson the lakes, otherthan data these two months, temperature was approxi­ reports for the individual surveys. mately 3°C (5°F) higher. Summer temperature conditions for west­ ern Lake Erie may be assumed to be indicative 4.5 Water Temperature of temperature modification by the other Great Lakes, although mid-lake modification is undoubtedly more pronounced since the is­ 4.5.1 Water Surface Temperature land itselfproduces some effect. Winter condi­ tions, on the other hand, may not be compara­ The oldest sources of water surface temper­ ble because other lakes have much greater ature data in the Great Lakes are the records depths and different ice-cover conditions. in a obtained at various marine structures, such as comparison of summer data for Fort William docks, breakwaters, and lighthouses. These and Caribou Island 075 miles away) made in stations were later replaced by the somewhat connection with a synoptic survey of Lake Su-. more sophisticated sites offered by the intake perior, Anderson and Rodgers14 show that air structures of the water treatment plants lo­ temperature on the island is much more stable cated around the lakes. Water temperature at than and differs considerably from that at the intake stations is obtained in the coastal Fort William, on the perimeter of the lake. waters, a few hundred to a few thousand me­ They state that measurements from the island tersoffshore, and at depthsof3 to 15 meters(l0 are extremely valuable since they represent to 50 feet) below the surface. These data obvi­ an entirely maritime situation, which is ously do not represent the temperature at the caused by modification oflow level air masses surface and require adjustments for open lake by the lake. However, there are only a few conditions. Initially, open lake measurements island stations measuring air temperatures were made by commercial vessels along their on the lakes and most are operated only during navigation routes, and more recently by re­ the navigation season. search vessels engaged in synoptic surveys of Measurement of air temperature on lake the lakes. The latest development in measur­ towers or buoys, and synoptic surveys by ves­ ing surface water temperatures involves the sels initiated in the late 1950s, provide more use of airborne infrared thermometers. The reliable data by eliminating possible island ef­ use of airborne radiation thermometers per­ fects. Based on synoptic survey data collected mits fast and regular observations of surface by the research vessel Porte Dauphine on temperatures over large areas. Information Lake Ontario, Bruce and Rodgers108 observed on surface temperatures is also provided by that air temperatures at 3 mOO ft) above the satellite imagery, but in the present state of water surface are much closer to water sur­ art this information cannot be used for quan­ face temperatures than to land temperatures titative temperature determination. at the lake perimeter (mean of temperatures Among the earlier studies ofthe water sur­ at Toronto and Rochester). Rodgers and An­ face temperatures in the Great Lakes were derson,875 utilizing these data in the energy those by Freeman 271 and by Horton and budget study of Lake Ontario, made similar Grunsky.376 In both studies water tempera­ observations and showed that air tempera­ ture records from daily observations at vari­ ture over water in June is about 6°C OO°F) ous harbor locations for the 1874-86 period 1 8t Appendix ~

TABLE 4-10 Comparison of Great Lakes Water Surface Temperature (Degrees Centigrade)

Jan. Feb. Mar. Apr. May June July Aug. Sept. Oct. Nov. Dec. Annual Lalte Superior Lake Survey (1944) 1904-431 0 0 0 3 4 8 12 11 8 5 1 4 Millar (1952) 1935-39 2 4 7 13 12 9 6 Richards & Irbe (1969) 1959-68 2 0 0 1 2 4 7 12 12 9 6 4 4 Lalte Michigan Lake Survey (1944) 1904-431 0 0 1 4 7 12 17 18 16 11 7 2 8 Hillar (1952) 1935-41 5 11 16 21 18 12 8 Lalte Huron Lake Survey (1944) 1904-431 0 0 1 3 6 12 18 19 17 12 7 2 8 Millar (1952) 1935-41 4 9 18 20 16 12 7 Richards & Irbe (1969) 1959-68 3 2 1 1 4 8 15 18 16 12 8 6 8 Lalte Erie Lalte Survey (1944) 1904-431 0 0 3 6 9 18 22 22 21 14 7 1 10 Millar (1952) 1937-41 10 17 21 23 19 IS 9 Richards & Irbe (1969) 1950-68 1 1 1 3 9 17 21 22 19 15 9 4 10 Lalte Ontario Lalte Survey (1944) 1904-431 0 0 2 5 9 14 18 19 17 13 7 1 9 Millar (1952) 1936-46 3 2 2 3 6 12 19 21 18 13 7 4 9 Richards & lrbe (1969) 1950-68 3 2 2 3 6 12 19 21 18 13 7 4 9 lperiod shown for Lalte Survey study indicates extreme limits and not actual length of data. were used as basic data. Monthly tempera­ Determinations of Great Lakes water sur­ tures for the individual lakes were derived by face temperatures, limited to a single lake, applying correction factors for time of obser­ were made by several investigators. vations and some adjustment for open lake Church 142.143.144 analyzed water temper­ conditions. atures for Lake Michigan, based on bathy­ The U.S. Lake Survey825 compiled monthly thermograph observations obtained during water surface temperatures for each lakefrom 1941-44 period. He showed that irregularities temperature data collected at various loca­ in the water temperature distribution are the tions by the field parties from 1904 through result of strong winds and upwelling, both of 1943. Derived values differ somewhat from which act to lower the water temperature at those given by Freeman and are considerably the surface. Another presentation of Lake different from Horton and Grunsky's values. Michigan temperatures for the summer Probably the best known and most often months is given by Ayers et a1. 29 Their values used Great Lakes water surface temperatures are based on synoptic surveys conducted in are those determined by Millar.s43 Millar's 1955. Ayers et al.28 also determined water study is based on the data obtained from con­ temperatures for Lake Huron from a similar tinuous recordings of water temperature survey conducted on thatlake in 1954. The last taken by thermographs installed on the con­ two studies are summarized by Ayers.25 denser intakes of steamships. Data collection Monthlywatertemperatures on Lake Erie are coversthe 193~6period for Lake Ontario and given by Powers et a1. 824 who present a com­ the 1935-41 period for all other lakes. Millar parison of long-term water intake records to developed temperature distributions on each offshore cruise data. lake by months and derived average monthly The most recentdetermination ofwater sur­ values for each lake. Due to restricted naviga­ face temperatures was made by Richards and tion, winter temperatures for lakes otherthan Irbe.851 Their study covers the 1950-68 period Ontario were either not available or gave in­ for Lakes Erie and Ontario, and the 1959-68 sufficient coverage to derive reliable monthly period for Lakes Huron and Superior. Monthly means. Many investigators in recent years temperatures determined for each yearon the have used Millar's temperatures, most fre­ individual lakes were based on available in­ quently to adjust surface temperatures de­ formation from airborne radiation thermome~ rived for various periods from the water in­ ter surveys, ship observations, water intake takes or other sources. Studies of this type stations, and subjective adjustments of mean include Hunt,39S Snyder,751 Rodgers and An­ lake temperatures based on mean air temper­ derson,87s and Richards and Rodgers.854 atures from shoreline stations. The subjective HlIdrometeorolof/1/ 88

adjultments of mean water temperatures IfM'lta'Ulf .C" .,ere used primarily during winter months for o 6 lakes lacking sufficient temperature meas­ 10 12 14 \6 l' 20 22 urements. . A comparison of the Great Lakes mean ('HIMNION i .ater surface temperatures (Table 4-10) -----~ ,howS that surface temperatures vary with 'HERMOCLINE latitude and depth of the lakes. The average annual surface temperatures vary from 4°C (40"F) for Lake Superior, the northernmost and deepest lake, to lOoC (600F) for Lake Erie, the southernmost and shallowest lake. Aver­ age annual surface temperatures on the other lakes, with intermediate latitudes and depths, amount to SoC (46°F) for Lakes Michigan and Huron and 9°C (48"F) for Lake Ontario. The average monthly surface temperatures vary from the winter lows of O"C (32"F) on Lake Su­ perior and 2°C (36°F) on Lake Ontario to the 100 summer highs of1aoC (56°F) on Lake Superior and 2SOC (73°F) on Lake Erie.

4.5.2 Temperature at Depth Many of the studies mentioned in the pre­ ceding discussion on water surface tempera­ FIGURE 4-96 Seasonal Changes in the Ther­ ture also deal with the vertical temperature mal Structure of • Large, Deep Lake of Mid­ distribution in the Great Lakes. Church 1~.143 latitudes showed that the annual temperature of Lake Michigan undergoes four basic seasonal cycles producing further development and deepen­ with distinct characteristics, namely, the ing ofthe thermocline. As the thermocline ap­ spring warming, summer stationary, autumn proaches its maximum depth, the mixing be­ cooling, and winter stationary. Because ofdif­ comes progressively less. The final deepening ferent latitudes and depths, the timing and occurs during summer storms with relatively duration in the lakes of these cycles are not little activity during calmer periods. synchronous, but all lakes display these four The summer stationary period is charac­ basic seasonal periods. Occurrence of the terized by nearly stationary lake surface tem­ periods is governed by the lake temperature­ peratures in their maximum range. At the be­ water density relationship. ginning of this period the thermocline de­ The seasonal changes of thermal structure scends to its maximum depth, where it re­ in large deep lakes of mid-latitudes, such as mains relatively stable. The establishment of the Great Lakes, are shown graphically in a strong thermocline at constant depth indi­ Figure 4-96, and discussed in Section 6. As the cates that there is only a negligible transfer of warming season progresses, the top layers of heat by conduction, either above or below the water absorb heat from the atmosphere, be­ thermocline. Anderson and Rodgers14 showed coming progressively warmer, and through that the maximum depth of the thermocline conduction of heat downward and mixing in­ during this period of peak heat content is duced by winds, the warm layer becomes about 15 m (50 ft) in all the Great Lakes, in deeper. The warm upper water (epilimnion) spite of marked difference in their transpar­ is separated from the cold deeper water ency, configuration, size, orientation, and (hypolimnion) by the thermocline. The latitude. epilimnion is less dense and literally floats on Lake cooling begins in the fall, with sub­ top of the hypolimnion. In the early stages of stantial net loss of heat resulting from the development, the thermocline is rather weak interaction of cooling and heating processes, and is easily broken down by wind action, such as radiation, evaporation, conduction, which readily mixes the thin surface layer of precipitation, and condensation. In the fully heated water with colder water below, thus developed cooling period of late fall, with 84 Appendi%.4 water surface temperature substantially tures. The coastal waters are well above 4°C at above that ofmaximum density, cooling ofthe the surface and vertically stratified, While homogeneous surface layer proceeds rapidly, those in mid-lake are less than 4°C and uni­ since only this layer is affected due to stability form in temperature from top to bottom in ofthe thermocline. At the same time, with the depths as great as 180 m (600 ft) (Church IQ increased frequency of storms in this season and Rodgers8 'l'2). The most extensive investi_ the upper layer becomes deeper as well. As the gation of this phenomenon, named the ther­ cooling continues, temperature of the upper mal bar, was made on Lake Ontario by Rod­ layer approaches that of maximum density, gers.lI13 As the warming season progresses the the thermocline becomes less stable, and the thermal bar zone moves lakeward from the wind-stirring may be complete from top to bot­ shores and eventually disappears. During the tom with the destruction of the thermocline. fan coolingperiodthe processis repeated, with Church1" indicates this critical temperature the cold temperatures moving offshore to­ of the homogeneous surface layer to be 6°C wards the center of the lake. (43°F) or slightly above. In the advanced In the above discussion ofseasonal temper­ stages of cooling, depth exerts an important ature changes within a column of water, no control on temperatures in the lakes. With the consideration was given to any movement of loss of the thermocline the entire column of water into or out of the column due to the water becomes isothermal and further cooling normal lake currents. Actually, water tem­ at the surface proceeds slowly because the en­ peratures and currents in a lake are closely tire water column loses heat. related. Any lake is subject to water move­ Duringthe winter stationaryperiodthe lake ment resulting from the inflow-outflow bal­ waters are again at less than maximum ance and the general water circulation in­ density, but in this case the surface is colder duced by winds and geostrophic forces. Thus, than 4°C. In coastal areas and lakes with shal­ once established, a thermocline seldom stays lower depths temperatures are generally still, but fluctuates in a wave-like motion (Sec­ isothermal vertically. In deep waters an in­ tion 6). Rodgers8'72 statesthatthe thermocline verse stratification develops, with colder moves up and down through a distance of10 to epilimnion and slightly warmer hypolimnion. 20 m (30 to 60 ft), with wavelengths of tens of This stratification is not as pronounced asdur­ kilometers. He also states that these internal ing the summer, because the downward mix­ waves are seldom evident to the casual ob­ ing process is aided by strong winds and con­ server and their detection requires continu­ vection produced by daytime heating during ous observation of temperatures at one or winter. Water temperature in the entire col­ more locations within the lake. umn or the deep upper layer, whichever the Periodically, strong winds blowing steadily case, approaches freezing point on the lakes from one direction may tilt the thermocline, with extensive ice cover, and stays at about deepening the epiIimnion on the downwind ZOC (3~F) on the lakes which are largely free of shore and reducing its depth on the upwind ice. Because great depths are affected, water shore. Prolonged strong winds pile up warm temperature changes during this period are surface water and produce sinking on the naturally slow. Thus, Lake Erie with its shal­ downwind shore, while at the same time they low depths freezes sooner and more often than remove warm surface water and produce cor­ Lake Superior, which because of its great responding upwelling of cold water from depths is capable of sustaining tremendous deeper layers on the upwind shore. Upwelling heat losses. By the same token, ice breakup on occurs frequently on the northwest shoreline Lake Erie is much faster. of Lake Ontario and the west shores of Lake Since each lake consists of a whole range of Michigan. Tilting of the thermocline can be depths, which generallyincrease from shallow observed from the water temperature records coastal waters to a maximum depth in mid­ of municipal water intakes located on the op­ lake, each lake contains water masses with posite shores of the lake. distinct thermal structures characteristic of their depth, particularly during warming and cooling periods. In the beginning stages ofthe 4.5.3 Air-Water Temperature Relationship spring warming period the shallow waters along the shore warm up much faster than The difference between air and water mid-lake areas, producing large horizontal surface temperatures is the primaryindicator temperature gradients near the boundary be­ of the atmospheric stability over the lakes. tween the warm and cold surface tempera- When the air is warmer than the water s~r- Hlldro1'MteoroloW 85

