Back

FACTORS INFLUENCING THE MAGNITUDE
OF THE NOCTURNAL INVERSION AND THE MINIMUM TEMPERATURE

By

J. Leith Holloway, Jr.

Submitted in Partial Fulfillment
of the Requirements for the Degree of Master of Science
at the Massachusetts Institute of Technology
May 25, 1953


Table of Contents

Abstract

      Two summers of wind, temperature, and cloudiness data taken at the Brookhaven National Laboratory are analyzed statistically. Empirical equations are derived for the maximum magnitude of the of the nocturnal ground inversion between the 410- and 37-foot elevation given the cloudiness and mean 12-hour 410-foot wind speed during the night. Also, equations are derived for predicting the fall of temperature from 1830 EST to minimum from the cloudiness and the 410-foot wind speed. In these equations the magnitude of the inversion and the fall of temperature to minimum are inversely proportional to the 410-foot wind speed and the amount of cloudiness. The effect of advection on the inversion magnitude and temperature fall is determined by examining the prediction errors of the equations classified according to wind direction. It is found that the magnitude of the inversion is increased by warm advection and decreased by cold advection. The fall of temperature to minimum is decreased by warm advection and increased by cold advection. Finally, the relative effect of low and high cloudiness upon the net long-wave radiation is estimated from a study of the equations derived in this thesis.

Thesis Supervisor:  James M. Austin
Title:  Associate Professor of Meteorology

ACKNOWLEDGMENTS

      The author wishes to thank Professor James M. Austin for his assistance as thesis adviser and Mr. Maynard E. Smith, Mr. Irving A. Singer, and the entire Meteorology Group at the Brookhaven National Laboratory and Mr. Raymond C. Wanta of the U.S. Weather Bureau Office at Upton, N.Y. for their guidance while the author was making his preliminary study at Brookhaven. Also, the author's appreciation goes to Dr. C. F. Brooks, Director of the Blue Hill Observatory, Milton, Massachusetts, for his help in locating important references and to Miss Cynthia Silver for typing the manuscript.

TABLE OF CONTENTS

      I.
Introduction
     II. History of the Subject
    III. Instrumentation, Data, and Description of the Site
    IV. Nocturnal Inversion
         A. Formation and Development of the Inversion
         B. Climatology of Inversions at Brookhaven
         C. Prediction of the Magnitude of the Inversion
      V. Prediction of the Minimum Temperature
     VI. Effect of Advection on the Inversion and the Fall of Temperature to Minimum
    VII. Conclusions and Recommendations

   Figure 1. The 420-foot Tower at the Brookhaven National Laboratory.
   Figure 2. A close-up view of the levels of the tower.
     Table I. Mean Temperature Differences at Brookhaven during 1951
    Table II. Mean Maximum, Minimum, and Daily Range during June 1951
   Table III. Mean Deviation of Data from Values Predicted by the Regression Functions

    Bibliography



I. INTRODUCTION
      The magnitude of the nocturnal ground inversion and the minimum temperature at a station are influenced by a great number of factors. Since the strength of the inversion and the minimum temperature are intimately related, factors influencing one are also important in controlling the other. The main factors influencing both are the wind speed, the net nocturnal radiation, and advection. Less important factors influencing the magnitude of the inversion and the minimum temperature at a given locality include the release of latent heat of condensation with the formation of dew and fog, the conductivity and moisture content of the ground, cold air drainage from a neighboring area, and the length of the night. The effect of all of the factors mentioned above has long been recognized, and forecasters consider them qualitatively when preparing minimum temperature predictions.
      The purpose of this study is to determine some quantitative relationships between the strength of the nocturnal inversions and the minimum temperature and the wind speed and the net nocturnal radiation estimated by cloudiness. These qualitative relationships are to be modified by a qualitative study of the effect of advection. The relations will be derived empirically. In the derivations the effects of factors not considered will either be minimized if the effects are significant or neglected if not.
      The practical value of these relationships is apparent. Accurate minimum temperature forecasts are valuable both in connection with the protection of agriculture from frost damage and in the prediction of radiation fog. The prediction of the magnitude of the nocturnal inversion is useful in connection with forecasting winds near the ground and industrial pollution at night and in the morning.

