THE UNIVERSITY OF MICIIGAN COLLEGE OF ENGINEERING Department of Atmospheric and Oceanic Science Second Annual Report AN INVEST IGA'IION OF Till': M:ETEI.OROLOGICAl., IMPACT OF MECI{ANICAL-DRAFT COOLING TOWERS AT THE PALISADES NUCLEAR PLANT Edward Ryznar Research Associate Dennis G. Baker Assistant Professor of Meteorology DRDA Project 320158 under contract with: CONSUMERS POWER COMPANY JACKSON, MICHIGAN administered through: The Division of Research and Development Administration June 1974

ABSTRACT The second year of research to determine the meteorological effects of mechanical-draft cooling towers at the Palisades Nuclear Plant is described. Work was primarily directed toward establishing a meteorological data base for the pre-cooling tower phase of the study. Data were collected by a network of 13 meteorological stations, network instrumentation was improved, data processing routines were developed, refined and applied, and data tabulations and analyses began. Results of these efforts are presented and discussed. ii

ACKNOWLEDGMENTS Our special appr-eciatioii is eextpressed to DIr. IHarry Moses of Argonne National Laboratory for his ideas and suggestions pertaining to all aspects of the work in the last year. Several people made important contributions to data processing and analysis and to portions of this report. They are Jeffrey Baron, William Snell, Robert Kessler, and Michael Weber, all of whom are or were graduate students; and Andrew Detwiler, Dennis Kahlbaum, and Dennis Dismachek, who are undergraduate students. Gary Goldman constructed the humidity calibration chambers. Paul Titus was responsible for much of the calibration work and coordinated the field program with our man in the field, Donald Pearson, who maintained the network data collection. Ms. Bonnie Beasley did her usual excellent job of typing the report. For the second year, very little vandalism occurred in the network, which is a credit to the community in the area near the Palisades Nuclear Plant. iii

CONTENTS Page ABSTRACTi ACKNOWLEDGMENTSiii LIST OF FIGURES v LIST OF TABLES vi I. INTRODUCTION1 Background Purpose and approach 2 Interpretation of work by others 4 II. DATA COLLECTION 12 III. DATA TABULATION AND ANALYSIS 14 Examples of tabulations and summaries 14 Natural variability 27 IV. DATA PROCESSING 36 Digitization 38 Data package 40 Final storage format 41 Example of input data 42 Final data plot 43 Accuracy of the analog to digital conversion 46 V. NETWORK INSTRUMENTATION 54 Calibration of hair hygrothermographs 54 Dew Point system 70 Precipitation gages 72 Visiometers 72 Wind systems 72 VI. WORK PLANNED FOR NEXT YEAR 74 REFERENCES 76 iv

List of Figures Page 1. Locations of network stations. 3 2. Annual wind rose for Muskegon County Airport. 7 3. Wind rose for station PO3A for February, 1973. 20 4. Wind rose for station PO7A for February, 1973. 21 5. Precipitation wind roses for Muskegon County Airport. 32 6. Data processing flow diagram. 37 7. Graf/pen GP-2 sonic digitizing system. 39 8. CALCOMP plot of variables for station PO3A 44 9. Spatial distribution of time averaged distance for 47 7 runs with digitizer. 10. Spatial distribution of time averaged standard 48 deviation for 7 runs with digitizer. 11. Humidity calibration chamber 56 v

List of Tables Page 1. Percent possible data recorded. 13 2. Visibility data for January 1973 for station P03A. 15 3. Visibility data for February 1973 for station P03A. 16 4. Daily and monthly precipitation totals for January 22 1973. 5. Daily and monthly precipitation totals for February 23 1973. 6. Daily maximum and minimum temperatures for January 24 1973. 7. Daily maximum and minimum temperatures for February 25 1973. 8. Percent of total hours with natural fog by month. 27 9. Monthly means of temperature and precipitation for 29 South Haven and Bloomingdale. 10. Average distance and standard deviation of data 50 points for digitizer. 11. Equivalent accuracy of digitizer in meteorological 52 units. 12. Results of matching thermometers. 59 13. Psychrometric and dew point hygrometer data. 61 14. Chamber humidity calibration data for four hygrother- 62 mographs. 15. Calibration repeatability test data for one hygro- 64 thermograph. vi

Page 16. Psychrometric and hygrothermograph humidity dif- 65 ferences for changing humidities. 17. Hygrothermograph calibration data obtained in 67 Ann Arbor and Benton Harbor 18. Hygrothermograph calibration data obtained after 69 transportation to Benton Harbor and return. vii

I. INTRODUCTION Background This is the second annual progress report for an investigation of the meteorological effects of mechanical-draft cooling towers at the Palisades Nuclear Plant. The study was initiated in 1971 by a request from the NOAA state climatologist for Michigan in cooperation with Consumers Power Company and Indiana & Michigan Power Company that the meteorological effects of nuclear power plant cooling systems in southwestern Lower Michigan be investigated. The investigation was to be for the Consumers Power Company's Palisades Nuclear Plant, which was to use mechanical-draft cooling towers, and for the Indiana & Michigan Power Company's Donald C. Cook Nuclear Plant, which was to use a once-through cooling system. Both cooling systems were under construction at the time and were scheduled for operation in 1974. Even though the meteorological effects of the two methods of cooling were expected to be different, a study of the effects of one system could supplement the other in many respects, since both nuclear plants were located on the Lake Michigan shoreline and separated by a distance of about 20 miles. The two investigations were set up as similar 5-year projects, therefore, and the work on them began in April, 1972. A 4-year measurement program, comprised of 2 years of measurements before the cooling systems began operation and 2 years afterward was planned. 1

2 Purpose and approach The goal of the Palisades cooling tower investigation is to determine (1) if the cooling towers will significantly affect meteorological conditions inland from the nuclear plant and (2) if so, how and to what extent several meteorological variables will be affected. Of major interest and concern is the possibility that the moist heated plumes emanating from the cooling towers and moving inland may, under certain conditions, increase atmospheric moisture enough to cause or enhance not only fog and/or icing at the surface, but also cloud growth and precipitation. An observational approach was taken, therefore, which was designed to provide basic information on possible effects on fog, precipitation, temperature, and atmospheric moisture. Because the nearest National Weather Service station which could provide adequate and representative information on these variables was at Muskegon County Airport located about 70 miles away, a network of 13 meteorological stations extending about 12 miles inland from the cooling towers was established to measure the variables listed above. The evaluation of meteorological effects will be based on (1) a statistical analysis of data obtained for 2 years prior to cooling tower operation and for 2 years during their operation and (2) case studies of plume effects. A description of the network stations, equipment and operations is given in the First Annual Progress Report (Ryznar and Baker, 1973). A map showing station locations is given in Figure 1.

SOUTH HAVE 196 GRAND JUNCTION f~~~~~~~~~~ [ PALISADES NUClE j>/ - ~ PLANT SITE BANGOR BANGOR 5 ~~~~~~~~~~~~~~~~~~~~~~~~~~~7 30 AVE. LAKE 10 MICHIGAN /// / / COVET 19) 12 44 13 1 2 3 4 5 MILES Figure 1. Locations of climatological stations

4 Temperature, relative humidity, and precipitation are measured at all stations. At two main stations (#3 and #7, called P03A and P07A in this report), wind velocity, visibility, and incident solar (direct plus diffuse) and total incoming (solar plus atmospheric) radiation are also measured. Dew point is measured at station #3. Station P03A is located in a flat field about 0.8 mile ESE of the cooling towers and near the Interstate 196 freeway. Measurements of visibility and the other variables at this station are especially significant in determining cooling tower effects on fog and freeway trafficability. Station P07A, because it is about 12 miles inland, is out of range of direct cooling tower effects and acts as a control station. Interpretation of work by others At the time of inception of the study, observational data on meteorological effects of mechanical-draft cooling towers, were not available nor had possible effects been assessed adequately by means of mathematical models. Huff, et al. (1971) and Ackerman (1971) had summarized most information available at that time. In addition, Koss and Altomare (1971) had evaluated possible environmental effects of cooling towers for the Palisades site. Within the last year, however, publications have appeared which describe observational studies of plume behavior and meteorological effects of plumes for mechanical draft towers. Several are listed below with an interpretation of some results of the studies for the Palisades cooling towers.

5 (1) Hanna, S. R. and S. G. Perry, 1973: Meteorological Effects of the Cooling Towers at the Oak Ridge Gaseous Diffusion Plant. Part I: Description of Source Parameters and Analysis of Plume Photographs and Hygrothermograph Records. NOAA Atmospheric Turbulence and Diffusion Laboratory, Oak Ridge, Tennessee. ATDL Contribution File No. 86, 40 pp. (2) Hanna, S. R., 1974: Meteorological Effects of the Cooling Towers at the Oak Ridge Gaseous Diffusion Plant. Part II: Predictions of Fog Occurrence and Drift Deposition. NOAA Atmospheric Turbulence and Diffusion Laboratory, Oak Ridge, Tennessee. ATDL Contribution File No. 88, 39 pp. (3) Hanna, S. R., 1974: Meteorological Effects of the Mechanical Draft Cooling Towers at the Oak Ridge Gaseous Diffusion Plant. NOAA Atmospheric Turbulence and Diffusion Laboratory, Oak Ridge, Tennessee. ATDL Contribution No. 89, 23 pp. (4) Meyer, J. H., 1974: Mechanical Draft Cooling Tower Visible Plume Measurement Program for Plume Modeling. The Johns Hopkins University Applied Physics Laboratory, 29 pp. (5) Eagles, T. W., and L. C. Kohlenstein, 1974: A Cooling Tower Visible Plume Prediction Model Based on Measurements. The Johns Hopkins Applied Physics Laboratory, 14 pp. The report by Hanna and Perry (1) is one of the first observational studies dealing with downwash effects of mechanical draft towers. Their observation program was conducted at the Oak Ridge Gaseous Diffusion Plant. Except for a larger drift rate and drift drop size, the design and operating characteristics of the Oak

6 Ridge cooling towers are similar to those for the cooling towers at the Palisades Nuclear Plant. Among the findings of lanna and Perry are the following: a. Downwash occurs about 50-60% of the time during the daytime in fall, winter, and spring. b. In summer, due to lighter winds, downwash occurs only about 30% of the time during the daytime. c. At wind speeds greater than about 3 or 4 meter/sec, downwash is likely to occur. d. When the wind direction is nearly parallel to the axis of the cooling towers, downwash will not occur even at wind speeds of 5 m/sec. These results can be applied to the cooling towers at the Palisades Nuclear Plant, which consist of two blocks of towers comprised of 18 cells each. The blocks are aligned in a westeast direction and are separated by about 200 yards. The orientation of the cooling towers is discussed below in relation to downwash characteristics, based on the annual wind rose for Muskegon and the findings of Ilanna and Perry. The annual wind rose for Muskegon compiled by Portman (1973) is given in Figure 2. Each line extends in the direction from which the wind blows, and its length is proportional to the percent of time that the wind is from that direction. In the figure, one inch is 4% of the time on an annual basis. The number at the end of each line is the average annual speed in meters/sec. If the 10% frequency of occurrence of a west wind and the 8% frequency of occurrence of an east wind (neither of which should cause downwash at wind speeds less than about 5 meters/sec) are

7 N N W N E 4 3 5 4 3 4 W- 5 4 3 6 6 4 SW SE S 1 inch = 4 % Figure 2. Annual wind rose for Muskegon County Airport. The numbers are wind speed in meters per second.

