{"Bibliographic":{"Title":"A test of the impact of NOAA-2 VTPR soundings on operational analyses and forecasts","Authors":"","Publication date":"1976","Publisher":""},"Administrative":{"Date created":"08-20-2023","Language":"English","Rights":"CC 0","Size":"0000056494"},"Pages":["OF COMMUNICA\nQC\n851\n.U6\nN5\nno.57\nNOAA Technical Memorandum NWS NMC-57\n*\n*\nwith\nSTATES\nOF\nRECEIVED\nA TEST OF THE IMPACT OF NOAA-2 VTPR SOUNDINGS\nON OPERATIONAL ANALYSES AND FORECASTS\nMAR 24 1976\nWilliam D. Bonner\nASSOC\nPaul L. Lemar\nRobert J. Van Haaren\nOF\nArmand J. Desmarais\nHugh M. O'Neil\nNational Meteorological Center\nWashington, D.C.\nFebruary 1976\nnoaa\nNATIONAL OCEANIC AND\nNational Weather\nATMOSPHERIC ADMINISTRATION\nService","NOAA TECHNICAL MEMORANDA\nNational Meteorological Center\nNational Weather Service, National Meterological Center Series\nThe National Meteorological Center (NMC) of the National Weather Service (NWS) produces weather anal-\nyses and forecasts for the Northern Hemisphere. Areal coverage is being expanded to include the entire\nglobe. The Center conducts research and development to improve the accuracy of forecasts, to provide\ninformation in the most useful form, and to present data as automatically as practicable.\nNOAA Technical Memoranda in the NWS NMC series facilitate rapid dissemination of material of general\ninterest which may be preliminary in nature and which may be published formally elsewhere at a later\ndate. Publications 34 through 37 are in the former series, Weather Bureau Technical Notes (TN), Na-\ntional Meterological Center Technical Memoranda; publications 38 through 48 are in the former series\nESSA Technical Memoranda, Weather Bureau Technical Memoranda (WBTM). Beginning with 49, publications\nare now part of the series, NOAA Technical Memoranda NWS.\nPublications listed below are available from the National Technical Information Service (NTIS), U.S.\nDepartment of Commerce, Sills Bldg., 5285 Port Royal Road, Springfield, Va. 22151. Price: $3.00 paper\ncopy; $1.45 microfiche. Order by accession number, when given, in parentheses.\nWeather Bureau Technical Notes\nTN 22 NMC 34 Tropospheric Heating and Cooling for Selected Days and Locations over the United States\nDuring Winter 1960 and Spring 1962. Philip F. Clapp and Francis J. Winninghoff, 1965.\n(PB-170-584)\nTN 30 NMC 35 Saturation Thickness Tables for the Dry Adiabatic, Pseudo-adiabatic, and Standard Atmo-\nspheres. Jerrold A. LaRue and Russell J. Younkin, January 1966. (PB-169-382)\nTN 37 NMC 36 Summary of Verification of Numerical Operational Tropical Cyclone Forecast Tracks for\n1965. March 1966. (PB-170-410)\nTN 40 NMC 37 Catalog of 5-Day Mean 700-mb. Height Anomaly Centers 1947-1963 and Suggested Applica-\ntions. J. F. O'Connor, April 1966. (PB-170-376)\nESSA Technical Memoranda\nA Summary of the First-Guess Fields Used for Operational Analyses. J. E. McDonell, Feb-\nWBTM\nNMC\n38\nruary 1967. (AD-810-279)\nWBTM NMC 39 Objective Numerical Prediction Out to Six Days Using the Primitive Equation Model--A Test\nCase. A. J. Wagner, May 1967. (PB-174-920)\nWBTM NMC 40 A Snow Index. R. J. Younkin, June 1967. (PB-175-641)\nWBTM NMC 41 Detailed Sounding Analysis and Computer Forecasts of the Lifted Index. John D. Stackpole,\nAugust 1967. (PB-175-928)\nNMC 42 On Analysis and Initialization for the Primitive Forecast Equations. Takashi Nitta and\nWBTM\nJohn B. Hovermale, October 1967. (PB-176-510)\nNMC 43 The Air Pollution Potential Forecast Program. John D. Stackpole, November 1967. (PB-176-\nWBTM\n949)\nWBTM NMC 44 Northern Hemisphere Cloud Cover for Selected Late Fall Seasons Using TIROS Nephanalyses.