tario, Hunt" used mean values from three

-8 perimeter stations (Toronto, Oswego, and Trenton) for the 1937-56 period to obtain aver­ age air temperature over the lake. Water sur­ face temperatures for this lake were obtained by adjusting values given by Freeman,ln U.S. Lake Survey,815 and Millar.l43 In the same study, for the northern Lake Michigan air temperature, Hunt used St. James and Beaver Island data from 1911 to 1956; water surface temperatures were obtained from rec­ ords in the vicinity of Beaver Island from un­ stated sources. In a study on Lake Erie, Hunt398 used the water surface temperatures given by Freeman, U.S. Lake Survey, and Mil­ lar, and the normal air temperatures from four perimeter stations (Detroit, Cleveland, Erie, and Toledo). In a second study for Lake Ontario, Rodgers and Anderson 875 used Mil­ lar's water surface temperatures and normal air temperatures from Toronto and Rochester. In all cases except Lake Michigan, air tem­ perature over the lakes was determined from perimeter stations and may be considerably different than at mid-lake, as pointed out by Rodgers and Anderson in their study. Because

.10 '-- --' land area is more sensitive toboth heatingand cooling, these air temperatures obtained at .. " M o N 0 perimeterstations would tend to intensify the FIGURE 4-97 The Relationship of Air-Water extreme conditions, being higher than at mid­ Temperature Differences in the Great Lakes lake during summer and lower during winter periods. Air temperature data for Lake Michi­ gan represent conditions over an island and face, the air will tend to be stable. Conversely, may be more representative. The water sur­ when the air is cooler than the water surface, face temperatures, except those by Millar, the air will tend to be unstable. Thus, other were also determined mostly from perimeter things being equal, the greater the positive stations and would tend to have the same de­ difference between the air and water surface ficiencies, but probably not to the same de­ temperatures, the more stable the atmos­ gree. Thus, the air-water relationships shown phere and smaller the opportunity for the oc­ probably over-accentuate the magnitudes be­ currence of overwater precipitation, high tween these temperatures. winds, and evaporation. On the other hand, The monthly values of air-water tempera­ the greater the negative difference between ture difference indicate that there are four the air and water surface temperatures, the periods of prevailing atmospheric conditions more unstable the atmosphere and greater over the Great Lakes. Duration and timing of the opportunity for the occurrence of the these periods vary for different lakes, depend­ above climatic processes. ing primarily on latitude. Generally, the air is Atmospheric stability over the Great Lakes normally warmer than water during spring varies appreciably during the year. The air is months of April, May, and June (also July in normally warmer than the water surface dur­ northern areas), and the atmospheric condi­ ing spring and colder for a somewhat longer tions are stable. During mid-summer months period in fan (Figure 4-97). Because of the ofJuly and August the air and watertempera­ shortcomings in data used for air-water tem­ tures are approximately the same, thus indif­ perature differences, their magnitude may ferent or adiabatic (occurring without loss or not be representative of the actual mid-lake gain of heat) equilibrium conditions exist in conditions, and considerable variation in the the atmosphere. From early fan to mid·winter magnitude shown at times for the same lake months(Septemberthrough February)the air seemstoindicatethisdeficiency. For LakeOn- is normally colder than water and the atmos- 86 Appenditr '" phere is unstable. However, during winter TABLE 4-11 Avenge Perimeter Humidity for months the presence ofice and snow cover has the Great Lakes (Percent) a significant modifying effect on the air-water Lake temperature relationship. Finally, duringlate Period Superior Michigan Huron Erie DDtario winter in March (also April in northern areas) January 77 76 81 77 78 the air and water temperatures are again February 76 73 79 77 77 about the same, and the adiabatic equilibrium March 74 73 76 74 74 AprH 69 69 73 70 69 conditions prevail. The adiabatic equilibrium May 68 66 70 69 68 conditions occur during relatively short, June 72 69 74 70 70 transitory periods and variation in the atmos­ July 74 71 74 71 68 August 76 74 77 74 72 pheric conditions during these two periods September 79 76 7. 75 74 may be considerable. In contrast, the spring October 76 73 79 75 74 stable conditions and the fall unstable condi­ Nov"'er 78 76 83 78 78 tions present long, well-established periods. Dec"'er 79 78 83 79 78 Annual 75 73 77 74 73 4.6 Humidity Values are based on mean data published in 1969 for the following stations: 4.6.1 Lake Perimeter Humidity Superior: Sault Ste. Marie. Marquette. Duluth. aDd :'bUDder Bay. Michillan: Milwaukee. Huskegon, and Green Bay. The influence of the lakes produces higher Huron: Alpena, Gore Bay. and Wianon. Erie: Toledo, Cleveland. Buffalo. aad London. and more stable humidity in the Great Lakes Ontario: Rochester, Syracuse, Trenton, and Toronto. area than at similar latitudes in the mid­ continent. Since warmer atmosphere is capa­ An estimate ofthe average monthly and an­ ble ofholding more watervapor, the amount of nual humidity values for the individual Great water vapor present in the atmosphere varies Lakes, based on data from perimeter stations, constantly with temperature and availability is given in Table 4-11. The perimeterhumidity of moisture. However, this discussion is con­ for all lakes increases from a low of approxi­ cerned with the relative humidity or the ratio mately 70 percent in the spring to a high of between the actual vapor pressure and the approximately 80 percent during the late fall. saturation vapor pressure at the same tem­ Average annual humidity varies from 73 per­ perature. Since all lakes provide large quan­ cent for Lake Michigan to 78 percent for Lake tities of moisture through evaporation, the Huron, with intermediate values for other upper Great Lakes with lower temperatures lakes. and corresponding lower dew points attain Examination of records for the individual somewhat higher values ofrelative humidity. stations aroundthe lakesindicates a dailyvar­ Prevailing winds and lake breezes are impor­ iation and a general northward increase in tant factors in raising or lowering humidity humidity. Based on four observations a day, values on land areas adjacent to the lakes. highest humidity normally occurs late at Humidity measurements on lake perimeters night and during early morning hours <1:00 are provided by the first order meteorological and 7:00 a.m. readings), while lowest humidity stations located around the lakes. Humidity occurs in the early afternoon (1:00 p.m. read­ values at these stations are generally pub­ ing). At most stations average annual relative lished for the four daily synoptic hours (1:00 humidity values for the night and morning and 7:00, a.m. and p.m.), but hourly values are readings range between 75 and 85 percent, also available. Data from these stations are with the maximum values occurring during the sole source ofcontinuoushumidity records summer. The afternoon readings range be­ for extended periods of time. Because of the tween 60 and 70 percent, and are lowest in the lake effect on adjacent land areas, various in­ spring and summer. At most locations average vestigators have utilized these data to obtain daily range in humidity is from 5 to 10 percent estimates of humidity over the lakes by av­ duringwinterand from 15 to20 percentduring eraging records from several perimeter sta­ summer. tions. Some of the more recent studies also employed correction factors to adjustthesees­ timates to overlake conditions. The correction .(.6.2 Overwater Humidity factors were derived from infrequent overwa­ ter measurements, similar to those used for The humidity records from perimeter sta­ winds, and are discussed later. tions contain both lake and land effects and H1IdrometeoroloVII 81

_ABLE 4-12 Lake-Land Humidity Ratios for TABLE 4-13 Average Perimeter Precipita- ..Great Lakes tion for the Great Lakes, 1937-1969 (cm) Lalr.e Richards & Fortin Jackson Period Superior Michigan Huron Erie Ontario (1962) (1963) 1959-1961 1959-1962 January 5.5 4.8 6.7 6.5 6.e ..riod F"bruarj 4.1 3.9 5.1 5.7 6.4 Karch 4.4- 5.0 5.3 6.9 6.5 -January 1.33 1. 25 April 6.0 7.3 6.5 8.5 7.2 1.30 1.24 Kay 7.6 7.6 7.0 8.2 7.5 february June 9.0 8.6 7.2 8.3 6.3 .rcb 1.21 1.22 July 7.2 7.5 6.7 7.7 7.2 April 1.14 1.04 August 8.7 7.7 7.5 8.1 7.4 .86 .89 September 8.6 8.4 8.2 7.1 7.0 MaY October 6.4 6.2 6.9 6.8 6.9 June .94 .94 November 6.8 6.3 7.7 7.4 7.5 July 1.09 1.10 December 5.5 4.8 7.3 6.5 7.0 1.09 1.10 August Arlnua1 79.8 78.1 82.3 87.7 83.7 September 1.11 1.09 October 1.15 1.14 November 1.15 1.13 !(ate: Based on data assembled "w the Lake Survey Center. December 1.31 1. 28 NOAA.

