NOAA Technical Report EDS 20 GATE Convection Subprogram Data Center - Analysis of Rawinsonde Intercom parison Data Washington, D.C. November 1976 U.S. DEPARTMENT OF COMMERCE National Oceanic and Atmospheric Administration Environmental Data Service NOAA TECHNICAL REPORTS Environmental Data Service Series The Environmental Data Service (EDS) archives and disseminates a broad spectrum of environmental data gathered by the various components of NOAA and by the various cooperating agencies and activities throughout the world. The EDS is a "bank" of worldwide environmental data upon which the researcher may draw to study and analyze environmental phenomena and their impact upon commerce, agriculture, industry, aviation, and other activities of man. The EDS also conducts studies to put environmental phenomena and relations into proper historical and statistical perspective and to provide a basis for assessing changes in the natural environment brought about by man's activities. The EDS series of NOAA Technical Reports is a continuation of the former series, the Environmental Science Services Administration (ESSA) Technical Report, EDS. Reports in the series are available from the National Technical Information Service, U.S. Department of Commerce, Sills Bldg. , 5285 Port Royal Road, Springfield, Va. 22151. Price: $3.00 paper copy; $1.45 microfiche. When available, order by accession number shown in parentheses. ESSA Technical Reports EDS 1 Upper Wind Statistics of the Northern Western Hemisphere. Harold L. Crutcher and Don K.. Halli- gan, April 1967. (PB-174-921) EDS 2 Direct and Inverse Tables of the Gamma Distribution. H. C. S. Thorn, April 1968. (PB-178-320) EDS 3 Standard Deviation of Monthly Average Temperature. H. C. S. Thorn, April 1968. (PB-178-309) EDS 4 Prediction of Movement and Intensity of Tropical Storms Over the Indian Seas During the October to December Season. P. Jagannathan and H. L. Crutcher, May 1968. (PB-178-497) EDS 5 An Application of the Gamma Distribution Function to Indian Rainfall. D. A. Mooley and H. L. Crutcher, August 1968. (PB- 180-056) EDS 6 Quantiles of Monthly Precipitation for Selected Stations in the Contiguous United States. H. C. S. Thorn and Ida B. Vestal, August 1968. (PB-180-057) EDS 7 A Comparison of Radiosonde Temperatures at the 100-, 80-, 50-, and 30-mb Levels. Harold L. Crutcher and Frank T. Quinlan, August 1968. (PB-180-058) EDS 8 Characteristics and Probabilities of Precipitation in China. Augustine Y. M. Yao, September 1969. (PB-188-420) EDS 9 Markov Chain Models for Probabilities of Hot and Cool Days Sequences and Hot Spells in Nevada. Clarence M. Sakamoto, March 1970. (PB-193-221) NOAA Technical Reports EDS 10 BOMEX Temporary Archive Description of Available Data. Terry de la Moriniere, January 1972. (COM-72-50289) EDS 11 A Note on a Gamma Distribution Computer Program and Graph Paper. Harold L. Crutcher, Gerald L. Barger, and Grady F. McKay, April 1973. (COM-73- 11401) EDS 12 BOMEX Permanent Archive: Description of Data. Center for Experiment Design and Data Analysis, May 1975. EDS 13 Precipitation Analysis for BOMEX Period III. M. D. Hudlow and W. D. Scherer, September 1975. (PB-246-870) EDS 14 IFYGL Rawinsonde System: Description of Archived Data. Sandra M. Hoexter, May 1976. EDS 15 IFYGL Physical Data Collection System: Description of Archived Data. Jack Foreman, September 1976. (Continued on inside back cover) NOAA Technical Report EDS 20 ^pATMOsq^ r MENJ Of GATE Convection Subprogram Data Center - Analysis of Rawinsonde Intercomparison Data Center for Experiment Design and Data Analysis Robert Reeves, Scott Williams, Eugene Rasmusson, Donald Acheson, Thomas Carpenter James Rasmussen, Scientific Officer GARP Activities Office, WMO Geneva, Switzerland Washington, D.C. November 1976 U.S. DEPARTMENT OF COMMERCE Elliot L. Richardson, Secretary National Oceanic and Atmospheric Administration Robert M. White, Administrator Environmental Data Service Thomas S. Austin, Director CONTENTS PART I Analysis of Temperature and Humidity Data 1. Introduction , 1 2. Description of the formal intercomparison data sets 2 3. Analysis 3 3.1 Comparison of observations near the trade inversion and tropopause . . . . . 3 3.2 Statistical analysis 4 3.2.1 Results from the formal intercomparison data 4 3.2.2 Results of the Anton D ohrn M60-VIZ intercomparisons 5 4 . Summary and conclusions , 6 PART II Comparison of Wind Data from U.S.S.R. and U.S. Systems 1. Introduction 2 . Results Page 2 .1 Oceano graph er - Vanguard Intercomparisons 9 2.2 U.S. -U.S.S.R. Intercomparisons 11 2.2.1 Vanguard -U . S . S . R . Intercomparisons 11 2.2.2 Omega and VLF-U.S.S.R. Intercomparisons 12 3. Conclusions 13 Appendix A - U.S.S.R. Upper Air System and Data Processing .. 15 Appendix B - U.S. Upper Air Systems and Data Processing 20 in TABLES 1. --Dates and locations of the GATE Intercomparison Periods 30 2. --List of formal intercomparison flights used in the analysis, 31 by date and ship , . ' . . , 3. --Precision and vertical resolution of archived data, by Nation and parameter 30 4. --Data used in the wind intercomparison analyses 32 5 . - - Oceanographer (hyperbolic) - Vanguard wind intercomparison statistics, data set 1 33 6. --Oceanographer (multisolution) - Vanguard wind intercomparison statistics, data set 2 , 34 7 . -- Oceanographer (hyperbolic) -U.S. S.R. wind intercomparison statistics, data set 3 . . 35 8. -- Researcher (Hyperbolic) -U.S. S ,R. wind intercomparison statistics data, set 4 36 9 . -- Oceanographer (multisolution) - Vanguard intercomparison 37 10. -- Oceanographer (multisolution) - Vanguard differences - U component for data set 2 38 11 . -- Oceanographer (multisolution) - Vanguard differences - V component for data set 2 39 12 . --Intercomparison statistics (40-mb averaged values) 40 IV FIGURES 1. --Series of temperature soundings for Intercomparison Period I. The soundings are plotted in relative time sequence 41 2. --Detail of the upper part of a temperature comparison between the Poryv and Researcher during IC III . 42 3. --Means and standard deviations of the temperature differences vs. pressure for the U.S.S.R.-U.S., U.S.S.R. -Canada, and U.S.S.R. -F.R.G. comparisons 43 4. --Frequency distribution of the temperature differences for selected pressure layers for U.S.S.R.-U.S., U.S.S . R. -Canada, and U.S.S.R. -F.R.G. comparisons 44 5 . --Comparison of day (29 pairs) and night (9 pairs) mean temper- ature differences vs. pressure for the U.S.S.R. -VIZ comparisons. Soundings from the U.S., Canada, and F.R.G. were combined.... 45 6. --Standard deviation of the temperature and vertical plots of the correlation coefficients for the U.S.S.R.-U.S., U.S.S.R. - Canada, and U.S.S.R. -F.R.G. comparisons 46 7. Means and standard deviations of the relative humidity differences vs. pressure for the U.S.S.R.-U.S,, U.S.S. R.- Canada, and U.S.S.R. -F.R.G. comparisons 4 7 8. --Frequency distributions of the relative humidity differences for selected pressure layers for the U.S.S.R.-U.S., U.S.S.R. - Canada, and U.S.S.R. -F.R.G. comparisons 48 9. --Detail of the lower part of the temperature and relative humidity profiles for the comparison between Professor Zubov and Quadra for 0900 GMT June 18 49 10. --Detail of the lower part of the temperature and relative humidity profiles for the comparison between Okean and Researcher for 1200 GMT August 16 50 11. --Standard deviations of the relative humidity measurements and vertical plots of the correlation coefficients for the U.S.S.R.-U.S., U.S.S.R. -Canada, and U.S.S.R. -F .R.G. comparisons 51 12. --Temperature and relative humidity profiles for the compar- ison between the F.R.G. M60 and U.S. VIZ sondes for 0800 GMT August 2 , . 52 v FIGURES (continued) 13 14 15 16 17 -a. Means and standard deviations of the temperature differences vs. pressure for the F.R.G, M60-U.S, VIZ comparisons , » 53 b. Standard deviations of the temperature measurements and vertical plot of the correlation coefficient for the F.R.G. M60-U.S. VIZ comparisons. 54 -Frequency distributions of the temperature differences for selected pressure layers for the F.R.G. M60-U.S. VIZ comparisons . 55 -a. Means and standard deviations of the relative humidity differences vs. pressure for the F.R.G. M60-U.S. VIZ comparisons 56 b. Standard deviations of the relative humidity measure- ments and vertical plot of the correlation coefficient for the F.R.G. M60-U.S. VIZ comparisons 57 -Frequency distributions of the relative humidity differences for selected pressure layers for the F.R.G. M60-U.S. VIZ comparisons 58 -Frequency distribution of differences, Oceanographer (M) minus Vanguard U Component (m/sec) 59 -Frequency distribution of differences, Oceanographer (M) minus Vanguard V Component (m/sec) , 60 19. --Standard Deviation (a) of differences, U component, Oceanographer minus Vanguard (m/sec) vs. averaging thickness. Curves labelled Oceanograp her (H) and Oceanographer (M) are for hyperbolic and multisolution methods, respectively. Circle with cross denotes result for Oceanographer (H) - U.S.S.R. intercomparison. Circle with dot is result for Researcher-U . S . S . R . comparison , 61 20. --Same as figure 19 for V-component 62 21 . --Correlation coefficient of U component Oceanographer to U component Vanguard plotted as a function of averaging thickness 63 22 . --Same as figure 21 for V-component 64 23. --Percent of observations less than given value vs. averaging thickness, for U component differences, Oceanographer (M) minus Vanguard . . 65 VI FIGURES (continued) 24. --Same as figure 23 for V-component differences, «■/-., 25. --Fifth and 95th percentiles of the distribution of component differences, Oceanographer (M) minus V anguard vs . averaging thickness 67 26. --Same as figure 25 for 2.5th and 97.5th percentiles. ......... 67 27. --Standard deviation (a) of differences, U and V components, Oceanographer (Multisolution) minus Vanguard (m/sec) vs. averaging thickness for three separate layers. , . , . , 68 28 . --Vertical profiles of U and V wind components for Okean, Researcher (H) , and Vanguard three intercomparison flights during IC II , 69 (a) Aug. 16, 1200 GMT (b) Aug. 17, 0300 GMT (C) Aug. 17, 1200 GMT Vll Digitized by the Internet Archive in 2013 http://archive.org/details/gateconvectionsuOOenvi GATE Convection Subprogram Data Center - Analysis of Rawinsonde Intercomparison Data Robert Reeves Scott Williams Eugene Rasmus son Donald Acheson Thomas Carpenter Center for Experiment Design and Data Analysis Environmental Data Service NOAA, Washington, D.C. James Rasmus sen Scientific Officer GARP Activities Office, WMO Geneva, Switzerland ABSTRACT. An intercomparison analysis of GATE rawinsonde temperature, humidity, and winds is described. Data from the formal Intercomparison Periods and Phases I and III from the United States, the Federal Republic of Germany, Canada, and the Union of Soviet Socialist Republics were used. Height -dependent differences between the measure- ment systems are presented for the temperature and humidity data. Comparisons were made between the U.S.S.R. winds derived from radar tracking, and the U.S. winds derived from radar, VLF, and OMEGA tracking. Means and standard deviations of the wind U and V component differences are computed for different averaging depths. PART I. ANALYSIS OF TEMPERATURE AND HUMIDITY DATA 1. INTRODUCTION The Convection Subprogram Data Center (CSDC) is responsible for preparing a validated set of international rawinsonde data acquired by the ships of the A/B, B, and C arrays during the operational phases of GATE. The first step in the process of international validation of these data consists of an analysis of observations taken by these ships during the GATE formal intercomparison periods. The formal intercomparison data were derived from a series of balloon launches during which two different instrument packages were attached to the same balloon, with simultaneous recording of the data by the two nations sup- plying the packages. All flights during the formal intercomparison periods were launched from U.S.S.R. ships. In several cases when a U.S. instrument was supplied for the comparison, more than one U.S. ship recorded the signal. This provided a check on the consistency of the receipt and recording of the signal at different U.S. ships. Analyses were performed on the rawinsonde temperature and humidity data from four nations: The Federal Republic of Germany (F.R.G.), the Soviet Union (U.S.S.R.), Canada, and the United States . The periods during which the intercomparison data used in these analyses were acquired are listed in table 1. Table 2 lists the flights from the for- mal intercomparison periods used for these analyses. A comparison between the U.S.S.R. ship Ernst Krenkel and the F.R.G. ship Planet (0100 GMT Sept. 22) was excluded because of an approximate 10°C temperature difference between the participating sondes throughout a large part of the atmosphere. In addition to the formal intercomparison flights, a special series of flights was launched from the F.R.G. ship Anton Dohrn during Phase 2, de- signed to obtain comparative data between the German M60 Registration sonde and the U.S. Beukers/VIZ sonde. This data set has also been included in our temperature and humidity comparative analyses. Additional observation phase data from the U.S. ships Vanguard and Oceanographer have been used in the wind analysis described in part II of this report. The data base for comparative analyses will be further increased by the addition of observation phase data as they become available. These consist of routine observations from the nearby ships Professor Vize and Oceanographer during Phase 2, and Professor Vize and Meteor during Phase 3. The objective of this report is to provide the GATE scientific community with early information on the compatibility of the rawinsonde temperature and humidity measurements. We hope this will lead to feedback from the GATE sci- entific community and will thereby aid the CSDC in preparing a well-integrated international rawinsonde data set. The analysis is presented in two parts: the first discusses the character and quality of the data and describes the processing procedures at the CSDC; the second is a quantitative comparison of data from pairs of ships. 