\nPhilip F. Clapp, December 1968. (PB-186-392)\nNMC 45 On a Certain Type of Integration Error in Numerical Weather Prediction Models. Hans\nWBTM\nOkland, September 1969. (PB-187-795)\n46 Noise Analysis of a Limited-Area Fine-Mesh Prediction Model. Joseph P. Gerrity, Jr., and\nWBTM\nNMC\nRonald D. McPherson, February 1970. (PB-191-188)\nNMC 47 The National Air Pollution Potential Forecast Program. Edward Gross, May 1970. (PB-192-\nWBTM\n324)\nWBTM NMC 48 Recent Studies of Computational Stability. Joseph P. Gerrity, Jr., and Ronald D. McPher-\nson, May 1970. (PB-192-979)\n(Continued on inside back cover)","QC\n851\n46\nN5\nno.57\nNOAA Technical Memorandum NWS NMC-57\nA TEST OF THE IMPACT OF NOAA-2 VTPR SOUNDINGS\nON OPERATIONAL ANALYSES AND FORECASTS\nWilliam D. Bonner\nPaul L. Lemar\nRobert J. Van Haaren\nArmand J. Desmarais\nHugh M. O'Neil\nNational Meteorological Center\nWashington, D.C.\nFebruary 1976\nLIBRARY\nJUN 1 7 1993\nN.O.A.\nUS Dept. of Commerce\nATMOSPHERIC\nAND\nINTERNATIONAL\nNOAA\nNational Weather\nNATIONAL OCEANIC AND\nUNITED STATES\nService\nATMOSPHERIC ADMINISTRATION\nDEPARTMENT OF COMMERCE\nGeorge P. Cressman, Director\nElliot L. Richardson, Secretary\nRobert M. White, Administrator\nUS\nCOMMENTS\nDEPARTMENT\nOF","","CONTENTS\n1\nAbstract\n1\n1.\nIntroduction\n2\n2.\nThe Operational Mode\nAnalysis/Forecast Cycle\n2\nA.\n2\nB.\nVTPR Data\n3\nC.\nBogus Reports\n3\n3. The B-mode\n3\nAnalysis/Forecast Cycle\nA.\n3\nB.\nBogus Reports\n4\n4.\nData Base\n5\n5. Analysis Comparisons\n7\n11 March (fig. 6)\nA.\n7\n(fig. 7)\n14 March\nB.\n7\n(fig. 8)\nC.\n17 March\n7\n(fig. 9)\n20 March\nD.\n8\n30 March (fig. 10)\nE.\n8\n(fig. 11)\nF.\n2 April\n8\n(fig. 12)\nG.\n4\nApril\n8\n7 April (fig. 13)\nH.\n9\n10 April (fig. 14)\nI.\n9\nJ.\nSummary\n6. Forecast Verification\n9\n11 March (fig. 16)\n16\nA.\n14 March (fig. 17)\nB.\n16\n(fig.\n18)\n16\nC.\n17 March\n19)\n20 March\n(fig.\n16\nD.\n20)\n30 March\n(fig.\n16\nE.\n(fig. 21)\n16\nF.\n2 April\n4 April\n(fig. 22)\n17\nG.\n(fig. 23)\n17\nH.\n7 April\nIO April (fig. 24)\n17\nI.\n17\nJ.\nSummary\n17\n7. Conclusions\n18\n8. Acknowledgments\n19\n9. References\n10. Figures\n20\niii","30 Mar76\nThe NMC Newsletter 76-1 for January-February 1976 carried the following\narticle on the \"Impact of Satellite Soundings\":\n\"An extensive test of the impact of satellite temperature soundings\nwas conducted from data sets collected in August and September 1975. Test\nanalyses were compared with analyses produced from an identical data base\nminus satellite soundings. A series of 84-hour 6L PE forecasts were made\nfrom these analyses. Results showed, once again, little or no improvement\nin forecast skill from use of satellite temperatures.\nAnother analysis=forecast cycle used a data set with radiosonde\nobse rvations removed (satellite soundings provided temperatures aloft).\nThese analyses agree remarkably well with operational NMC analyses on CO\nrespon ding days.\"\nDuane, I don't really grasp the significance of the second paragraph --\none could jump to a number of conclusions like (a) the models seem to\nbe relatively insensitive to new data inserted to up date them or (b) the\nsatellite soundings work as well as radiosones -- which may just be another\nway of saying (a).\nIndine\nOr,\nModels and tests should accept\n(c) thicknesses as input,\n(d) thermal windsas input,\nDSC","A TEST OF THE IMPACT OF NOAA-2 VTPR\nSOUNDINGS ON OPERATIONAL ANALYSES AND FORECASTS\nWilliam D. Bonner, Paul L. Lemar, Robert J.\nVan Haaren, Armand J. Desmarais and Hugh M. O'Neil\nDevelopment Division, National Meteorological Center\nNational Weather Service, NOAA\nWashington, D.C.\nABSTRACT. The impact of vertical temperature profile\nradiometer (VTPR) data from the NOAA-2 satellite is\nexamined by comparing numerical analyses and forecasts\nmade with and without these data. VTPR soundings are\npart of the normal data base for the operational analyses\nat the National Meteorological Center (NMC). For 30\ndays in March and April 1973, parallel analyses were\nproduced using the same analysis and forecast codes\nbut with VTPR soundings excluded from the data base\nfor the parallel mode. Forecasts to 48-hr were made\nfrom parallel-mode analyses on 9 of these days.\nComparisons between these forecasts and the correspond-\ning operational forecasts show at best a very slight\nimprovement in forecast skill from use of VTPR soundings.