8 excluded, the wind rose shows that a component of the average wind direction is transverse to the long axis of the blocks of cooling towers about 80% of the time. The average wind speed for all directions, furthermore, is at least 3 meters/sec, with the highest average speeds (about 5 meters/sec) associated with onshore winds. Since the shoreline is oriented approximately 200-020 degrees, a wind direction between 210 clockwise through 010 degrees is considered to be an onshore wind. Onshore winds occur about 50% of the time (excluding a west wind). Based on these data and the results of Hanna and Perry, therefore, downwash conditions with onshore winds occur an appreciable amount of the time. Regarding fog occurrences, Hanna (3) calculated that about 300 additional hours of fog per year could occur at distances less than 3 km in the dominant downwind direction. His calculations were based on a heat dissipation rate of 1000 MW and a water vapor injection rate into the atmosphere of 4.6 x 105 gram/sec. As he points out, however, there are several reasons why the additional hours of fog are probably too high. The reasons are given verbatim from his text. First, because of plume lift-off, a plume which initially suffers downwash may rise to heights of several hundred meters. Consequently our assumption that effective plume rise is 10 meters during downwash conditions will lead to overestimates of ground fog occurrence. Second, when condensation occurs in a plume, latent heat is released and the plume follows the thermodynamic laws of cloud physics. Thus the Gaussian plume model, which assumes that the substance being modelled is inert, is no longer valid. On a typical day when the plume is visible for great distances, it looks like a long, narrow stratus cloud with a well-defined constant cloud base.

9 Third, we have never observed the visible plume at the ground at distances greater than 0.5 km from the towers. Hanna and Perry also found that cooling tower plumes initiated cloud development about 10% of the time. Neighboring plumes from the cells in a bank merged after a distance of about 50 meters. Plumes from the three banks of towers, which were initially separated by 100 meters, merged about 500 meters downwind. Total plume rise averaged about 200 meters. Hanna (reference 3) is studying precipitation records for the Oak Ridge area to determine if the presence of the cooling towers has significantly increased local precipitation. One example of an effect on precipitation is given in Culkowski (1962), who discusses a light snowfall which extended several kilometers downwind of the Oak Ridge towers. Reference (4) describes results of a study of the 8-cell cooling tower at the PEPCO Benning Road Generating Station, Washington, D. C. The study began in the Fall of 1973 and included photographic and visual observations of the visible plume, accompanied by measurements of meteorological variables at the surface and at higher altitudes. Aerial transects of the plume were made to determine its internal temperature and humidity structure. Eighty case studies were made between 25 October 1973 and 26 February 1974. Data obtained in the study described above were used in Reference (5) to modify existing models of predicting plume lengths and heights. Based on 50 case studies, the authors

10 found that 92% of the cases showed less than 50% difference between predicted and observed plume heights and 83% showed less than 50% difference between predicted and observed plume lengths. Observed plume heights ranged from 30 meters to 360 meters and plume lengths ranged from 25 meters to 1600 meters. In summary, based on a review of publications describing observations of meteorological effects of mechanical draft cooling tower plumes, it may be expected that the effects of the Palisades cooling towers will be more pronounced than those described above when the nuclear plant is operating at full capacity. Operating at full capacity implies that the 36 cells comprising the two blocks of cooling towers are in operation and that the nuclear plant is generating about 700 MW of electrical power. If a 32% efficiency is assumed, the cooling towers will have a heat dissipation requirement of about 1500 MW. At the time of Hanna'sexperiments at Oak Ridge, the total heat dissipation was about 250 MW which is a fraction of its normal value and a factor of about 6 less than that possible for Palisades. The PEPCO cooling towers were dissipating about 400 MW of heat during the observation program described above. It is known that the more heat that the cooling towers are required to dissipate, the more heated water is injected into the atmosphere. Hanna's calculation that for a heat dissipation rate of 1000 MW there are 4.6 x 105 grams/sec or about 7200 gal/min. injected compares well with the 400 MW of heat dissipated and 2400 gal/min. injected for the PEPCO towers. On this basis,

11 about 10,000 gal/min. will be injected by the Palisades towers at full capacity. The likelihood of a significantly greater meteorological impact than that observed in the publications reviewed is evident.

II. DATA COLLECTION The recording of meteorological data by the network of 13 stations during the past year was maintained on a continuous basis as much as possible. The percent possible data recorded for each month for the period 1 April 1973 through 31 March 1974 are given in Table 1. The percentages given for precipitation, temperature, and relative humidity are averages for the 13 stations. Those for solar radiation (direct plus diffuse), total incoming (solar plus atmosphere) radiation, wind direction, wind speed, and visibility are for each of the two main stations as indicated. The dew point system was installed at station PO3A on 13 February, 1974. The reasons for major periods of missing data are also given in the table. The comparatively low (45-60%) recoverability of temperature and humidity data for April, May, and June of 1973 resulted from several hygrothermographs being calibrated and rotated with field units during these three months. Data losses due to calibrations were eliminated by procurement of spare units. 12

13 Table 1. Percent possible data recorded for the Palisades network 1973 1974 Apr May Jun Jul Aug Sept Oct Nov Dec Jan Feb Mar Precip. 83 94 98 98 96 93 97 99 95 95 94 99 Temp. 62 46 49 97 86 86 94 91 80 85 91 97 Rel. Hum. 59 45 50 98 85 91 95 91 80 85 84 97 Solar Rad. PO3A 100 100 100 96 100 100 100 100 100 100 100 97 4 PO7A 0 64 85 100 100 100 100 100 97 11 62 88 Total Rad. PO3A 100 100 100 80 97 100 100 100 97 92 88 89 PO7A 100 100 91 47 95 95 99 100 100 24 61 85 Wind Dir. PO3A 83 100 100 92 100 100 100 74 44 100 97 97 1 5 PO7A 97 100 100 100 61 0 0 52 100 97 100 100 Wind Speed PO3A 82 100 100 91 100 100 92 100 60 80 89 81 1 P07A 58 100 100 100 82 0 21 93 95 100 100 100 2 3 2 Visibility PO3A 0 74 100 94 100 100 100 99 100 0 56 0 2 2 P07A 0 53 97 100 61 61 100 100 47 93 97 0 Dew Pt. P03A 52 55 1. Wind system decommissioned for wind tunnel calibration. 2. Visiometer reconditioned by manufacturer. 3. Defective high voltage power supply. 4. Pyranometer malfunction. 5. Transistor failure.

14 III. DATA TABULATION AND ANALYSIS Examples of tabulations and summaries In the past year, a substantial amount of network data were processed and tabulated. The tabulations, along with the other data, are being analyzed to determine meteorological conditions near the shoreline and their variations with distance inland which existed prior to the operation of the cooling towers. This work is being carried on for the network data obtained to date. A separate report containing data tabulations and an analysis of conditions for this phase of the study is in preparation and will be submitted at a later date. This section of the progress report presents and describes some examples of the tabulations, summaries and analyses being prepared. Several variables measured in January and February, 1973, which were the first two months of relatively complete network data, are discussed. Visibility. The visiometer data for station P03A for visibilities less than 3 km are given in Tables 2 & 3. The criterion used here to define an episode of low visibility is a visibility less than 3 km. As discussed below, this value represents the greatest visibility for which the visiometer data are reliable. The onset time of an episode shown in the table means that the visibility has remained greater than 3 km for at least 1/2 hour prior to decreasing to less than 3 km. The ending time means that the visibility has remained greater than 3 km for at least 1/2 hour.

Table 2. Visiometer data for station PO3A for visibilities less than 3 km Dates of Recording: 1-28 January 1973 Total Hours of Data: 672 Date < 3 km <1 km -< 0.5 km Obstruction Onset End Total Onset End Total Onset End Total Min. vis. to (EST) Hours (EST) Hours (EST) Hours (km) Vis. 3 Jan 1847 1940 0.20 None 2.3 Rain, fog 2108 2109 5 Jan 1812 1814 2035 2142 0.18 None 1.5 Snow 2324 2326 6 Jan 0616 0640 0.06 None 2.2 Snow 9 Jan 0656 0657 0.01 None 2.9 Snow and blowing snow 11 Jan 0817 0821 0947 0948 0.06 None 1.9 Snow and blowing snow 14 Jan 1942 1952 2126 2400 2.75 None 1.3 Fog 15 Jan 0000 0508 0617 0748 4.78 0.653 0723 0.50 0656 0718 0.37 0.2 Fog 2054 2116 18 Jan 1327 1329 1434 1505 0.14 None 2.0 Rain, fog 26 Jan 0702 0931 2.25 0739 0856 0.52 0740 0841 0.21 0.3 Fog Summary: Percent of recorded data Rain & fog fog snow <3 km 0.05 1.4 0.05 <1 km 0 0.15 0 <0.5 km 0 0.08 0

Table 3. Visiometer data for station POJA tor visiblities less than 3 km February, 1973 Dates of recording: 4-9 and 13-24 Total hours of data: 387 Date <3 km < 1 km <0.5 km Obstruction Onset End Total Onset End Total Onset End Total Min. vis. to (EST) Hours (EST) Hours (EST) Hours (km) Vis. 7 Feb 1116 1227 0.17 None 2.4 Snow, fog 8 Feb 0740 0742 0.04 None 2.3 Snow and blowing snow 9 Feb 0143 0218 0.11 None 1.9 Snow 14 Feb 0458 0700 1707 2116 4.27 None 2.0 Fog o% 15 Feb 0305 0336 0.51 None 1.3 Snow 20 Feb 0031 0438 3.79 None 1.4 Fog 21 Feb 0210 0705 4.88 0288 0558 0.49 0.6 Snow 22 Feb 1129 1251 1.16 None 1.4 Snow 23 Feb 0153 0351 1.96 None 2.5 Haze Summary: Percent of recorded data Fog Haze Snow & fog Snow < 3 km 2.1 0.5 0.04 1.7 < 1 km 0 0 0 0.12 <0.5 km 0 0 0 0

17 For each episode of low visibility the percentages of each hour that the visibility is (1) less than 3 km, (2) less than 1 km, and (3) less than 0.5 km are shown. If the obstruction to visibility is fog, a visibility less than about 0.5 km is classified by National Weather Service as heavy fog. The percentages are not computed if data are missing for all or part of a given hour. The number of hours with complete data for a given month are given at the top of the table and the percent of complete data in each visibility category is summarized at the bottom. For each episode shown, the minimum visibility and the obstruction to visibility are also given. Information on the type of obstruction is necessary because (1) it is likely that the cooling towers will exert a different effect on fog than on precipitation and (2) the visiometer does not give as accurate a measure of visibility in precipitation as it does in obstructions comprised of smaller-sized particles. A preliminary analysis of visibility data obtained during snow, for example, has shown occasional significant reductions in visibility but no measured precipitation associated with them. The reasons for this behavior, which occurs mainly during gusty winds, appear to be that (1) the visibility may indeed be reduced but sufficient snow has not entered the gage because of the gusty wind and/or the water equivalent of the snow did not exceed the 0.01 inch sensitivity of the gage or (2) the actual visibility is greater than that indicated by the visiometer, which may overrespond to snow. Either reason or a combination of both could be valid, and both are being studied.