Annual 1.14 1.12 ference in time is related to the number of daily observations, and indicates that four ob­ servations a day are not sufficient to obtain are not necessarily representative of the ex­ the daily humidity distribution. tensive water areas included in the Great Lakes. The major controlling factor of humid­ ity is the air temperature, and temperatures 4.7 Precipitation over lake and land areas differ. Overwater humidity data are obtained either from direct overwatermeasurements, which are available 4.7.1 Lake Perimeter Precipitation only on an intermittent basis, or from empiri­ cal relationships derived from those meas­ Precipitation includes all forms of moisture urements. Such relationships combine many deposited on the earth surface from the at­ of the differences between overwater and mosphere. The principal forms ofprecipitation overland conditions into a single correction include rain, hail, sleet, and snow, all ofwhich factor, a lake-land humidity ratio (Richards are readily measurable. Of particularinterest and Fortin,650 Jackson 420). The monthly in the Great Lakes Basin is the precipitation humidity ratios derived in these studies are measured at lake perimeters. Investigations shown in Table 4-12. They indicate that on an of precipitation distribution indicate that annual basis overwater humidity is some 10 to perimeter stations show marked variation 15 percent higher than overland humidity at from precipitation further inland, and since perimeter stations. During spring, overwater direct overwater measurements are generally humidity is approximately 10 percent lower not available, it is assumed that perimeter ob­ than perimeter humidity, but during the rest servations are sufficiently representative of ofthe year overwater humidity is higher, with overwater conditions (e.g., Freeman271). a maximum difference of approximately 30 Estimates for the average monthly and an­ percent in the winter. nual precipitation on individual lakes during The average daily variation ofthe humidity the 1937-69 period are shown in Table 4-13. ratios presented in the studies shows high The annual precipitation varies from 78 cm humidity ratios during the night and low (30.8 inches) for Lake Michigan to 88 cm (34.5 ratios during daytime hours. The nighttime inches) for Lake Erie, with an overall average maximum occurs generally between 1:00 and for all the lakes of 81 cm (32.0 inches). Annual 4:00 a.m. and the daytime minimum occurs precipitation increases from north to south around noon. Richards and Fortin, based on and from west to east. The southward increase four daily observations, indicate lowest in precipitation is climatic, since warmer at­ humidity ratios at 1:00 p.m., while Jackson, mosphere is capable of sustaining more mois­ using eight daily observations, shows the low­ ture, while the eastward increase is caused by est ratio at 10:00 a.m. Inspection of their the lake effect, since additional moisture is diurnal variation curves shows that the dif- supplied to the atmosphere by the lakes as 88 Appendix '" ......

4··I ---i---::---).'-4-_-...A• *.,

N· ... n·

FIGURE 4-98 Mean Monthly Number of Days with Measurable Precipitation (open bars) and Thunderstorms (shaded) in the Great Lakes Basin From U.S. Weather Bureau. 1959

they are exposed to the prevailing westerly measurable precipitation and thunderstorms winds. at perimeter stations is shown in Figure 4-98. The seasonal precipitation pattern shows Highest thunderstorm frequencies occur well-distributed and abundant precipitation along the western shore ofLake Michigan and throughout the year, although a larger por­ the southern shore of Lake Erie. tion of the annual supply falls during the During the winter months precipitation in summer months, a characteristic ofcontinen­ the Great Lakes Basin is largely in the form of tal climates. The relatively high summer rain­ snow. In the northern areas it generally con­ fall is especially pronounced on the western sists of snowfall exclusively, with permanent and northern lakes. The average monthly pre­ snow cover throughout the winter. In the cipitation increases from a winter low of 4 cm southern areas precipitation alternates be­ to 5 cm (1.6 to 2.0 inches) in the upperlakes and tween snowfall and rainfall, with intermittent 6 cm to 7 cm (2.4 to 2.8 inches) on the lower snow cover on the ground. lakes to a summer high of8 cm to 9 cm(3.1 to 3.5 inches) on all lakes. The mean number of days per month with 4.7.2 Overwater Precipitation measurable precipitation at perimeter sta­ tions increases from the windward to the lee Observations have indicated that large sides ofthe lakes, and also increasesgenerally bodies ofwater, such as the Great Lakes, mod­ from summer to winter months. Exceptions to ify the atmosphere above them, including pre­ this seasonal distribution occur along the cipitation patterns. One theory explainingthe western and northern shores of Lakes Michi­ reduction of overwater precipitation in the gan and Superior. The number of days with ·summeristhatthe watercools the air above it. Hrdrometeorolon 89

. r precipitation measurements on the lake-land precipitation ratios, which indicate n(~il to confirm this in all cases. Horton that precipitation on Lakes Superior, Michi­ GI nsky 376 and Verber848 suggested that gan, and Huron (Beaver Island) averages 93 e::: overwater precipitation in the sum­ percent of that measured at shore stations ay be due to less thunderstorm activ­ during winter months, and 94 percent during ;':is results from the cooling effect of the summer months. For Lakes Erie (North Bass • which produces a more stable atmos­ Island) and St. Clair their overwater precipi­ I, over the water than over surrounding tation amounts to 84 percent of perimeter pre­ ~reas.Byers and Braham 117 .ma.de a simi­ cipitation for winter and 85 percent for sum· observation about Lake MIchIgan and mer months. However, Day205 suggests that ~ed that in the winter the warm~n~ef~ect of the differences in island and shore precipita­ til lakes encourages greater precIpItatIOn on tion are due to wind reduction of precipitation e lee shore than on the windward shore. gage catch at the more exposed island sites, ~arson 599 from a study of precipitation ob­ rather than any real deficit in precipitation on rvatio~s by radar, found that the formation the lakes. =air-mass showers over Lake Michigan in the To study overwater precipitation a storage­ .ummer was inhibited. type precipitation network was established in The elements affecting overwater precipita· 1952 on a number ofislands in northern Lake tion in the winter are less definite. Itis appar­ Michigan by the U.S. Lake Survey and the ent that the warming effect of the lakes en­ U.S. Weather Bureau. Based on twice-a-year courages snow flurries, but whether the pre­ precipitation records from this network and cipitation is less, equal to, or more than that monthly records from Beaver Island and adja­ falling on the adjacent shorelines is a matter cent shore stations, Hunt395 concluded that of controversy. Light precipitation belts on annual overwater precipitation on Lakes Mich­ windward sides and heavy precipitation belts igan and Ontario averages 79 percent of that on the lee sides of the lakes are well estab­ measured at perimeterstations, with monthly lished, but the quantities recorded at these values varying from 60 percentin Augustto 91 locations do not necessarily yield representa­ percent in November. Hunt assumed precipi­ tive overwater precipitation. The observed tation at the smallest, exposed island to be heavy snowfall on the lee shores of the lakes true overwater precipitation. Kohler"l ques­ might be confined to the lake perimeters and tioned that assumption and indicated thatthe would then represent an accumulation ofpre­ relative catch ofthe island gages is highly cor­ cipitation resulting from the lake-land in­ related with windiness, and that virtually all teraction. The elevation of the air mass as it the differences in precipitation catches could moves from the water to the land surface, be explained by relative windiness atthe gage coupled with the air movement from the warm sites. In another precipitation study of the water to the cold land during winter, combine northern Lake Michigan island network, to cause more precipitation on the lee shores. Kresge, Blust, and Ropes 476 agreed with This may apply to the islands as well. Winter Kohler and used the higher measured measurements are also less accurate because amounts at both island and land gages as true snow is more sensitive to wind, increasing the overwater and shore precipitation and cor­ effects of exposure, so the gages do not ade­ rected the other gage records for gage expo­ quately measure winter precipitation. Freez­ sure. The annual overwater precipitation was ing of the lakes complicates the process fur­ about the same (102 percent) as precipitation ther. from shore stations. Seasonally, overwater - Recognizing the shortcomings of perimeter precipitation ranged from 3 to 10 percent less observations for estimating overwater pre­ than perimeter precipitation in the summer, cipitation, several investigators derived rela­ depending on the offshore and onshore winds, tionships ofoverwater to perimeter precipita­ respectively; and 9 percent more inthe winter. tion, utilizing data from islands to represent overwater conditions. However, island data In a Lake Michigan precipitation study may not be reliable for this purpose, and re­ based on records from the Four-Mile Crib in sults of the studies are often contradictory. the southern tip of the lake and land gages in Based on records from Beaver Island in the the Chicago area, Changnon137 determined a Lake Michigan and North Bass Island in Lake lake-land precipitation relationship similar to Erie, Horton and Grunsky 376 concluded that that derived by Hunt. Changnon's lake-land precipitation on the lakes was lower than at precipitation ratios show monthly variations perimeter stations. They calculated seasonal ranging from 78 percent in October to 95 per- 90 Appendiz 4