\nDifferences between analyses produced with and without\nthe satellite retrievals are, in general, quite small.\nThis appears to be due, at least in part, to a lack of\nretrievals from the meteorologically active baroclinic\nzones.\n1. INTRODUCTION\nVertical temperature profile radiometer (VTPR) data from NOAA satellites\nhave been used operationally at the National Meteorological Center (NMC) since\nlate December 1972. In March and April 1973, we conducted a test of the\nimpact of VTPR data on operational analyses and forecasts. In this test,\nwe\nestablished a parallel or B-mode 12-hr analysis/forecast cycle--similar\nto\nthe operational or A-mode cycle at NMC, except that VTPR data were excluded\nfrom the B-mode. B-mode operations were run on the CDC 6600 system at NMC\nand on the IBM 360/95 at the Goddard Institute for Space Studies, New York\n(GISS) via a remote terminal located at NMC. 1\nMention of a commercial product does not constitute an endorsement.","The test was conducted in two different stages. In Stage 1, we ran the\nparallel-mode system on a near-real-time basis during two test periods,\nproducing two series of analyses which were independent of VTPR data. B-mode\nanalyses were mapped at selected levels and compared both visually and\nstatistically with the corresponding operational or A-mode analyses. Period\n1 extended from 9 March to 21 March, Period 2 from 27 March to 13 April. In\nStage 2, we selected 9 \"forecast days\" and made predictions to 48-hr from\nparallel-mode analyses on each of these days. A- and B-mode forecasts were\nmapped and compared. Both sets of forecasts were verified against operational\nanalyses. Four of the days--11 March, 14 March, 17 March, and 20 March--are\nfrom Period 1. Five days--30 March, 2 April, 4 April, 7 April, and 10 April--\nare from Period 2. Selection of days was not random. Eleven March, 20 March,\n30 March, and 4 April, especially, were chosen as days when relatively large\ndifferences appeared between A- and B-mode analyses.\nThe principal goal of the test was to determine whether use of VTPR data\nincreased the skill of operational forecasts. Significant differences in\nforecast skill can be expected only if there are fairly large differences\nbetween analyses produced with and without VTPR data. Analysis differences\nare discussed in Section 5, forecast verifications in Section 6. Results\nmust be interpreted in terms of the analysis/forecast system, the method of\nassimilating VTPR data, and the amounts and types of data available during\nthe test. These are discussed in Sections 2, 3, and 4.\n2. THE OPERATIONAL MODE\nA. Analysis/Forecast Cycle\nThe sequence of analysis/forecast operations used by NMC at the time of the\ntest is summarized in figure 1. Analyses were produced at 00 and 12 GMT.\nData cut-off times for these analyses were 0325 and 1525 GMT. First-guess\nfields for the 12 GMT operational analysis were provided by a 12-hr forecast\nfrom the 00 GMT analysis. There was a special \"final\" analysis of 12 GMT\ndata with a late cut-off time of 22 GMT. A 12-hr forecast from this final\nanalysis gave first guess values for the analysis at 00 GMT.\nThe method of analysis was based upon the \"successive correction\" technique\n(Bergthorssen and Doos 1955; Cressman 1959). Analyses were produced on the\noctagonal grid in figure 2, at 10 pressure levels from 1000- to 100-mb.\nPredictions were made with the six-layer primitive equation model developed\nby Shuman and Hovermale (1968).\nB. VTPR Data\nSatellite soundings received from the National Environmental Satellite\nService (NESS) were made within areas that are roughly 600 km on a side.\nRetrievals were made only over the oceans. Soundings were derived by the so-\ncalled Minimum Information Technique (McMillin et al 1973), which uses short-\nperiod NMC temperature forecasts as a first guess for each retrieval.