2. —DESCRIPTION OF THE FORMAL INTERCOMPARISON DATA SETS Each data set reflects the measurement and processing methods employed by each nation. Table 3 gives the precision and vertical resolution of data in the archive by nation and parameter. The F.R.G. determined significant levels using the World Meteorological Organization (WMO) standard procedures, then interpolated linearly between significant points to produce data at 5-mb interval levels. Documentation accompanying the Meteor and Planet intercom- parison data states that the U.S. VIZ sonde was employed for those flights, and that no corrections have been applied to the data. The U.S.S.R. data were determined for mandatory and significant levels. Moisture was reported as relative humidity, and temperature was corrected for radiation-induced errors: (see appendix A). The Canadian and U.S. systems employed the VIZ radiosonde, the soundings being processed and archived with 5-mb resolution. A timelag correction was applied to the U.S. thermistor measurements, and timelag and temperature cor- rections were applied to the hygristor readings; (see appendix B) . Further processing of the National Processing Center (NPC) archived products was required in order to produce comparable data sets. Since the U.S.S.R. temperature and humidity were reported at significant levels only (with heights to the nearest 10 m) , 5-mb data points had to be obtained by linear interpo- lation. Concurrently, the heights were recomputed hydrostatically to the nearest meter using pressure, temperature, and humidity. Several different humidity parameters were reported; the U.S. recorded speci- fic humidity to the nearest 0.01 g/kg , Canada and the U.S.S.R. reported relative humidity to the nearest 1%, and the F.R.G. reported dewpoint to the nearest 0.1°C. Relative humidities were recomputed for the analysis. Surface data are not used in this analysis (although they influence the lowest few points through lag effects or interpolation), but the user should be aware that the surface values reported are those of the tracking ship and not of the launching ship (which in some cases was many miles away) . 3. ANALYSIS 3.1 Comparison of Observations Near the Trade Inversion and Tropopause Observations during the first and third intercomparison periods (IC I and IC III a and b) were obtained near 13°N latitude, while those acquired during IC II were taken near 8°N. IC I was characterized by a well-developed trade regime with a strong trade inversion, while IC II and IC III a and b were more typical of the ITCZ regime with moisture penetration through a greater depth of the troposphere. The M60 - VIZ comparisons were made during August 4-15 when the Anton Dohrn was stationed at the Equator, exposed to the influence of the southeast trades. The trade inversion and the tropopause are particularly well-suited for ex- amining the response of different radiosonde systems to sharp vertical changes in the temperature and/or humidity. A time series of the temperature sound- ings through the trade inversion during IC I is shown in figure 1. Included with the radiosonde observations are four soundings taken by the Boundary Layer Instrument System (BLIS) aboard the U.S. ship Researcher . Except for the U.S.S.R. ship Professor Zubov , there is general agreement among the several systems on the pressure at the base of the trade inversion. At all five com- parison times, however, the Professor Zubov data show the pressure at the inversion base from 5 to 15 mb lower than the other observations. This leads to some large temperature differences within the inversion layer. Since obser- vations from the second IC I U.S.S.R. ship Ernst Krenkel are in agreement with the U.S. and F.R.G. ships on the height of the trade inversion, those observa- tions suggest a ship-dependent problem. The picture is complicated by discre- pancies in the identification of the top of the trade inversion, which in some cases appear to result from different way?; of determining significant levels. Similar discrepancies between the U.S.S.R. and U.S. appear in the vicinity of the tropopause. Twenty-eight cases were selected from the comparative data set in which the level of minimum temperature could be identified unambiguously for both sondes. Of these, the U.S.S.R. sondes reported the minimum tempera- ture at a pressure 5 mb lower in 19 cases, and at a higher pressure only twice This problem is illustrated in figure 2, which shows the upper part of the soundings of the Poryv and Researcher for 1200 GMT September 21. Examination of the Anton Dohrn comparative soundings indicated only small differences between the VIZ and M60 sondes in identifying the tropopause and trade inversion base and no systematic differences such as those observed in the flight pairs involving the U.S.S.R. sondes. 3.2 Statistical Analysis For statistical comparison, the flights were grouped by participating nation. Since all flights from the formal intercomparison periods were launched and tracked by the U.S.S.R. ships, there were 22 U.S.S.R. -U.S. flights, 11 U.S.S.R. Canada flights, and 11 U.S. S.R. -F.R.G. flights. The U.S., Canada, and the F.R.G. used the VIZ sonde, so that differences between comparison groupings should reflect only sampling and processing procedures. These results are discussed in section 3.2.1. The independent M60 - VIZ comparison performed on 23 pairs is discussed separately in section 3.2.2. The statistical computations summarized below were performed for 50-mb aver- ages using data for all pressures below 1000 mb. 3.2.1 Results From the Formal Intercomparison Data The mean temperature differences vs. height for the U.S.S.R. comparisons are shown in figure 3. The difference profiles are similar for all comparative sets, i.e., the U.S.S.R. temperatures are cooler in the lower troposphere and warmer above relative to the other three nations. Most of the differences at individual levels are not significantly different from zero by Student's t test. However, when the entire profile is viewed, there can be no question that the differences display a strong height dependence. Frequency distribu- tions of the differences for selected layers are shown in figure 4. Note that for the Canadian and U.S. comparisons, all the U.S.S.R. soundings were warmer in the 200-250 mb layer. The differences change sharply near the tropopause, with the U.S.S.R. soundings becoming cooler above. The Poryv -Researcher com- parison of figure 2 is another example of this feature. The observed tropos- pheric height-dependence difference in temperatures cannot be attributed to U.S.S.R. solar radiation corrections, which reduce the temperatures at all levels by an amount that increases with height; (see appendix A). Figure 5 compares day and night U.S. S.R. -VIZ mean temperature differences. That com- parison is consistent in the upper levels with the expected change, i.e., lowering of U.S. S.R. -VIZ differences from night to day, but inconclusive in the lowest 200 mb where radiation corrections are small. These results must be considered tentative, since only 8 sounding pairs were available for com- puting the nighttime average. Lag corrections applied to the U.S. temperature data account for only a small part of the differences, since they typically lower the U.S. values by about .1°-.2°C near the surface and about .3°^. 4 C in the upper troposphere. Figure 6 shows the height distribution of correlation coefficients together with sample standard deviations for the three comparisons. If the natural variability of temperature during the intercomparison periods is typical of the variability during the regular observation phases, then the results must be regaraed as disappointing. Only for the lowest few kilometers of the U.S.S.R.-U.S. and U.S.S.R.-F.R.G. comparisons are the correlation coeffi- cients consistently above 0.7. At other levels it appears that the natural variability of the atmosphere constitutes less than half the total variance of the temperature data. (See part II, p. 45 for a discussion of the correlation coefficient) . Relative humidity differences vs. height appear in figure 7. The relative humidity distributions shown in figure 8 are much flatter than the correspond- ing temperature distributions and are clearly not Gaussian. The dryness of the U.S.S.R. soundings in the 950-1000 mb layer appears to be caused in part by inability to resolve the humidity profile from the significant level values. An example of this is given in figure 9, which shows the lower portion of the Professor Zubov-Quadra comparative flight pair of 0900 GMT June 18. The first significant point above the surface on the Professor Zubov humidity sounding appears in the trade inversion. Linear interpolation between significant points fails to preserve the detail of the boundary layer humidity profile typical of the trade regime, and yields a mean lower humidity for the bottom part of the Professor Zubov sounding. This difficulty is by no means confined to IC I with its strong trade regime, as the height curves for the mean rela- tive humidity differences during IC II and IC III (not shown here) are similar in shape. For example, figure 10 shows the higher values of humidity observed by the Researcher in its comparison with the Okean at 1200 GMT August 16. The U.S. soundings are significantly drier in the upper half of the tropos- phere, where the mean specific humidities are lower than 0.5 g/kg. The correlation coefficient and sample standard deviations for the relative humidity comparisons appear in figure 11. The correlations are remarkably high for the U.S.S.R.-U.S. comparisons up to the 300-mb level, and despite the biases previously discussed, indicate that the random errors of the two systems are not large enough to prevent them from responding in like manner to the variations in moisture content. 3.2.2 Results of the Anton Dohrn M60-VIZ Intercomparisons This comparative data set was processed by the F.R.G. NPC. Five-mb data were obtained by linear interpolation from significant and standard levels. No further processing of these data was performed by the CSDC . The overall agreement of the data is good, as illustrated by the sounding for 0800 GMT August 2 shown in figure 12. Note the good agreement of the humidity profiles. Mean temperature differences vs. height are shown in figure 13a. VIZ sonde temperatures averaged .5°C warmer over all layers and all flights. The VIZ sonde recorded higher mean temperatures for 20 of the 23 flights, a feature clearly reflected in frequency distributions for selected layers shown in figure 14. The increased differences above the 300-mb layer may be due to increased insolation error in the VIZ sonde. A Student's t test performed on the mean for the layer between 1000 and 300 mb , assuming only 23 independent samples, indicates that the mean 0.4°C difference is non zero at the 95% confidence level. The correlation coeffi- cients and sample standard deviations are also shown in figure 13b. The cor- relation coefficients are high only in the layer from 800 mb to 400 mb and, inexplicably, in the vicinity of the tropopause where the maximum variability of temperature was observed. The relative humidity differences shown in figure 15 are small between the surface and 500 mb, but large above 500 mb where the mean specific humidity is less than 1 g/kg. The correlation coefficient drops off sharply just above 500 mb, where the magnitude of the humidity and its variations become small compared to the random errors of measurement. The frequency distributions of figure 16 complete the analysis. 