TABLE 4-14 Lake-Land Precipitation Ratios years used in their derivations. The Upchurch for Lake Michigan and Lake Erie data are not comparable directly with other data in the table, because Upchurch used re­ Lake 1I1ch1&an Lake Erie mote inland gages while all others used lake Irelae I 1 perimeter gages. The inland stations, how­ Hunt Chanlnon Et aJ. Upchurch Derecki (1959.) (191)1) (1963) (1970) (191)4) ever, indicate more correctly the lake effects .l5 years 11 years 30 years 6 years 13 years Period 1911-56 194~-51> 1906-62 1963-68 1920-46 on precipitation distribution throughout the year. January .88 .95 1.13 1.39 .95 February .85 .81> 1.13 2.26 .89 llarch .81 .84 1.07 1.05 1.03 April .78 .81 1.01 .89 1.04 llay .75 .87 .96 .83 1.07 4.7.3 Weather Radar June .73 .79 .93 .49 .92 July .69 .82 .90 .70 1.04 August .60 .87 .91 .64 1.03 Present methods ofmeasuringprecipitation September .74 .83 .98 1.10 .95 October .89 .78 1.05 .96 .89 have many shortcomings. such as use of point November .91 .79 1.10 1.62 1.02 measurements to represent an area, varia­ Occ~1Ilber .89 .89 1.13 1.32 1.03 tions in gage catch, accuracy, effects ofwindi­ Wlr.ter ness and exposure on the catch, and access or (Nov-Apr) .85 .86 1.09 1.42 .99 SUIl"'JDer installation difficulties in remote areas on (llay-Oct) .73 .83 .96 • :9 .98 large bodies of water. A potentially powerful Annual .79 .84 1.02 1.10 .99 tool for eliminating some of these problems and obtaining more truly representative pre­ IUpChurch ratios are ba.ed on inland stations located 160 km west cipitation data may be the use of weather of the lake (thiS appendix). radar. Weather radar is applicable to both land and water areas, but it is of particular cent in January, with an annual average of84 interest for large lakes because ofgage instal­ percent. lation and access difficulties. The lake-land precipitation ratios calcu­ The use of weather radar for obtaining 214 lated for Lake Erie (Derecki ) show consid­ quantitative precipitation data requires erable monthly variations, without a definite climatological analysis of photographed pre­ seasonal trend. Overwater precipitation was cipitation echo patterns. The process involves based on records from Pelee, South Bass (Put­ use of a grid overlay on radar photographs, in-Bay), and Catawba Islands, and land pre­ counting echo occurrences, and correlating cipitation on records from surrounding with measured precipitation. Current use of perimeterstations. The resultanf"annual ratio radar precipitation observations, although was 99 percent with monthly ratios that var­ promising, is not advanced sufficiently to pro­ ied from 89 percent in Februaryand October to vide usable, quantitative data. Poor perform­ 107 percent in May. ance is attributed mainly to weak radar In 1963 the U.S. Lake Survey, in cooperation equipment, which displays a decrease of echo with the U.S. Weather Bureau, modified their occurrences per volume of rainfall outward existing precipitation network in northern from the radar location, thus limiting the ef­ Lake Michigan by replacing the storage-type fective range (Bruce and Rodgers 108). More gages, which were read twice a year, with pre­ powerful radar equipment is required. Fur­ cipitation recorders producing hourly read­ ther developments in radar technology in ings. Similar gage networks were established combination with high-speed computers may in western Lake Erie in 1964 and eastern Lake provide an ideal answer to the overwater pre­ Ontario in 1969. Using recorded data from cipitation measurement problem. Lake Michigan islands and land stations lo­ cated some 160 km (100 miles) west ofthe lake (Figure 4-18), Upchurch 808 derived lake-land 4.8 Evaporation ratios which vary from 49 percent in June to 226 percent in February, with an annual aver­ age of 110 percent. 4.8.1 Determination of Evaporation The monthly precipitation ratios derived in the above studies are tabulated in Table 4-14. Evaporation from the lakes is the loss of Monthly ratios given by Hunt 395 and by water from the lake surface to the atmosphere Kresge et a1. 478 represent values from in the form of water vapor. Considering lake smoothed annual graphs, while those ofothers and land areas of the Great Lakes Basin, two­ are arithmetic averages for the number of thirds ofthe watersupplied by precipitation is Hydrometeorology 91 lost by evaporation. Evaporation losses from for any appreciable period of time are almost the water surface of the lakes, where the sup­ exclusively restricted to the perimeter land ply of moisture is continuous, are substan­ stations, which may be and often are not rep­ tially higherthan from theland and amountto resentative of open-lake conditions. Varia­ approximately three-quarters of the overwa­ tions in air stability that affect both wind and ter precipitation. Thus, it is readily apparent vapor pressure are essentially diurnal in that evaporation has a great effect on the character over land and seasonal over water. availability ofwater and on the heat budget of The required adjustments for perimeter data, the lakes, since evaporation is basically a cool­ or lake-land ratios for wind and humidity have ing process. been made in recent years and are being im­ There is no direct method to measure evap­ proved as more overwater data are collected. oration from large water bodies. Since the ac­ Thus, the mass transfer method ofcomputing tual evaporation losses are dependent directly evaporation from the Great Lakes holds great on meteorological factors, it is possible to de­ promise for the future. Its primary advantage velop methods that use hydrologic and is the elimination ofthe main objections ofthe meteorologic data and permit determination water budget method, namely, uncertainties of evaporation losses from the lakes with ac­ with respect to ground water and dependence ceptable accuracy. These methods include ofcomputed evaporationon large factors, such water budget, mass transfer, energy budget, as inflow and outflow from the lakes. evaporation-pan observations, atmospheric The energy budget method requires deter­ humidity budget, and momentum transfer. mination of the energy exchange between a The first four ofthese methods have been used body of water and the atmosphere, which in­ to compute evaporation from the Great Lakes. cludes such factors as the net solar and atmos­ The waterbudget method consists ofsolving pheric radiation, conduction of sensible heat the water budget or mass-balance equation, to the atmosphere, energy utilized by evap­ described at the end of this section, for the oration, net advective energy, and energy unknown evaporation component. All other storage within the body ofwater, disregarding major components ofthe water budget neces­ some minor heat sources or sinks (chemical, sary to compute evaporation frem the Great biological, exchange with bottom sediments). Lakes are either measured directly or can be The water loss is determined from the related estimated from related measurements. This is energy or heat loss by evaporation. Detailed the only direct method ofcomputing evapora­ discussion of various terms comprising the tion estimates and has been used in various energy budget and its practical application is studies to provide control for other methods, presented by Anderson.16 However, in the which require determination ofempirical con­ Great Lakes this method has been used in­ stants. Evaporation, as determined by the frequently because of the difficulty in obtain­ water budget method, is a residual of several ~ng data on energy components. large factors and includes the errors of these A convenient and inexpensive method ofob­ factors, which may affect the computed evap­ taining evaporation estimates, although fre­ oration values considerably. Care must be quently questioned on theoretical grounds, is exercised to reduce these errorsto a minimum that of evaporation-pan observations, which by using all available data and considering the utilize observed water losses from evapora­ effects ofthe lakes on some ofthem, such as on tion pans and experimentally determined overwater precipitation. pan-to-lake relationships. The ratios of pan The mass transfer method of computing evaporation to lake evaporation, or pan coeffi­ evaporation is a modified application of Dal­ cients, vary depending on pan characteristics. ton's law, where evaporation is considered to An extensive investigation of the relation­ be a function of the wind speed and the differ­ ships between pan and lake evaporation was ence between the vapor pressure of saturated reported by Kohler et a1. 462 air at the water surface and the vapor pres­ sure of the air above. A summary of the theoretical development of the mass transfer 4.8.2 Evaporation from Lakes method and a review of the equations de­ veloped by various investigators is given by To compute evaporation from the Great Anderson et a1. 17 The more promising ofthese Lakes various investigators have used one or equations were tested by Marciano and Har­ more ofthe four methods described above. The beck.512 The problem in applying this method water budget and masstransfer methods were to the Great Lakes is that climatological data used most frequently. There are only two 9£ Appendix.+ known studieswhich utilize theenergybudget mass transfer equation by introducing approach. The evaporation-pan observation monthly wind and humidity ratios for adjust­ method has also been used infrequently. Re­ ing data obtained from land stations. In previ­ sults obtained by various investigators often ous studies wind data from perimeterstations differ considerably, especially for shorter, were being adjusted to overwaterwinds based monthly periods. Some ofthis variation is nat­ on seasonal periods. Because of data limita­ ural, reflecting variability of evaporation be­ tions, most studies mentioned above employed tween the various periods of record involved, inconsistent periods of record to determine but some of it, especially in extreme cases, is various factors ofthe mass transfer equation. due to computation procedures and inac­ In some ofthem, the average evaporation val­ curacies ofbasic data. More recent determina­ ues are based on only a few years of record, tions use better basic data and should be more which is much too short to establish reliable accurate. long-term trends. The first mass transfer de­ Among the earliest methods used to deter­ termination of evaporation with consistent mine evaporation from the Great Lakes is that periods of record for all factors was made for of evaporation-pan observations. Henry841 Lake Ontario by Richards and Rodgers.654 concluded that the ratio of evaporation from This studywasextended to include other lakes floating pans to land pans was approximately borderingon Canada by Richards and Irbe.651 0.5 and used this ratio to estimate lake evap­ Evaporation studies by the energy budget oration. Hickman 355 described experiments in method conducted on the Great Lakes are lim­ which water temperature in the pan was ited to Lake Ontario (Bruce and Rodgers,lOS maintained at the lake surface water temper­ and Rodgers and Anderson675). The results of ature, and the pan-lake ratioor pan coefficient these studies are practically identical. The au­ was assumed to be 1.0. The latest estimates of thors concede that their estimates are high evaporation from the lakes by this method due to inaccuracies ofdata used, and they dis­ (U.S. Weather Bureau828) give pancoefficients cuss the possible errors. Continuing research ranging from approximately 0.75 in the south­ on interaction between the atmosphere and ern areas ofthe Basin to 0.80 in the northern lake surface should enable more direct evalua­ areas. The above studies give annual evapora­ tion of the energy exchange factors, reduce tion estimates, without. breakdown into sea­ their dependence on empirical relationships, sonal or monthly amounts. and improve the accuracy of evaporation es­ The water budget is one of the traditional timates determined by the energy budget methods of determining lake evaporation method. (Russell,692 Freeman,271 Pettis,60S Hunt,395 Evaporation from the Great Lakes varies Brunk,lll and Derecki 213.214). In some studies with latitude and depth. The warmer, lower (Hunt, Derecki) precipitation from perimeter latitudes provide greater evaporation oppor­ stations was adjusted by lake-land ratios to tunity, and the lake depths govern heat stor­ represent overwater conditions. Others used age capacity. The influence of lake depths is unadjusted perimeter precipitation. Later mainly seasonal. Deeper lakes warm and cool studies showed that Hunt's reduction was more slowly, retarding the seasonal low and probably too extreme, resulting in somewhat high evaporation losses. The depths of the lower evaporation than indicated by most of Great Lakes coincide with latitude; Lake Su­ the other recent studies for Lake Ontario. Be­ perior is deepest (410 m) and Lake Erie is the cause Lakes Michigan-Huron have a common shallowest (65 m), while the centrally located outlet (St. Clair River), water budget determi­ lakes have approximately similar inter­ nations for these lakes can be made only for mediate depths (230 to 280 m). Thus, evapora­ both lakes as a single unit. tion from the lakes increases from north to Probably the first mass transfer determina­ south, being lowest for Lake Superior and tion of Great Lakes evaporation is that given highest for Lake Erie. The centrally located by Freeman 271 who used a relatively simple lakes, Michigan, Huron, and Ontario, have in­ formula similartothose used in all subsequent termediate evaporation rates. Great Lakes mass transfer studies. Other The average evaporation values for individ­ evaporation studies, mostly of Lake Ontario, uallakes, Obtained in some ofthe better known which employed mass transfer methods are or more recent studies mentioned in the pre­ those by Horton and Grunsky,376 Hunt,395 ceding paragraphs, are listed in Table 4-15. Kohler,461 Snyder,751 Bruce and Rodgers,IOs The table also shows the methods used, the Richards,us Richards and Rodgers,654 and source, and the periods ofrecord, although, for Richards and Irbe.651 Richards648 modified the methods other than water budget and the last H 1Idrometeorolog71 98

TABLE 4-15 Comparison of Great Lakes Evaporation (Centimeters)

Jan. Feb. Mar. Apr. May June July Aug. Sept. Oct. 1Iov. Dec. Annual

1M! SU1'1II0Il tam IIUIIGET rr...-a (1926) 1921-25 8.9 7.9 6.6 5.8 0.3 0.3 1.5 -0.3 3.3 5.3 7.9 9.7 57.2 Derseti (1965) 1939-59 8.9 6.6 5.8 -0.8 -1.0 -1.0 -0.5 2.8 5.8 6.9 9.7 9.7 52.8 IMSS TIAIl5Fn rr_ (1926) 1900-24 9.1 8.4 6.9 2.5 0.0 0.0 0.0 3.0 4.3 7.1 6.9 8.1 56.4 Sny....r (1960) 1921-50 7.6 6.4 4.8 1.8 -1.8 -5.8 -7.4 -0.5 5.6 6.4 6.9 8.4 32.3 IichArds , Irbe (1969) 1959-68 11.7 9.1 5.8 1.5 1.0 -3.6 -8.1 -4.1 3.3 6.4 10.4 12.2 45.7