Back to Table of Contents

II. HISTORY OF THE SUBJECT
      The phenomenon of the nocturnal temperature inversion has long been observed by means of the thermometers either on towers or at various elevations on hill slopes (see references: 4; 6; 7; 8; 12; 21). However, no one until now to the knowledge of the author has attempted to derive an empirical equation for forecasting the magnitude of the inversion. Nevertheless, it will be worthwhile to review what previous investigators have observed concerning the influence of wind speed on the magnitude of the inversion. Durst (3) in 1933 suggested that the breakdown of the ground inversion at night was caused by the development of excessive wind shear between the top and bottom of the inversion as the magnitude of the inversion increased. It should be noted in connection with this thesis that with calm wind at the ground, the wind speed at an elevated level is equivalent to the wind shear between this level and the ground. Smith (19) has observed at Brookhaven that a higher wind speed is required to eliminate an existing inversion than is required to prevent one from forming initially.
      In contrast to the lack of empirical formulas for forecasting the inversion magnitude, the literature contains a multitude of formulas for predicting the minimum temperature. The reason for this greater emphasis on minimum temperature formulas is doubtless that a forecast of the minimum temperature is of greater practical value than a prediction of the magnitude of the inversion. Also, the minimum temperature is considerably easier to observe than the inversion magnitude.
      Most of the minimum temperature formulas are derived empirically. Generally they are based on the afternoon surface temperature, dew point and relative humidity. A number of such formulas are contained in Supplement 16 of the Monthly Weather Review( 18). Also, Ellison (5), and Sutton (20) discuss minimum temperature formulas at length. Some of the formulas may be modified for differences in wind speed and cloudiness but only by qualitative or semi-qualitative means. That is, cloud amount may be specified as overcast, partly cloudy, or clear, and wind speed as calm, light, moderate, or strong. In the hands of a trained forecaster who knows his region the formulas, Ellison claims, should predict the minimum temperature within 3F on most nights. Brunt (2) gives a theoretical formula whereby the minimum temperature may be computed from the net radiation, and the conductivity, specific heat, and density of the ground, and the length of the night. In the derivation of his formula, Brunt assumes that all the heat lost by radiation at night comes from the ground. knighting (14), considering this assumption unrealistic, derives a rather complicated theoretical formula for predicting the fall of temperature to minimum by considering that the heat lost by radiation comes from both the ground and the air. However, these theoretical formulas are of no practical value since the parameters contained in them, such as soil properties, are variable from night to night and are never known accurately.
      This review of the subject indicates that empirical equations for forecasting the magnitude of the inversion are needed, and that equations for the minimum temperature depending directly upon cloudiness and wind speed probably would be useful. Therefore, this study was undertaken to derive such equations.

Back to Table of Contents

III. INSTRUMENTATION, DATA, AND DESCRIPTION OF THE SITE
      This study was made from data taken during the summers of 1950 and 1951 at the Brookhaven National Laboratory, Upton, Long Island, New York. At Brookhaven there is a 420-foot micrometeorological tower on which aspirated resistance thermometers (Thermohms) are mounted on booms at eight levels. See figures 1 and 2. These levels are 9, 18, 37, 75, 150, 300, 355, and 410 feet.

Figure 1. The 420-foot micrometeorological tower at the Brookhaven National Laboratory. The plume of smoke comes from a stack extending to the 355-foot level on the tower. Diffusion of this smoke is studied at Brookhaven.
Figure 2. A close-up view of the 18- and 37-foot levels of the Brookhaven tower. Note the Aerovanes on the tops of the booms and the housings for the Thermohms extending downward from the ends of the booms.