\n2","Heights and temperatures from the VTPR soundings were \"updated\" to either\n00 or 12 GMT before they were used in the analysis codes. Updating could be\nfor periods as long as 6-hr; for example, satellite observations taken between\n06 and 18 GMT were used in the 12 GMT final analysis. Updating was accomplished\nby using the 12-hr tendency from the previous model forecast (Desmarais 1972).\nThe current 1000-mb analysis was used as a reference level for converting the\nVTPR temperatures to heights at constant pressure levels.\nC. Bogus Reports\nSubjective or bogus \"observations\" are created routinely as part of the\nnormal forecast operation at NMC. Analyses at 300-mb and at the surface are\nmonitored by meteorologists who have the power to delete or correct \"bad\" data\nand to create \"observations\" in relatively data-free regions. This process,\nused at most operational centers, is intended primarily to improve the quality\nof the numerical analysis by incorporating subjectively the information from\nsatellite pictures and from careful hand analysis of the relatively dense\nsurface ship network. Observations are created only over the ocean areas and\nonly at the surface and at 300-mb. However, they influence the analyses at\nall levels because of the vertical consistency imposed within the numerical\nanalysis routines.\n3. THE B-MODE\nA. Analysis/Forecast Cycle\nThe B-mode used the same procedures and codes as the NMC operational or A-\nmode system. However, to simplify the B-mode, we eliminated the 1200 GMT\noperational analysis from the analysis/forecast cycle. This analysis is not\nused to privide first-guess fields (fig. 1), and its elimination meant only\nthat all B-mode test forecasts beyond 12-hr were made from B-mode analyses\nat 00 GMT.\nB-mode operations are summarized in figure 3. Analysis codes were run at\nNMC; initialization and forecast codes were run at GISS using the remote\nterminal installed for this purpose at NMC. Analyzed fields and processed\ndata were transferred between GISS and NMC using a tape-to-tape unit available\nat NESS.\nB. Bogus Reports\nThe bogusing operation posed a problem in the design of the B-mode. A-mode\nbogus was created with full knowledge of VTPR reports. If we used operational\nbogus in the B-mode analyses, the B-mode would not be completely independent\nof VTPR data. The bogusing operation is an important part of the NMC analysis/\nforecast system, and we decided not to exclude this operation from the parallel\ncycle. Consequently, we established a separate and completely independent\nbogusing operation as part of the B-mode. This operation was made as similar\nas possible to the normal NMC operation. It was carried out by experienced\nforecasters operating under realistic deadlines with, as nearly as possible,\nthe same data available to A-mode and B-mode bogus teams except for the VTPR\nreports.\n3","Including separate bogus data in A- and B-modes makes it difficult to\nassess directly the impact of VTPR data on the operational analyses. However,\nthe experiment as designed does answer the fundamental question: \"What\ndid\nwe gain operationally through use of VTPR data?\" The B-mode simulates the\nsystem as it would have existed if there were no VTPR reports.\n4.\nDATA BASE\nThe data base common to both modes consisted of surface synoptic reports,\nsurface-based upper air data, and winds derived from satellite imagery (Hubert\nand Whitney 1971) . The average numbers of upper air reports available to\nboth A- and B-modes at each analysis level are summarized in tables 1 and 2.\nTable 3 shows the average numbers of VTPR reports available to A-mode analyses\nduring the period of the test. Typically, about 40 bogus reports of 300-mb\nheights and winds were created in both A- and B-mode operations.