18 The type of obstruction to visibility is determined on the basis of (1) hourly weather observations made at Benton IHarbor Airport between 0630 and 2030 each day and at Muskegon and South Bend on a 24-hour basis, (2) limited observations made by the manager of South Haven Airport, and (3) measurements of precipitation and other variables within the meteorological network. Wind speed and direction. For tabulation, digitized wind data are reduced to hourly averages of wind speed and direction. A finite-difference scheme is used which involves making linear interpolations between digitized values for every 3 minutes. These values are then averaged for each hour and tabulated. From the hourly averages, joint occurrences of wind direction and wind speed in assigned categories are counted. The categories for wind speed in miles per hour are: calm (less than 1 mph), 1-3, 4-7, 8-12, 13-18, 19+, and missing. Wind direction is defined by the direction from which the air is moving. It is tabulated in two ways: (1) by every 10 degrees and (2) by 16 standard compass points (N, NNE, NE, etc.), plus categories for calm and missing data. In addition, a category for a variable wind direction is included. A wind direction is classified "variable" if the range of wind directions during an hour exceeds 180 degrees. Joint frequencies for each month are determined by dividing the number of joint occurrences in each pair of categories by the total number of hours of data. From the joint frequency data, two wind roses for each main station are plotted, one using 10degree increments and one using compass points.

19 Wind roses for station PO3Aand P07Afor February 1973 are given in Figures 3 and 4. The wind roses shown are for each 10 degrees. North is toward the top of the figure. The length of each line is proportional to the percent of time that the wind was from that direction. One inch equals 2% of the time. The number at the end of each line is the average wind speed in miles per hour for that direction. The percent of time that the wind was calm, as well as the percent of time the wind direction was variable, are also indicated. The sum of the percentages for (1) the 36 directions, (2) the percentage of calms, and (3) the percentage of variable winds equals 100%. Precipitation. Daily and monthly totals of precipitation for all stations are given in Tables 4 and 5 for January and February, 1973. For these two months, twenty days, or 34%, had measurable precipitation. Variations in precipitation among the stations are evident both on a daily and monthly basis, but an isopleth analysis did not reveal a significant trend with distance inland. Some data were lost in early January because a severe snowstorm prevented the observer from servicing the stations for several days. Temperature. Daily maximum and minimum temperatures for January and February, 1973, are given in Tables 6 and 7. The reason for the missing data early in February is that initial delivery of the hygrothermographs was on 2 February and they were compared and adjusted prior to installation. Two units on loan provided the data for stations PO3A and PO7A, given in Table 6

20 N NW NE 7.2 6.4 SW SE s. Calm 4.96% 5.1 5.0 7.1 5.755.9 Calm. 4.6% Based on 552 hours of data Figure 3. Wind rose for station P03A for February, 1973. The numbers are wind speed in miles per hour.

21 N NW NE 9.9 6.6 9.1. 7.7 7.1 5.1 as 8 6. 4al 85.8 L7 1 io.1 V4.1 Figure 4 Wind rose for station P07A for February, 1973. 5The 6numbers are wind 4.0speed in miles per hour. 60) 5.2 5.8 o50 SW S E S Calm 8.8% Variable 0.1% 1 inch = 2% Based on 624 hours of data Figure 4. Wind rose for station PO7A for February, 1973. The numbers are wind speed in miles per hour.

Table 4.Daily totals of precipitation for January 1973 for the Palisades network DAY PO1A PO2A PO3A PO4A PO5A PO6A PO7A PO8A PO9A P1OA P11A P12A P13A 1 0 0 0 0 0 0 0 0 0 0 0 0 0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 3 0.66 0.64 0.65 0.61 0.64 0.65 0.55 0.60e 0.65 0.69 0.70 0.63 0.63 4 * * * 0 0 0 0 0 5 * * * * * * * ** - * - 0 6 * * * * * * * * * - * - 0 6 7 * * * * * * * * * - * - 0 8 * * * * * * * * * - * - 0 9 (0.10) (0.10) (0.06) (0.06) (0.03) (0.03) (0.04) (0.03) (0.03) - (0.07) - 0 10 0 0 0 0 0 0 0 0 0 0 0 - O 11 0 0 0 0 0 0 0 0 0 0 0 - 0 12 0 0 0 0 0 0 0 0 0 0 0 - 13 0 0 0 0 0 0 0 0 0 0 0 - 0 14 0.06 0.05 0.05 0.06 0.04 0.06 0.06 0.06 0.06 0.05 0.04 - 0.03 15 0.01 0.06 0.05 0.04 0.02 0.01 0.06 0.05 0.02 0 0.03 - 0.04 16 0 0 0 0 0 0 0 0 0 0 0 - 0 17 0 0 0 0 0 0 0 0 0 0 0 - 0 18 0.21 0.26 0.29 0.26 0.33 0.30 0.50 0.27 0.29 0.28 0.28 - 0.42 N 19 0.06 0.06 0.08 0.09 0.17 0.13 0.08 0.05 0.08 0.10 0.12 - 0.12 20 0 0 0 0 0 0 0 0 0 0 0 - 0 21 0.21 0.16 0.17 0.14 0.15 0.13 0.13 0.16 0.16 0.19 0.15 - 0.10 22 0.17 0.23 0.16 0.15 0.23 0.21 0.20 0.18 0.18 0.21 0.20 - 0.20 23 0 0 0 0 0 0 0 0 0 0 0 - 0.06 24 0 0 0 0 0 0 0 0 0 0 0 - 0 25 0 0 0 0 0 0 0 0 0 0 0 - 0 26 0 0 0 0 0 0 0 0 0 0 0 - 0 27 0 0 0 0 0 0 0 0 0 0 0 - 0 28 0 0 0 0 0 0 0 0 0 0 0 - 0 29 0 0 0 0 0 0 0 0 0 0 0 - 0 30 0 0 0 0 0 0 0 0 0 0 0 - 0 31 0 0 0 0 0 0 0 0 0 0 0 0 0 1.48 1.56 1.51 1.41 1.61 1.52 1.62 1.40e 1.47 1.55e 1.59 - 1.60 e estimated missing data ( ) The number in parenthesis is the total of the amounts for days marked with asterisks.

Table 5. Daily totals of precipitation for February 1973 for the Palisades network DAY PO1A P02A P03A P04A P05A P06A P07A P08A P09A P1OA PllA P12A P13A 1 0.15 0.13 0.14 0.12 * 0.10 0.11 0.13 0.11 0.15 0.10 0.11 0.07 2 0.21 0.24 0.24 0.23 * 0.26 0.29 0.24 0.24 0.28 0.25 0.25 0.29 3 0 0 0 0 * 0 0 0 0 0 0 0 0 4 0 0 0 0 * 0 0 0 0 0 0 0 0 5 0 0 0 0 (0.35) 0 0 0 0 0 0 0 0 6 0 0 0 0 0 0 0 0 0 0 0 0 0 7 0.08 0.09 0.04 0.05 0.05 0.06 0.05 0.06 0.03 0.07 0.04 0.05e 0.03 8 0.09 0.12 0.10 0.08 0.10 0.13 0.05 0.09 0.08 0.09 0.09 0.08 0.10 9 0.03 0.02 0.04 0.03 0.02 0.01 0.01 0.01 0.01 0.03 0.03 0.03 0.02 10 0 0 0 0 0 0 0 0 0 0 0 0 0 11 0 0 0 0 0 0 0 0 0 0 0 0 0 12 0 0 0 0 0 0 0 0 0 0 0 0 0 13 0 0 0 0 *0 0 0 0 0 0 0 0 14 0.03 0.04 0.04 0.04 * 0.06 0.08 0.04 0.03 0.03 0.03 0.03 0.05 15 0.12 0.13 0.08 0.10 * 0.09 0.10 0.11 0.06 0.11 0.09 0.07 0.06 16 0 0 0 0 * 0 0 0 0 0 0 0 0 17 0 0 0 0 * 0 0 0 0 0 0 0 0 18 0 0 0 0 * 0 0 0 0 0 0 0 0 19 0.02 0.02 0.02 0.02 * 0 0 0.02 0.03 0.01 0 0 0.01 20 0.08 0.11 0.08 0.08 (0.22) 0.06 0.07 0.09 0.10 0.09 0.10 0.05 0.05 21 0.27 0.26 0.24 0.26 0.23 0.25 0.23 0.26 0.22 0.26 0.25 0.23 0.23 22 0.02 0.03 0 0.01 0.02e 0.01 0 0.02 0.03 0.02 0.02 0.01 0 23 0 0 0 0 0 0 0 0 0 0 0 0 0 24 0 0 0 0 0 0 0 0 0 0 0 0 0 25 0.05 0.05 0.03 0.03 0.02 0.02 0.04 0.05 0.05 0.04 0.04 0.06 0.02 26 0 0 0 0 0 0 0 0 0 0 0 0 0 27 0 0 0 0 0 0 0 0 0 0 0 0 0 28 0 0 0 0 0 0 0 0 0 0 0 0 0 1.15 1.24 1.05 1.05 1.01 1.05 1.02 1.12 0.99 1.18 1.04 0.97e 0.93 e estimated missing data ( ) The number in parenthesis is the total of the amounts for the preceding days marked with asterisks.