The temperatures measured are recorded every 5.8 minutes on a Leeds and Northrup, triple range, twelve point, Micromax indicating recorder. The temperature difference between 410 and 37 feet is recorded on a Leeds and Northrup single point, Model R recorder. Standard Bendix-Friez Aerovanes are installed on all levels of the tower except at the 9-foot level. Wind speeds and directions indicated by these Aerovanes are recorded on Esterline-Angus recorders. The temperatures indicated on the Micromax recorder are accurate to within 0.10. However, the values of the hourly average temperature differences determined graphically from the Model R traces may be in error as much as 0.30 due to accumulated instrument and human errors. The wind speeds indicated by the Aerovanes are accurate to within two or three-tenths of a meter per second except at speeds less than five meters per second where they are somewhat less accurate. For further information on the instrumentation at Brookhaven see reference (15).
      Data from the tower instruments along with regular surface observations have been compiled by the Meteorology Group at the Brookhaven Laboratory since 1948. However, mainly only 1950 and 1951 data are used in this study because of their accuracy, completeness, and convenient tabulation.
      The ground in the vicinity of the tower and the Meteorology Group's building is essentially level. To the south of the tower lies a flat field covered with tall grass and weeds and to the north, a pine forest about twenty-five feet high. The standard U.S. Weather Bureau thermometer shelter where the minimum temperatures are observed is located on the ground near the Meteorology Group's building about 700 feet to the southwest of the tower.

Back to Table of Contents

IV. NOCTURNAL INVERSION
A. Formation and Development of the Inversion
      A brief discussion on the mechanisms producing and maintaining the ground inversion should aid the reader in interpreting the results presented in this thesis. The ground, being for all practical purposes a black body, is a better long-wave radiator than the air. Therefore, at night the ground cools much more rapidly by radiation than the air. Thus, the air in contact with the ground cools faster at night than the free air. The free air will be defined here as the air several hundred feet above the ground. As long as there is a temperature decrease with height in the lower atmosphere during the evening, the cooling at the surface will be distributed upward readily by turbulent mixing, and the free-air temperature will fall almost as rapidly as the surface temperature. However, when inversion is formed, the stratification in the lower atmosphere becomes stable, and convective heat exchange is damped out. Therefore, the free air is virtually isolated from any further surface cooling. One might say that the free air is then insulated against the coldness below at the surface by the greatly reduced eddy conductivity of the inverted lower atmosphere. It has been found in this study that almost simultaneously with the formation of the inversion, the free-air temperature as a rule decreases its rate of fall noticeably and remains nearly constant throughout the rest of the night whereas the surface temperature generally continues to fall steadily until morning. However, if there is cold advection, the free-air temperature will continue to fall even after the inversion has formed. This decrease in temperature results mainly from the horizontal influx of cool air from a colder region, not from vertical transport of cool air from the surface. On the other hand, with warm advection the free-air temperature may rise during the inversion while at the same time the surface temperature is continuing to fall due to radiation.
      After the inversion is formed, its development during the night is controlled principally by the interaction of two opposing forces, namely, radiation and wind. Radiation tends to increase the strength of the inversion by indirectly cooling the surface air, whereas the wind tends to decrease the inversion by causing vertical mixing of the air near the ground. In addition to radiation and wind speed, a third variable which affects the magnitude of the inversion is the sign of the advection. Cold advection decreases the strength of the inversion by transporting colder air to higher levels faster than to the surface. On the other hand, warm advection increases the inversion.

B.
Climatology of Inversions at Brookhaven
      Large inversions and low minimum temperatures are favored at Brookhaven by the surface of sandy soil prevalent here. Sandy soil cools rapidly by radiation because of its low conductivity. Therefore, 100 inversions and summer minimum temperatures below 45F are common here. The maximum inversion recorded was 17.8C between the 9- and 355-foot levels in the early morning of 28 December 1950. The temperature at the 9-foot level was -25.6C while the 355-foot temperature was -7.8C. An almost equally strong inversion of 17.7C occurred on the morning of 17 December 1948. At this time the surface temperature was -13.4C. A minimum temperature of 33F wass reported on the morning of 9 September 1952. However, inversions at Brookhaven probably appear large only because detailed observations in the lowest several hundred feet of the atmosphere are still relatively rare especially in rural areas where the nocturnal inversions are most pronounced and the minimum temperatures lowest. Data from other meteorological towers and soundings show that large inversions are also common in a number of other areas (7; 9; 16; 21).
      The nocturnal ground inversion is generally rather shallow. On most nights the 420-foot Brookhaven tower extends above the top of the inversion. On many nights eighty per cent of the temperature difference in the inversion occurs below the 150-foot level.
      Of twenty-three large inversions (410- to 37-foot temperature difference greater than 11C) reported from January 1950 to September 1952, nearly half occurred in September and October. It has been discovered elsewhere (8) that both the most stable and unstable atmospheric conditions occur in the summer. Probably warm advection off the Atlantic Ocean in autumn is instrumental in shifting the period of maximum stability at Brookhaven from summer to fall.
      The mean diurnal variations of surface and free-air temperatures are illustrated in tables I and II. Nine- and 355-foot temperatures at 0300, 0700, 1100, 1500, 1900, and 2300 EST are averaged for the entire month of June 1951. These data are presented in
Table I. This table shows that in the mean the 355-foot temperature tends to become steady at night whereas the 9-foot temperature continues to fall throughout the night. Also included in this table are the mean temperature differences between these two levels at these six times and the mean temperature for the month at each level. The mean maximum and minimum temperatures and the mean daily temperature ranges at the two levels for June 1951 are tabulated in Table II. Notice that the mean daily temperature range at 355 feet during June 1951 was only 54 per cent of the mean range at 9 feet. This decrease in the amplitude of the temperature wave with height has been observed at a number of other stations (1; 4). For comparison, the average daily range of temperature at the base of the Eiffel Tower in July is 9.20 whereas at the top of this 990-foot tower the diurnal range of temperature is only 6.90 (4). Since the data in tables I and II are a mean picture of the month, the effects of advection and differences in radiation are approximately averaged out. For further information on climatology at Brookhaven see references (17) and (19).