\nTable 1. Average upper air data counts for 00 GMT A-\nand B-mode analyses. Data cut-off 0325 GMT\nPressure Level (mb)\n100\n250\n200\n150\n400\n300\n850\n700\n500\nData Type\nRaob, Rawin\n379\n429\n522\n409\n468\n555\n523\nand Pibal\n583\n572\nReconnais-\nsance\n0\n0\n0\n1\n0\naircraft\n0\n0\n5\n0\nAircraft\n4\n0\n162\n296\n194\n33\n0\n1\n4\nwinds\nSatellite-\nderived\n0\n0\n17\n17\n17\n10\n0\nwinds\n51\n51\n4","Table 2. Average upper air data counts for 12 GMT A-\nand B-mode analyses. Data cut-off 22 GMT\nPressure Level (mb)\nData Type\n850\n700\n500\n400\n300\n250\n200\n150\n100\nRaob, Rawin\nand Pibal\n713\n693\n618\n590\n573\n435\n531\n422\n486\nReconnais-\nsance\naircraft\n0\n0\n1\n0\n0\n0\n0\n0\n0\nAircraft\nwinds\n0\n1\n9\n33\n143\n307\n135 a\n2\n0\nSatellite-\nderived\nwinds\n48\n48\n8\n0\n11\n11\n11\n0\n0\nTable 3. Average numbers of VTPR reports in A-mode analyses\nAnalysis\n850\n700\n500\n400\n300\n250\n200\n150\n100\n108\n108\n0 GMT\n108\n108\n108\n108\n108\n108\n108\n12 GMT\n93\n93\n93\n93\n93\n93\n93\n93\n93\n5. ANALYSIS COMPARISONS\nof the many maps and statistics produced, we will show only a few--which we\nbelieve are sufficient to support the major conclusions of this section.\nFigures 4 and 5 show root-mean-square difference between 500-mb heights on\nA- and B-mode analyses over North America, the Atlantic, and the Pacific for\nPeriods 1 and 2. Regions are defined in figure 2. Differences have been\n5","weighted to take into account variations in the area on the earth that each\ngrid point represents.\nNumbers at the bottom are the total number of VTPR reports available to\neach A-mode analysis over the Atlantic and Pacific separately. VTPR\ntabulations in this case are for the entire ocean areas north of 20° N --\nnot the regions shown in figure 2.\nThe B-mode in each period began with the operational or A-mode first guess\nso that differences on day 0 result strictly from the VTPR and bogus reports\nat analysis time. Thereafter, the B-mode cycled on its own prediction.\nWhere the observational network is relatively dense so that the effect of\nthe first guess is slight and where the same observations are available to\nboth modes -- as over North America we would expect the analyses to be\nessentially the same. Largest differences should occur over the Atlantic\nand Pacific where the A-mode has VTPR and each mode has a different set of\nbogus reports. Where observations in either mode are scarce, first-guess\nfields assume greater importance and the difference between analyses should\nincrease with the number of days into the period.\nAs expected, differences are largest over the Pacific and smallest over\nNorth America. Over the Pacific, the average value of the root-mean-square\ndifference is 25.3-m for all days in Period 1, 27.2-m for Period 2. The\naverage root-mean-square difference is less than 5-m over North America in\nboth periods. The largest difference observed in either period is 57-m at\n00 GMT, 4 April.\nDifferences do not increase with the number of days into each B-mode cycle.\nThere is a slight indication of growth with time in Period 1; however,\ndifferences decrease with time in Period 2 -- probably because the period\nextends into late spring, when flow patterns over the Pacific are less\nperturbed. The effect of the first guess is apparent only when the height\ndifference is very large -- as on 4 April. Even then, differences diminish\nquickly. By 00 GMT, 5 April (fig. 5), the RMS difference has dropped from\n57.2-m to 27.5-m, approximately the normal value for Period 2. Apparently,\nthere is enough information available to the B-mode so that \"errors\" or\ndifferences once introduced are quite quickly removed. This information may\nbe derived from actual data; for example, from aircraft winds. Or it may be\nrelated to the subjective analysis and creation of bogus reports.\nOnly a very small percentage of the variations in RMS height difference\nwithin the Atlantic and Pacific in figures 4 and 5 can be explained in terms\nof the number of VTPR reports in each A-mode analysis. Correlation\ncoefficients between height differences and VTPR totals in the Atlantic and\nPacific are 0.15 and 0.04, respectively.