4. SUMMARY AND CONCLUSIONS An analysis of temperature and relative humidity was performed on the GATE rawinsonde data from the three formal comparative periods and from a special series of soundings launched from the F.R.G. ship Anton Dohrn . Much useful information on the character and compatibility of the several national systems has been gleaned from the analysis despite the limited sample size. The major conclusions are: a. A possible ship-dependency was identified from the series of IC I soundings taken through the trade inversion. The Professor Zubov soundings reported the trade inversion base consistently 5 to 15 mb lower than either the cor- responding VIZ soundings or soundings from the other IC I U.S.S.R. ship, Ernst Krenkel . b. The tropopause pressure reported by the VIZ soundings was consistently higher than that reported by the U.S.S.R. soundings. This suggests a systematic difference between U.S.S.R. and VIZ height or pressure measurements. c. The VIZ-M60 comparison showed no significant differences in the heights of the trade inversion or the tropopause. d. The U.S.S.R. -VIZ comparisons revealed a height -dependent temperature difference, with the VIZ sonde relatively warmer in the lower troposphere, and cooler in the upper troposphere. Neither solar radiation corrections applied to the U.S.S.R. data nor lag corrections applied to the U.S. data can account for the observed effect. e. The VIZ temperature was warmer than the M60 temperature by a small but statistically significant 0.4°C in the layer 1000 to 300 mb. The difference increased above 300 mb. f. The only significant differences in relative humidity in the lower tropos- phere were found in the U.S.S.R. -VIZ comparisons for the 1000-950 mb layer. The drier U.S.S.R. values may be due in part to loss of vertical resolution arising from the relative humidity criterion, which permits departures of up to 15% without establishing another significiant level. g. Significant relative humidity differences were found for the upper tropos- phere in both the U.S.S.R. -VIZ and M60-VIZ comparisons. Above 500 mb, the U.S.S.R humidity values are 10% to 20% higher than the VIZ humidities, which were, in turn, about 10% higher than the M60 humidities. However, since these differences occurred at specific humidities of less than 1 g/kg, they are of lesser Importance for budget studies than would be the case for a 10% difference in the lower layers. All the comparisons show higher correlation coefficients in the lower troposphere. The valuable set of comparisons performed from the Anton Dohrn established compatibility between the M60 and VIZ temperature and humidity measurements. However, the small but statistically significant mean temperature difference shows the value of including as part of the F.R.G. documentation the type of sonde (VIZ or M60) used for each ascent. These results have been presented with only limited speculation on the reasons for observed differences. As work progresses and more data become available, we hope to suggest simple models for the biases, and for normaliz- ing the data from the various observation system so they may be combined into an international data set of sufficient quality to meet the requirements of the GATE Convection Subprogram. PART II. COMPARISON OF WIND DATA FROM U.S.S.R. AND U.S. SYSTEMS 1 . INTRODUCTION The results of the comparison of U.S.S.R. vs. U.S. rawinsonde wind data are described in this report. The data were acquired during the formal GATE inter- comparison periods, and by the ships V anguard and Oceanographer during Phase I; (see table 4). Statistics were computed for four sets of comparative data. (a) Data Set 1 — These data consist of winds computed from simultaneous obser- vations of 14 rawinsonde flights from the O ceanographer and Vanguard during Phase I. All radiosondes were launched from the Oceanographer . Only the hyperbolic solution was used in obtaining Oc eanographer winds for Data Set 1 . (see appendix B, sec. 3). (b) Data Set 2 — This set was obtained from the same flights as those used for Data Set 1. The Oceanographer winds in Data Set 2 were determined by the multiple solution technique, with most winds ultimately being computed from the three-transmitter elliptical solution; (see appendix B, sec. 3). (c) Data Set 3 — These data were obtained on flights during intercomparison periods I and Ilia between the U.S.S.R. ships Zubov o r Krenkel and the Ocean- ographer . Twelve flights launched and tracked from the U.S.S.R. ships were also tracked by the Oceanographer . Oc eanographer winds were computed by only the hyperbolic method since lack of precise information at this time on the geographical coordinates of the launch point, i.e., the U.S.S.R. ships, pre- cludes the use of the multisolution method. (d) Data Set 4 — These data consist of 14 comparisons between a U.S.S.R. ship and the Researcher during Comparative Periods I, II, and Illb. Flights were launched and tracked from the Zubov or Krenkel during IC I, from the Okean during IC II , and from the Poryv during IC Illb: Three of the Researcher and Okean comparision flights during IC II were also tracked by the Vanguard radar. The winds from the Researcher were processed by only the hyperbolic method. These represent the most definitive data sets available for comparing the rawinsonde wind data. Additional studies are planned upon receipt of the observed data from the A/'B, B, and C arrays of ships. 2 . RESULTS The statistics presented in tables 5 to 8 are for 40-mb-layer averages for each of the comparative data sets. In these tables, the letter H in parentheses in the header indicates the hyperbolic solution and the letter M the multisol- ution. All the tables contain the following elements: Col. 1 ~ Layer for which data apply. Col. 2 - Number of values used in the computations. It is apparent from these figures that shortage of data exists in the lowest 50 to 100 mb and above 100 mb. Cols. 3 & 6 - Average value of U (zonal) and V (meridional) wind component , obtained by averaging data from both observation systems. Cols. 4 & 7 - Difference between average values of U and V components, respectively, for the two observa- tion systems. Cols. 5 & 8 - Standard deviation of differences between U and V components, respectively, of the two systems. The bottom row (sum) combines the results for all layers, 2.1 Oceanographer -Vanguard Intercomparisons In evaluating the comparisons between the Oceanographer VLF winds and Vanguard radar winds (tables 5 and 6) , it appears reasonable to consider the Vanguard winds as the standard for comparison; (see appendix B) . As a first approximation, all Oceanographer-Vanguard differences are interpreted as errors in the Oceanographer winds. A comparison of the Oceanographer winds in tables 5 (hyperbolic method) and 6 (mult isolut ion method) offers strong evidence for the superiority of the multisolution method of computation. (1) The standard deviations of the differences are significantly less for the multisolution method (1.52 vs. 2.86 m/sec for the U component, and 1.37 vs. 2.57 for the V component), and (2) The average absolute difference of the layer mean values, excluding layers above 120 mb because of sparse data, is smaller for the multisolution method (0.71 vs. 0.82 m/sec for the U component and. 0.52 vs. 0.76 for the V component) The average difference in the V component for all layers is also small (0.21 m/sec) for the multisolution method. The average difference in the U component for all layers (-0.39 m/sec) is, however, larger than that obtained from the hyperbolic method, and also larger than would be expected by chance from a normally distributed sample of 223 values whose standard deviation is 1.52 m/sec. Figures 17 and 18 show that the distributions of 5-mb Oceanographer-Vanguard differences for the U and V components obtained from the multisolution method have a greater number of extreme values than would be the case for a. normal distribution. Accord- ingly, significance tests based on the assmuption of normality may not be valid. Nevertheless, figure 17 also gives evidence of a general negative bias in the U component differences. The possibility of a small systematic error in the multisolution values should be examined when more data become available. Quality control tests were applied in the course of both the hyperbolic and multisolution processing in order to eliminate obviously bad data; (see appendix B) . Approximately 7% of the data processed by the hyperbolic method were eliminated in the multisolution processing, suggesting that this techni- que more clearly discriminates between acceptable and obviously bad data. Figures 19 and 20 show the standard deviations (SD) of the Vanguard - Oceano - grapher wind component differences vs. averaging thickness. Following the 10 convention used in tables 5 and 6, the curves marked Oceanographer H were ■ obtained from the hyperbolic and the Oceanographer M curves from the multi- solution technique. The irregularity in the curves is due to slight but unavoidable inconsistencies in the computations arising as a result of dif- ferent combinations of the 5-mb data points in the variable layer averages, the effect of missing data which at times leads to slightly different inter- polated values, and a decrease in the amount of data as the thickness of the layers increases. These curves again demonstrate the clear superiority of the multisolution technique in reducing the random error of the differences. Sigma values of around 2.5 m/sec are computed for the 5-mb data. Values decrease to around 1.5 m/sec when values are averaged through 50 mb. For thicknesses greater than 40 mb, the decrease in SD is approximately proportional to 1//n7 where N is the number of 5-mb points used to obtain the layer average. The rate of decrease in SD as the layer increases from 5 to 40 mb is significantly less than 1/vNT This suggests a positive correlation between wind errors in adjacent layers up to 40 mb thick but little correla- tion of errors between layers more than 40 mb thick. It is largely for this reason that we chose to present the statistics of tables 5 to 8 in terms of 40-mb layers. The indication of positively correlated errors between adjacent shallow layers was expected, partly because the 5-mb values are obtained by fitting a second-order polynomial to 2 min of data centered at the level of interest; (see appendix B) . Two min of data represent a layer about 60 mb in thickness near the surface and 30 mb near 500 mb. Also shown as a function of averaging thickness are the correlation coeffi- cients between the Oceanographer -Vanguard wind component variations (figs. 21 and 22). It should be noted that these computations were performed using departures from the layer mean value in order to eliminate correlations arising simply as a result of vertical variations in the mean profile. The higher correlation coefficients obtained for the multiple solution values would be expected from the previous results. The distribution of Oceanographer (M) -Vanguard wind component differences have also been examined as a function of averaging thickness. Figures 23 and 24 show isolines of the percentages of differences equal to or less than the abscissa value, vs. averaging thickness. The irregularities in the curves are again the results of factors previously noted in the discussion of figures 19 and 20. Plots of the 5th and 95th percentiles for the U and V differences are shown on figure 25 and of the 2.5th and 97.5th percentiles in figure 26. The tails of these distributions are of interest. There is a rapid decrease in the span containing 90% or 95% of the data as the layer average is increas- ed from 5 to 40 mb, with slower and more irregular decreases for greater thicknesses. It is apparent from figures 25 and 26 that the distribution of U differences is broader than for V differences. The V tails and the positive