~ MICHIGAN IWlS TRANSFER Pr_n (1926) 1900-24 7.6 7.1 4.1 2.3 1.5 1.8 7.9 9.9 8.6 6.6 6.6 73.2 Snyder (1960) 1921-50 7.6 6.6 4.1 1.0 -2.3 -3.0 3.0 7.9 8.6 6.9 7.6 57.9 ~ MICHIGAN-BURON WATIlll lIIlIlCET rreeaan (1926) 1915-24 8.4 2.8 4.3 3.6 -0.3 -0.3 0.5 4.8 10.4 10.2 10.7 64.8

~ lIUR01' IWlS TRANSFER Pre_ (1926) 1900-24 7.9 7.9 5.1 2.0 2.0 2.5 7.9 10.2 7.1 6.1 7.1 74.9 Snyder (1960) 1921-50 7.9 6.9 4.3 1.3 -1.5 -2.8 3.3 8.3 8.1 6.9 8.4 61.2 Richards , Irbe (1969) 1959-68 11.9 8.9 4.6 -0.8 0.0 -2.0 0.5 5.1 11.4 11.7 11.7 11.6 ~!!!! WATIlll BllDGET Preesan (1926) 1951-24 4.6 2.8 12.7 -3.8 0.0 0.8 4.8 H.4 13.7 15.0 12.2 9.1 83.3 Derecki (1964) 1937-59 5.3 1.5 1.5 0.0 1.0 2.8 9.4 13.7 16.0 13.7 12.2 7.1 84.3 MASS TRANSFER Pre...n (1926) 1900-24 6.6 2.~ 3.6 1.9 17.0 18.3 16.0 13.5 7.1 6.1 106.4 Snyder (1960) 1921-50 5.1 2.5 1.3 2.0 9.4 12.2 12.7 10.9 9.7 8.9 85.9 Richards , Irbe (1969) 1950-68 4.1 1.3 -3.0 5.8 7.1 11.2 14.2 15.7 13.7 7.6 90.9 ~~ WATEll. BtllGI:T Hunt (1959a) 1934-53 6.1 3.0 1.8 1.0 0.5 -0.5 2.8 7.9 9.7 11.4 10.4 9.7 63.8 Morton , ~berl (1959) 1934-52 8.1 5.8 2.0 2.0 1.3 0.5 4.6 10.2 10.7 11.7 11.4 11.2 79.5 MASS TRANSFEll. Pre_ (1926) 1900-24 6.6 6.6 4.1 2.5 0.8 2.8 9.4 13.0 11.7 10.9 6.1 6.1 80.5 Hunt (1959a) 1937-52 7.1 5.6 5.6 2.0 0.0 0.3 4.8 6.4 9.7 8.1 7.1 6.1 62.7 Snyder (1960) 1921-50 7.9 6.9 4.3 1.5 -1.5 0.3 5.8 10.2 10.7 9.4 6.6 7.6 69.6 Richards' lrbe (1969) 1950-68 9.7 7.4 4.3 -1.8 1.0 0.3 4.8 8.1 10.9 9.1 8.9 8.9 71.6 ENERGY BllDGET Rogers' Anderson (1'61) 1958-60 12.4 8.9 5.1 0.5 -1.8 0.0 11.7 8.9 10.2 9.4 10.9 10.7 86.9

NOTE: EnerBY budget and _liS transfer IItudiell (except IUcbards , lrbe. 1969) uad vadable periods of record for various factors involved. mass transfer study (Richards and Irbe,651) from the Great Lakes amounts approximately these periods are only approximations. Re­ to 65 cm (25 cm (25 in.) The following are esti­ sults of most studies show reasonable agree­ matesofannualevaporation for theindividual ment between different computations. lakes: Superior, 55 cm (21 in.); Michigan­ The average long-term seasonal variation of Huron, 65 cm (26 in.); Erie, 85 cm (38 in.); and evaporation from the Great Lakes is shown in Ontario, 70 cm (28 in.). These estimates were Figure 4-99, presented as smoothed graphs based on several studies and should be consid­ based on studies indicated in Table 4-15. ered as indicators of the long-term average Lakes Michigan and Huron are included in a value. During individual years annual evap­ single graph, because ofcommonwaterbudget oration may vary considerably from the long­ computations and limited determinations by term values listed above. other methods. The available information in­ Seasonally the low evaporation normally oc­ dicates that the evaporation from these lakes curs in the spring, when the water tempera­ is similar and compares with that of Lake On­ ture is close to or even below the dew point tario. In view of the latitude and depth dis­ temperature ofthe air. These low evaporation tributions ofthese three lakes, this similarity rates vary from slight evaporation to conden­ appears to be quite reasonable. sation. With rising water temperatures evap­ The long-term average annual evaporation oration increases until it reaches the high 94 Appendix"

ground-water zone. Although interflow

II> reaches the streams with more delay than di­ ANNu"t f:VAPORATION rect surface runoff, the two are difficult to dis­ /- • sUPUIOlil-S~c:fftl'''" l tinguish, and together are referred to as sur­ I. MltH - HURON -65c:ftl,(26,n 1 I\ face runoff. The infiltration capacity of soils ERIE - esc,," (331ft) ONtARIO -1Oc",(28lftl / \\ varies widely depending upon precipitation '2 / \ factors such as intensity, duration, and type of .Il~\ precipitation, and soil factors such as antece­ 10 "'/ ...... ', dentsoil moisture, type ofsoil, shape and slope of the basin, and vegetation. During winter ", /~/ >\;. '" months, the ground is usually frozen, and / 1'1 " 3 i: " \~ ~ interflow is reduced to the periods of ground "\ lotI' / I \ thaws, but surface runoff still occurs during ,\. \/ Id I\ intermittent snowmelts. Basin hydrology is \ \\ \/ 'i! I considered in Appendix 2, Surface Water Hy­ \ ~ \ "J: I \ .~ \ I II I drology. " J<' Water which infiltrates into the soil is \ \i I / k transmitted downward to the ground-water ,,\')/ l' If EVAPOIATION table and becomes a part ofthe ground-water 0 0 CONOENSATiON ~ I reservoir. Ground-water flow is discussed ..... _".,,' later in this section. -2 . . -I The storage of water on the drainage basin JAN MAR MAY JUt SIP NOV JAN in whatever form, be it ground water, channel MONTHS storage, or snowpack, is a fundamental hydro­ FIGURE 4-99 Evaporation from the Great logic factor of runoff delay. Natural regula­ Lakes. The smoothed Jines are based on several tion by upland lakes and artificial regulation studies from 1926-1969. by man-made reservoirs and control struc­ tures are also common in the basins of all the evaporation season, which for most lakes is in Great Lakes. Lake regulation delays, pro­ the fall, when the water temperature is con­ longs, and diminishes the peak runoff by stor­ siderably above the dew point temperature of ing water during periods of high runoff. the air. Because of its great depth and tre­ Runoff is measured at gaging stations on mendous heat storage capacity, highest evap­ many of the tributary streams and together oration from Lake Superior occurs in the late with measured flows in connecting rivers is fall and early winter period. The long-term av­ the most accurate data in the hydrologic cycle. erage values of the highest monthly evapora­ Unlike measurements of other components, tion vary from approximately 9 cm (3.5 in.) for which sample only points within an area, Lake Superior to 15 cm (6.0 in.) for Lake Erie. gaged runoff effectively integrates the entire These high evaporation rates, coupled with area about the point of measurement. How­ low air temperatures, cause rapid dissipation ever, runoff data contain some uncertainties, of heat from the water surface. With the but these are considered to be random. Errors sharply falling water temperatures, evapora­ may be introduced in the measurement of tion begins to decrease. runoff, particularlyduring winter months due to ice effects, and by extrapolating the gaged runoff to the nearby ungaged areas to obtain 4.9 Runoff from Drainage Basin coverage for the entire drainage basin. Routine computations of the total average surface runofftothe lakesare not made by any 4.9.1 Surface Runoff agency associated with the Great Lakes, but runoffestimates have been made in the course Water enters the lakes from the drainage ofhydrologic studies. The values ofrunoffpre­ basin mainly through the tributary streams. sented in this report were obtained by direct When rainfall exceeds the infiltration capac­ areal extrapolation of the available gaged ity of the soil the excess water reaches the streamflow records to the nearby ungaged tributary streams either as direct surface areas, and summation of the total runoff runoff or as interflow. Interflow is defined as amounts from individual basins of the tribu­ that water which percolates through the soil tary streams within each lake basin, which in above the phreatic or permanently saturated tumwerecombinedtoobtain totalrunofffrom Hydrometeorology 95

TABLE 4-16 Runoff·Precipitation Ratios, TABLE 4-17 Average Runoff into the Great 1937-1969 Lakes, 1937-1969 Lake Runoff in cm depth on lake surface Period Superior Michigan Huron Erie Ontario 1 2 Period Superior Michigan Huron Erie Ontario January .49 .47 .43 .55 .55 .50 .66 .411 .68 .55 February January 3.4 .61 .62 .75 .1l6 .95 4.3 5.5 8.2 13.2 March February 2.9 5.2 1.01 .61 .93 .63 1.13 5.1 8.9 12.5 April March 3.8 6.4 8.4 .85 .42 .67 .35 .60 14.1 22.8 May April 8.6 .46 .26 .37 .19 .34 9.2 12.7 12.8 30.3 June May 9.8 7.2 10.2 7.2 17.7 July .36 .24 .27 .12 .20 June .18 .08 6.5 5.2 6.1 4.2 9.2 August .26 .18 .15 July 4.5 4.2 .18 .16 .08 .16 3.9 2.4 5.7 September .28 August 3.6 3.0 .40 .28 .28 .13 .25 3.1 1.5 4.4 October September 3.6 3.2 3.0 1.3 4.6 November .44 .32 .33 .20 .34 October 3.7 3.7 4.1 2.0 6.5 December .52 .44 .43 .40 .48 November 4.2 4.1 5.3 3.3 9.8 December 3.7 4.2 6.0 5.9 12.8 Annual .52 .39 .44 .35 .48 Annual 58.3 59.6 73.7 71.8 149.5