\nWithout examination of individual maps, it is difficult to envision the kind\nof differences that these statistics imply. For this reason, and to provide\nbackground information concerning initial analyses on which forecasts were\nbased, figures 6 through 14 show A- and B-mode analyses of 500-mb heights on\neach forecast day. Shown as well are the differences (A-B) between these\nanalyses, and the locations of VTPR reports available to the A-mode. Each\n6","analysis is for 00 GMT. Notice first that A- and B-mode analyses are\npractically the same over the continents. Difference centers, except on\n14 March and 4 April, are restricted to the ocean areas.\nOn each day, VTPR reports were available from at least three orbits of NOAA-\n2 over the Pacific and two over the Atlantic. There are gaps in the data\nbetween orbits to the south of 49°N (see McMillin et al 1973), and gaps\nwithin orbits where retrievals were not made or soundings were rejected.\nA. 11 March (fig. 6)\nA-mode heights are higher than B-mode heights in a broad region across the\ncentral Pacific. Differences exceed 100-m in a small area south of Alaska.\nThere are two VTPR reports within the 60-m difference contour in this area,\nsuggesting that differences observed are a direct result of current VTPR\ndata. Comparing figures 6a and 6b, the major difference between the two\nanalyses is in the amplitude of a short-wave trough moving into the ridge to\nthe south of Alaska. This system is slightly stronger in the B- than in the\nA-mode analysis. There is a second 100-m difference center in the Atlantic\nnear 15°N -- to the south of the Atlantic subregion in figure 2.\nB. 14 March (fig. 7)\nDifferences in the Atlantic are extremely small (figs. 4 and 7c). There is\na 100-m difference center in the eastern Pacific between 45°N and 50° N that\nappears in figures 7a and 7b as a difference in the trough structure in the\ncentral and eastern Pacific. There are two VTPR reports within the 60-m\ndifference contour that directly influence the A-mode analysis.\nC. 17 March (fig. 8)\nLargest height differences occur between 15°N and 20° N in the Atlantic and\nin a broad zone to the south of the main current in the central Pacific. The\nridge between the two troughs in the central and eastern Pacific (figs. 8a and\n8b) is more pronounced in the A- than in the B-mode.\nThe shape and location of the 60-m contour in the Pacific suggest a direct\nconnection with VTPR data.\nD. 20 March (fig. 9)\nThe 60-m difference contour in the central Pacific reflects differences in\nthe curvature of the flow downstream from a long-wave trough (figs. 9a and\n9b). VTPR reports obviously affect the shape of this contour. However,\nmaximum differences occur where there are gaps in the data.\n2In both cases, differences are due to slight differences between first-guess\nvalues that allowed a \"marginal\" observation to be accepted in one mode and\nrejected in the other.\n7","E. 30 March (fig. 10)\nDifference centers in the eastern Pacific are associated with differences in\nthe amplitude of a short-wave trough extending southeastward from Alaska.\nThis trough is considerably stronger in the B-mode analysis and extends into\nlower latitudes. In the A-mode, the wind maximum in the trough is located\nnear 45 N. In the B-mode, this wind maximum is shifted southward by about\n5 degrees.\nThe negative difference center appears to be clearly related to current\nVTPR reports. Positive differences are located along a satellite orbit;\nhowever, most of the retrievals in this area are missing and the largest\nheight differences -- more than 120-m -- occur within a gap in the data.\nF. 2 April (fig. 11)\nA-mode heights are slightly lower in the northern portion of the major trough\nin the central Pacific. There are six VTPR reports on or inside the 40-m\ndifference contour and the small differences are probably related to current\nVTPR data.