C. Prediction of the Magnitude of the Inversion
      One of the purposes of this study is to determine a quantitative relationship between the wind speed and the magnitude of the nighttime ground inversion. The question arises as to what would be good criteria for the strength of the inversion and the wind speed. The criterion chosen for the strength of the inversion is the maximum hourly average temperature difference between 410 and 37 feet. An hourly average temperature difference is chosen in order to eliminate the effect of sporadic fluctuations. In regard to a suitable wind criterion, obviously a wind speed near the the surface would not be a good parameter, for the surface wind almost invariably becomes calm during inversions. From a forecast standpoint, the mean geostrophic wind speed for the night would be the best parameter since this could be estimated from a prognostic pressure chart. In a study of old data such as this, the geostrophic wind speed could be computed from actual surface pressure maps. However, geostrophic wind is difficult to estimate accurately from synoptic charts especially in a region such as Brookhaven where the pressure patterns are complicated by land-sea effects. Therefore, the mean wind speed from 1800 to 0600 EST at the 410-foot level is used in this study as an estimate of the average geostrophic wind speed during a given night. At Brookhaven the 410-foot wind very seldom becomes calm at night. In fact, the wind speed at this level generally increases and more nearly approaches the geostrophic speed during the night. This phenomenon of the reversal of the diurnal variation of the wind speed at levels above about 200 feet elevation has also been observed at other places (10; 11). On the nights studied the mean 410-foot wind speed ranged from 2.1 to 11.4 meters per second.
      Sixty-eight fogless, rain-free nights in the summer months (June through September) of 1950 and 1951 are studied. These nights include those not having inversions as well as those that do. The data are analyzed statistically by the method of least squares. The regression function for the magnitude of the inversion with respect to the one variable, wind speed is

I = 8.6 - 0.7 W,          (1)

where
I is the maximum hourly average temperature difference between 410 and 37 feet in centigrade degrees between 1800 and 0600 EST (positive during inversions), and
W is the mean 410-feet wind speed in meters per second from 1800 to 0600 EST.

Note: Values of this hourly average 410- minus 37-foot temperature differences are tabulated for each hour by the Meteorology Group. The maximum value for each night is taken as I. Perhaps slightly higher values of I could have been obtained on many nights by taking means over hour periods not beginning on the hour.