1 Inc1u3es diversions into Lake Superior (about 5cm/yr - the entire drainage area of the Great Lakes. 2140 m Is). See DIVERSIONS. Similar methods were employed in previous Excluding drainage area of Lake St. Clair. studies (Freeman,271 Horton and Grunsky,376 Pettis,60li Hunt39li). Results obtained by vari­ piration. Similarly, the Lake Superior basin, ous investigators generally differ somewhat with the lowest precipitation, has a high runoff because of different periods of record and because of the reduced water losses in the varying coverage ofthe gaged tributary area, north. High runoff in the spring accounts for which increases steadily. Peakrunoffto all the 30 to 40 percent of the annual amount. lakes usually occurs in the early spring as a The importance of surface runoff to the hy­ result of melting snow augmented by rainfall drology ofthe Great Lakesdependsnot onlyon and lack ofgrowing vegetation. The low point the absolute magnitude offlow but also on the usually occurs in the late summer. relative magnitude of runoff with respect to To facilitate comparison with other compo­ surface area of each lake. Runoff represents nents of the hydrologic cycle, such as precipi­ one-third to one-halfofthe overland precipita­ tation or evaporation, runoff values are fre­ tion and is almost equal to overwater precipi­ quently given in units ofdepth overrespective tation on all the lakes except Ontario, where areas, and this practice is retained in this re­ runoff is almost twice as high. The average port. Runoff distribution on the drainage runoff values for monthly and annual periods, basin is shown in Figure 4-19. Comparison of expressed in centimeter depth on the lake sur­ runoff with the corresponding overland pre­ face, are listed in Table 4-17. Average annual cipitation for both monthly and annual runoffduringthe 33-year period (1937-69) var­ periods, expressed as runoff-precipitation ied from 58 cm (23 in.) for Lake Superior to 150 ratios, is shown in Table 4-16. These ra!ios cm (59 in.) for Lake Ontario, 60 cm for Lake indicate that average annual runoff dUring Michigan, 74 cm for Lake Huron and 72 cm for the period from 1937 to 1969 represents from Lake Erie. The annual value for Lake Supe­ 35 to 49 percent ofthe overland precipitation, rior includes approximately 5 cm (2 in.) from while average monthly runoffs vary from a the Ogoki and Long Lake Diversions, which high of 113 percent in the spring to a low of 8 are channeled and measured in the tributary percent during the summer. The substantial streams, while Lake Erie runoff excludes differences in the retention ofprecipitation on streamflow tributary to Lake St. Clair, since it the basins of the individual lakes are due is measured in the Detroit River and thus be­ primarily to higher evapotranspiration comes part ofthe inflow to that lake. The aver­ losses in the south, but other factors such as age monthly runoffvaried from a highof30 cm infiltration capacity of soils, forest cover, land (12 in.) during April on Lake Ontarioto a low of cultivation, and urbanization, also affect the 1.3 cm (0.5 in.) during Septemberon Lake Erie. relationshipbetween runoffand precipitation. Although precipitation on Lake Erie is among the highest and is comparable with that of 4.9.2 Underground Flow Lake Ontario, runoff into Lake Erie is low be­ cause ofthe high water losses by evapotrans- Underground flow from ground water 96 AppeMiz.4 reaches the lakes by percolation. either di­ reaches the lakes mainly through ground rectly through the lake bottom or through water and not through surface streams. They tributary streams. Since most streams in the also state that there are numerous instances Great Lakes Basin derive their base flow from ofwatershed leakage from one drainage basin ground water. the underground flow reaching to another and probably from the higher-lying the lakes through tributary streams is mea­ drainage basins directly into the lakes. Based sured and considered with the surface runoff. on these considerations, they concluded that Thus, direct contribution from ground water the summerrunoffto the lakesestimated from to the lakes is of primary interest to the lake measured streamflow may be less than the hydrology. Depending on the geohydrologic actual runoffby an amountexceeding 13 cm (6 characteristics of the system and the water in.) on the lakes, but in their study they as­ levels in lakes, the underground flow could be sumed the underground flow to be zero. Hunt either into or out of the lakes. acknowledges a possibility of a considerable There is very little information on the un­ underground flow between Lakes Erie and derground flow in the Great Lakes Basin. par­ Ontario. due to the very large potential head ticularly on direct ground-water supplies, and between them. He investigated this matter knowledge in this field should be expanded. and concluded that there is no appreciable The evaluation ofdirect underground flow re­ ground-waterflow into Lake Ontario. and that quires ground-water profiles. which can be de­ what flow might exist is probably steady rived from a network of observation wells lo­ throughout the year. cated around the lakes. There is only a handful In some hydrologic studies underground of wells located within 15 km (10 mi.) of the flow was estimated by various forms ofwater lakes, and wells located further inland may be budget computations. Russe1l6l12 estimated poor indicators ofground-water flow. In addi­ monthly amounts ofwater entering the lakes tion to lake perimeters. special attention from underground sources, which resulted in should also be concentrated on any areas the annual values 65 em (26 in.) on Lake Supe­ where, because of geologic structure, rior, 83 cm (33 in.) on Lakes Michigan-Huron, ground-water divides may be significantly dif­ 51 cm (20 in.) on Lake Erie, and 130 cm (61 in.) ferent from topographic divides. on Lake Ontario. However, Russell did not dis­ There are many geological studies of tinguish between ground-water flow into ground-water conditions for specific areas of streams and ground-water flow directly into the Basin. However, these studies cover rela­ lakes, so values listed above include base flow tively small areas (one section ofa lake basin) of streams. Pettis605 estimated direct un­ and are not related directly to the un­ derground flow into the upper lakes as 42 em derground flow into or out of the lakes. They (16 in.) to Lake Superior and 68 cm (23 in.) to are approached from the geological point of Lakes Michigan-Huron. Pettis also states that view and consider areal geology, water use. a considerable part of a large underground economics of water development. and the water body in the northern part ofthe State of usual ground-water data obtained from ob­ Ohio has a definite motion towards Lake Erie. servation wells, pumping tests. and chemical In contrast to the high underground flow analysis. Thus. they reveal only limited infor­ claimed by the above investigators, Berg­ mation, which is not readily adaptable to large strom and Hanson62 computed inflow to Lakes areas encompassing one or more lake basins. Michigan-Huron from the ground-water Large area hydrologic studies dealing with sources to be approximately 0.5 cm (0.2 in.). ground water are more important for the un­ They suggest that the actual discharge along derground flow aspect. Because ofthevarying the shore could be several times that given geologic and climatic conditions over such above, butthe amount would still be relatively large areas, estimation of the underground small. and probably less than the error inher­ flow into anyone lake is very difficult, and ent in extrapolation of runoff, lake evapora­ ground-water data are not sufficient to de­ tion, and precipitation. Snyder7S1 indicates termine this flow by direct methods. that there are underground outflows from In most hydrologic studies underground Lake Superior and Lakes Michigan-Huron flow was considered negligible and assumed to that amount to approximately22 cm (9 in.) and be zero (Freeman,271 Horton and GrunskY,376 5 em (2 in.) on these lakes. respectively. He also Hunt.395 Morton and Rosenberg,561 Der­ suggests that there are underground inflows ecki 214). Horton and Grunskystate thatthere is to Lakes Erie and Ontario ofapproximately 19 a marginal belt around the perimeters of em (8 in.) and 8 cm (3 in.), respectively. Snyder Lakes Michigan and Huron. from which water bases his indicated groundwater flow on the Hydrometeorology 97

differences in evaporation estimates com­ extremely important in studies of the hy­ puted by mass transfer and water budget drologic cycle. methods. Widely divergent opinions on the amount and direction ofunderground flow to the Great 4.10.2 Outflow Lakes leave the subject in a state of con­ troversy. For example, Pettis605 claims the The outflow ofeach ofthe Great Lakes is the underground inflow to Lake Superior is 42 cm quantity of water flowing from a given lake on lake surface per year and Snyder 751 indi­ through its natural outflow river and through cates an outflow from the same lake ofapprox­ man-made outlets. As a factor in the hy­ imately 22 cm.1tis apparent from the cautious drologic cycle of the lakes the outflow of any wording and the conflicts that exist in the pre­ lake, including that through man-made diver­ ceding studies that the estimates are little sions, represents the water yield ofthe entire more than conjectures. basin above the point of outflow measure­ ment. The outflow from Lake Superiorthroughthe St. Marys River has been artificially con­ 4.10 Lake Inflow and Outflow trolled since 1922 by a gated dam structure, which allows diversions for the generation of power. Release of water through the control 4.10.1 Inflow structure and for the generation of power is made in accordance with a regulation plan, The inflow to any of the Great Lakes is the designed to maintain the level of Lake Supe­ quantity of water supplied by the lake above, rior within specified limits. modified by the local inflow to the connecting The outflow from Lakes Michigan-Huron river. Local inflow is usually less thanone-half through the St. Clair and Detroit Rivers, percent of the total flow and generally is dis­ which together with Lake St. Clair constitute regarded in computing lake outflows and the the natural outlet of these lakes, is controlled corresponding inflows to the lakes below. largely by the level of Lake Huron at the head However, the Lake Huron outflow and Lake ofthe St. Clair River and the levelofLake Erie Erieinflow normally differ by approximately 2 at the mouth of the Detroit River. No man­ percent and occasionally the difference is made control ofthis flow exists, except for the much higher. Therefore, flow atthe head ofSt. fixed remedial control provided by the dikes Clair River is used to determine outflow from constructed in the lower Detroit River to com­ Lakes Michigan-Huron, while the flow of the pensate for the effects of deepening the navi­ Detroit River is used to obtain inflow to Lake gation channels in that river. These compen­ Erie. The importance of inflow to the lakes sating controls do not regulate lake levels. increases progressively in descending order However, they are designed to provide the through the lakes. The smaller lower lakes are same net discharge capacity in the rivers as affected by these flows to a much greater ex­ existed before the improvements, so that the tent than the upper lakes. Inflow to Lakes levels upstream are maintained. The naviga­ Michigan-Huron is of the same order of tion improvements in the St. Clair River, in­ magnitude as the overwater precipitation, but cluding some works recently completed, cause in Lakes Erie and Ontario, it is an order of a lowering of the levels of Lakes Michigan­ magnitude greater. During the period of hy­ Huron, and remedial works are being evalu­ drologic study, 1937-69, the average annual in­ ated. Inthe winter periodice normally slightly flow was equivalent to 60 cm (24 in.) for Lakes reduces the open water flow of the St. Clair Michigan-Huron, 640 cm (250 in.) for Lake and Detroit Rivers. Erie, and 927 cm (365 in.) for Lake Ontario. The The outflow from Lake Erie through the variation in the annual inflow during this time Niagara River is controlled largely by the had a range of approximately 40 cm (15 in.) for level of Lake Erie at the head ofthe river and Lakes Michigan-Huron, and approximately the level of the Chippawa-Grass Island Pool 240 cm (95 in.) for Lakes Erie and Ontario. above Niagara Falls. Ifthe diversions ofwater Thus, in the lower lakes the variation in the from the Chippawa-Grass Island Pool for the annual inflow is three times greater than the generation of power are not compensated for, average annual overwater precipitation. Be­ they can lower the levels of Lake Erie. How­ cause of the magnitude and fluctuation ofthe ever, a submerged weir was constructed dur­ inflows to the lower lakes, accuracy ofinflow is ing the period 1942-47 at the downstream end 98 Appendix ..