Table 6. Daily Maximum and Minimum Temperatures (00-2400EST) for January 1973 for the Two Palisades Main Stations Maximum Minimum Day PO3A PO7A PO3A PO7A 1 29 M M M 2 29 M 15 M 3 M M 18 M 4 M M M M 5 M M M M 6 26 M 11 M 7 24 M 10 M 8 25 M 10 M 9 M M M M 10 19 17 6 5 11 19 19 16 16 12 22 21 18 9 13 32 32 18 19 14 30 31 28 28 15 30 M 22 M 16 43 M 32 M 17 51 49 40 40 18 52 51 45 45 19 48 48 28 28 20 30 29 20 22 21 34 34 20 22 22 42 M 32 M 23 M M M M 24 36 36 22 20 25 46 51 26 28 26 51 54 28 32 27 45 48 32 33 28 34 33 24 23 29 27 M 14 M 30 30 M 18 M 31 35 35 20 23 Ave. 36.5 36.8 24.4 24.6 Extreme 52 54 6 5 Ave. Average for days with data for both stations

Table 7. Daily maximum temperatures (00-2400 EST) for February 1973 Day P01A P02A P03A P06A P07A P09AP12A 1 M M 44 M 44 MM 2 M M 45 M 44 MM 3 M M 35 M 34 MM 4 M M 44 M 48 MM 5 M M M M M MM 6 M M M M M MM 7 35 35 35 35 34 3535 8 24 23 24 24 22 2223 9 25 24 24 23 23 2323 10 26 28 28 29 27 2727 11 26 28 26 28 27 2627 12 33 35 34 35 34 3334 13 43 43 42 42 41 4241 14 37 36 36 36 36 3535 15 34 33 33 33 31 3233 16 11 13 12 11 12 1011 17 22 24 22 22 21 2220 18 32 31 31 32 30 3030 19 35 35 35 35 35 3535 20 38 36 36 37 37 3536 21 33 32 32 31 31 3131 22 32 32 32 30 30 3131 23 38 37 36 36 35 3536 24 33 32 32 33 32 3232 25 31 29 30 30 30 2929 26 32 31 31 31 30 3130 27 35 27 38 36 35 3536 28 45 45 45 45 44 4544 Ave: 31.8 31.8 31.5 31.5 30.8 30.7 30.9 Ave. Average for days with data for all stations.

Table 7. (cont) Daily minimum temperatures (00-2400 EST)for February 1973 Day POlA PO2A PO3A PO6A PO7A PO9A P12A 1 M M 35 M 35 M M 2 M M 32 M 32 M M 3 M M 30 M 31 M M 4 M M 30 M 31 M M 5 M M M M M M M 6 M M M M M M M 7 24 22 22 22 22 2222 8 17 16 16 15 15 15 15 9 15 14 13 14 15 13 13 10 10 14 13 12 13 13 10 11 6 8 9 8 10 8 9 12 14 14 14 14 14 13 14 13 20 19 20 18 19 18 18 14 34 33 33 31 33 30 31 15 10 9 9 8 8 9 11 16 -6 -4 -9 -9 -6 -7 -2 17 -10 -10 -13 -11 -9 -9 -10 18 10 12 13 12 7 12 12 19 32 31 31 31 30 30 30 20 32 31 31 31 30 30 31 21 23 22 21 21 19 21 21 22 21 20 20 19 18 M 20 23 22 24 20 18 18 20 22 24 22 24 20 18 18 20 22 25 25 23 24 24 23 23 23 26 21 20 20 20 20 19 19 27 10 9 9 9 8 9 M 28 12 15 13 16 17 15 19 Ave: 16.7 16.9 16.0 15.7 15.8 15.8 16.5 Ave. Average for days with data for all stations

27 The average temperatures given in these summaries are obtained only for those days where data were available for all stations. Such an average allows comparisons among stations without the averages being biased by unequal samples. The data show that in January, there is no evidence of any significant change in temperature across the network. In February, however, an average decrease of about 1~ F from near the Nuclear Plant to about 12 miles inland is evident in the maximum and minimum temperatures. Natural variability Fog. Table 8 shows the average percent of time natural fog may be expected to occur each month near the Palisades site. Table 8. Average percent of total hours with natural fog by month expected for the Palisades area J F M A M J J A S 0 N D 15 10 8 8 9 7 7 9 5 9 7 15 The percentages are from an analysis of the number of hours with fog observed at the National WeatherService station at Muskegon County Airport located about 70 miles north of Palisades and about 3.5 miles inland (Koss and Altomare, 1971). It is the nearest station with hourly observations on a 24-hour basis whose long-term visibility data are complete and considered to be representative of average conditions a few miles inland from the cooling towers.

28 It is evident that fog occurs most often in January and February, each of which has fog about 15% of the time, and least often in September, which has fog about 5% of the time. If the data are further divided into winter months of December, January, and February; spring months of March, April, and May; summer months of June, July, and August; and fall months of September, October, and November; fog occurs about 14% of the time in winter, 8% in spring, 8% in summer, and 7% in the fall. Except for summer, these frequencies compare well with those observed for Midway Airport in Chicago in spite of differences in location, distance with respect to Lake Michigan and other factors expected to cause differences in fog frequencies. Chicago has fog 14% of the total hours in winter, 8% in spring, 4% in summer, and 7% in fall (Huff and Vogel, 1973). It is likely that an urban heat island effect is primarily responsible for the lower frequencies of fog in summer for Chicago. Temperature and precipitation. Information on the natural variability of temperature and precipitation across the inland extent of the network was obtained from climatological records for South Haven, located on the Lake Michigan shoreline, and Bloomingdale, located 16 miles inland and about 8 miles northeast of station PO7A at Bangor. Climatological summaries for the period 1940-1969 were used (Climate of Michigan, 1971). Monthly means of temperature and precipitation for two stations are shown in Table 9. Differences in temperature and precipitation between the stations are also shown.

Table 9. Monthly means of temperature and precipitation for South Haven and Bloomingdale, Michigan Temp. F J F M A M J J A S 0 N D (1)South Haven 26.3 27.7 35.2 46.6 56.0 66.4 70.7 69.8 63.9 54.3 41.7 30.7 (2)Bloomingdale 24.8 26.6 35.1 48.1 58.3 68.1 71.8 70.3 63.1 53.0 39.9 29.1 diff.(1)-(2) +1.5 +1.1 +0.1 -1.5 -2.3 -1.7 -1.1 -0.5 +0.8 +1.3 +1.8 +1.6 Precip. (in.) (1)South Haven 2.07 1.66 2.40 3.36 3.18 3.52 2.95 2.99 3.05 3.07 1.65 2.29 (2)Bloomingdale 2.71 2.16 2.88 3.56 3.41 3.83 2.61 3.06 3.05 2.95 3.23 2.84 diff. (1)-(2) -0.64 -0.50 -0.48 -0.20 -0.23 -0.31 +0.34 -0.07 0 +0.12 -1.58 -0.55

30 The average temperature decreases with distance inland for the months between September and March and increases inland between April and August. The maximum decrease (about 1.6~F) occurs from November through January and the maximum increase (about 2~F) occurs in May and June. Both results can be explained by the fact that Lake Michigan lags in cooling in late fall and early winter and thereby is warmer than the land and then lags in heating in late spring and early summer and thereby is colder than the land. Unlike the seasonal variations of temperature differences, no pronounced seasonal variations were found for precipitation differences for the two stations. Except for July and October, average monthly precipitation at South Haven is consistently lower than at Bloomingdale. A similar analysis performed for Benton Harbor, located near the Lake Michigan shoreline, and Eau Claire, located 14 miles inland, showed greater precipitation at Benton Harbor from November through March and less from April through October. The latter result is more in keeping with what would be expected, since as air moves from over a cold (lake) to a warm (land) surface convective activity generally increases with distance downwind, which in this case is inland, on the average. The wintertime result implies that for onshore winds, vertical motions are enhanced by (1) thermal instability (cold air over warm water) and (2) greater surface roughness along the shoreline and act in concert with air near saturation to produce more snowfall near the shoreline.

31 Variations of precipitation with season and wind direction. One of the statistical analyses to be made in studying meteorological effects of the cooling towers involves a comparison of effects with onshore and offshore winds. To determine if differences in precipitation occur in onshore and offshore flow, a study was made using data for Muskegon. Hourly precipitation totals and wind observations were obtained for the 4-year period from 1960 through 1963. The limited extent of this period was determined by the availability of a homogeneous data set on magnetic tapes from the National Climatic Center. A total of 35,064 hourly observations were used, a preliminary analysis and interpretation of which are given below. The graphical output of the analysis is shown in Figure 5. Wind roses are given in the first row. The number at the end of each direction is the frequency, not the mean speed as given in other wind roses shown in this report. In the upper left box, directions for onshore, offshore, and along shore winds are shown for reference. The data were further categorized in terms of the four seasons as shown, with winter on the left and fall on the right. The wind roses show that for all seasons, onshore winds occur more frequently than offshore winds. Changes in frequencies are evident from one season to the next. For example, the prevailing wind direction in winter is WNW, while in the other seasons it is SSW. Winds from NNE and SSE occur least frequently in all seasons.

32 WIND Rost /St.. 0 6 I6 II 6 C ZI 5,. 7'1 / s1 ~~~~10 ~~6 7 77 2 POlABILITY 0 2.01404 P~ICIIATI3N WATIR F A G VN WID DAR NCT I INCHIS 0I 6 3 5, 7 LAND-I O R 2 4 W Figure 5. Wind roses and Roprecipitation wind roses forou Muskegon by season. The number at the end of 13 6I~~~~~~~~~ 6 5n 3 I uv 7 ~ —--— ~ —-77 13 s frequency of occurrence in 12 2 2 3 4 4 7 C I ~3 2 Cat2e 3 FOI A GivIT WIND Da0ICTO u 7 9 7 - 7 it \ I / 12 4o 33 33 3 6 2 7 9 9 2 P A 6S r~j\\ 2 IP 3 /13 5 4 DI STRIIB ION OF HoumLY PIECIPITAT2ON ToTAL6 IN ONSNOIa AND OnnsNoms WINDS 33- 33 40 - /' 2.o.o.3RO B I L T 0.3.32.3.40 a..20.3 C.P0 T. f O.3 Figure 5. Wind roses and precipitation wind roses for Muskegon by season. The number at the end of each direction is frequency of occurrence in percent.

33 The second row in Figure 5 gives precipitation roses, which are wind roses for a subset of the above data consisting of the number of hours, converted to frequency in percent, when 0.01 inch or more of precipitation was recorded. If precipitation and wind direction were completely uncorrelated, the precipitation and wind roses would be quite similar. Distinct differences are apparent in the patterns shown, however, indicating that precipitation is correlated with the wind direction. In winter there are maxima for the W and WNW directions which are probably associated with frequent polar outbreaks accompanied by snow to the lee of Lake Michigan. The high frequency for east winds could be due to precipitation with low pressure systems passing the south of Muskegon. Like the wind rose, in winter a frequency minimum is generally apparent for the NNE and SSE directions. In spring, the maximum associated with a west wind in winter disappears and is replaced by a maximum with an ENE wind. In summer, SW and SSW are preferred directions for precipitation. The precipitation roses discussed above are useful for studying the natural frequency of precipitation for winds from various directions. They are inadequate, however, for determining changes from the natural frequency of precipitation. For instance, a wind direction which has a low frequency of occurrence in the wind rose usually has a low frequency in the precipitation rose also. One technique for examining the changes is to calculate the probability that precipitation will occur, given a particular wind direction.