The correlation coefficient for this regression function is -0.52. The correlation is high enough to indicate that a good inverse relationship exists between the wind speed at 410 feet and the magnitude of the inversion between 410 and 37 feet. Nevertheless, the relation is not good enough to be useful for predicting the maximum strength of the inversion from a forecast of the 410-feet wind. It is noteworthy, however, that equation (1) states that a ground inversion can develop at Brookhaven even when the mean twelve-hour wind speed at 410 feet is greater than ten meters per second.
      In addition to wind speed, a second major factor influencing the magnitude of the nighttime ground inversion is the amount of net long-wave radiation. Therefore, in order to obtain a better equation for the magnitude of the inversion, the effect of differences in radiation is added to the statistical analysis. Since nocturnal radiation is not measured at Brookhaven, its intensity has to be estimated by means of a meteorological variable which is observed. Wet nocturnal radiation is controlled mainly by cloudiness, and surface temperature and humidity. In this study the intensity of the net radiation is estimated by the amount of cloudiness. The sixty-eight nights chosen for this study are selected only from summer months so that surface temperatures and humidities would be approximately the same for all nights. A tendency toward reduced radiation at the end of the summer due to lower surface temperature is probably approximately compensated by longer nights and lower humidities occurring then. Also, nights with fog reported at any time are excluded from this study since fog complicates the development of the inversion in two ways. First, the formation of fog releases latent heat of condensation which slows the fall of temperature at the surface. Secondly, the presence of fog reduces the net nocturnal radiation. It is hoped that by eliminating nights having fog, those with heavy dew fall would also be excluded from this study. The presence of haze during some of the nights studied is disregarded since measurements of nocturnal radiation by other investigators (13) has shown that haze has no appreciable effect upon net radiation.
      For convenience in the statistical analysis clouds are divided into only two groups, low and high clouds, instead of the three groups common in synoptic meteorology. Low clouds in this study are clouds classed as low or middle clouds in present synoptic codes, and high clouds are the same as in synoptic meteorology. This seems to be a logical division especially since the cloud observations used were taken at night when low and middle clouds cannot be easily distinguished. The number of tenths of low and middle cloudiness reported at each of the twelve hourly observations from 1830 to 0530 EST are totaled to obtain for a given night a number representative of the total amount of low cloudiness as defined here. Thus, the maximum number obtained (for ten-tenths low cloudiness all night) would be 120. The number representing the amount of high cloudiness is obtained in the same manner. In the application of this scheme, no hour is allowed to have a total of low, middle, and high clouds greater than ten-tenths. The tenths of lower clouds always takes precedence over the tenths of higher clouds. For instance, if six-tenths low clouds and nine-tenths high clouds were reported one hour, hour-tenths of low and high clouds would be computed on the basis of six-tenths low clouds and only three-tenths high clouds.
      When cloudiness variables are used in addition to the 410-feet wind speed, the resulting regression function for the magnitude of the inversion is

I = 12.7 - 0.06 L - 0.01 H - 1.02 W,      (2)

where
L is the number of hour-tenths of low cloud from 1830 through 0530 EST,
H is the number of hour-tenths of high cloud during the same period, and
I and W are the same as in (1).

The multiple correlation coefficient for this regression function is 0.78.

Note: The multiple correlation coefficient is always taken positive in sign regardless of the signs of the regression coefficients.

Furthermore, the mean prediction error of (2) is only 1.5C. In this thesis the prediction error of a regression function will be defined as the absolute value of the difference between the observed and predicted values. Therefore, (2) is a useful equation for predicting the maximum development of the nocturnal inversion at Brookhaven in the summer.
      The introduction of cloudiness into the analysis increased the absolute value correlation coefficient by 0.26. However, this increase in correlation due to the addition of cloudiness is only half the degree of correlation obtained between wind speed and inversion magnitude in (1). This fact should indicate that cloudiness is less important in controlling the magnitude of the 410- to 37-foot inversion than is the 410-foot wind speed provided cloudiness and wind speed are independent of each other. This supposition was checked by computing the regression function for the magnitude of the inversion with respect to only the two variables, low and high cloudiness. This regression function is

I = 4.2 - 0.03 L - 0.007 H .      (3)

The multiple correlation coefficient for (3) is 0.31 which is significantly smaller than the absolute value of the correlation coefficient of (1). Therefore, the preceding supposition is correct. It also follows that for all practical purposes wind speed and cloudiness are, indeed, independent of each other at Brookhaven in the summer.