\nG. 4 April (fig. 12)\nThe very large differences apparent in figure 5 are concentrated within a\nbroad area in the eastern Pacific where A-mode heights are as much as 220-m\nlower than those for the B-mode. Both analyses (figs. 12a and 12b) show a\nsharp ridge in the central Pacific and a trough to the north of Hawaii. The\ntrough is about 2 degrees further to the west in the A-mode analysis. The\nmajor difference between analyses is in the strength of the current in the\nsouthern portion of this trough. In the A-mode analysis, contours dip\nsharply southward in the trough. Maximum winds, as indicated by the height\ngradient, are located near 30°N. In the B-mode analysis, the height gradient\nin the trough is much weaker and the zone of maximum winds appears between\n45°N and 50 N.\nThe difference center in figure 12c is located over the subpoint track for\nthe second of three NOAA-2 orbits over the Pacific (fig. 12d). However,\nnumerous retrievals from this orbit are missing -- partly as a result of\nextensive cloud cover near the trough -- and the association between height\ndifferences and current VTPR reports is not clear. Further examination\nrevealed. that the extreme southward displacement of the jet in the A-mode\nanalysis resulted from a combination of VTPR and bogus reports. The general\npattern of the A-mode analysis in the eastern Pacific was forced by the VTPR\ndata; however, because of the missing retrievals, subjective data were created\nwhich may have overemphasized the southward penetration of cold air.\nH. 7 April (fig. 13)\nTwo small 60-m difference centers in the Pacific reflect slight variations\nin the position of a trough and the intensity of the upstream ridge near 35°N.\nBoth centers appear to be clearly related to VTPR data. A 60-m difference in\nthe Atlantic is associated with one irregular contour on the A-mode analysis\nto the southeast of Greenland. The shape of this contour appears to be forced\nby nearby VTPR reports.\n8","I. 10 April (fig. 14)\nSeparate difference centers in the Atlantic and the Pacific reflect\ndifferences in the position and orientation of 500-mb troughs. The center\nin the Atlantic encloses two VTPR reports. Differences in the Pacific,\nhowever, are located between two satellite tracks.\nJ. Summary\nDespite the presence of an average of about 100 VTPR reports in each\noperational analysis, differences between analyses with and without VTPR data\nwere quite small. Root-mean-square differences between A-mode and B-mode\n500-mb height analyses averaged about 15-m in the Atlantic, 25-m in the\nPacific. These differences are less than the errors in a typical 12-hr\nforecast and probably not much larger than differences we would expect from\nthe bogusing alone. Large differences were observed only on 4 April and\ndifferences in this case were due largely to A-mode bogus reports.\n6. FORECAST VERIFICATION\nForecasts to 48-hr from A- and B-mode analyses were verified over North\nAmerica, Europe, the eastern Atlantic and eastern Pacific. The discussion here,\nhowever, will be limited to the verification scores for forecasts of height,\ntemperature, wind, and height gradients over North America.\nForecasts were verified against the operational analyses. For short-period\nforecasts of 12- and perhaps 24-hr, this probably introduced a bias in favor of\nthe A-mode. Statistics computed include the mean errors, standard deviations\nof error, root-mean-square errors and the correlation coefficients between\nobserved and forecast values. Height gradients were verified using the S1\nscore (Teweles and Wobus 1954). This score, computed routinely at NMC, is\nsimply the magnitude of the vector error in forecast height gradient divided\nby the magnitude of the observed or forecast gradient -- whichever is larger.\nThus, the lower the S1 score, the better the forecast.\nRoot-mean-square errors in heights, temperatures and winds for the pooled\nsample of nine forecast days are summarized in tables 4 through 6. Root-mean-\nsquare errors are area weighted. The pooled root-mean-square error is simply\nL\n9\n(RMSE)\ni\nRMSE =\ni=1\n9\nwhere RMSE is the value on day i.