U tail behave similarly as the averaging thickness increases, although, as previously noted, the 95% and 97.5% values for the U differences are larger than the corresponding V values. The behavior of the negative U tail is also 11 similar for thicknesses less than 30 mb, but for greater thicknesses the values decrease much more slowly. Although possibly only a feature of this particular data set, this suggests that large negative errors in the Oceano - grapher U component tend to occur through thick layers, a possibility that will be investigated in future intercomparison studies. The results thus far described combine data from all levels. The distribu- tion of errors with height has also been examined by computing the standard deviation of U and V differences as a function of averaging thickness for 3 separate layers (table 9 and figure 27) . An increase in the standard devi- ation of the differences with height is apparent. There should be no height or slant range dependence in the VLF signal phase measurement, or wind compu- tation. Vanguard radar measurement accuracies do depend on slant range, but we believe the Vanguard radar wind errors are very small, i.e., fractions of a meter/sec, at the slant ranges encountered during GATE. We believe the increase with height of the standard deviation of the differences most likely arises from 403-MHz telemetry problems which increased with increasing range between ship and radiosonde during GATE. The distributions of 5-mb differences for each 50-mb layer are given in tables 10 and 11. The modal value for the U differences is generally between and -1 m/sec, and the layer-to-layer variation is similar to the variation of layer mean differences presented in table 6. The distribution of errors broadens significantly with height, particularly above 600 mb for negative differences, but the bias does not change with height. The modal value for the V differences (table 11) is clearly negative in the lowest layers, becoming generally positive above 850 mb, suggesting a posi- tive bias through most of the column. The changing value of the bias in the layers below 850 mb may be due to sampling differences arising as a result of the rapid decrease in the number of samples in the lower layers. There is again a broadening of the error distribution with height. 2.2 U.S . -U. S. S.R. Intercomparisons 2.2.1 Vanguard -U . S . S . R . Intercomparisons Three of the Okean-Researcher comparative soundings launched during Inter- comparison IC II were also tracked by the Vanguard radar. These are the only comparative data available for the U.S. S.R. ship radar and Vanguard radar data Figures 28a, b, and c are plots of these 3 wind profiles and include the Researcher Omega data as well. A thorough statistical evaluation is not merited because of the small amount of data. It is clear that the Okean- Vanguard data compare quite favorably over most of the soundings although there are major differences, particularly above 300 mb. Two points on the Okean profiles of figure 28c are clearly in error. Also, the Researcher U component is unusually smooth and therefore questionable above 600 mb on figure 28a. 12 2.2.2 Omega and VLF - U.S.S.R. Inter comparisons Statistics for comparisons between the U.S. -VLF ( Oceanographer ) or Omega ( Researcher ) system and the U.S.S.R. system derived from data acquired during the formal intercomparison periods are summarized in tables 7 and 8. Note that the Oc eanographer and Researcher data in these tables are obtained from the hyperbolic solution. Consequently, observation phase data from the Oceano - grapher , which are obtained from the multisolution technique, can be expected to agree more closely with U.S.S.R. data. Intercomparison and observation phase data from the Researcher system should, however, be similar since the hyperbolic solution is used to obtain Researcher winds during both phases. The U.S.S.R. comparative data received at the CSDC were for mandatory and significant levels. It appears that the significant levels were usually determined by temperature and humidity criteria, rather than by wind shear. (See appendix A.) Five-mb data points were obtained by linear interpolation between the U and V values at the mandatory and significant levels. The data in tables 7 and 8 exhibit the following features: a. The component mean differences are less than a/vN, i.e., not sifnifi- cantly different from what would be expected by chance. There is no evidence of a system bias. b. In our previous analyses, we assumed the Vanguard wind data to be the most accurate. Given this assumption one would expect the standard deviations of the differences for the "sum" in table 5 ( Oceanographer (H) -Vanguard inter- comparison) to be smaller than the comparable values in table 7. That this is not observed may be due in part to the fact that the Oceanographer hyper- bolic data which was compared with Vanguard data included 3 soundings with questionable data above 300 mb. This is reflected in table 12 by the high sigma values of the U component differences in layers above 320 mb. Note also that these data were rejected in the multisolution processing (compare "Number" values in tables 5 and 6) which more clearly discriminates between good and bad data. Even with these data removed, the SDs for the Oceano - grapher (H)-U. S .S ,R. wind component differences are similar to those obtained from the Oceanographer (H) -Vanguard comparisons, again suggesting that the U.S.S.R. wind data compare favorably with that collected by the Vanguard . Additional information on the relation between the U.S. and U.S.S.R. data are presented in table 12, which lists the variances, covariances, and correlation coefficients for each of the 4 pairs. The variances of the measured wind components arise as a result of the natural variability in the signal and the errors in the measured values, e.g. , IL» U + e, M where U w is the measured value of U and e is the error in the measurement. M Assuming signal and errors of measurement to be uncorrelated, VAR U M = VAR u+ VAR e . 13 Since the Vanguard wind measurements are assumed to be the standard for comparison, their variances should be less than those of the Oceanographer . This is observed in all cases except the V component obtained from the multi- solution technique, in which case the SDs have approximately the same value. Furthermore, the smaller Oceanographer variances for the multisolution techni- que as compared to those for the hyperbolic technique are consistent with our previous conclusions concerning the superiority of the multisolution technique, The expression for the correlation coefficient can be written R = =-^ ; =— - (1) 1 + where o is t^e standard deviation of the signal, i.e., the true natural w variability, and £.. , e„ are the errors in the measured values from systems 1 and 2. It can be seen from (1) that the sample-to-sample variation in the correla^ tion is a result of variations in the signal-to-noise ratio, e/a . Thus the w correlation coefficient is not a particularly good parameter for comparing the quality of measurements taken under different atmospheric conditions. For instance, the high correlation coefficient (0.95) computed for the Researcher (H)-U.S .S.R. U component comparison largely reflects more variable atmos- pheric conditions as demonstrated by the large variances of the measured values from the two systems. Including a few questionable Oceanographer data in the Oceanograph er (H) - Vanguard comparison has previously been noted. This accounts at least in part for the small correlation in the Oceanographer (H) -Vanguard U data com- pared with that obtained for the Oceanographer (H)-U.S. S.R. 3. CONCLUSION a. Over 50-mb layers, differences between VLF wind components and those determined by radar tracking have standard deviations of about 1.5 m/sec. This result, obtained from the most sophisticated processing strategy avail- able, approaches the requirement for the GATE B-scale analysis. b. For smaller vertical layers the result is less satisfactory. In 5-mb layers the standard deviations of the differences between VLF and radar wind components are about 2.5 m/sec. This result will not be adequate for all C- and D-scale objectives of GATE. c. The winds from the MTF -Omega ( Researcher ) appear to compare less favorably with radar tracking than do the VLF winds. The data recorded on the Omega system are not compatible with the most sophisticated processing strategy, i.e., the multisolution technique being used for the VLF data from the Oceanographer , Dallas , and Gilliss . 14 d. Comparisons to date of Omega and VLF data with the U.S.S.R. ship winds have not included the mult isolut ion strategy. The results of only nine flights indicate that the U.S.S.R. wind data compare favorably with that ob- tained by the Vanguard system. The MTF Omega -U.S.S.R. comparisons produce a standard deviation of differences somewhat larger than the VLF-U.S. S.R. comparisons. Much work has yet to be done to document the quality of the rawinsonde wind data. The simultaneous flights launched from the Vanguard and Oceanographer during Phase I and from the Vize and Oceanographer during Phase II will be of prime importance in later studies. 15 APPENDIX A U.S.S.R. Upper Air System and Data Processing 1. UPPER AIR SYSTEM This summary is based on our understanding of material extracted from the following publications kindly provided by the U.S.S.R. : Instructions to Hydrometeorological Stations and Posts , Issue 4, Aerological Observations at Stations, Part 3a, Temperature-Wind Sounding of the Atmos- phere with the "Meteorit" - RKZ System; Gidrometeoizdat , Leningrad 1973, 256 pp. Instructions to Hydrometeorological Stations and Posts , Issue 4, Aerological Observations at Stations, Part 3b, Temperature-Wind Sounding of the Atmos- phere by the "Meteorit" -RKZ System on Ship Stations; Gidrometeoizdat, Leningrad 1974, 133 pp. (translated). Acc uracy of Data from Aerological Sounding at Sea ; V. M. Linkin, V. N. Bakin, V, I. Nikonov, V. V. Udalov; TROPEKS-72, Trudy Mezhvedomstvennoi geofiziches- koi ekspeditsii po programme natsional 'nogo Alanticheskogo tropicheskogo eksperimenta, M. A. Petrosiants (chief editor); "Gidrometeoizdat," Leningrad, 1974, pp. 59-597. The U.S.S.R. ships which participated in GATE were equipped with "Meteor it - R" surface equipment, and used the RKZ-2 sonde. The U.S.S.R. sounding system has the following features. (1) The radar antenna is mounted on a stabilized platform. (2) Slant range, azimuth, and elevation angle are measured. Slant range is measured by the time difference between an interrogation pulse and receipt of a transponder signal from the sonde. (3) The carrier frequency is nominally 1982 MHz. (4) The sounding pulse tracking frequency is 833 pps. (5) The pulse power of the transmitter is 200 kW. (6) The pulse length of the transmitter is 0.8 us. (7) Antenna disk diameter is 1.83 m. (8) The automatic tracking range is 150 km. (9) The mean range error of angular coordinates is 0.12 degrees. 16 (10) The mean range error is not more than 40 m. (11) Provision is made for automatic input of ship's heading so that true azimuth is reported. (12) The recording cycle time is 5 sec for meteorological data and 30 sec for time and position data. (13) The recording precision (on printed paper tape) is 0.06 degrees for angle measurements, 10 m for slant range and 1 Hz for frequency. The RKZ-2 sonde has the following features. (1) The carrier frequency is nominally 1782 Hz. (2) The reference frequency is 2000 Hz. (3) The temperature signal frequency range is 500 to 1950 Hz. (4) The relative humidity frequency range is 1500 to 1950 Hz. (5) The length of pauses in the pulse code is 50 to 300 us. (6) Signal commutation is by a baroswitch. (Pressure is not reported.) (7) The temperature sensor is a rod thermistor (size not given) . It is exposed to the air on the outrigger. (8) The relative humidity sensor is a membrane of animal skin mounted on a circular frame and exposed in a duct at the top of the sonde. Three sets of calibrations are provided with each sonde: (1) An 18-point oscillator calibration of (sensor-signal frequency)/ (reference frequency) with standard resistances in place of the sensor. (2) A temperature-resistance calibration of the thermistor included with the sonde. (3) A relative humidity-resistance calibration of the humidity unit includ- ed with the sonde. The oscillator calibration is normally combined with the sensor calibrations for convenience in the data reduction. Instructions are given for preparing new combined calibrations if sensor replacements are necessary in the field. During the pref light baseline test, a two-point calibration check is made on the sonde oscillator and one-point checks are made on the temperature assembly and humidity assembly. Several criteria are specified, which, if not met, are reason for rejecting the sonde, temperature assembly, or humidity assembly. Small temperature and humidity errors discovered in baselining are corrected in processing by adding fixed corrections (offsets) to all derived 17 temperature and/or humidity values. The maximum allowable correctable errors appear to be 2°C for temperature and 20% for relative humidity. 