TABLE 4-18 Average Flows in Connecting The range in the average monthly flows was Rivers of the Great Lakes, 1937-1969 (m3/s) approximately 700 m 3/S (25,000 ds) for the St. Marys and Niagara Rivers, and 1,300 m 3/s St. St. St. (45,000 cfs) for the other rivers. The difference Period Marys Clair Detroit Niagara Lawrence between the high and low annual flows during January 1,980 4,420 4,620 5,240 6,:!30 the 33-year period varied from approximately February 1,980 4,250 4,420 5,240 6,230 1,400 m 3/S (50,000 cfs) on the St. Marys River to !'larch 1,940 4,810 4,960 5,320 6,400 April 2,020 5,130 5,270 5,640 6,830 2,000 m 3/S (70,000 cfs) on the Detroit River. The !'lay 2,180 5,240 5,350 5,920 7,020 relatively large variations in flow of the St. June 2,310 5,350 5,410 5,950 7,190 July 2,450 5,410 5,490 5,830 7,140 Clair and Detroit Rivers, in comparison with August 2,570 5,440 5,490 5,720 7,000 other rivers, are due primarily to ice retarda­ September 2,550 5,380 5,440 5,580 6,770 tion of winter flows. October 2,510 5,320 5,380 5,470 6,570 November 2,430 5,270 5,300 5,470 6,460 The importance ofoutflow to lake hydrology December 2,110 5,130 5,240 5,470 6,460 increases progressively with downstream lakes. The average annual outflows (river Annual 2,250 5,100 5,200 5,570 6,690 flows and diversions), expressed in units depth on the lake surface, represent 86 cm (34 in.) on ofthe pool, as the initial phase ofthe remedial Lake Superior, 139 cm (55 in.) on Lakes works designed in part to counteract this ef­ Michigan-Huron, 706 cm (278 in.) on Lake Erie fect. Presently, a gated control structure ex­ and 1,077 cm (424 in.) on Lake Ontario. Varia­ tending partially into the river from the tion in the annual outflow duringthe period of Canadian shore provides compensation for the study had a range of approximately 50 cm (20 power diversion, which was increased by the in.) for Lakes Superior and Michigan-Huron, 1950 treaty between the United States and 190 cm (75 in.) for Lake Erie, and 300 cm (120 Canada. This second control structure is lo­ in.) for Lake Ontario. cated downstream and runs parallel to the Numerous studies of the connecting river weir. flows have been made. These studies have The outflow from Lake Ontario through the analyzed the effects on flow equations of reg­ St. Lawrence River since 1958 has been imen changes, formation ofice, weed retarda­ largely controlled by the release of water tion, and water temperatures. Detailed dis­ through the Iroquois Dam near Iroquois, On­ cussion of outflows is given in Appendix 11, tario. Beginning in April 1960 the release of Levels and Flows. Flow equations and the water has been made in accordance with a means ofrevisingthem whenchannel changes regulation plan, which provides for weekly occur, and turbine ratings used to compute flow changes throughout the year. Thus, Lake flows through power structures are generally Ontario levels are fully regulated and are in­ well established. As a result, data on the flows dependent of any channel changes or diver­ in the connectingrivers ofthe Great Lakes are sions. Prior to the construction ofthe Iroquois considered to have a greater accuracy than for Dam in 1958, the Galop Rapids, a short dis­ any other hydrologic factor, except the change tance downstream from Ogdensburg, New in lake storage. York, constituted a natural weir, the flow over which was controlled substantially by the level of Lake Ontario. 4.10.3 Diversions The average monthly and annual flows of the outflow rivers for the 1937-69 period are Water diversions in the Great Lakes Basin listed in Table 4-18. Prior to 1957 the flows of may be broadly divided into two types: outside the St. Clair River were based on published rec­ diversions, which take water into or out ofthe ords of combined St. Clair-Detroit River system; and inside diversions, which retain flows, since the flows in both rivers were con­ water entirely within the system. Diversions sidered to be essentially equal. During the of water into the Basin have the effect ofrais­ period of study, annual flow from the St. ing water levels of the lake into which the di­ Marys, St. Clair, Detroit, Niagara, and St. verted water is discharged and the levels of Lawrence Rivers averaged 2,250 m 3/S, 5,100 the lakes downstream through which the di­ m 3/S, 5,210 m 3/S, 5,580 m 3/S, and 6,680 m 3/S (79, verted water must pass on its way to the sea. 180, 184, 197, and 236 thousand cfs), respec­ Diversions of water from the Basin have the tively. Low flows occur in the winter and high converse effect on the levelsofthe lakes at and flows in the summer, with a progressive delay downstream from the point of diversion. Di­ of the summer highs in the upstream rivers. versions from one point to another within the r Hydrometeorology 99

Basin may have no effect on the lake levels if outside diversions, the inputs to Lake Supe­ within the same lake basin; or, if not compen­ rior and the output from Lake Michigan, is to sated for, diversions may lower the levels of increase the suppliesto Lake Michigan-Huron the lake upstream and temporarily raise the and the downstream lakes by approximately levels downstream. The temporary rise in 54 m 3/S (1,900 cfs). levels downstream is due to increased dis­ The diversion ofwater through the WeIland charge rates while the levels ofthe lake above Canal, from Lake Erie at Port Colborne to drop to adjust to the larger outletcapacity due Lake Ontario at Port Weller, includes water to the diversion. The rate of adjustment in used in the DeCew Falls power plant, which lake levels decreases exponentially with time, amounts to most of this diversion, plus diver­ and the period of adjustment depends on the sions for navigation purposes. The total WeI­ size ofthe lake involved and the capacity ofits land Canal diversion amounts to approxi­ outlet. For example, Lakes Michigan-Huron mately 198 m 3/s (7,000 cfs), which is equivalent reach 90 percent adjustmentin approximately to approximately 25 cm (10 in.) on the Lake seven years and Lake Erie in less than one Erie surface per year. year (Bajorunas 3lia). The New York State BargeCanalwithdraws There are five major diversions in the Great water from the Niagara River at Tonawanda, Lakes Basin (Figure 4-1): the Ogoki and Long New York, for navigation purposes, but the Lake Projects divert water into Lake Supe­ water diverted into the canal is returned to rior; the Chicago Sanitary and Ship Canal di­ Lake Ontario at Oswego, New York. The verts water out of Lake Michigan; and the Barge Canal diverts approximately 31 m 3/s Welland Canal and the New York State Barge (1,100 cfs) during the navigation season. Since Canal divert water from Lake Erie and Niag­ 1956 there has been no diversion during win­ ara River into Lake Ontario. All other diver­ ter months. sions presently in existence on the St. Marys, The average monthly and annual flows dur­ Niagara, and St. Lawrence Rivers divert ingthe period ofstudy, 1937-69, for the major water from one point to another within the diversions described above are listed in Table same river and with remedial structures have 4-19. Annual values listed in the table are no effect on lake water supplies and lake somewhat different from the normal values levels. presented in the discussion becauseofperiodic The Ogoki and Long Lake Projects divert variations from the normal. water into Lake Superior from the Albany River drainage basin in the Hudson Bay wa­ tershed. The Ogoki diversion diverts water 4.11 Lake Level Fluctuations from the Ogoki River into the Nipigon River. The Long Lake diversion diverts water from Long Lake at the head of the Kenogami River 4.11.1 Lake Levels to the Aguasabon River. These diversions have increased the supply of water to Lake The elevations of the water surface of the Superior by an average rate of approximately Great Lakes are tied to the mean sea level at 142 m 3/s (5,000 cfs), which is equivalent to ap­ Father Point, Quebec, on the Gulf of St. Law­ proximately 5 cm (2 in.) on the lake surface per rence. This plane of reference, established year. especially for the Great Lakes in 1955, is called The Chicago Sanitary and Ship Canal, along the International Great Lakes Datum. The with the Calumet Sag Canal, a branch which average monthly and annual lake levels for connects with Lake Michigan south of Chi­ the 33-year period, 1937-69, are given in Table cago, diverts water from Lake Michigan 4-20. Theselevels are based on mean lake level through the Des Plaines and Illinois Rivers to tabulations published by the Lake Survey, the Mississippi River. This diversion, com­ and represent records from master gages, monly referred to as the Chicago diversion. each lake having a single master gage located represents the amount ofwater diverted from at a strategic point. Approximate water sur­ Lake Michigan for navigation purposes and face elevations ofthe lakes are 183 m (601 ft.) for domestic use by the City of Chicago. The for Lake Superior; 176 m (578 ft.) for Lakes total diversion from the lake at Chicago Michigan-Huron; 174 m (570 ft.) for Lake Erie; amounts to 88 m 3/s (3,100 cfs), which repre­ and 75 m (245 ft.) for Lake Ontario. sents an annual amount ofwater exceeding 2 Of primary interest in lake hydrology are cm (about 1 in.) on the surface of Lakes the variations of lake levels caused by the Michigan-Huron. The net effect of all three changing volume of water in the lakes. These 100 Appeftdiz 4.

TABLE 4-19 M~or DiveraioRB in the Great Lakes Buin, 1937-1969 (tubic Meter. per Second)

Ogoki Long Lake Chicago WeIland N. Y. State Project Project Diversion Canal Barge Canal into into out of from L. Erie from Niagara R. L. Superior L. Super~or L. Michi~an to L. Ontario to L. Ontario Period 1943-691 1939-69 1937-69 1937-694 1937-695

January 94 35 90 160 9 February 75 35 88 162 9 March 65 29 85 162 4 April 67 28 95 170 22 May 148 54 99 177 31 June 216 65 107 178 31 July 152 48 110 172 31 August 121 39 114 181 31 September 111 34 105 180 31 October 105 33 91 182 31 November 120 36 85 181 31 December 111 35 98 170 16 Annual Average 115 39 97 173 23