34 Probabilities of precipitation for various wind directions are given in the third row in Figure 5. They show a marked change with season,with winter having the highest probability for all directions than the other seasons. There is a distinct mimimum for the south quadrant, probably resulting from the fact that low pressure systems which move to the west of Muskegon generally do not produce much precipitation at Muskegon in winter. It is usually not until a cold frontal passage associated with the systems occurs that snow begins, which could account for the maximum in the NW quadrant. In summer the probability of precipitation is about one-half to one-third that of winter for a particular direction, reflecting the brief convective nature of summertime rains. In spring, the probability of precipitation for an onshore wind is much lower than that for an offshore wind. The difference results from the stabilizing effect of Lake Michigan in spring - an effect that was not evident from the precipitation roses. In summer, the probabilities are more evenly distributed. In fall there is a distinct variation in probability with direction, although this variation is apparently not related to whether or not the wind is onshore or offshore. To determine if the onshore or offshore winds have a pronounced effect on the intensity of precipitation, hourly precipitation data were divided into precipitation categories of 0.05 inch (except.04 in the first box) and are shown in the fourth row. They are further divided into onshore and offshore winds. The totals per hour in winter are much smaller than in summer, for which there is a

35 distinct skewness of the distribution to the right. Significantly, there does not appear to be a difference for onshore and offshore directions.

IV. DATA PROCESSING The University of Michigan 360/67 digital computer is used extensively to process the 96 items recorded by both meteorological networks. The conversion from analog format on strip charts to digital format on magnetic tapes is accomplished with a Graf/pen digitizer and a complementary set of computer programs. Figure 6 gives a flow diagram showing how the data from both meteorological networks are handled. The blocks at the top of the diagram show the 9 variables recorded on analog strip charts. The number of stations per data type varies from 2 for dew point to 26 for temperature, relative humidity and precipitation. The variable recorded, the number of stations for both the Palisades and Cook networks, and the interval between chart changes are given in each block. As indicated in Figure 6, when the strip chart records are received from the field, (1) they are screened for any obvious errors due to instrument malfunction, (2) dates and times are entered, (3) the analog recordings are digitized, and (4) the digitized data are read into disk files of the University of Michigan Computer Facility. Frequent checks of these data are made by reproducing the original data on a cathode-ray tube display scope. The processed data are then merged onto a magnetic tape. Steps in this procedure are discussed below and examples of processed data are presented. 36

Variable Precip. Temp. Humidity ew Point Visibility Sol. Rad. Tot. Rad. Wind Dir. Wind Speed # stations 26 26 26 2 3 44 4 chart changed once/week once/week once/week once/40 days once/40 days once/mo. once/mo.. once/2 wk once/2 wk. COMPUTER DATA REDUCTION PROGRAM COMPUTER DATA REDUCTION PROGRAM NUMBER 1 NUMBER 2 1. check for missing data 1. check for missing data 2. convert to proper units 2. convert to proper units 3. correct on-off times 3. correct on-off times 4. make skewness correction 4. make calibration corrections based 5. make calibration corrections on linear interpolations from 6. interpolate to midnight and 2 calibrations on either side of create new data point data points 7. separate data into one day 5. interpolate to midnight and create data lengths new data point 6. separate data into one day data lengths PLOCOMPUTERT SELECTED DATAEDUCTION PROGRAM COMPUTER DATA REDUCTION PROGRAM correct or eliminate make fina check of orrect or eliminate data data data correct-or eliminate^ ______ \, make final check ofcorrect orrect or eliminate data data_____}____\ __ data / MERGE ALL DATA ONTO FINAL MAGNETIC TAPE Figure 6. Flow diagram for meteorological data handling.

38 Digitization Figure 7 shows the digitizer and peripheral equipment. The main component is a Graf/Pen GP-2 Sonic digitizer manufactured by Science Accessories Corporation. The digitizer measures the interval of time it takes sound to travel from a spark on the pen shown in Figure 7 to two microphone sensors separated by 90 degrees. The time interval is then converted electronically to X and Y distances with an accuracy of 0.01" and punched onto paper tape using a FACIT 4070 Tape Punch. The sensor portion of the tablet has what is called a menu at the top. This is a portion of the sensing area which is reserved for special instructions. By placing the pen on a particular pre-assigned point, the person digitizing is able to (1) identify the station, (2) indicate the variable, (3) give on-off times, (4) indicate end of data, (5) indicate mispunched data, and (6) indicate missing data. Rules have been established for digitizing the various types of data. For most data, the greater the variability of the parameter recorded, the more the number of values digitized. A greater number of values of solar radiation, for example, are obtained on a partly cloudy day than on a clear or uniformly overcast day. In addition, if there are sudden excursions in the trace of a recorded variable, a decision must be made to determine if they are real or if they are caused by equipment problems. An example is those which occur occasionally in the visiometer data during precipitation, in which case they would be expected, and those which occur in the absence of precipitation, in which case they

i na Lin~~~~~~~~~~~~~~~aear Tap ~~~~~~~~~~~~~.==.::....I Punch~~~~&s: ~~~~~~~~~~~~~~2~~~~~~~~~~~~1;k~~~~~~~~~~~~...........,;~':~%~ Data' Figure 7. Graf/pen Model GP-2 Sonic Digitizing System.

40 are due to electrical noise and are not digitized. In addition to these and other rules which are applied in the digitizing process, the fact that the meteorological variables are recorded on a different type of strip charts requires different ways of handling the data. For example, precipitation is recorded on a curvilinear chart with arcs for the time scale concave to the right. Temperature and relative humidity are recorded together on a double curvilinear chart, also with the time arcs concave to the right. Wind speed and direction charts are curvilinear with the time arcs concave to the left. Radiation and visibility charts are rectilinear, with the radiation chart 6 inches wide and the visibility chart 12 inches wide. Data package Because the strip charts which contain a week of recordings of precipitation, temperature, and relative humidity are less than 12 inches long, and those which contain recordings of radiation, wind and visibility for longer time periods are 100 feet long, two different computer programs are needed for data reduction. In Figure 6 they are shown as "Program #1" for the former charts and "Program #2" for the latter. Both programs accomplish similar tasks; only the details are different. Digitized data are read into the computer in a sequence called here a data package. A package is defined as a body of digitized data which is handled throughout the computerized operation as a single compact data file. Each data package is initialized by a prologue which identifies it uniquely. The prologue provides the station name and data type, the starting date and time and other

41 information which establishes a proper coordinate system. The computer programs are designed to be as general as possible. For example, a package can begin on any date and at any time of day and, in most cases, can end on any later date and time of day. Diagnostic routines have been implemented to correct for chart interruptions caused by power failures, clock stoppages, time adjustments, chart changes, and calibrations. The computer programs are designed, furthermore, so that for unique or unusual eventswhich cannot be anticipated, human intervention and manual correction of the data are possible. In addition, comparisons of the computer output with data processed manually are made occasionally. Final storage format The processed data are stored on magnetic tape in a series of files, each of which contains one month of data. Within each file, the data are ordered by day in a pre-assigned sequence. All digitized data points for one item, one station and one day (e.g., precipitation on January 3, 1973, at CO1A) are stored on one record. Since there are nearly 100 items of data, a file consists of about 3000 records. A record can be variable in length. The organization of the first 15 4-byte words of the record establishes its identity and is given below: Word No. Item 1 year 2 month 3 day 4 data type

42 Word No. Item 5-8 station call letters 9 priority number 10 creation date 11-14 storage of special data 15 data accuracy From the 16th word on, the data are stored in 4-byte word pairs. The first word gives the time and the second gives the value of the variable. Where necessary, values of the times 0000 EST and 2400 EST are interpolated from the given data and treated like other data points. Example of input data In an example given below, visibility data are used to illustrate the data reduction process. Data are followed from the recording stage to their final presentation. Visibility is recorded at stations PO3A and PO7A of the Palisades network, and station number CO3A of the Cook network. The visibility measuring system is a MRI visiometer and recorder. The voltage output from the visiometer is recorded on a MRI 0-5 volt strip chart recorder at a chart speed of one inch per hour. A decreasing visibility produces an increasing voltage. A visibility data package must be accompanied by a calibration file before it can be completely processed. The file consists of (1) a calibration voltage for the visiometer as input to the recorder and (2) a recorder calibration. Recorder calibration factors are determined by the number of inches the recorder pen moves up scale in response to voltages up to 5 volts applied in 1-volt increments. Linearity is assumed only for the 1-volt increments and, as a result, each digitized value is calibrated

43 for pen movements upscale corresponding to 0, 1, 2, 3, 4 and 5 volts, respectively. There is a system calibration for the beginning and end of the package as well as any which is made between the beginning and ending dates of the package. The computer program applies the calibration data and converts the digitized data to visibility values. It accomplishes this by taking the data from the calibration file and setting and adjusting the recorder and sensor calibration parameters one day at a time by interpolating between the calibrations nearest to that day. Voltage is converted to visibility by the equation: Vis(km) = 0.39/voltage The data are questionable for voltages less than about 0.1 volt, which corresponds to a visibility of at least 4 km. Any calculated visibility greater than 5 kilometers, therefore, is set equal to 5 kilometers. Final data plot An example showing data for station PO3A for the period 28 June - 2 July 1973 which have been plotted by the CALCOMP 780/763 digital plotter is given in Figure 8. Time is listed across the bottom and increases from left to right. Vertical lines are given for every 6 hours. Plots of the variables are arranged vertically. From top to bottom, the variables are visibility, in kilometers; total incoming radiation (solar plus atmospheric), in

44 - -';Visibility (Km) - Moximn i Posible issibility m iorT - 1~ Minimuln Possible Visibility from Visiom r F Fo R & Fog F. /.I S\,klscbingl Solar Radiation (Ly/ a Unat |. T Incoming Atmospheric Radidtion (LyMin) o I n c o ming" A t mpher7c tRWind Direction (Deg 9) Fi g u I t l I ind if r (Mtion Trmaper tur. (F) Dew I oint (F) MISSING DRTF |'I~~~~~~~ ~Relative H midity (%). MISSIrG DRTF I A cccumulted Precipitation (Inches) 00-56 12 18 0 6 12 18 06 12 18 00 6 12 18 00 29 June 30 June Time (EST) July 2 July Figure 8. CALCOMP plot of variables for station P03A.

45 langley*/min; solar radiation (direct and diffuse), in langleys/ min; atmospheric radiation (total incoming minus solar), in langleys/min; wind direction, in degrees clockwise from north; wind speed, in miles/hour; temperature and dew point, in degrees F; relative humidity, in percent; and precipitation, in inches. In the plot of solar radiation, the upper curve is the computed solar radiation arriving at the top of the earth's atmosphere and the lower curve is the measured values. The values of dew point shown are calculated from temperature and humidity data. The value of such a plot is that it gives a visual display of all the variables measured and thereby allows their interdependence to be examined. The data from a main station are not plotted in this way on a continuous basis, but occasionally for monitoring the output data and for case studies, such as those involving fog episodes. It is expected that a significantly greater amount of data will be plotted in this way once the cooling towers begin operation. * A langley is equivalent to one gram-calorie per square centimeter.