2. TEMPERATURE AND HUMIDITY DATA PROCESSING Data processing was normally semiautomated on the GATE ships, with input data being manually selected and punched onto paper tape for input to Minsk- 22 computers. The manual is not specific on this point, but it appears that sounding input data were from manually selected significant points. Radiosonde temperatures are routinely corrected for insolation error by an amount dependent on sonde altitude and solar elevation angle as illustrated by the following sample values: Table A-l. --Selected Values of Insolation Correction for Soviet Rawinsonde Temperatures. (Values are degrees Celsius.) Height -3 Elevation angle (degrees) (Km) 30 60 90 1 0.0 0.0 0.4 0.5 0.5 5 0.0 0.1 0.5 0.7 0,7 10 0.0 0.2 0.7 1.0 1.0 15 0.0 0.3 1.3 1.8 1.9 20 0.1 0.4 2.1 2.5 2.9 25 0.3 0.9 3.1 3.7 4.2 3. WIND DATA PROCESSING The following coordinate times are used for wind computations. a. Every 30 seconds up to the third minute after launch. b. Every minute from the third through the tenth minutes. c. Every 2 minutes from the 10th through the 40th minutes. d. Every 4 minutes from the 40th minute to the end of observation. Wind speed and direction are computed for each of the above layers and the wind assigned to the mid-time and altitude for the layer. Before slant range can be read (i.e., the first minute or so of flight) sonde height is determined by extrapolation of the time-altitude curve of 18 portions of the sounding. Tracking in elevation is apparently good before slant range can be measured. An alternative method of determining sonde height before slant range can be measured is mentioned but we are unclear as to what this method is. Ship movement is taken into account in computing the winds. Instructions are given for correcting height computations for Earth curva- ture and radio-wave refraction. However, it is not clear when or whether these corrections were applied. Wind speed and direction for reported points of the sounding are obtained by linear interpolation in height. 4. SELECTION OF MANDATORY AND SIGNIFICANT LEVELS 4.1 Mandatory Levels Examination of the U.S.S.R. data and data documentation indicates that data values are being given for a. Standard heights (m) of 200, 310, 500, 610, 910, 1000, 1500, 2000, 3000, 4000.... to the top of the sounding. b. Standard pressures (mb) of 1000, 900, 850, 800, 700, 600, 500, 400, 350, 300, 250, 200, 150, 100, 70, 50, 30, 20, 15, and 10. 4.2 Significant Levels Selected points are included in accordance with criteria specified in the WMO Manual on Codes (WMO No. 306, FM 36-V, TEMP / SHIP ) . These criteria can be summarized as follows. Pressure - surface and top of sounding. Temperature - a. maximum and minimum b. Base and top of inversions and isothermal layers of at least 20-mb thick- ness provided the layer occurs below the 300-mb level. c. Temperature obtained by linear interpolation between adjacent significant levels shall not depart from the observed temperature by more than 1°C below 300 mb, and 2°C above. Relative Humidity - a. Maximum and minimum. b. Relative humidity obtained by linear interpolation between adjacent signi- ficant levels shall not depart from the observed relative humidity by more than 15%. 19 Temperature and Relative Humidity Significant levels selected shall, insofar as possible, be the levels at which the prominent changes in lapse rate of temperature or relative humidity occurs. Winds - a. Direction - the deviation between two previously selected significant levels shall not exceed 10°. b. Speed - the deviation between two previously selected levels shall not exceed 5 m/sec . c. maximum winds - shall be above the 500-mb surface and greater than 30 m/sec. - shall be recorded if they meet the above criteria or if they exceed the adjacent minima by 10 m/sec or more. Tropopause - a. The first tropopause is defined as the lowest level at which the lapse rate decreases to 2°/km or less, provided also the average lapse rate between this level and all higher levels within 2 km does not exceed 2° /km. b. A second tropopause is defined if above the first tropopause the average lapse rate between any level and all higher levels within 1 km exceeds 3°/km. This tropopause may be within or above the 1 km layer. U.S.S.R. Criteria Documentation accompanying the U.S.S.R. data indicates the following signi- ficant level criteria were used, in addition to the WMO criteria. Temperature - 0°C level Level of cloud base 20 APPENDIX B U.S. UPPER-AIR SYSTEMS AND DATA PROCESSING ■ 1. U.S. UPPER AIR SYSTEMS 1.1 NAVAID Systems The U.S. vessels Dallas , Gill is , and Oceanographer were outfitted with Beukers upper air systems equivalent to the Beukers W-3 system. The U.S. equipment consisted of Beukers W-2 systems that had been refurbished and had an additional Ampex tape recorder. The system software was also equivalent to the Beukers W-3 software used by several other countries in GATE, but interfaced with the somewhat different configuration of recorders in the U.S. system. These systems could use either OMEGA or VLF radio transmissions for determining winds. the VLF mode was used for nearly all soundings. The sondes used with the Beukers systems were manufactured by VIZ Manufacturing Company, Philadelphia, Pa., Model 1232-300 LO-CATE sondes were designed to retransmit VLF signals in 10-25 KHz band and were equipped with premium meteorological sensors that required no base-line check prior to launch. The MTF-OMEGA system, developed at the Mississippi Test Facility (MTF) of the National Aeronautics and Space Administration, was the unit used in the pre-GATE Wallops Island Tests and the GATE International Sea Trials (GIST). Only one prototype system was constructed and it was installed on the U.S. ship Researcher . The MTF equipment logged preprocessed OMEGA and meteorologi- cal (met) information on magnetic tape, but required a separate computer for data reduction. This system was restricted to using OMEGA radio transmissions for windfinding. The sondes used with the MTF system were manufactured by VIZ Manufacturing Company (Model 1224-300 designed to retransmit 13.6-KHz OMEGA NAVAID signals) and were also equipped with premium meteorological sensors that did not require a base-line check before launching. 1.2 Vanguard System The bulk of the Vanguard data acquired during GATE consisted of FPS-16 (C band) radar tracking data recorded on magnetic tape and met data (pressure, temperature, humidity) from a VIZ Model 1298 radiosonde, recorded on a strip chart . 2. OPERATIONAL EXPERIENCE Two major problems affected the quality and quantity of the U.S. NAVAID data. First, the 403-MHz telemetry link was the cause of considerable diffi- culty in both the acquisition and processing of data. Frequency drift during flights was a common complaint of operators. One result of the drift was radio interference at high altitudes, apparently from nearby sondes. Careful monitoring and adjustment of the receiver coupled with frequent changes in antenna direction helped minimize this problem but did not eliminate it. 21 A second problem was cross-talk between the meteorological and VLF or OMEGA transmissions from the sonde. This problem was partially corrected during the in-port period prior to Phase III by alterations to the electronic circuitry on each sonde and has been treated with some success during the processing of the data. 3. POST-FIELD PHASE DATA PROCESSING This discussion is divided into four parts. Section 3.1 describes the pri- mary automatic data processing (ADP) step necessary to produce 5-mb data for the Researcher , Gilliss , Dallas , and Oceanographer , and section 3.2 does the same for Vanguard . Section 3.3 outlines the steps taken to clean up many of the errors in the Gilliss , Dallas , and Oceanographer 5-mb data output by the primary processing step. Section 3.4 describes the manual review and data validation step through which 5-mb data from all ships pass to generate the archive data files. The primary production software processed upper -air data recorded at sea by shipboard systems designed and manufactured by Beukers Laboratories Incorpo- rated (BLI) and radiosondes designed by BLI and VIZ Manufacturing Company (VIZ) and manufactured by VIZ. With minor modifications, the software can also process data recorded on a shipboard system designed and fabricated by the NASA Mississippi Test Facility (MTF) and used on the Researcher during GATE. Because of the large volume of data and the limited time available for processing after the experimental phase of GATE, the^ production software was designed to process data recorded at sea without preliminary examination of the data or any manual intervention. Diagnostics identify the source of failures which helps to expedite reprocessing and making corrections to each sounding . The correction software accepts the output of the primary production soft- ware, namely 5-mb data ready for archiving (except for occasional, objective- ly identifiable errors) plus all input to the winds computation routine. With some manual assistance it also interpolates through or deletes nonsense. Our limited experience has shown that the loss of radiosonde-to-ship 403-MHz telemetry was the most common source of trouble with pressure, temperature, wind and humidity data. Remaining problems with wind data are misbehavior of the NAVAID transmitters, operator error, and hardware design deficiences (other than telemetry loss) . The processing strategy for the wind data from the NAVAID systems is a mix- ture of automatic and manual processing steps. The VLF and OMEGA winds determined in the primary production step of the processing are computed using a hyperbolic coordinate system for the trans- mitter phase measurement. In this computation scheme, data from three transmitters are required. The Beukers upper air systems and the ship navigation data on the Gilliss , Dallas , and Oc eanographer provide the capability to also determine winds using a computation based upon two-transmitter elliptical solutions. The term elliptic arises from the fact that a phase measurement of any one transmitter 22 locates the sonde on the surface of an ellipsoid whose foci lie at the ship and the transmiter. Two such ellipsoids, defined by phase measurements from transmitter 1 and 2, 1 and 3, or 2 and 3, intersect in a curve. The inter- section of this curve with the z surface corresponding to the radiosonde height defines the position of the radiosonde in space. When three ellipsoids are intersected with the z surface, the position is overdetermined . Since only two transmitters are needed to determine location, the redundant data in the three-transmitters case can be used to obtain another solution for position through least-square fitting. The mixture of hyperbolic and elliptical solutions permits a selection of the best of the 5 wind solutions, based on the following criteria: a. If one of the three transmitters is disabled or providing unrealistic values of phase, only one wind solution is possible. b. When acceptable phase measurement data are available from all three trans- mitters, each of the four elliptical solutions is compared with the hyperbo- lic solution for the entire flight, and mean differences are tested for stati- stical significance. Any elliptical solution passing this test is eligible to replace the hyperbolic solution if it is smoother, according to objective criteria, than the hyperbolic. Among the four elliptical solutions, three- transmitter solution has the highest priority, followed by the smoothest of the two-transmitter solutions. An accurate determination of the geographical coordinates of the surface launch point is required for processing VLF phase data by elliptic geometry. We have as yet been unable to establish the position of the Soviet launch ships with the accuracy required for elliptic geometry. Thus the formal com- parative data acquired by the U.S. ships having Beukers equipment ( Dallas , Gilliss and Oceanographer ) have been processed by only the hyperbolic method. The vertical resolution of the processed wind data is dependent in a com- plicated way on the processing techniques. The 5-mb wind values are computed from phase data spanning 2 min centered on the time the sonde traversed each pressure level. The processing software includes an editing and filtering step designed to remove or flag suspicious or obviously bad data (Acheson^ 1975) . This filtering process results in vertical resolution of roughly 500 m for the VLF and OMEGA data, (200 m for Vanguard ) . Vertical resolution is further degraded in parts of soundings where interpolated data have been introduced . 