Back to Table of Contents

V. PREDICTION OF THE MINIMUM TEMPERATURE
      Since the minimum temperature is very closely associated with the strength of the ground inversion, and high correlation between the magnitude of the inversion and the wind speed and cloudiness indicates that an equally high correlation can be expected to exist between the minimum temperature and wind speed and cloudiness. Furthermore, a good empirical equation for predicting the minimum temperature would be of definite practical value. However, the value of the minimum temperature itself would not be a good variable to use in this study since it is so much a function of the minimum temperature of the preceding day that day to day variations in the temperature maxima would obscure the influence of wind speed and cloudiness on the minima. Therefore, in order to avoid the influence of day to day temperature changes on the minima, the fall of temperature from near the time of sunset to minimum is used in this study as a measure of the lowness of the minimum.
      The temperature fall to minimum is analyzed statistically using data from the same 68 nights used in the inversion study. The regression function for the fall of temperature in respect to the mean 410-foot wind speed is

F = 21 - 0.9 W,      (4)

where
F is the fall of temperature in Fahrenheit degrees from 1830 EST to minimum observed at the instrument shelter.

The correlation coefficient of (4) is -0.30. This is significantly smaller in absolute magnitude than the correlation coefficient of (1). Thus, it appears that the wind speed at 410 feet is not as good a predictor of the temperature fall to minimum as it is a predictor of the magnitude of the inversion between 410 and 37 feet. The explanation for this paradox probably lies in what occurs below the 37-foot level during an inversion night. On some nights, the temperature stratification below the 37-foot level is nearly isothermal whereas on other nights the shelter temperature may be as much as 10F colder than the 37-foot temperature. During most inversions the 37-foot wind becomes nearly calm no matter what the wind speed at 410 feet may be. Therefore, the 410-foot wind speed has little effect on the temperature stratification below 37 feet. The thermal structure below 37 feet is probably mainly influenced by the net radiation and surface dew point. Thus, the wind speed at 410 feet plays a larger part in determining the strength of the inversion between 410 and 37 feet than it does in controlling the fall of surface temperature to minimum because the temperature to minimum because the temperature fall at the ground is to a large extent influenced by what occurs in the layer of air below the 37-foot level whereas the inversion magnitude by definition here is independent of this layer. Undoubtedly, if the criterion for the strength of the inversion had been chosen as the temperature difference between 410 feet and a level nearer the ground than 37 feet, the correlation between the 410-foot wind speed and the magnitude of the inversion defined in this manner would have been less than is obtained in (1). It should be remarked that the 410- to 37-foot temperature difference is recorded at Brookhaven rather than the temperature difference in a layer extending closer to the ground because it is considered that the 410- to 37-foot temperature difference is more representative of the stratification in the lowest 400 feet of the atmosphere than a temperature difference affected by the thermal structure in the surface layer below 37 feet.
      Cloudiness is next added to the analysis of the temperature fall to minimum. The regression function for the temperature fall with respect to the three variables, 410-foot wind speed, and the low and high cloudiness is

F = 32 - 0.014 L - 0.08 H - 1.6 W.      (5)

The multiple correlation coefficient for (5) is 0.62 which is nearly as high as that of (2). Also, the mean prediction error of equation (5) is only 4F. Thus, (5) is good enough to be useful in forecasting the fall of temperature to minimum at Brookhaven on summer nights without fog. If in actual practice (5) predicts a minimum temperature far below the dew point at sunset, this indicates that fog will probably form. Therefore, in such situations (5) should not be expected to be accurate.