46 Accuracy of the analog to digital conversion Since all analog data collected in the field are converted to digital data, it is important that the accuracy of this conversion be established. The accuracy depends upon the precision of the Graf/pen digitizer, where the precision is the repeatability of a measurement. The computer programs are designed so that they depend not upon the absolute distance measurement of the apparatus but upon the distance measured relative to a reference distance determined by placing the pen on preassigned points on the chart being digitized. A statistical routine was developed to establish the inherent precision of the Graf/pen digitizer in relation to the x, y position on the tablet, the age of the digitizer and the person digitizing. A test pattern of 36 points on the 14" by 14" grid on the digitizing tablet is digitized ten times. A computer program was written to calculate the average location of each of the 36 points, the mean distance between adjacent points, the standard deviation S of each distance, and the standard deviation S of each mean distance computed from the repeated digitization of the grid points. Figures 9 and 10 show the grids with a reduced scale. The standard deviation S is the best statistical estimate of the inherent precision of the digitizing process. It is defined by Equation 1. 21/2 Z (xi - x) s= ( ) 1) n - 1

47 a. x-direction Col 1 Col 3 Col 5 + 3.984 + +3.977 + + 3.984 + +3.980 + + 3.972 + +-3.979 + + 3.980 + +3.975 + +3.984 + +3.983 tt 3.978 -+ -3.982 t + 3.981 + + 3.975 + +3.984 f +3.979 + t3.972 + +3.982 + Ave. 3.981 3.975 3.983 b. y-direction 4 4- +4 ~+ + - Ave. Row 3.980 3.984 1 3.977 3.975 3.975 3.983 3.979 + -+ + t+ + + + + Row 3.974 3.975 3 3.978 3.972 3.975 3.981 3.976 +- + + + + + 4- + + +- - Row 3.984 3.990 5 3.987 3.983 3.992 3.987 3.987 + + + +- + Average for grid is 3.980 Scale 012345 inches Figure 9. Spatial distribution of time-averaged distances for 7 runs with digitizer.

48 a. x-direction Col 1 Col 3 Col 5 +.0166 + +.0190 + +.0187 + +.0185 + +.0152 + +.0161 + +.0177 + +.0157 + +.0172 + +.0165 + +.0158 + +.0168 + +.0194 + *.0184 + +.0160 + +.0195 + +.0175 + +.0154 + Ave..0180.0169.0167 b. y-direction ++-+ + + + Ave Ave Row.0210.0176 1.0196.0212.0247.0186.0204 + + + + + + + + + + + Row.0222.0184 3.0201.0230.0257.0192.0214 + + + + + + + + + + + + Row.0213.0189 5.0175.0234.0252.0187.0225 + 4 + + - + Average for grid =.0193 Scale 012345 Figure 10. Spatial distribution of time averaged standard deviation for 7 runs with digitizer.

49 where n = sample size xi = elements of sample - 1 x = sample mean = - Ex. n 1 The accuracy of a sample mean increases with sample size and with the inherent accuracy of a process. In this case, the standard deviation Sn of the mean distance between any two points, based on the given sample, is referred to as the adjusted standard error. It is defined by Equation 2. S - 2) n /n The dates when the 36-point test grid was digitized are shown in Table10. Run #5 revealed that left-handed individuals could not use the digitizer in its present arrangement, since the position of the left hand holding the pen changed the travel time of the sound from the pen to the transducers (labelled "linear microphones" in Figure 7 ). For accurate, reproducible results, all digitizing is now done with the pen held in the right hand. Results of the use of the test grid are also given in Table 10, which gives the spatial average of all independent distances shown in Figures 9 and 10and S during each use of the test pattern. Although the distances in the x and y directions were measured by a ruler to be 4.00 inches, the distance measured by the digitizer was 3.98 inches. Since data processing programs are designed to be independent of actual distance measurements, the difference between a ruler measurement and a digitizer measurement is unimportant.

TABLE 10 Average distance and standard deviation, S, of the data points in inches X Y Run $ Date Col 1 Col 3 Col 5 S Row 1 Row 3 Row 5 S 1 5/2/73 3.972 3.971 3.973.0167 3.962 3.967 3.977.0182 2A 4/17/73 3.977 3.977 3.976.0130 3.978 3.974 3.977.0141 2B 5/17/73 3.977 3.975 3.976.0116 3.973 3.976 3.979.0109 3 11/1/73 3.988 3.976 3.983.0276 3.981 3.946 3.983.0291 4 3/2/74 3.989 3.981 3.991.0112 3.982 3.986 3.989.0121 6 4/29/74 3.977 3.978 3.994.0186 3.987 3.985 4.012.0190 o 7 4/30/74 3.989 3.967 3.986.0219 3.990 3.991 3.992.0361 L A Average.0172.0199

51 The standard deviation S of the various distance measurements is slightly different for each separate use of the test grid, with the values in Table 10ranging from 0.011 inches to 0.036 inches. The variations observed appear to depend more on the person using the digitizer than the apparatus itself. The average S for the x direction (corresponding to time on meteorological charts) is 0.0172 inches, with the average S in the y direction (corresponding to the magnitude of a variable) of 0.0199 inches. Thus, assuming a normal distribution, 67% of all data digitized are within these limits. Table 11gives the equivalent accuracy of a 0.02-inch variation in the digitized data in terms of meteorological units. Except for precipitation, the accuracy of the analog to digital data conversion is better than the accuracy of the meteorological instruments as given by the manufacturers. For precipitation, the accuracy of the digitization process is less than the measurement accuracy claimed by the manufacturer. However, considering the overall accuracy of the precipitation gage and its shortcomings in measuring precipitation during snow and/or strong winds, the accuracy of the digitization is sufficiently accurate. The time average of the digitized distance between each pair of points separated by 4.00 inches (ruler distance) is performed to ascertain whether variations in distance occur across the tablet. The variations in distances measured by the digitizer are generally less than the adjusted error (approximately 0.007 inches) and are not significant. However, a trend across

52 Table 11 Equivalent accuracy in meteorological units to +.02" in strip chart reading Variable Time (hours) Accuracy Temperature +.30 +.64 F Relative Humidity +.30 + 1.28% R.H. Precipitation +.30 +.02 inch Visiblity +.02 see note 2 Wind Direction +.006 + 2.4 degrees Wind Speed +.006 +.12 miles/hour Short-wave Radiation +.02 +.06 ly/min Long-wave Radiation +.02 +.06 ly/min Dew point +.02 see note 3 Notes 1. This is the accuracy before any smoothing by the man digitizing 2. The accuracy is +.01 volts which corresponds to +.64 km at 5 km and +.01 km at 0.1 km 3. The accuracy is +.01 which corresponds to around + 0.4 F at -20 Fdew point to +.24 Fat 20F dew point.

53 the grid may be indicated by a smaller measured distance across column number 3 in comparison with distance measured across the other two columns, and a larger measured distance across row number 5 than across the other two rows. This trend is too slight to be important in the data processing. A spatial display of standard deviation is shown in Figure 1Q. FigurelOa shows a steady increase in the variability of measurements from the top to the bottom of the test grid, and Figure lObshows an increase from right to left. The magnitude of the standard deviation in the x direction is about 24% greater on the lower left side of the grid than on the lower right side. For the y direction, the standard deviation shows a 35% increase from lower right to lower left. In digitizing meteorological charts, this increase is unimportant except, possibly, for precipitation charts. It was found that since the variation of standard deviation across row number 3 is much less than across row number 5, the digitization of precipitation charts could be improved by placing the charts on the upper two-thirds of the digitizer tablet. In conclusion, it has been found that use of a digitizer test grid has been of value in ascertaining the accuracy of the digitization procedure. The use of the right hand in holding the pen and the placement of the precipitation chart on the upper two thirds of the tablet have contributed to better analog to digitial data conversion. The test grid indicates that the digitization accuracy is slightly less than +.02 inches.

V. NETWORK INSTRUMENTATION Calibration of hair hygrothermographs In the planning stages of the study when equipment for continuous measurements of temperature and relative humidity was being considered, it was realized that the main shortcoming of the hygrothermograph was the uncertain accuracy of the hair hygrometer for measuring relative humidity. Although Belfort Instrument Company claimed an accuracy of indicated relative humidity within + 4% of true, other researchers, as well as our own experience, suggested that to achieve such an accuracy, the hygrothermographs had to be routinely calibrated using a standard humidity measuring device. For the first several months, therefore, a psychrometer, consisting of matched mercury thermometers was used to obtain field measurements of relative humidity which were then compared with those indicated by the hygrothermographs. For windy and overcast conditions, during which the humidity did not change significantly, the method gave reliable results. The method was unreliable, however, for conditions in which the humidity was varying significantly because the hair element had a slower response to changes in humidity than did the psychrometer. To obtain a controlled range of steady relative humidities, a humidity calibration chamber was built, following the principles and design discussed by Haegele and Matthews (1964). This section describes the humidity chamber, summaries the calibration procedure, and presents results of basic tests made on the hygrothermographs. 54

55 Principle of operation. Many saturated salt solutions, when placed in a closed container, cause the relative humidity within that container to reach a value which depends upon the particular salt solution and its temperature. Some salt solutions produce relative humidities which are nearly independent of their temperature, and most produce humidities which show a minor temperature dependence (Wexler and Hasegawa, 1954). If various salt solutions are placed one at a time in a chamber containing (1) the hygrothermograph to be calibrated and (2) a standard instrument for determining the true relative humidity, therefore, it is possible to determine the difference between the true relative humidity as measured with a psychrometer and that indicated by the hygrothermograph for a large range of relative humidities. Physical description of the chamber. A photograph of the chamber is shown in Figure 11. It is 16" high by 16" long by 12" wide, and weighs about 35 pounds. It is made of 1/2" thick transparent plexiglass and has two hinged access doors. A large door is for passage of the hygrothermograph and a small one is for the salt solution container. The permanent seals of the chamber (those edges not containing a door) are made using an acrylic solvent and are air-tight. The access doors are rimmed with a soft rubber gasket material and the ports have either gasket material or rubber stoppers. Although a completely air-tight chamber was not achieved, for a given salt solution the chamber did maintain a fixed relative humidity over many hours of operation. Fans are installed at both ends of the chamber to facilitate airflow at the salt solution - air interface and to circulate the air throughout the chamber, thus preventing stratification. The

56 ii.*i 4 ire ui it!t F r 11. " IH I a * *a ti ch ab e

57 driving motors for the fans are placed outside the chamber walls to minimize internal heating. The hygrothermograph shelf is placed above the salt solution container so that the solution can be changed with the hygrothermograph in place. The hygrothermograph shelf consists of a track of two 1/2" thick rails that run the length of the chamber about 1/3 the way from the bottom to the top. The use of two rails instead of a solid platform permits additional circulation within the chamber. Because the relative humidity produced by each salt solution is approximate, the true relative humidity,which is compared to that indicated by the hygrothermograph, is measured with a psychrometer. The psychrometer consists of matched wet- and dry-bulb thermometers with 0.1~C graduations. In the chamber, the wet-bulb thermometer is placed with its wick about 2" downstream from the upper fan. The dry-bulb is placed about 2" downstream from the wet-bulb and 2" lower. Each thermometer is held in place by a one-hole rubber stopper that is set into the top of the chamber. Auxiliary and preoperational tests. The use of a psychrometer for humidity measurements requires matched thermometers, since even a small difference in the two thermometers can produce significant errors in relative humidity, especially for low temperatures. In order to find matched pairs of thermometers, 10 thermometers (mercury-in-glass) graduated to 0.1~C with a range of -1~C to 51~C were compared. The 10 thermometers were suspended in a large vessel containing water and, initially, ice. The water was constantly