3.1 Primary Production Software 3.1.1. Raw Data Base The raw data base on magnetic tape from which the archive data were derived consists of four met data words each 0.8 sec and 12 or more NAVAID words each 10 sec. The four met data words represent the frequencies generated by a resistance- controlled oscillator whose input is switched by a solid-state commutator each 0.2 sec between pressure, temperature, and humidity sensors, and a refe- 23 rence resistor. The resistances of the thermistor and carbon element humidity sensor (hygristor) are functions of temperature and relative humidity. The pressure sensor is an aneroid capsule linked to an arm driven across a commut- ator by the changing pressure. This commutator has three interwoven combs of electrically conducting material laid down upon a nonconducting substrate, constituting a baroswitch. Four distinct frequencies are associated with the baroswitch. When the arm rests on a nonconducting region of the commutator, a second reference frequency of about 1000 Hz is generated. The three inter- woven combs of conducting material generate frequencies of about 1200, 1400, and 1600 Hz when contacted. The three different types are used to produce a contact pattern cyclic in 5 contacts with a subpattern cyclic in 15 contacts. The manufacturer furnished a calibration with each baroswitch that establish- es the pressure corresponding to the leading edge of each contact. The refe- rence frequency is obtained when the solid-state commutator switches zero resistance into the oscillator. The frequency produced, about 2000 Hz, then represents the internal resistance of the oscillator circuitry. A minimum of 12 NAVAID data words are available. For each NAVAID trans- mitter in use, there are four words — the phase of the transmitter signal received and retransmitted by the radiosonde, a signal quality flag for this phase measurement, the phase of the same transmitter received directly at the base (shipboard) station, and its signal quality flag. Signals from a mini- mum of three transmitters were used during GATE (see table A-l) . In addition to these data base words, time to the nearest 0.1 sec was record- ed on magnetic tape with each frame (4 met words or 12 or more NAVAID words) . Table B-l . —NAVAID Transmitter Used During GATE Code Type Location Frequency (Hz) A OMEGA Norway 13600 B OMEGA Trinidad 13600 C OMEGA Hawaii 13600 D OMEGA North Dakota 13600 H OMEGA Forestport, NY 13600 I VLF Cutler, Me. 17800 K VLF Balboa, Canal Zone 24000 L VLF Rugby, England 16000 24 3.1.2 Steps in Met Data Processing a. Correction of pressure, temperature, and humidity frequencies by the 2000- Hz reference. b. Extraction of 1000-Hz reference frequencies from the pressure frequencies. c. Detection of baroswitch contact leading edges at the time at which a trans- ition from 1000 to 1200, 1400, or 1600 Hz occurred. d. Correction of temperature and humidity frequencies by the 1000-Hz reference. e. Computation of thermistor and hygristor resistances from the temperature and humidity frequencies and the known frequency-to-resistance transfer func- tion of the oscillator. f. Computation of temperature from the known resistance-to-temperature trans- fer function of each thermistor lot number and 30°C resistance of each thermis- tor. g. Editing of temperatures and hygristor resistance to remove wild data points and other outliers. (Acheson, D. T., "Data Editing-Subroutine EDITQ", NOAA TM EDS CEDDA-6, June 1975, pp. 12). h. From step (c) an "observed" baroswitch pattern of 1000 to 1200, 1000 to 1400, and 1000 to 1600 Hz transition was obtained along with the transition times. This observed pattern was compared with the known pattern and cali- brated pressures, the correspondence made, and the time and pressure at each baroswitch contact established. i. Logrithmic interpolation of pressure from the contact time and calibrated pressure done to compute a pressure for each temperature and humidity (each 0.8 sec) . j. Radiosonde ascent rate computed from the hydrostatic equation in the form p (dz/dt) - (dp/dt)/g Q p = density (kg/m 3 ) z = geometric height (m) t = time (sec) p = pressure (pascal) g = 9.80 (m/sec 2 ) k. Lag correct temperature using the equation* T = T T (dT/dt) + T c T 25 T = lag corrected thermistor temperature (°k) T_, = 9.77/(p dz/dt) -^ 3 (sec) 1 T = uncorrected thermistor temperature (°k) j-r— = slope of a least-squares straight line fitted to 11 values of T centered on time t, (sec) 1. Correct the hygristor temperature for lag by solving the equation* x H (dT H( ,/dt) + T HC - T c T„ n = hygristor temperature corrected for lag t tt = 52.0/(p dz/dt)°- 7 1 T - thermistor temperature corrected for lag t = time m. Computation of relative humidity from hygristor resistance and tempera- ture using the equations: 2 3 4 RH = 100 - l/(a n + a n x + a_x + ax + a.x ) RH > 33% 12 3 4 — RH = 33.0 - (3 Q + B^x + 3 2 x 2 + ^x 3 + 6 4 x 4) RH where x + ( T(J + x^ + x^* + x 3 T HC 3 ) ln e (R/R33) x., i = 0, 3 are known coefficients R = hygristor resistance R = calibrated hygristor resistance at 33% -J -J a., i = 0,4 and 6., i = 0,4 are hygristor lot number dependent coefficients RH - observed relative humidity ^Williams, S. L. and D. T. Acheson, "Thermal Time Constants of United States GATE Radiosonde Sensors" (To be published) . 26 The forms of the equations and the values of the coefficients are such that the equations are asymptotic at 100% and 0% RH. The distortion forced by this procedure is small at the high RH end not exceeding 1.5% RH at 95% RH, which is nearly the maximum observed before any correction. However, at the low end below 10% RH, the departures from average calibration data are significant Use of the asymtotic equation is justified in view of the uncertainty of hygristor calibrated data in the region, hysteresis of the hygristor combined with the extreme sensitivity of a nominally accurate transfer function at very low humidities, the desirability of preserving relative variations at low humidities instead of truncating computed negative values. n. Correct the observed water vapor pressure for hygristor temperature. e = RH e (T, ) , c w he where e = corrected water vapor pressure and e (T, ) = saturation vapor pressure at temperature T, w he he . o. From pressure, temperature, and water vapor pressure, compute specific humidity and geometric height. For the height computation, use Lambert's equation for the acceleration of gravity as a function of latitude and height. At this stage in the processing, we have still retained the original met data time base; e.g., values for pressure, height, temperature, and specific humidity exist at 0.8-sec intervals throughout the flight. No corrections have been applied for radiation heating or cooling of the thermistor or hygristor . Archived met data at each pressure level P. were obtained by extracting all 0.8-sec height (Z) , temperature (T) , and humidity (q) values lying between (P. - 2.5) and (P + 2.5 mb) and least-squares fitting these with three second- order polynominals. The polynomials at level P were archived as Z. , T , , and q and the square roots of the residual variances from the temperature and humidity polynomial fits archived as temperature and humidity error estimates. These error estimates reflect only random variations and do not include any systematic errors or biases from any source. 3.1.3 Wind Data Processing The 5-mb winds were computed from phase data spanning 2 min (13 10-sec data frames) centered on the 5-mb times. Since the computation of winds was essentially a geometric transformation from one coordinate system to another, a similar transformation was, in effect, done on the phase error estimates to obtain the wind error estimates. Thus, wind error estimates reflect only ran- dom scatter and do not incorporate any measure of systematic error. 3.2 Vanguard Data Processing The bulk of the Vanguard data acquired during GATE consisted of FPS-16 (C band) radar tracking data recorded on magnetic tape and met data (pressure, temperature, humidity) from a VIZ Model 1298 radiosonde recorded on a strip 27 chart. The met data were extracted manually from the strip chart and pro- cessed to provide pressure, height, temperature, and specific humidity at each baroswitch contact. The only source of time for met data was the strip chart abscissa, which was unreliable and was not used. 3.3 Correction Software Output at 5-mb intervals from the primary production processing is not suit- able for archiving. During intervals of radiosonde-to-ship telemetry dropout, values of temperature, humidity, and usually winds are erroneous, albeit with large errors. Heights, integrated upward from the surface, are greatly in error. The reprocessing software accomplishes the following in the order given. a. The 5-mb data output from the production processing is scanned for temper- atures _> 99.9°C or temperature error estimates a > 1.0°C. The specific hum- idity q and its error a are scanned for either a > 0.99 g/kg, or a / q > 5.0 q q — q — or RH > 105%, or the temperature flagged as erroneous. b. If the temperature is flagged, a value is obtained by linear interpolation between two acceptable values, and o is set to the negative of the average o of the two acceptable values. If interpolation is not possible because flagged values reach to the top of the flight, the flight is truncated at the last level having an acceptable temperature. c. Flagged humidity data are interpolated linearly if possible, and a is set to - (average of a ' s of the two acceptable values used for interpolation) g/kg. If interpolation is not possible at the top of a flight, q and a are both set to 0.00 g/kg. q d. Height is recomputed from the first level for which either temperature or humidity or both has been interpolated or extrapolated upward to the end of the flight. e. Winds are recomputed as below, using data cards specifying the t. of the occurrence of a transmitter jump and/or the time span (t to t ? ) of a trans- mitter outage as evidenced from the phase plots output by primary production processing : (1) If phase data have not been deleted on either side of a jump, the magnitude of the jump is recomputed by least-square fitting straight lines to 13 frames (120 sec) of radio range data on both sides of the jump and using the gap between fitted lines at time t. as the jump magnitude in place of that estimated from the phase plot. (2) All data used to compute winds*during the primary processing at each 5-mb level are recorded. Radio ranges are corrected for jumps and deleted between times t and t for the transmitter that is out. (3) These radio range data, now free of all observable systematic errors, are used to compute all possible combinations of winds among 28 three-transmitter hyperbolic three-transmitter elliptic three-transmitter elliptic (transmitters 1+2, 1+3, 2+3). (4) A variety of statistics are computed, the most significant of which are the mean differences between elliptical solutions and the hyperbolic solution and a smoothness function S defined as: N L s= w (N-i) l < u r u i-i )2 i N = Number of 5-mb levels u - = u component at level i v - = v component at level i An alternative to the three-transmitter hyperbolic wind solution is sought among the four elliptic wind solutions. Of the elliptic solutions, where mean differences (u and v) from the hyperbolic solution is statistically small, the elliptic solution is chosen if S .... . . < S, , -, . with the elliptic elliptxc hyperbolic solutions given priority as a. Three-transmitter elliptic b. The two-transmitter elliptic solutions with smallest value of S (5) All wind solutions for the entire flight, a statistical summary, the wind solution chosen as "best," and other informative data are output to the line printer and reviewed. The reviewer may alter the choice of "best" or delete erroneous winds at any level. 