Back to Table of Contents

VI. EFFECT OF ADVECTION ON THE INVERSION AND THE FALL OF TEMPERATURE TO MINIMUM
      The sign and magnitude of deviations of data from the values predicted by the regression functions (2) and (5) are definitely a function of wind direction. This fact indicates that advection caused a large part of these deviations. Since the multiple correlation coefficient of the temperature-fall regression function (2), it appears that the sign of the advection has greater influence on the temperature fall than on the magnitude of the inversion. This seems paradoxical since one would think advection would have little effect on the surface temperature during inversions due to accompanying light surface winds. However, advection is quite instrumental in controlling the surface temperature before the inversion is formed in the evening. At this time substantial wind exists at the surface. Furthermore, the effect of advection upon the magnitude of the inversion is relatively small because the inversion is not modified directly by advection but by a differential of advection between the top and bottom of the inversion.
      Table III gives the mean deviation of the data from the values predicted by the regression functions classified according to wind direction. The wind direction chosen for this analysis is the geostrophic wind direction determined from the 0130 EST surface pressure map for the night in question. It is thought that the direction obtained in this way would be more representative of the type of advection present than a wind direction obtained from the tower observations. On thirteen of the nights studied the pressure gradient was so weak that the geostrophic wind direction could not be determined with certainty. This table show that larger than normal inversions occur at Brookhaven with west and northwest winds, and that inversions smaller than expected develop on nights with north and northeast winds. The effect of southwest winds on the strength of the inversion appears to be neutral. On the other hand, temperature falls were less than the regression function predicted with west and southwest winds and greater than predicted when the wind was from the northwest, north, and northeast. Not enough data are available for winds from the south, southeast, and east for conclusions to be made on the effect of wind from these directions upon the inversion or the temperature fall. Therefore, minimum temperature and inversion magnitude forecasts made using (5) and (2) should be modified qualitatively by means of an estimate of the effect of advection based on the forecast wind direction and the results shown in Table III.
      The result that in the mean northwest winds are accompanied by larger inversions than (2) predicts, probably does not indicate that northwest winds give warm advection at Brookhaven, but that northwest winds bring drier air over the area and increase the net outgoing radiation for a given amount of cloudiness. This influence of the drier air probably affects the tendency for cold advection with northwest winds to decrease the strength of the inversion. The sign of the advection for different wind directions is better determined from the deviations of data from values predicted by (5) than by (2), for advection produced deviations are larger in the case of the equation for temperature fall. Therefore, it is concluded that at Brookhaven in the summer warm advection generally occurs with west and southwest winds, and cold advection with northwest, north, and northeast winds. The sign of the advection with south, southeast, and east winds is undetermined for lack of data.
      Since advection caused a great deal of the deviation of data in (5), a better equation for the temperature fall can be obtained by minimizing the effect of advection. The strength of the advection is proportional to the wind speed and the strength of the temperature gradient. Thus, the influence of advection can be reduced by restricting the study to nights with light wind. Therefore, from the original 68 nights studied, 32 are chosen during which the mean 12-hour 410-foot wind speed did not exceed 7.5 meters per second. Since the range of wind speeds is restricted, the effect of differences in wind speed on the temperature fall is thereby minimized. Thus, wind speed can be regarded as constant in the study of these 32 nights especially since wind speed is not a very good predictor of temperature fall in (4). The resulting regression function for the temperature fall to minimum with respect to low and high cloudiness is

F = 22 - 0.12 L - 0.05 H.      (6)

The multiple correlation coefficient for (6) is 0.80, and the mean prediction error is only 3F. Therefore, (6) should be very useful for forecasting minimum temperatures at Brookhaven in the summer. The advantage of this equation is that as long as it can be predicted that the mean 12-hour 410-foot wind speed will not exceed 7.5 meters per second on a given night, the minimum temperature can be predicted from from (6) without need of a more detailed 410-foot wind speed forecast. Only a cloud forecast is needed. As with (5), equation (6) should not be expected to be accurate when the minimum it forecasts is far below the evening dew point, for then fog would probably form.
      The deviation of data from the values predicted by (6) may come from the effect of differences in wind speed or advection not completely filtered out of the data by restricting the study to nights having only light wind. However, more probably the deviations come mainly from errors introduced into this work by assuming that the same net radiation always occurs with a given amount of cloudiness. Also, the deviation may be caused by the multitude of minor factors which had to be neglected from this study. For instance, the moisture content of the soil influences the fall of temperature at night by affecting the conductivity of the soil. The presence or absence of cold air drainage from higher areas on the northern shore of Long Island during a particular night could also affect the temperature fall at Brookhaven. However, a thorough analysis of any of these minor factors is beyond the scope of this thesis.
      The reader may have noted that the ratio of the regression coefficients for low cloudiness to that for high cloudiness is different in each of the regression functions containing cloudiness. This ratio varies from six in (2) to one and three-quarters in (5). Therefore, it is difficult to derive from these regression functions any conclusions concerning the relative effects of low and high cloudiness upon the strength of the inversion, the fall of temperature to minimum, or the net nocturnal radiation. However, since (6) has the highest correlation coefficient and contains only cloudiness variables, it probably most accurately indicates the relative effect of low and high cloudiness upon the net radiation. Thus, low cloudiness, as defined in this thesis, is probably about two and one-half times as effective in reducing net radiation as high cloudiness.