1period of record starts in July 1943. Since 1945 total amount of Ogoki and Long 2Lake diversions has averaged 142 m3/s (5,000 cfs). 3Period of record starts in July 1939. 3 Since 1938 total diversion (navigation plus domestic pumpage) has averaged 88 m Is (3,100 cfs). However, higher flows were authorized on two occasions by the U. S. 4Supreme Court. 3 Since 1950 total diversion (navigation and hydropower) has averaged 198 m /s 5(7,000 cfs). 3 Since 1929 during navigation season this diversion has amounted to 31 m /s (1,100 cfs). Since 1956 there has been no diversion during winter months. volumetric changes are generally referred to TABLE 4-20 Average Levels of the Great as lake level fluctuations and apply to the en­ Lakes, IGLD (1955). 1937-1969 (Meters) tire lake. They involve time periods of suffi­ cient duration to allow absorption of any local Superior lUclligan-lluron Erie Ontario at at at at short-period variations, so that entire water Period Marquette Barbor Beach Cleveland 08mo surface can be assumed to be level. The local Ja1lWlry 183.00 176.01 173.66 74.40 short-period variations, classified as water February 182.93 176.01 173.68 74.42 level disturbances, do not involve volumetric March 182.89 176.02 173.76 74.50 April 182.92 176.09 173.92 74.71 changes but displacement of water level Hay 183.04 176.19 174.03 74.85 caused primarily by winds and variations in June 183.13 172.26 174.07 74.92 July 183.20 176.32 174.06 74.88 barometric pressure. Water level disturb­ August 183.23 176.30 174.00 74.77 ances are discussedin Section6, while detailed Septl!lllber 183.23 176.25 173.90 74.64 discussion of lake levels is given in Appendix October 183.20 176.19 173.79 74.51 Novlllllber 183.15 176.13 173.70 74.44 11, Leve18 and Flow,. Ileclllllber 183.08 176.08 173.68 74.42 Annual 183.08 176.15 The water level fluctuations represent stor­ 173.85 74.62 age ordepletionofwaterin the lakes. Seasonal fluctuations undergo a relatively regular cy­ 4.11.2 Change in Storage cle; high levels usually occur in the summer and low in the winter. The change in storage on the lakes for any , HJldromeuorologJl 101 TABLE 4-21 Average Chap il. Storage on reduce possible effects of short-term water the Great Lakes, 1937-1"9 (Centimeters) level disturbances (Quinn,lalla Quinn and' Todd-b). Michtaau- The coordinated average change in storage 1 3 4 p~riod Superior Huron 2 Erie Ontario for monthly and annual periods on each lake during the 1937-69 period is shown in Table January -6.7 -2.1 0.6 0.9 February -5.2 0.0 2.4 3.4 4-21. Since the change in lake storage is Karch -2.1 4.0 14.0 15.2 primarily a seasonal phenomenon, the long­ April 9.1 11.3 15.5 21.3 term annual values should be small due to KaY 11. 3 8.2 6.4 11.0 balancing of rising and falling lake levels, as June 8.8 6.7 1.8 0.3 July 4.0 0.9 -4.3 -7.6 indicated in the table. The average seasonal August 1.8 -3.4 -8.5 -12.8 change in storage varies with latitude. The September -1.8 -5.8 -10.7 -13.4 lower lakes have rising lake levels duringwin­ October -4.9 -6.7 -9.4 -10.7 ter and spring, and falling lake levels during November -5.8 -4.6 -4.9 -4.3 December -8.5 -5.8 0.0 -1.8 summer and fall. This distribution is delayed by approximately one month on Lakes Annual 0.0 2.7 2.9 1.5 Michigan-Huron and by a full season (3 months) on Lake Superior. The highest aver­ Note: Change in storage determined from 10-day age monthly rise was approximately 11 cm on means (5 at end and 5 at beginning of fol­ Lakes Superior and Michigan-Huron, 16 cm on lowing month) by averaging records from the Lake Erie, and 21 cm on Lake Ontario; the 1 following gages: highest average monthly decline was approx­ Thunder Bay, Duluth, Hichipicoten, Marquette. imately 8 cm for Lake Superior, 7 cm for Lakes 2and Pt. Iroquois. Milwaukee. Ludington. Mackinaw City. Harbor Beach. Michigan-Huron, 11 cm for Lake Erie, and 13 3Thessalon, and Godericb. cm for Lake Ontario. During individual years 4Cleveland and Port Stanley. the variationin annual and monthlychange in Oswego, Kinaston. Cobourg. Toronto, Port Weller, and Rochester. lake storage may be considerable. In extreme cases this range may exceed several times the highest average monthly change in storage on given period is determined from the change in each lake. lake levels. Mean lake levels for several days The change in storage discussed above in­ are used for the determination of beginning­ cludes volumetric changes, which are affected of-period levels. This minimizes the effect of by water density variations. A mass of water external forces such as winds or barometric expands or contracts as it is heated or cooled. pressure. The mean level ofa lake at any given The amount of expansion or contraction de­ time is determined by averaging recorded pendson the change in temperature and depth levels of several gages. situated at points to which this change becomes effective (ther­ around the lakes in a pattern selected to pro­ mocline depth). Investigations of the thermal vide good approximation of the whole lake expansion ofwaterin the Great Lakes indicate level. that thermal expansion is insignificant and In recent years, the gage patterns used for may be disregarded (Hunt,3lI5 Derecki214). determination oflake storage have been coor­ dinated by the Lake Survey and Canadian 4.12 Heat Budget agencies to provide consistent values in both countries. These gage patterns consist of five The interaction of the various climatic and gages for Lake Superior, six gages for Lakes hydrologic elements results in heating and Michigan-Huron and Ontario, and two gages cooling processes within the lakes. Some proc­ for Lake Erie. Each determination is based on esses take place at the lake surface and are ten days of recorded levels (five at the end of transmitted through the water body while one month and five at the beginning of the others produce heat changes by mixing of the next month). This determination period is water masses. Meteorological factors such as rather long for the beginning-of-month levels. radiation, air temperature, precipitation, and In other determinations four (two plus two) or evaporation affect surface temperature, while two (one plus one) days were normally used. winds contribute to the deepening of the sur­ The most recentdetermination is basedon two face layer. Hydrologic factors such as runoff, days of recorded levels (one at end and one at inflow, and outflow cause temperature beginning of month), employing more gages changes by horizontal movement of water weighted by the Thiesson polygon method to mass. lOt Appendi:r"

"AT AlS_EO CIIl lIlIT IT +400 Qh = conduction of sensible heat to or LAIlE OlIITAIlID from the atmosphere Q., = energy utilized by evaporation o Qt = energy storage within the body of water.

- 000 The heat budget for Lake Ontario, the only lake for which such determination has been +000 made (Rodgers and Anderson,8'75 Bruce and . Rodgers,tOl and Rodgers872), is presented in i o .. Figure 4-100. The largest energy change is ! produced by the absorption and loss ofheat by LONG-WAVE - ZOO IlAllIATION LOSS = the lake water mass; the lake gains heat dur­ ing spring and summer months and loses heat ! in the fall and winter. The radiation processes •o o ... produce both gain and loss of heat; the lake absorbs heat from solar radiation and loses heat through the longwave radiation ex­ change between water surface and atmos­ o phere. Evaporation cools the water surface -2110 and produces heat loss, except during spring when slight condensation produces small heat gain. The transfer of sensible heat to the at­ IJIFI .. lal .. IJIJlalsloINIDI mosphere results from the air-water tempera­ ture differences; the lake surface is cooler FIGURE 4-100 The Heat Budget of Lake On­ than air and gains heat in the spring and tario summer, and the process is reversed in the fall From Rodgers, IH9 and winter. The net effect of all these proc­ esses is to produce heat gain during the spring-summer period and heat loss during The heating and cooling processes are sum­ fall and winter months. marized in the heat budget ofthe lakes, which The heat budgets for the other lakes would represents the amountofenergygained orlost follow generally similar patterns, although by the lakes during various temperature the amounts of energy contained in various changes. There are five basic energy or heat processes would differ depending on the hy­ processes affecting the Great Lakes. The four drometeorological conditions on each lake. major processes include energy produced by The accuracy of heat budget presented for radiation, sensible heattransfertoor from the Lake Ontario may be sufficient to indicate atmosphere, heat loss by evaporation, and general trends for various energy processes, energy storage within the lake. A fifth process but evaporation studies show that accuracy of net advected energy may be important lo­ should be improved for successful application cally, especially at the mouths of the inflow to the solution ofpractical problems. This was rivers and near the effluents of sewage dis­ one ofthe objectivesofthe International Field posal or cooling water from power plants. Year for the Great Lakes, an intensive field However, this process has very little effect on observation program conducted on Lake On­ the total heat content of the lakes because tario in 1972. such inflows with substantial difference in temperatures are relatively small. The energy exchange may be expressed by the equation: 4.13 Water Budget O. + Qv = ~ + Qh + Q., + Qt where Q. = net solar radiation (incident 4.13.1 Water Budget Computations minus reflected) Qv = net advected energy (heat due to The water budget of the Great Lakes is an water input minus output and accounting of all incoming and outgoing wa­ snow melt) ter, such as inflow and outflow by the rivers, Qb = net terrestrial radiation (emitted supply from and storage in the ground, over­ minus atmospheric) water precipitation, evaporation, and varia- , TABLE 4-22 Aver.,e Water Budget. 1937-1969 (Centimeters)

Balance Water Supply Water Loss Storage Needed Lake p R I 0 E t£, B

Superior 80 58 0 86 55 0 -3 Michigan-Huron 80 67 60 139 65 3 0 Erie 88 72 640 706 85 3 6 Ontario 84 150 927 1,077 70 2 12

P + R + I - 0 - E- 6S ... ±B P precipitation on the lake surface R runoff from drainage area (surface and underground) I = inflow from the upstream lakes o outflow to the lake below E evaporation from the lake surface f)S change in storage of water in the lake B balance needed

NOTE: Diversions are included in runoff, inflow, or outflow, where applicable. Evaporation values are the long-term estimates, not necessarily applicable to this 33-year period.

tion ofwater storage in the lakes.These water tion of these factors for each lake. The differ­ budget factors are interrelated in the hy­ ences needed for balancing of the major fac­ drologic cycle, which is composed of a per­ tors represent a combination of any possible petual sequence of events governing the de­ ground-water flow and cumulative errors in pletion and replenishment ofwater in the Ba­ estimating other factors. For most lakes these sin. The Great Lakes water budget may be differences are quite small for the average an­ expressed by the equation: nual values, and are well within the limits of error. The largest difference, for Lake On­ + + + ::t P R I = 0 E as tario, is approximately equal to 15 percent of where P = precipitation on the lake surface precipitation or evaporation, 8 percent of R = runoff from drainage area (sur­ runoff, or 1 percent of inflow or outflow. For face and underground) shorter monthly periods and for individual I = inflow from the upstream lakes years, the percent differences should increase o outflow to the lake below significantly, because the effect ofcompensat­ E = evaporation from thelake surface ing reduction of random errors would be as = change in storage ofwater in the smaller. lake (plus if storage increases, Further studies pertaining to the water minus if decreases) budget of the Great Lakes should be directed In practical applications the water budget towards elimination of existing gaps in pres­ equation may be modified by eliminating all ent knowledge, improvementofdatacollection factors that are negligible or not applicable to networks, comparability of measurement ac­ individual lakes (e.g., inflow for Lake Supe­ curacies for various factors, development and rior). Factors other than those listed may also implementation of new measurement meth­ be included. For example, runoff and ground ods, and closer coordination of these efforts in water may be treated separately, and diver­ both countries. sions may be included as a separate factor. The average annual water budget for the 1937-69 period is shown in Table 4-22, which 4.13.2 Importance of Water Budget contains groupings of water !SupplYJ water losses, lake storage, and algebraic aceumula- Lake levels and outflows ofthe Great Lakes 104 Appendiz"

effectively integrate all other components of studies and investigations (e.g., Freeman,ITI the water budget and are of primary interest Horton and Grunsky,n. U.S. Congress­ ll7 IIJ to lake users. However, growth of the popula­ Senate • ). Knowledge of the magnitudes tion and economy of the area has resulted in and variations ofthe individual water budget an increase in and diversification of demands components is needed for the improvement of for lake water, and the competition for its use forecasts of lake levels and outflows, for the is increasing rapidly. Use ofthe lakes for navi­ refinement of lake regulation plans, and for gation, water power, municipal and industrial determination ofthe effects ofdiversions into water supplies, sanitation, irrigation, fish and and outofthe system. Because ofthe vastness wildlife, recreation, and other riparian inter­ ofthe Great Lakes,changes in lake levels take ests frequently resultsin conflietingdemands, place rather slowly and advanced information some of which are detrimental to water qual­ on the expected stages is of great interest to ity. To provide optimum utilization and pres­ navigation, hydropower, and for shore protec­ ervation ofthe lakes, a thorough understand­ tion. For Lakes Superior and Ontario, the only ingofthe entire hydrologiccycle ofthe system lakes presently regulated, accurate forecasts is necessary. are even more important to permit planning The importance ofthe water budget to lake for the most beneficial operation of the reg­ water resources has been recognized in many ulating structures.