3.4 Manual Review and Data Validation During manual review, the following actions were taken: a. If the pressure was in error, the flight was reprocessed with primary pro- duction software modified to allow flight-by-flight corrections to the baro- switch data. b. Temperature, humidity, and u and v wind components were corrected, flagged, or deleted. Corrected meant that data .were interpolated linearly with respect to pres- sure between adjacent acceptable values or deleted if extrapolation would be required at the top of a flight. All interpolated data were flagged by re- placing the error estimates of the interpolated data with the negative of the average of the error estimates of two adjacent acceptable data values. The negative sign flagging recomputed winds is removed. 29 Our motivation in adopting a conservative review and validation philosophy is based on our observation that the 5-mb data resolve a large amount of the variability in the tropical atmosphere that is unfamiliar to those accustomed to conventional upper -air observations; our desire to preserve as much data as possible consistent with reality; and our belief that the individual user is often better equipped than we to accept or reject data values for this purpose. REFERENCES Acheson, D. T., 1974, "OMEGA Windfinding and GATE," Bulletin of the American Meteorological Society , Vol. 55, No. 5, May 1974, pp. 385-398. Acheson, D. T., 1975, "Data Editing— Subroutine EDITQ," NOAA Technical Memo - randum, EDS, CEDDA-6 , National Oceanic and Atmospheric Administration. 30 Table 1. --Dates and locations of the GATE Intercomparison Periods Period Dates Location IC I June 17-19, 1974 13°N, 21°W IC II August 16-18, 1974 7°45'N, 22°12'W IC Ilia September 21-23, 1974 13°N, 21°W IC Illb September 21-23, 1974 12°N, 21°W Table 2. --(See next page.) Table 3. --Precision and vertical resolution of archived data, by Nation and parameter Pr ecision of recorded values Vertical tesolution Temperature Relative Humidity Specific Humidity Dewpoint USSR 0.1°C 1% Sig. level US 0.1°C 0.01 gkg" 1 5 mb Canada 0.1°C 1% 5 mb FRG 0.1°C .1°C Sig. level Table 2. --List of Formal Intercomparison Flights Used in the Analysis, by Date and Ship 31 U.S.A. U.S.S.R. Canada F.R.G. Date IC I IC II IC Ilia Sep. 21 IC Illb Sep. 22 Sep. 23 Time &> a bo o c o ■3 IS] PJ 1/5 cti • ■P <- t— i <4H trt Cfl i— i O G tLi 03 M f-i ^ Q D- UJ o > o 1200 1800 0600 1200 1800 0600 -J c a! June 17 1800 2100 X X X X June 18 0000 0900 1200 1800 2100 X X X X X X X X X X X June 19 0300 0900 1200 1800 2200 X X X X X X X X X X X X X X Aug. 16 1500 > X X Aug. 17 0300 0900 1500 X X X X X A Aug. 18 0000 0400 0600 1200 X X X X X X X X Aug. 18 1800 X X Sep. 21 1500 1800 2100 X X X X X X X X Sep. 22 0100 0600 0900 1500 1800 X X X X X X X X X X X X Sep. 23 0000 0600 0900 X X X X X X X 32 Table 4. --Data Used in the Wind Intercomparison Analyses Date Time > o t-J <4-l O u c CD H G ',- UJ c as CD O > y> o a, 0) u M O c aj June 18 1200 X xi 2100 X X X IC I June 19 0300 1200 2200 X X X X X X X X X Aug. 16 1200 x- X X u; II Aug. Aug. 17 18 0300 1200 0400 1200 X X 3 X X X X X X X X Sep. 21 1500 2100 X 4 X X X Sep. 22 0600 X X IC Ilia Sep. 23 0900 1500 1800 0600 0900 X X X X X X X X X X Sep. 2 2 0600 X X IC 1 1 lb Sep. 23 1200 0600 X X X X June 30 0000 0300 1800 X X X X X X July 01 0000 1300 1500 2100 X X X X X X X X July 02 0600 0900 1200 X X X X X X PH/ VSE I July July 03 04 1500 2100 0000 0300 0600 0900 1200 1500 1800 2100 0000 X X X X X X X X X X X X X X X X X X X X X X Winds SFC-590MB not used. 2. Winds above 700 mb not used. 3. Winds between 310-255 mb not used, 4. Winds above 710 mb not used. 33 T3 n •H iH "-— ■M 13 o fH 01 03 3 rt ba 4-> c Cd cd a > i •* / — » ^-> o o •H o ^H to o ■\ Xi B u o £ 10 o X -H ^— -* ■P ^ tn — 1 •P X fi rt hH H 00 I vO l> > 3 lO O v -' 3 to ■§ 3 2 •i 1 LO 1 1 '*oo **> .-H.-lOOOOOOOOCMOOOOi-HOOOOOOtO o I OvOOlOlMatCTiNMOHMNvOOtOl^N^^HrNl I (NJOOOOO'— 1>— ICslCMOOOO'— I HH Mffl -* O i- I H I I I I I+ + + + + + + + + + + + + + + + + + I NrtMMONMMOfO^H;rjHMN\OK)Nr\lUTO I NMNO>\oo>MooLOO>nwcriMi/)NOi/imoofO i • I HMHHHOtOHHHKlNHNNNtOTt-KJMNiH ! l/)L^H-\OH;COCO(NH;Ln(X3mLnONLO(Nlt\0(^MO o I oooooooo + + I i I I + I o o o o + + i + ! + + O i— I o I + + + I 00vDOltT-^OvDHt0H;vDOH-OHt000l^00K)K!HH I I HNHHONinOOCTlM^Nm^OOOOOOTtNO^tM I+ + + +I I I I I I I I II lrHlrHr-1,— Ifsjl I I I I I I I MOHb0WMM"*(NNH;(NHHOtT>OHMH00LnH ■ I I I I I I ~-J _.J ■ • _ J ■ 1 I . 1 I o o o o \£> (Nl 00 rt oi ai oo oo oo I I I I o o o o O vD CJ 00 00>ffiOOCOCONN\DvO\Ol/)lOH- o o CO o o CM o ■to O -i" -o o o c xC o CM O o -f o "St o o o O IN o oo o o o o '■D O o oo oooooooo OONM^O^OtNOC ^}- tO M (N CNi M I I I I I I o o o o o o <3- O \£> . 34 "3 r c byC c crj o > a i o +J i — i 3 t/i 3 -M H +J P- C/3 ~ U c bi o w C •H crj :^ o rt o Ph O 5 i C i u h 'O a; +^ 0) •~ r-H 1— 1 -Q rt H 'Xi > I vO |> I CD ■Q H CD >, crj cm r>- OHtWWNHOH^LnffiH'OLnOlOHH^ l/)NhNt0OC0t00>MNvDaMO(000CHO>O to t— I O O t— I O O I— It— I O l— I r- I CM r— ii— I t— I r— I i— ( i— 1 i-H O CM C N LO Ol I s CO \D r- l(NOt(7i«NvO(NI^^O 0\0{7lNOO\OONvOMt s -NinHO>0>^'vOvONOlO t— lOOOOOt-Ht— It— li— I CM CM t- 1 i— l O <— It— I CM r— I i— I i— lC^l/1(Nl^r^rHr^^CTlt00000K)r^K5L0OCMCJ>0000L0 O vO r— I (O h 1/1 \D t- inoNMLO^CNi- ICMt— lOOrHO-* r-lOOOOOOOOrHOOOOOCMCMt— I O O i— I O ^t I I I I I I I I I I I I I I I HOiOO'+Crn/lMmvOHO^HMOlCONOvONWa CMCMCMCMOt— IWOOOlOONNl/lvDNOiaiHrrNNIOtO i—l rH i— I rH I I I I I I I I I I I I I I I I I I I ^CTlOM'jrJWtlNKN^NHOOOOOOOOOCOvOM ooooooooooooooooooooo NM^O*(N00<*O\0M00^O\0M00^Oi£l(MOO OiooooooNis\o0 N00 ^fOvOCNOOTfOvOCMOO O^O^OOOOOOr^r^vOvOvOLOLO'^t'^-'^-tO^OCMCMCMt— It— I CM LO to o I o I to CM CM t/1 a3 35 c • H tO • d> PC CO CO Oj CO CC • c CD u o •H Ph rt as O aj o o I +J CO § V) H Sh o o r- l> i ^ l> LO I to l:o CM ■i MOOOOvOWHWOONMftCftMOvONN^tftOifl^' HOHHNMtN(NTtHOHH(N(NHHHNH'JNH IO\OWt s >vDOOOWN'!j , inNS\0^ , tO , *N\OflfON\0 O.— li-HOOOOi— lOOOOOOOOOOOOi— I O O I+ + + + + + + +! I l+l I « I 1+ + +I I OOHvOMhvOvONOOOOMIO^NvOI/IXAvOLOHO MNNM^fNOO^-OOLO C^.— IOOOOO.— lOOO©0>-4000000rHOO IPII I+III+ + + + + + + + + +I+ + OOWHHMvDHWOl/lNOOMMnH^OlN^WOi (OtLnONOOOlCNHHOOOOCOOOWOlOOOOaiMHvO i— I rH i— I .— I i-H i— I I— I I I I I I I I I I I I I I I I I I I I I I I ^\DN«CTiO)CriooooooooooooooooNi^Nh-t^N>£iLn to to o + to o + (T, co OOOOOOOOOOOOOOO \©CMOOTtO\OC\IOO*3-0\OC\100'TfO o^moocooorvN\o»o\ouiLn^-'*'5t I I I I I I I I I I I I I I I OOOOOOOOOOOOOOO OvOM00'*Ov0N00'*O\DN00^t- OmCTlOOOOOONNvD^vOI/ll/l^-^f o o o o o o o vOCMOO'^tOvOCNlOO tOtOOJMNHHCO^- I I I I I I I I I ooooooooo to o o I 00 I o 36 -o ^r ■ H M S CD U) os c3 +^ c, rt • c^ W * 3 , — > i u /-•, o 'j t/5 • H ■^^ r- | g C v — ' .c; :- 'J. (,> '.:■ > •H I" 1 tfl -— ' •H Fn 4- ^O to ^3 i a- r^vOHN tOlWHNCMOMtOWl/lTtfMHNCOHNvOOO rH in ^ M N N H M H (N (N (O MM NCMCM^fvOMlOtO CM CM I-~ O LO MN^WOO(N<*r l(Ni— I CT> v£> LO O f^ \£> r- 1 N O LO *rt LO "3" r^-OCCi— lN\ONCTlN*CMMr)(MfMrtl-O^NO o rH r-~ cm Hi/iHvoooffiHONo^vocom^^vOtnvo tO fl ^ rf N r- li— IOi— I i— I N (M r- 1 H N p- li— ( O O r- I I— I I I I I I I I I I I I I I I I I II ^•rj-tMHcriTj-cM^-Hh-LnrfvOHbO'*'*^ en OrHOOOOOOOOOrHOOOOOOOO"— ( r— IOO II III II I I I I III OOOOOOOOOOOOOOOOOOOO \D(N00'xfOv0(NC0'st-Ov0CN00<*O\CN00^tO I I I I I I I I I I I I I I I I I I I I OOOOOOOOOOOOOOOOOOOO O^OlNOOrt-O^OMQO'd-OvOCMOO'^O^NOOt OOlCTl00C000NNvD^\0U1Lnrt^"*MmNM o o \D CM O i— I r- I 00 I I i o o o O v£> CM (NHH K> CM rH O O rH O O O O <— lO>— lOOOOOrHrHCMrHOMOO I II II I I I I I I I I I CT> hod |x fs O C7> NCTi Lf) OK) i_o ^•C^(NNHH(NHHHCMI/)M^MHtO(N rH CJ rH C\] rH rH CtNCOWMOlHHMMMmeOHMfflHhOMOWW O O I totHooo)ooooTi-ino«)K)0)ooTtoo'*OHvOM»ai«) o rH rH I— I rH I I I I I I I I I I I I I I I I I I I I I I I L)">l-0OrHrHrHrHOOOOrHrHrHCNirNl(NrHC000I^-VDL0K) rH CM CM (A u CD ct! 37 c o if) •H u o3 & O o rH +-> t3 rH oj bO C a > i o •H +-> 3 r— I o tfl •H CD rC Ph 03 rH M O e c3 CD O o I CXi CD i— I X> 03 H '"> > o > v — ' ft> Ph 3 X r-^ £ > > o 1 LO O tO :z> * — ' c h > CD v -* X :d o I LO o to ^3 1 o ro LO ^D . O 2 '~> > 1 O 'h > CD V — J X co 03 rH X ^ s > •=> c i LO c v£> D o CO LO Oi . O Z M i/) C en •H CD &o e ^ 03 ^ X) ^ o e CD -H ^- ' > X < 4-> oo \C ct> r-- o CTi vO oo to «vf- r- to CTi ^r i— i o h> t^ *-J" o CTi CN 00 to LO o C\l LO CTi to LO . — 1 CM r^ to O *£> r-H 00 r^ ■y CTi o \D to to r-H en ■<* r-~ i— i LO oo £> LO vO CN i— 1 r^ vO «=fr to to to CM cn **- CN i— 1 H r— 1 to CN r-H LO oo to LO CTi to CN «3" oo CN CN CN i — i rH rH r— 1 r-H r— 1 r— 1 rH ct LO vo yQ r-- to CTi r-~ o to o LO o oo h- to . — 1 to CT> CN OO vo co CN HO o CN T— 1 oo O CTi CTi CTi r ) 00 rH 00 o CTi CTi 00 r— 1 t— 1 r-H r~4 rH O O o o o o to to CTi CN tO CN CO O LO o oo 1— 1 -* LO ^f vO to to CN CT) \0 oo CTi to CTi oo t~v \D -r r^ to i— 1 rH o O o o O o o o o CN to "tf LO ^O r-. 00 CTi o 38 'ft dJ o c CD u <4H m •H Ci c X3 cO fn ,CJ 03 4J 3 M CD C ra CO ca CD ^-s > +-» rH CJ 1 M ■—1 c O U rSZ •H 4= 3 (S cO 35 owoo>^) ocT>Lnooooa>^£>t— ocm ooio>ooa>oio>wcTiai OOlOOKlOiOOOQ^MOiH oaioioio>oi(jim(iimoooi o o o o OlNNOl/1NOONo'>cr>CTCTiO">CT>oc>cr> ooi-ovDLnoLnvor-cM\£>vOoo cncr>a>ooo^cnc7»cT>CT>oooooo OOOt^ a)OON(OHN"stO«ONCOH(M OOCTtC^OOaiCJimoOOOOlCTiCnoONCfloO aoii/imoivomffiiONinHomMo^vo OOtJlO^aiOlOiCOOOOO^NCOCTlOOOONNN NHO*0«0(N\OOOOOHH;OOhO>«(J> OOCT)00000>cJ|OONN \OxONNNCxD\D\0\0 LOt-OvOi— I vD O >0 »OtO 00 (O N CO r- I vO LO CN CT) NNl/lvOt^h-vD^-xD'st^Ull/lh-Nt/lini/) tO rt CM HIOWOOON^HHVONOOCNVO ^■tOtM(N(OtO(OHr--'>ocN\o\ooi— ii-hooooo N H r-H CNCNHCNCNr)-T]-M(NlrO o>coHTf^(NMHcoorMOOH«ono r-Hr- IrHrHrHtOCNJCNlCNCNl (NcNjoooorHooovovOt^toor^ooo^i- COOONN\OvOlOl/)^rtMtOMNHH 1 I I I I I I I I I I I I I I I I I oooooooooooooooooo OLnoLOOLnotnoLOOLooLnoLooi^ OOlOlOOOONNxDvOtnW^-^tOMNNH 39 CO CD O c CD U CD U-i <+-l •H a -a H rt 3 bO C CM rt > +J i CD r — s VD c o rt •H ■p +-> rt 3 c i — i o Fh (/) O •H l+H +■> i — 1 ■l-i 3 C 2: CD v — ' C Jn P DO O c rt o o 1 1 r- 1 4=> rt E- fl rt 43 4-1 CD CO 0) /■">. ^H u CU cn 'Si 00 c 4= •H j-i T3 •H rt & a) 43 c3 4-> C ^ B -c 3 H 4-1 o O CJ ■u c 01 a h cu a. o CD CO vO m to eg i bO i LO vO I i 4^ 6 CD X ^■l/)l/)l/)HH>OHin\OCntONHM(MCNN o o r- O O C7> C7i OH t OO) CT) o o o o oo a> ui r--o^or^t^-io O O CT> 0> O CT> O 00 CT> 00 C7> OOOO^LOOr^OOLOvOLOO C7> O^ OOlOlOlOOimClCTiMOl O NOMOCCl/lNON^HCHrtOH o mocriocj>aia>ciCTimaiooaiccoo C7> MvOONifiMNMJlCOO^OOt CT> OlOCTlO^QOiaoOOlOOOONOONN t^-OCT>rHv0O001-0t^t^L000r3-CTi\0L0rH OlO00©00Ol0000C0NC0hi£)Lr)iO^\0 OONMHmK)*OCOWiDCOtOOON^^vO 00)MNN\£)N\D\DNvOvD»0't^-Lnul^t LOOiffiooc7)MCNa>LOi/iNt/i(NiaiMmoo^t l/KOCO^MNHONOO^lflNNOtl/l^ NCMHHHH«M(N«(NMNCMH«M(N NNNCNMlOlfl^^Lnr- 1 N ^ in i- IfMLOLO N H i— I i— I CM r- 1 CM r- 1 i— I i— ICMCNr— I OOHtOHHHrtNOOinrM/l'CftCONlfllOOO r-l i— I i— I i-H H (N r— I'tHOOOHOlOONMfllOlO^OOlM (Ncsiooooooi-nr^i— i m cm h m \on m Or- IOOOOOO"d-vOr-(K3i— I O CN ^ CO (N OOOOOOOOCNJ^rHtO^HOCNtOvOt— I oooooooo LOOLOOlOOLOO ^^■tOtOCNNHH I I I I I I I I oooooooc ounoLooLnoi-n OmOiC000NN\0\0Lni/)rf^K;K)(MCMH oooooooooo LOOLOOLT)OLOOLOO Cf>C7>OOOONI^OvOLOlO I I I I I I I I I I oooooooooo OLOOLOOLOLOLOOLO 40 o - —I > CD C^j > ■§ o O ■H ■M '/. •H ■M 03 4J to c o (/] •H & o u fn o H > CN to o r~- ■«* i— I r-« en v0 r^ u • • CN to oo lo cn i — i .—i o» T— 1 oo > • • • • O 00 r-» LO to o i— i to CN a> CN CN cn to to vD u • • • • nS Ol 00 vD r-\ > CN CN i— i CD E 3 e R o +J E R o +-> to CN co oo CTi O TJ T3 u fH rf c3 3 3 6C M g C cti o3 > > to CTi oo CO O CO CO o to o lO oi 00 CTi o 1— 1 CN —1 co — 1 o to LO CT> \0 oo 1 — 1 rH o 1 — 1 CM i— ( to CT, o CN CM to CM r— 1 LO i—l oo ^r "* CT> rH «tf CN vO CM CN CN r— | CN OS CO CO ' — ' v — ' *— ' Jh fH (h ST CO CO CO v — ' c^ r— ,C P. P. P- fH rt rt 03 bfl bO bfi a o O o U c c c o3 c3 03 cd O CU CO CO I/) u u o CO o o o Pi 41 _o d O E o 03 CO i>* ****' N O O CO o o o ^j , . 8 ^^ o — I — o CD X3 • C < O CO g CO CC N O o s- — CM o o o o lO o O) 0> o o o .y o LO o o o CD ■a cc O N°> O T_ o a) CO c > o 3 -3 1 o -9 1 N ■ *4-_ 1 O !—■ i-l- CD I CO 1 c k— LU 0) 4-1 u ca CO c •H TJ C o co 01 ,c H -a c •H H CU PM C aj O o CO s •H d >-i a cd - a a; 6 co O o cu >H g 0) -H 4-1 4J c t-H CU O 4-1 *4-i nj rH en ai c •H TJ o CO cu 4-1 cd u V p o CD cu ■H U CU C/3 I I QJ U cc •H '— (qui) 9jnss9Jd (qiu) 9jnss9Jd 42 /■'' 1- 1 1 /■■ ° 1 9 /' CM 1 / CM ' ' /■■ "• E 2 J t '•• 1 ■. Septe Poryv Resea / / 1 ". /• s /■ \ /; \ : /• \ : 1 /■■ 1 1; \ •..:■. f \ ,-. / "*\ ' .