Back to Table of Contents

VII. CONCLUSIONS AND RECOMMENDATIONS
      A useful empirical equation can be derived for the magnitude of the nocturnal ground inversion or the fall of temperature from sunset to minimum based on the amount of cloudiness and the wind speed several hundred feet above the ground. Deviations of the inversion magnitude and the temperature fall from the values predicted by the equations are to a large extent a function of the sign of the advection. Therefore, minimum temperature and inversion forecasts made from equations presented in this thesis should be modified qualitatively according to the sign of the advection. The empirical equations derived for Brookhaven in this study do not necessarily apply to other localities. Studies similar to this should be made in other areas to see how much modification the equations in this thesis must undergo for them to be useful elsewhere. In fact, a more thorough study at Brookhaven, with emphasis upon other seasons beside summer, would be worthwhile.

Bibliography












Back to Table of Contents

BIBLIOGRAPHY



 1.     Ballard, J. C., 1933:  The diurnal variation of free-air temperature and the temperature lapse rate.
            Monthly Weather Review, Vol. 61, No. 3, pp 61-80.

 2.     Brunt, David, 1941:  Physical and dynamical meteorology. London, Cambridge University Press, 124-46.

 3.     Durst, C. S., 1933:  The breakdown of steep wind gradients in inversions.
            Quarterly Journal of the Royal Meteorological Society, 59: 131-36.

 4.     Eiffel, G., 1900:  Travaux scientifiques exécutées a la tour de trois cente mètres de 1889 a 1900. Paris, 116-7.

 5.     Ellison, Eckley S., 1928:  A critique on the construction and use of minimum-temperature formulas.
            Monthly Weather Review, 56, 485-95.

 6.     Ferguson, S. P., 1900:  Annals of the astronomical observatory of Harvard College.
            Vol. 10, part I, Appendix C, 131-7. (Blue Hill data)

 7.     Flower, W. D., 1937:  An investigation into the variation of the lapse rate of temperature in the
            atmosphere near the ground at Ismailia, Egypt. Geophysical Memoirs No. 71. Meteorological Office, London.

 8.     Geiger, Rudolf, 1950:  The climate near the ground. Translated by Milroy N. Stewart and others.
            Cambridge, Mass., Harvard University Press, 49, 80-5.

 9.     Gifford, Frank, Jr., 1952:  The breakdown of a low-level inversion studied by means of detailed soundings with
            a modified radiosonde. Bulletin of the American Meteorological Society, 33: 373-9.

10.    Haurwitz, B., 1941:  Dynamic meteorology. New York, McGraw-Hill Co., 213-5.

11.    Hellmann, G., 1941:  Uber die Bewegung der luft in den untersten Schichten der Atmosphare. Meteor. Z., 34:273.

12.    Johnson, N. K. and G. S. P. Heyword, 1938:  An investigation of the lapse rate of temperature in the
            lowest hundred metres of the atmosphere. Geophysical Memoirs No. 77. Meteorological Offices, London.

13.     Kimball, H. H., 1918:  Nocturnal radiation measurements. Monthly Weather Review, 46: 57.

14.     Knighting, E., 1950:  A note on nocturnal cooling.
            Quarterly Journal of the Royal Meteorological Society, 76: 173-81.

15.    Mazzarella, D. A., 1952:  Operational and research facilities in meteorology at Brookhaven National Laboratory.
            Paper presented at the 117th National Meeting of the American Meteorological Society,
            Buffalo, N.Y., 1-5 July 1952.

16.     Myers, R. F., 1952:  A low-level temperature sounding system for routine use.
            Bulletin of the American Meteorological Society, 33: 7-12.

17.    Singer, I. A. and M. E. Smith, 1953:  Relation of gustiness to other meteorological parameters.
            Journal of Meteorology, 10: 121-6.

18.    Smith, J. Warren and others, 1920:  Predicting minimum temperatures from hygrometric data.
            Monthly Weather Review Supplement No. 16.

19.     Smith, M. E., 1951:  The forecasting of micrometeorological variables.
            Meteorological Monographs, Vol. 1, No. 4, pp 50-55.

20.    Sutton, O. G., 1953:  Micrometeorology. New York, McGraw-Hill Co., 178-89, 209, 229-30.

21.     Young, Floyd D., 1921:  Nocturnal temperature inversions. Monthly Weather Review, 49: 138-48.


Back to Table of Contents


Back