58 stirred by a magnetic stirrer spinning at the bottom of the vessel (about 6" below the thermometers' bulbs.) The temperature of the water was changed by applying heat to the bottom of the vessel. The heat was turned off and the entire contents of the vessel were allowed to come to equilibrium while the stirring continued. After 5 minutes with no change in temperature of any of the thermometers, they were read to the nearest 0.05~C. Thermometer readings were compared at four temperatures: 8.5, 15, 17.5 and 20~C. The pairs which matched the best and their temperatures are given in Table 12. It is evident that Pair I had the best match, with a maximum difference of only 0.05~C occurring at only one point. Pairs II and III had a maximum difference of 0.05~C at two and three points, respectively. The results for Pairs IV and V were not as good, varying from a minimum difference of 0.05~C to a maximum difference of 0.20~C. Because of these large differences, pairs IV and V were not used. The maximum error introduced into the psychrometric measurements of relative humidity is around + 0.5%, using Pairs I, II and III. This error is considerably less than other sources of error in the calibration procedure. Relative humidity: psychrometric vs. dew point. In the initial testing, two nearly independent methods were used to determine the relative humidity within the chamber. Relative humidities obtained from the wet- and dry-bulb thermometers were compared with those obtained with an EG&G Model 880 Dew Point Hygrometer.

59 Table 12. Thermometer Matching Results Pair # Thermometer Temp. 1 Temp. 2 Temp. 3 Temp. 4 (serial #) (~C) (~C) (~C) (~C) I 48243 8.40 15.30 17.50 19.90 6194093 8.45 15.30 17.50 19.90 Difference 0.05 0.00 0.00 0.00 II 6194095 8.45 15.35 17.50 19.85 6194097 8.40 15.35 17.50 19.90 Difference 0.05 0.00 0.00 0.05 III 48138 8.40 15.35 17.55 20.00 6194052 8.45 15.30 17.50 20.00 Difference 0.05 0.05 0.05 0.00 IV 6194125 8.40 15.20 17.35 19.80 6194200 8.45 15.30 17.45 19.85 Difference 0.05 0.10 0.10 0.05 V 6194141 8.30 15.20 17.45 19.90 6194185 8.55 15.40 17.55 20.00 Difference 0.15 0.20 0.10 0.10

60 Table 13 presents twenty-five sets of readings taken over a range of relative humidities between 20 and 94%. The rms difference between the relative humidities determined from the psychrometer and from the dew point hygrometer was + 1.4%, with an average difference of + 0.1%. 88% of these readings differed by 1% relative humidity or less. Of the three readings that were larger than 1%, 2 were taken under ambient conditions of about 20% relative humidity. For these low humidities, the discrepancies may have been caused by evaporation from the wet bulb which affected the dew point sensor. These results indicate that the relative humidity within the chamber was determined with wet- and dry-bulb thermometers to an accuracy of + 1%. Calibration procedures. A calibration begins with measurements at the highest humidity and proceeds in a stepwise fashion to the lowest. Once the lowest humidity is reached, measurements at progressively higher humidities are again obtained. Consequently, measurements at every humidity but the lowest are obtained twice per calibration. Table 14 shows the results of calibrations obtained for 4 hygrothermographs. The arrow to the right of the solution type gives the direction of progression of the calibration. It is evident that differences occur, depending on whether the relative humidity is increasing or decreasing. Repeatability of laboratory calibration. The repeatability of a calibration was studied by subjecting one instrument to three cycles of humidities in rapid succession. Tables 15 andl6 jive the results.

61 Table 13. Difference between the psychrometric and dew point hygrometer methods of determining relative humidity in the humidity calibration chamber. All values are relative humidity in percent. Psychrometric Method Dew Point Method Difference 21.5 19 2.5* 23 18 5 25 24 1 54.5 55 -0.5 55 58 -3 56 57 -1 78.5 78 0.5 79 79 0 81 81.5 -0.5 81 81.5 -0.5 81.5 82 -0.5 82 81 1 82 83 -1 82.5 82 0.5 82.5 82.5 0 84 85 -1 85 85 0 85.5 84.5 1 86 86 0 86.5 86.5 0 90.5 91.5 -1 91 92 -1 93 92.5 0.5 93.5 93 0.5 94 93.5 0.5 average difference is 0.1% rms difference is 1.4% *exceeds the rms difference

62 Table 14. Results of chamber calibration of four hygrothermographs Serial # Chamber Hygrothermograph Psychrometric Difference Conditions Relative Humidity Relative Humidity (Hygro-Psychro) (%) (%) 6599 Ambient 22 21.5 +0.5 K2SO4 (+) 88 91 -3 (+) 90 90.5 -0.5 ZnSO4 () 82 84 -2 (+) 82 82 0 NaCl () 79.5 81.5 -2 (+) 80 82 -2 Na2Cr207 52 54.5 -2.5 4985 Ambient 28 19 +9 K2SO4 () 90 91 -1 (+) 92 92.5 -0.5 ZnSO4 () 80 83 -3 (+) 84 85 -1 NaCl (4) 76 80 -4 () 80 82 -2 Na2 Cr207 53.5 55 -1.5 7478 Ambient 26 24.5 +1.5 K2SO4 (+) 89 92 -3 (+) 88.5 93 -4.5 ZnSO4 (+) 82 85 -3 (+) 81 83 -2 NaCl (+) 76.5 81 -4.5 (+) 77 79 -2 Na2Cr207 51 54.5 -3.5 Ambient 18 15 +3 K2SO (+) 90.5 90 +0.5 () 92.5 91 +1.5

63 Table 14. (cont.) Serial # Chamber Hygrothermograph Psychrometric Difference Conditions Relative Humidity Relative Humidity (Hygro-Psychro) (%) (%) ZnSO4 (4) 82.5 84 -1.5 (+) 86.5 85.5 +1 NaC1 (+) 79.5 80 -0.5 (+) 79.5 80 -0.5 Na2Cr207 49.5 55.5 -6

64 Table 15. Repeatability test on hygrothermograph #4985 Solution Hygrothermograph Psychrometric Difference Rel. Humidity Rel. Humidity (Hygro-Psychro) (%) (%) (%) Ambient 28 25 +3 K2S04 90 93 -3 ZnSO4 81.5 85 -3.5 (NH4)2S04 74 78.5 -4.5 NaCl 78 81 -3 Na2Cr207 54 55 -1 NaCl 80 81 -1 (NH4)2SO4 77 79 -2 ZnSO4 84 86 -2 K2SO4 92.5 93.5 -1 ZnSO 84 86.5 -2.5 NaC14 80 82.5 -2.5 Na Cr2O7 55 56 -1 NaCl 80 82.5 -2.5 ZnSO4 84 85.5 -1.5 K2SO4 92.5 94 -1.5 Ambient 24 23 +1 K2SO4 90 91 -1 ZnSO4 80 83 -3 NaCl 76 80 -4 Na2Cr207 53.5 55 -1.5 NaCl 80 82 -2 ZnSO4 84 85 -1 K2SO4 92 92.5 -0.5 Ambient 28 19 +9

65 Table 16. Difference between psychrometric measurements of relative humidity and hygrothermograph #4985 readings for increasing relative humidity; decreasing relative humidity, and overall. Approximate Chamber Difference Hygro-Psychro Average Difference Relative Humidity Relative Humidity Relative Humidity (%) (%) (%) Increasing Relative Humidity 95 -1, -1.5, -2.5, -3 -2.0 90 -0.5, -1 -0.8 85 -1, -1, -1.5, -2 -1.4 80 -2, -2.5, -3 -2.5 Average -1.7 Decreasing Relative Humidity 85 -2.5, -3, -3.5 -3 80 -2, -2.5, -4, -4.5 -3.2 55 -1, -1, -1.5 -1.2 Average -2.5 Overall 95 -1, -1.5, -2.5, -3 -2 90 -0.5, -1 -0.8 85 -1, -1, -1.5, -2, -2.5 -2.1 -3, -3.5 80 -2, -2, -2.5, -2.5, -3 -2.9 -4, -4.5 55 -1, -1, -1.5 -1.2 Less than 30 +3, +9 +6 Average -2.0

66 As can be seen in Table 16 the hygrothermograph is about 1% lower, on the average, for decreasing humidity than for increasing humidity. Overall, it averages about 2% low throughout the entire range, regardless of the direction. The data further indicate about a 2% difference in relative humidity from one calibration to another. This is equivalent to one chart division and exceeds the accuracy claimed by the manufacturer. Effects of transportation. It was originally thought that the chamber could be used to calibrate the hygrothermographs in the field. Tests with the chamber, however, indicated that using it for on-site calibrations would be impractical and it was concluded that the most convenient location for the calibrations would be at the meteorological laboratory in Ann Arbor. This meant, however, that the hygrothermographs would have to be transported about 150 miles, with the risk of a possible calibration change in transit. A study was made to determine the effects of transportation by calibrating four hygrothermographs in Ann Arbor and transporting them by car to Benton Harbor, where two of them (hygrothermographs 2984 and 7478) were recalibrated. Although the trip itself took about 4 hours, for practical reasons the calibrations occurred 4 days apart. The results of the calibrations are shown in Table 17. For hygrothermograph 4985, the calibration in Ann Arbor differed from that in Benton Harbor by 2% or less. The comparison of the calibrations for hygrothermograph 7478 is almost as good, with only the ambient reading differing by more than 2%.