••' • / \ \ f w \ / w ( ! _ 1 1 o I O in o o o 1-1 3 O o cv o O in oo u u o o r, o 4-1 1 c a) a) 4J ai 43 G o co O •H o n *■** CO 03 ft o 3 6 o o 1— 1 «5 CD a 1— i 1 03 CD (-H a E 5-i 3 CJ 1 — 1 CD ■U h- CO hf, 5-< 0) •H ft J-i b *1 CD T3 (qui) a.mssaJd u CO ft n a) ft ft, 3 a) 4= o •H CO 4-1 OJ Q I l < CO i o o o to o o 03 O o o (qui) 3-inssa.y erf CO CO 0) X. u 05 a to a) u c QJ 0) U •H I S-i f- 4_j CO a: cd u cu ex e QJ 4-1 0) CO TJ O C cd co U C " o ca co > a) cc -a 03 CO CO c CO co c nj a) I i co (U W ■H Pn 44 i C3 DC M. I, D-. CO CO CO :~> i DC CO CO 3 cz JH "T o o o o o o o CO *t CM CO «tf CM O O CO I I o o «* CM cE. _ c£ r io A o e, * !: i < . <* oooo ©oo o CO tf CM > o o o c CO "3- CM 3 O O O O CO «* CM II t=E OOO ooo o o v CO ^ CM CO 't CM IT CM —» o o ^ CM "- LD O CM II 2 CZ c CM CM II 2 CM CM r io A L_ pg ■ — • i r I CM O CM o O o "^ CM o-- Si , E oo OUT) CM CM o '3 o c .<> , £ OO OU5 ooo CO «* CM E CO OO OLD COCO o o ■3- CM i E OO Otf> 00 CO o cc o -3 o CM I V £ l o oo ino CD *— CO co O CO 5-1 CU >> cfl lO 1 (U V J-i V) T3 0) 4J CJ <1) i-i 0) CO o CO •H f-l CO Ph E o o CO QJ a C a) 5-i cu 4-1 m •H T3 Li, ! CC en co -a c cd cu 5-i 4J 2 - 5-< ^ e 2 J_l ™ 0) m o C o •H CO co CO CO ■H U c a) cr cu 5-i cu 5-i 00 •H Cu 45 USSR-VIZ 200- 2 400- E LlJ r> CO C/> LU £ 600- a m 800 1000 A #NIGHT A DAY "3 -2 AT (°C) 3 Figure 5. — Comparison of day (29 pairs) and night (9 pairs) mean temperature differences vs. pressure for the U.S.S.R.-VIZ comparisons. Sound- ings from the U.S., Canada, and F.R.G. were combined. 46 to" ~ ~ ~ ' ~ ' -' ^ ~ - - ^ ^ ooen^ifluiNoiiDi- cm ro «— lt> r- r-. *- cj (/)<-^^i-^'<-^<-* 't-^ ' *^ cm cm «-^ «-^ *-^ CO to 2. u & en C/3 d b en o ■* o oo ITS CO o ■JO en CO <* U-5 o i>4 en Q co CO en CM CD CO CO to a l*x CO L£5 CM «~ r ~ ^* t~ *- •" CO o O n a CNI oo en ■5? U5 ■it en CO CO en CO e= CM CD ■=T CO en CO 4-> C a> •H o •H M-l 4-1 a) o a c o •H •u cO rH cu o u cu J2 CO c o to •H CO a, 6 o O O En CO 1 4J o ^ rH tx CO LD H W © 03 .-! o P •H i) "T3 la H c CU cO > T3 CO o e 'S cd 5 c 0) cO n u 3 '. CO Pi M • cu en a • E w cu • 4-1 1=) CO miMooi-^iBinin^stooNoaaotSi-N ■ i-nisooo>ai>qNnsinaoor|F|in4 ai-naiois^Oi-LDCMa oaoiN>n>flc9ooi-ai-ifl to D b d (qui) 9.inssaJd cu CO l # C Pi •Sen* 4-1 cO to > cu M CO t3 fl CO 4J I I • CU u bO •H 47 o o o CO _ o CM c£ LL CC CO CO • • o & <=> x CC o < 7 o CM OOOOOOOOOT-t-r-r- O l_ CO CO o CI co c ro o CC CO CO zi • • © © x o < 5» •3-CDl^CO 0000)00 OO r-N(M N(M(M CO D CC CO CO o h cm o M O M O CM - O o X CC o < i o CM o o CM o o o o CO O o CC' O "- CO o u o 4-1 0) CO CO a) > • CO co c CD O CJ CO C -H Cj CD 4-1 4-1 cj) d e o •h a •H •H CU CO > • •H CO U CC !=> QJ cd T3 4-1 CO O C rci CO C_) d i o • ■H Qi CO CO •H • > CO CD • X) £> TD * CCJ CO d ^> CO I 4-i CO pi T3 CO d • cd co CO JD fi cti CJ ! I a) u 3 M ■H (quj) 3Jnss9Jd 48 }-i o M-l *— r— *■■ r— II t™ r" <^ CO q z II z pf II r ii z in S cu cz CM A _ o CM „ O b a) H V- 01 S-i LL l DC CO 4 1 - o o " T X DC < 3 CO , _L CO 0) L cz S-i ft . CO c c o - CM CO D 1 cu o I 1 I ■ 1 ■ 1 • * r «— - CM 4-) CO CJ -H I 1 1 1 i i 1 1 1 1 i i i cu ^ o o o c ) o o o o o o o © o o V H CD '■D * CM 6^ CO o .Q 3 CD N - / / \ -/- / / / •.•• / / / x / s / / s / / o o © o d Lf) CM O 6 CM e o •H u CO ex S o O . 0) •c E 53 o o O o_ 0) 1_ 3 <-■ ro L* CD a £ o o c\i o o 00 o m oo o o o in . -a •H e > •H cd ~ o — cu m m o m C3 Jh QJ ccj H 3 3 C 4J CO "c3 0) cj ex o ai ,13 o iH •H CO cu ON CU oo ■U -O 3 0) M ,13 o ■f. -f, a> m o u ex ex '- 50 / ~r / / / / / / + / y o o CM SO 3 3 < // o L. TO / CD re CD O CC ^y ^r / / / s \s. o o q in 1^. q o O O CN C o CO •H >-l cd a, e o o CD )-l o 2- en > m £ ° 3 ex cd > re CD CC CM >^ 4-1 •H •H E CD > •H 4J cd ■H CD C cd cd 4-J vO to CJ O o CM U o o CO rfi S-l ° a; e 4-1 o CM 0} 3 +-> TO j, CD £ CD q o o co o m CO o o o in 0) o o o CO o o CN (qiu) 3Jnss8JcJ U £ 13 5- « S CO § 4J CD CO O ex C I S ^ 43 CD ■u iH •H cd 4-1 0) ? I CU S-i 00 ■H 51 q to CC c/i c/j D to n i- n o co oo in cm to o en r-_ so ■=r i-^ in <} i^ n in od CO *— «— CM CM CM *— a> cm eo «* in ^ i>- o> *■- ^ en 9 m ** oo is si in in n 4 is oiseooeo<3- 0) i— 1 0) 5-1 o a in •H M-l f-i O CO e ■U o o o o < o to CC CO c/j r> to iMcoisainoaannmoosiio *in^ifloei*08NiBisio'^iri P« N O) r^ CM CO 0> CM CO «- !B IS B i- rt IO in in r^ is co in <. C/j D to a: cyj e/j D to cMr»inmcMe-cn<— encor»eo^-cMO«— en isinen^ceaaoocMincoar»r«cecMCM i-nneovr-ioinedoiiBr-i^^n^in »-*-«—»— r- CM CM CM i- <— i> N N N N N caincofxcnrxco-CMinc- oo to o »— «* r- co t-lsO^OLINOlMUii-NBlNONin eco^'iricdoocnincocMcorseriencnrsoo U co OS o •H u. 4-1 1 5-1 . cu cuS > • CO U) "3 C CO T— n3 . 3 en 4-1 T3 C C •> . 4J Di •H o T3 CO ■H CO 3 . X 3 01 - > • •H CO 4-1 rt :d t-H 1 cu . o H pi f- CU CO .fl • 4-1 CO 4-1 3 o T3 C o T3 -H H U rt .h Td 14-| G M-i rt cu 4-J O CO o (qui) 8-insseJd 01 3 •H 52 q d o £3 1 I r— TD •, -^ o O r 1 lO £ 6 . j K) i 1 o © -a cd •i \ £ fa 3 .? ,/ / y< q in 9 N Relative 7een tb .^"^i "JS - XT --J d CM q d 4-1 0) ^ ' C 1 i / o •H 1- 1 : / ca cm Q i i : / o a O W r 1 "■* O 00 o / o 0/ 0) 3 o cm" ^> / ^ s — q < cc c/> /" o u o «4 < LL D ^__ . q d files r 080 / CM 1 o o U M-l / ex c/1 4-1 T3 ^ -H C o re (°C humid IZ so / d 3 ;> /■■ <* 1 £ > 03 -H /•' a 4-i £ 03 CD t! • ^ u (3 CO V / q d i OJ V / J-4 3 i / 1 4-1 CO U CD i / 0) (^..^ H 1 1 o 1 i 1 1 1 1 i i i i d 2 i OJ o o O o o o o o o c 1 — 1 o o o o o o o o o c 3 ' CD v m CM CO <* in co en 3 S-i 3 (qui) ajnssaj^ 60 ■H fa 53 400- E tr UJ cr a. 600 800 1000 1.54 1.59 .82 .82 .78 .99 .77 .98 .90 .90 .93 .99 1.10 1.21 1.05 1.07 .73 .70 .66 .81 .74 .67 .93 .87 .73 .81 .97 1.07 .98 1.00 .71 .64 .50 .52 .41 .45 r 0.5 .0 Figure 13b. --Standard deviations of the temperature measurements and vertical plot of the correlation coefficient for the F.R.G. M60-U.S. VIZ comparisons. M60 " VIZ 55 200- 25 0mb 400- 45 0mb 600- 650 mb 800" 850 mb 950- lOOOmb % % % % /o 60- 40- 20_ 0- 60_ 40- 20- 0- 60- 40- 20- 0_ 60- 40- 20- 0- 60- 40- 20- D =GL i — r n ri ZL P : -5 -4-2 2 AT(°C) N=23 N = 23 N=23 N = 23 = 23 1 1 4 >5 Figure 14. --Frequency distributions of the temperature differences for selec- ted pressure layers for the F.R.G. M60-U.S. VIZ comparisons. 56 200- 400- -Q £ UJ cr =) co CO UJ cr: 600^ 800- 000 M60-VIZ -30 -20 -10 10 20 ARH(%) 17 23 23 23 23 23 23 23 23 23 23 23 23 23 23 23 23 23 30 Figure 15a. --Means and standard deviations of the relative humidity differences vs. pressure for the F.R.G. M60-U.S. VIZ comparisons. 57 Om6 OviZ 200 X> 400 6 "-*' L'J q: O to CO Ul on knn a. DUU 800 1000 3.87 5.03 3.67 7.43 4.03 7.80 3.45 7,43 2.92 6.80 2.77 6.63 3.76 6,85 7.34 7.41 24.19 15.64 25.07 21.30 19.28 17.28 18.99 14.69 20.92 18.89 23.12 22.33 20.48 21.01 12.88 12.86 8.81 8.21 3.98 4.77 .0 Figure 15b. --Standard deviations of the relative humidity measurements and ver- tical plot of the correlation coefficient for the F.R.G. M60-U.S. VIZ comparisons. M60 " VIZ 200- 250 mb % 400" 40- 450 mb % 20- 0- n-n N=23 X=L 600" 65 mb % 60- 40- 20- o r-Thr. N=23 Lka 800" 8 5 m b % 60- 40- 20- — 60- 95 0" 40- OOOmb % 20- n ri N=23 P N=23 q- O-r— r < '25-20 -10 10 20>25 ARH (%) Figure 16. — Frequency distributions of the relative humidity differences for selected pressure layers for the F.R.G. M60-U.S. VIZ comparisons 59 CD _ «* _0> a E CO IS c to a V) - CM U o 1 D < CM I 10 o oo o CO o o tN O o o CO o CO o O O CM suoijBAJQsqo 40 J9quin|\j 4-1 c C O a o u T3 U > •H e s-i W3 o a cd QJ a O •H o •H 4-1 3 .-a M 4-1 Cfl ■H (J 0) — 0) Fn I I 0) too •H 60 to _ *t ■u G CD a o ex e o u _ o - oo T3 cti 3 00 c CO > ._ to T E TO in o M- c CD a. to 1 "3- ° E CM I > CJ c cu j-i cu M— t •H X) (O o •H ■u J3 .= o cm T u ■u CO •H >> a a; =) a* cu ^ l l o 00 o CO o o CM o o © CO o CO o o o CM suojjeAjasqo jo J8qiun|\| 00 cu S-i G OC •H 61 in 5f Lf) CO CO O r~ r*. CO ^t O) x r «s- co >* O 00 *r CM CM o ^ 2 0") o CM 10 00 «* CO i— in CO r" w m ^ CM CO in o t - «— CM 1 ' -T- - i i / X L. cu ^x" - a CO k. ai o CO / ' CD X o X O y^ / ® - ^ § k. 0) S. ~"^™~ ^^^^ £ _ a CO u cn o i i i Ocean CJ •H • • rH M 3 CO O CU o > .£> £J CO >-i C- •H ^ 0) CO u O 3. u cfl 01 >> 00 a co ,c o 6 -~~^ CJ o e u CO CJ w o CU <4-4 CJ • t3 O Pd u cu • crj M U CO 3 CO o • 00 "4-1 CO s ^— > • cd s 4-1 3> > *s_^ rH 1 u a CO CU CO u 3 ,£ CU CU 3 CL M -3 •h co 6 n co M 60 CU CO M o 4-J CU CU CJ o en x: CO CJ cu {X CU CU Pi CO CJ T3 }-i o S-i 60 CO O O T3 CO "4-1 s 3 o cd cd U 4-1 cu CJ H CJ , — > 3 o aa jr co ^w^ 4-J CU - S-i •H S-4 4-1 CU & C J= CO cu ex CU -H 3 « H O V4 CJ 4-1 a go S-i o e o •H id o a u a cc £ u cu 4J 0) 33 CJ • -H --. o ^ & E n H co -a cu cu > cu cu > i-i D a .h •H O i 3 H 4-J 5-i o D cu cu CJ «H S-i 4= CU CJ — CU CO a "4-4 iH CO 4-1 cu • •H CO s-4 3 T3 (1) o > •> CO 4-1 U CO -H O 3 T3 S-I CJ O CO /*-s x: a b w B v— ^ cu c CO E cj 3 co S-i o cu CJ CU •H C O 4-1 4-1 & •H C co a 4-J >H •H >H P > 43 i—l CU 4-) O Pi a CO • 00 •H CO -a c 4J )-< -H i—l CO CO CO 3 • T3 CO 6 3> 3 M l co cu "0 /-v 4-1 > 3 re CO cd CO >— ' (qui) sseu^oiLji 6ui6cj3av ct- cu U 3 oo Pn 62 I oo cm CD CO CD > I c > o •H • T-l CO O > XI S-i o ex -, CO X w o M CO 3 60 d > u CO 0) , d ex •H CO B ^ 00 o ti cd a) cj o d O CO •H M CO ex e o o pi O en S w en 5-i a) ex CO 00 o a CO 0/ o o iH P 3 1 CO U 0) a) S-i X CJ CO J-i QJ CO 4-1 a) O CO d cu CD Pfj T3 u CD o * >-i 4J a) C ja ai Cu c cO 5-i ex 00 B O c u ct3 0) > o o CO T3 a> at a h C > & H CU CU > H •H O 4-) 5-i CJ -H 0) o ex (quu) ssau>)3iqj Bui6BjeAV "H CO T3 CU > U-l 5-1 o d CJ O CO d co o cu •H C 4-1 ^5 co a ■H -H > X CU 4-) p 00 T3 d 5-1 -H CO 00 T3 CO (3 5^ CO CU 4-> > c/3 co 3 o CO •H 5-1 CO cx e o CJ 5-1 01 4-J ti •H O Pi CO • •H CO 4J H c/1 3 • e p i o tN CD 5-4 3 00 •H 63 M cc T3 4J 4J rH ex X) "1 f.r, a > 4-1 c o e c o 4-1 u CD -C PL, tt W) c c: rt QJ a c M W) oJ '-- CO > cvS 14-1 O (qui) ssau>p!Lji BmBejaAy M-l 4-1 r. 0) •H c ■r-i o a c a >H -i nj H CD o u I ! a) l-i :■' 3 64 (qiu) ssaiopiq; BuiBbjgav o M o cfl 1— T3 a) 4-1 ■u o rH Ou T3 M CO 3 bO ti ca o > cri •U d a> C o a e o O > o • 4_) (/) c/1 OJ o J-< c CO a) r^- 4= CJ a ■ H 03 4^ !-i 4-> bO O GO C C > "3 ■H 00 o fn o 4-1 > a rt 0) a ^ o o o r-.' a B G o o O -H 4-> > O C <4-( Zj O '-w 4-i nJ C a> •H CJ •H 14-1 q M-l U> QJ O U ti O ■H 4-1 aj rH OJ S-i !-i O 1 j CNI > CN tu U 3 bo ■H pL< 65 •rl 4J C 0) o ex e c a o !1> '/I 03 a) c CJ •H 4-1 d H •H TO ca m 0) 03 > > . 3 W C > - i 0) i-H ^. 03 w > in •H 03 C 2 ct3 C ,£! rt 4_J C/ CJ 03 C C/3 QJ - (qui) ssau>p!i|} 6ui6ej8AV X C o •H 4-1 > ai CO o M— I O u G (U U S-i (D a. CXI CD S-i 60 ■H 66 4-1 C CU c o 6 o u u o 4-1 en cu M CJ •H -C • 4-1 T3 0) CO co bo to c a) aJ £ > cd to to e cu ° 5 CO ^ > C cu > •H CO O c re cu o en o en cu I— I I/) CD tn o (qui) ssau>p!qi BuiBejaAV a o •H 4J CC ■ > 0) 03 o 4-4 O 4J c cu CJ S-i cu p-l CN CU u 00 •H 67 80 60 .q E « 40 s: +■< o> c a) < 20 V(95%) I U(95%) V(5%) <^ 5th and 95th percentiles I I i -1 2 -2 3 -3 5 -5 6 (95%) -6 (5 %) m/sec -4 Figure 25. — Fifth and 95th percentiles of the distribution of component differ- ences, Oceanographer (M) , minus Vanguard vs. averaging thicknesses. 80 60 - -e E c ■S 40 |- c > < 20 - \ M / \ K \\ \ )V > ' 2.5th and 97.5th J percentiles l y \ / / S >^U(97.5%) ^V*. ^ x ^S^v X\ 1 1 12 3 4 5 6 (97.5%) -1 -2 -3 m/sec _4 -5 -6(2.5%) Figure 26. --2.5th and 97.5th percentiles of the distribution of component dif- ferences, Oceanographer (M) minus Vanguard vs. averaging thickness. 68 (quu) ssau>p!m 6ui6ej8A\/ < \-y- •••■ • V ¥ - / / 950-650 i i i i 5-1 CU cu CU XI 5-i C/5 ta- rt 4-1 E u O 5-( O CM > > CO o > o o CO cu n "—-* *M b onents g thic and V comp s . averagin O ences, U m/sec) v *fr ) of differ Vanguard ( CO n (a inus CO o B •H . E 4J /-x CO C CO 5-1 '~ ' •H O CU - — ■ > Tl >> > CU 4J CO CM i 73 3 rH o 73 O CD D 5-1 CO •U CO «H CO h 73 4-1 5-i a h CO cO 3 o. +J £ CD C/l ^ l CO CN CD 5-1 3 60 •H Ph o o o CO o CO o o CM (qw) ssau>|0!q_L 6UJBBJ8AV 70 100 r 200 • 300 - 400 - 500 - 3 vt I 600 700 - 800 - 900 1000 -te-^r^ *"% y / I ' 4r f j s s , — I? r I y ' 1) : ( CD 600 Q_ 700 800 900 1000 c > \ I U August 17, 1974, 03 GMT Okean > Researcher Vanguard 30 20 10 m/sec -10 -20 -30 -40 -50 Figure 28 . --Vertical profiles of U components for Okean and Researcher (H) , and Vanguard three intercomparison flights during IC II. (b) August 17, 0300 GMT 73 100 200 300 400 | 500 03 k> 3 £ 600 Q- 700 800 900 1000 V August 17, 1974, 03 GMT Okean — Researcher Vanguard JL 40 30 20 10 m/sec -10 -20 -30 -40 Figure 28. --Vertical profiles of V components for Okean and Researcher (H) , and Vanguard three intercomparison flights during IC II. (b) August 17, 0300 GMT 74 100 200 300 400 500 a ■-, 3 t/i IT. 03 600 Q 700 - 800 900 iooo 30 20 August 17, 1974, 12 GMT Okean Researcher — Vanguard x j 10 -10 -20 -30 -40 m/sec Figure 28. --Vertical profiles of U components for Okean and Researcher (H) , and V anguard three intercomparison flights during IC II. -50 (c) August 17, 1200 GMT 75 100 200 - 300 - 400 SI 500 t 0} i_ 3 (/»