67 Table 17. Calibration data for two hygrothermographs calibrated in Ann Arbor and Benton Harbor Hygrothermograph Approximate Chamber Location Difference Serial Number Relative Humidity (Hygro-Psycho) (%) Relative Humidity (%) 4985 Ambient (25) Ann Arbor +6 (ave) Benton Harbor +4 93 Ann Arbor -1.7 (ave) Benton Harbor -1 93 Ann Arbor -1 (ave) Benton Harbor +1 80 Ann Arbor -3.1 (ave) Benton Harbor -2 80 Ann Arbor -1.8 (ave) Benton Harbor -0 55 Ann Arbor -1.5 (ave) Benton Harbor -0 7478 Ambient (25) Ann Arbor +1.5 Benton Harbor -2 93 Ann Arbor -3 Benton Harbor -3 93 Ann Arbor -4.5 Benton Harbor -2.5 85 Ann Arbor -3 Benton Harbor -4 85 Ann Arbor -2 Benton Harbor -3 80 Ann Arbor -4.5 Benton Harbor -6 80 Ann Arbor -2 Benton Harbor -2 55 Ann Arbor -4 Benton Harbor -4

68 Hygrothermographs 6599 and 7064 were transported to Benton Harbor and returned to Ann Arbor about 1 1/2 weeks later where they were again calibrated. The results of the calibrations are given in Table 18. Hygrothermograph 6599 exceeded the 2% difference for only two humidities, one of which was the ambient reading. Hygrothermograph 7064 also exceeded the 2% difference for only two humidities, and again, one of these was the ambient reading. In all cases, the greatest difference occurred for comparatively low ambient humidity conditions. As mentioned above, this is most likely due to the change in humidity within the chamber when the wet-bulb thermometer is added, and not to transportation. These results indicate that accuracy is not lost by transporting the instruments from the field to Ann Arbor for calibration and back to the field. Consequently, all calibrations are currently being performed in Ann Arbor. Calibration and maintenance of field units. The procedure for using the chamber to calibrate a hygrothermograph which has been returned from the field is given below: 1. The unit is subjected to the humidities listed. A set of data is obtained for both decreasing and increasing humidity. This procedure takes about 3 days, since about one hour is required for humidity conditions in the chamber to come to equilibrium with a salt solution. 2. The unit is then removed, cleaned and all moving parts are lubricated. It is reinstalled in the chamber and subjected to the extreme humidities of 35% and 90%.

69 Table 18.. Calibration data for two hygrothermographs before and after transportation from Ann Arbor to Benton Harbor and return Hygrothermograph Approximate Chamber Date Difference Serial Number Relative Humidity (Hygro-Psycho) (%) Relative Humidity (%) 6599 Ambient 9 +0.5 19,20 -3 93 9 -3 19,20 -4 93 9 -0.5 19,20 -4 85 9 -2 19,20 -- 85 9 -0 19,20 -1 80 9 -2 19,20 -3 80 9 -2 19,20 -2 55 9 -2.5 19,20 -3.5 7064 Ambient 9 +3 19 -0.5 93 9 +0.5 19 -1 93 9 +1.5 19 +1 85 9 -1.5 19 -- 85 9 +1 19 0 80 9 -0.5 19 -5.5 80 9 -0.5 19 -2 55 9 -5 19 -7

70 3. Any adjustments on the link and lever assembly necessary to make the unit conform as closely as possible to the psychrometric measurements at these humidities are made. 4. Step (1) is repeated so that data showing true versus indicated humidity are obtained for the unit as it received from the field and again after reconditioning. 5. The unit is returned to the field. The construction of four chambers was completed in December. They allow four hygrothermographs to be calibrated simultaneously. The time required for 4 units to be returned to Ann Arbor for calibration and returned to the network is about one month. Since there are 26 units altogether in the Palisades and Cook networks, each unit is reconditioned and calibrated at least once every 6 months. Enough spare units are on hand so that recordings at a station are not interrupted when a unit is removed for calibration. Dew point system A Cambridge Systems Model 880 Dew Point Hygrometer together with a sampling and recording system were procured, tested, and installed at Station P03A in February 1974. The dew point hygrometer measures the dew point of an air sample by (1) cooling a small polished gold surface exposed to the air sample until the temperature of the surface is below the temperature of the air sample and condensation occurs on it, (2) optically detecting the water film, and (3) thermoelectrically maintaining a condition in which the amount of water on the surface does not change. The

71 dew point is the state of dynamic equilibrium in which the rate at which water molecules leaving the water surface is equal to the rate at which water molecules enter the surface. The accuracy of the sensor is given by the manufacturer as + 1.5~F. Its response to changes in moisture is orders of magnitude faster than the hair hygrometer. For use in the field, the complete dew point system consists of the sensing unit and electronics, a pump and tubing for continuously drawing air past the sensor, and a recorder for the voltage signal from a thermistor in the sensing unit. The thermistor measures the temperature of the metal surface. Its output signal therefore, is proportional to the dew point. The recorder is a Leeds and Northrup Speedomax W strip chart recorder with a range of 0-40 millivolts on a 10-inch chart. In working with the complete system and comparing recorded values of dew point with those measured with a psychrometer, we have found that the accuracy better than the claimed accuracy of + 1.5~F. The dew point system supplements the hygrothermograph measurements of relative humidity at main station P03A. The faster response and greater sensitivity of the dew point system will prove important in cases of downwash of the cooling tower plume, for example, if it descends in the vicinity of the station.

72 Precipitation gages Complete calibrations of the precipitation gages by a National Weather Service technician were made in June 1973, and December 1973. In addition, weekly checks are made by the field observer by subjecting each gage to a weight equivalent of 2 inches of water. Visiometers The performance of the visiometers was satisfactory after they were reconditioned by the manufacturer in April 1973. Missing data were caused by recorder problems and a defective high voltage supply in the unit at station P03A in January 1974. The unit was repaired and placed back into operation in February. To insure optimum performance during the operation of the cooling towers, both visiometers were returned to the manufacturer in March, 1974, for calibration and incorporation of new components. Both units were placed back into field operation on May 8. Wind systems Wind tunnel calibrations of the wind speed systems for both main stations were made at the University of Michigan micrometeorology wind tunnel in August and December, 1973. Linearity checks of the wind direction sensors and translators were made in the field in February 1974 and were found to be the same as when installed. Spare wind speed sensors were recently obtained so that recordings

73 will not have to be interrupted in the future while wind tunnel calibrations are made.

VI. WORK PLANNED FOR NEXT YEAR The following five tasks will be worked on in the next year: 1. Data analysis. It is expected that the cooling towers will begin operation soon. Network data recorded for the past two years and until the cooling towers begin operation comprise the basic pre-operational data for analysis. Comparisons of data for individual network stations and groups of stations for both the pre-operational and post-operational phases will be made. One of the tools used in the analysis will be a tabulation prediction technique adapted for this study by Dr. Harry Moses. 2. Monitoring and preliminary evaluation of cooling tower effects. A limited number of case studies will be conducted when the cooling towers are in operation and the wind is onshore. Special attention will be given to monitoring downwash, which will have direct effects on conditions close to the cooling towers and possibly on the leeward side of the sand dunes near 1-196. Both visual observations and time lapse photographs will be made. Plans are being made to install one and possibly two time lapse cameras: one pointed inland in line with the blocks of cooling towers and one near station P03A pointed in the direction of the cooling towers. The photographs will occasionally be supplemented by observations of the plume by means of aircraft and personnel employed at the nuclear plant. Information will also be obtained on the infrequent but possibly important cases in which a nocturnal transport of the plumes 74

7 lakeward by a land breeze occurs, followed by a daytime return of the moist air to land by onshore flow. It is reasonable to expect that under certain conditions, the moisture content of the thermally stable lake air moving onshore, which is already high, may once again be increased by the cooling tower plumes and enhance meteorological effects that much more. 3. Continued operation and maintenance of the 13 meteorological stations. The measurements presently being made will continue, along with the schedule of calibration and maintenance of equipment. 4. Processing, tabulation and reporting of meteorological data. Data processing will continue. A report consisting of data tabulations and summaries will be prepared. It will include a computer printout of data using a format similar to that shown in Quarterly Progress Report 320158-5-L. 5. Fog studies. It is expected that a paper will be prepared and submitted which describes results of a preliminary fog climatology for western Michigan along with an assessment of the present knowledge of fog formation.

REFERENCES Ackerman, W. C., 1971: Research Needs On Waste Heat Transfer From Large Sources Into the Environment. Report to National Science Foundation, Grant GI-30971, Illinois State Water Survey, Urbana, 37 pp. Climate of Michigan by Stations, 1971: Michigan Weather Service, East Lansing, Michigan. Culkowski, W. M., 1962: An Anomalous Snow at Oak Ridge, Tennessee, Monthly Weather Review, vol. 90, pp. 194-196. Eagles, T. W. and L. C. Kohlenstein, 1974: A Cooling Tower Visible Plume Prediction Model Based on Measurements. The Johns Hopkins University, Applied Physics Laboratory, 14 pp. Haegele, C. and D. Mathews, 1964: A Controlled Humidity Chamber. U. S. Dept. of Commerce Weather Bureau Instrumental Engineering Division, Report No. 5, Washington, 8 pp. Hanna, S. R. and S. G. Perry, 1973: Meteorological Effects of the Cooling Towers at the Oak Ridge Gaseous Diffusion Plant. Part I: Description of Source Parameters and Analysis of Plume Photographs and Hygrothermograph Records. NOAA Atmospheric Turbulence and Diffusion Laboratory, Oak Ridge, Tenn., ATDL Contribution File No. 86, 40 pp. Hanna, S. R., 1974: Meteorological Effects of the Cooling Towers at th O Rie Oak Ridge Gaseous Diffusion Plant. Part II: Predictions of Fog Occurrence and Drift Deposition. NOAA Atmospheric Turbulence and Diffusion Laboratory, Oak Ridge, Tenn., ATDL Contribution File No. 88, 39 pp. 76

77 Hanna, S. R., 1974: Meteorological Effects of the Mechanical Draft Cooling Towers at the Oak Ridge Gaseous Diffusion Plant. NOAA Atmospheric Turbulence and Diffusion Laboratory, Oak Ridge, Tenn. ATDL Contribution No. 89, 23 pp. Huff, F. A., R. C. Beebe, D. M. A. Jones, G. M. Morgan, Jr. and R. B. Semonin, 1971: Effect of Cooling Tower Effluents on Atmospheric Conditions in Northeastern Illinois. Circular 100, Illinois State Water Survey, Urbana, 37 pp. Huff, F. A. and J. L. Vogel, 1973: Atmospheric Effects from Waste Heat Transfer Associated with Cooling Lakes. Report to National Science Foundation, Grant GI 35841, Illinois State Water Survey, Urbana, 89 pp. Meyer, J. H., 1974: Mechanical Draft Cooling Tower Visible Plume Measurement Program for Plume Modeling. The Johns Hopkins University Applied Physics Laboratory, 29 pp. Koss, T. C. and P. M. Altomare, 1971: Evaluation of Environmental Effects of Evaporative Heat Dissipation Systems at the Palisades Plant. NUS-785, Environmental Safeguards Division, NUS Corporation, Rockville, Maryland, 68 pp. Portman, D. J., 1973: Atmospheric Diffusion Potential for Michigan's Lower Peninsula. Report for Consumers Power Company, 30 pp. Ryznar, E. and D. G. Baker, 1973: An Investigation of the Meteorological Impact of Mechanical-Draft Cooling Towers at the Palisades Nuclear Plant. First Annual Progress Report, ORA Project 320158, The University of Michigan, 39 pp.

78 Wexler, A. and S. Hasegawa, 1954: "Relative Humidity-Temperature Relationships of Some Saturated Salt Solutions in the Temperature Range of 0~ to 50~C". Jour. Res. of Nat. Bur. Stand., 2512 USS, #1, pp. 19-26.