\n9","Table 4. Root-mean-square errors (meters) in forecast heights.\nPooled sample. North America (Region 1 in fig. 2).\nAn asterisk denotes the better of the two forecasts\nForecast Length (hr)\n36\n48\nLevel\n12\n24\nA\nB\nA\nB\n(mb)\nA\nB\nA\nB\n32.2*\n33.0\n43.6*\n44.0\n50.3*\n50.5\n1000\n21.0*\n21.7\n22.3\n29.1*\n30.6\n39.2*\n40.1\n46.5*\n47.3\n850\n22.1*\n33.2\n41.8*\n43.6\n49.3*\n52.2\n700\n23.4*\n23.8\n30.6*\n55.1\n60.3*\n65.2\n500\n27.9*\n28.5\n36.8*\n40.6\n52.1*\n66.5*\n69.5\n82.3*\n83.8\n300\n34.3*\n35.7\n49.8*\n53.0\nRoot-mean-square errors (deg. C) in forecast temperatures.\nTable 5.\nPooled sample. North America (Region 1 in fig. 2).\nAn asterisk denotes the better of the two forecasts\nForecast Length (hr)\n48\n36\n24\n12\nLevel\nB\nB\nA\nB\nA\n(mb)\nA\nB\nA\n4.3*\n4.7\n2.9*\n3.0\n2.7*\n2.8\n1.7\n1.7\n850\n3.3\n3.3\n2.7\n2.7\n2.1\n2.1\n700\n1.3\n1.3\n2.7\n2.6*\n1.7\n2.2*\n2.3\n1.3\n1.7\n500\n1.3\n2.4\n2.2*\n2.4\n1.9\n2.3\n1.9\n1.9\n1.9\n300\n10","Table 6. Root-mean-square vector errors (m/s) in forecast winds.\nPooled sample. North America (Region 1 in fig. 2).\nAn asterisk denotes the better of the two forecasts\nForecast Length (hr)\nLevel\n12\n24\n36\n48\n(mb)\nA\nB\nA\nB\nA\nB\nA\nB\n850\n4.9\n4.9\n5.6*\n5.8\n7.2*\n7.4\n7.7*\n7.8\n700\n4.6*\n4.7\n5.7*\n5.9\n7.1*\n7.6\n8.3*\n8.6\n500\n5.3*\n5.4\n7.4*\n7.6\n9.4*\n9.8\n10.8*\n10.9\n300\n8.0*\n8.2\n10.1*\n10.2\n13.3*\n13.8\n15.8*\n16.0\nFor the sample as a whole, there is essentially no difference between A-\nand B-mode temperature errors. Forecast errors in heights and winds are\nslightly but consistently lower for the A- than for the B-mode.\nSurface and 500-mb S1 scores and 1000- and 500-mb height errors on\nindividual days are shown in tables 7 through 10.\nProbably the most legitimate indication of a statistically significant\ndifference between A- and B-mode verification scores is the number of days on\nwhich any one mode is better. As shown in tables 7 to 10, the numbers of\ncomparisons \"won\" by A- and B-modes are approximately the same. Except for\n2 and 4 April, the 9 forecast days are at least 3 days apart. If these days\ncan be viewed as independent tests with two possible outcomes, P values for\nrejecting the null hypothesis that the skill of both modes is actually the\nsame can be determined from the binomial distribution (see, for example,\nBrownlee 1960). The differences in forecast skill are statistically signifi-\ncant at the 5-percent level only if one mode is better on at least 8 of the 9\ndays. Neither mode produced better verification scores in any category on\nmore than 6 days.\n11","Root-mean-square errors (meters) in 500-mb heights for A- and\nTable 7.\nB-mode forecasts on each forecast day. N is the number of\ncomparisons in which the A-mode is better. North America\n(Region 1 in fig. 2)\nForecast Period (hr)\nForecast\n48\n24\n36\n12\nMode\nDay\n79.8\n25.1\n49.2\n70.1\n11 March\nA\n48.5\n67.0\n73.2\n25.9\nB\n47.8\n63.1\n76.9\n22.4\n14 March\nA\n73.6\n93.0\n25.7\n55.2\nB\n70.0\n44.9\n59.4\n25.9\n17 March\nA\n42.9\n59.7\n68.8\n28.5\nB\n46.4\n48.1\n45.7\n32.1\n20 March\nA\n44.5\n46.0\n41.2\n29.0\nB\n60.6\n36.4\n51.2\n23.9\n30 March\nA\n43.7\n62.5\n79.1\n29.0\nB\n30.1\n34.2\n40.5\n25.3\n2 April\nA\n44.1\n35.6\n39.1\n25.3\nB\n70.2\n33.8\n53.7\n24.0\n4 April\nA\n42.2\n52.6\n60.3\n27.9\nB\n46.8\n52.6\n25.4\n25.6\n7 April\nA\n55.4\n27.5\n45.9\n24.9\nB\n41.0\n46.9\n33.0\n26.3\n10 April\nA\n39.8\n32.6\n50.3\n24.2\nB\n51.8\n60.6\n27.1\n37.0\nA\nMean\n63.4\n28.1\n39.7\n53.9\nB\n4\n5\n5\n5\nN\n12","Table 8.\nRoot-mean-square errors (meters) in 1000-mb heights for\neach forecast day. N is the number of comparisons in\nwhich the A-mode is better. North America (Region 1\nin fig. 2)\nForecast Period\nForecast\n(hr)\nMode\n12\n24\n36\n48\nDay\n17.9\n41.3\n59.4\n62.3\n11 March\nA\n24.0\n39.2\n54.3\n57.5\nB\n12.8\n35.9\n51.2\n69.0\n14 March\nA\n14.4\n40.1\n56.0\n73.5\nB\n44.9\n44.4\n17 March\n23.7\nA\n37.3\n26.8\n37.0\n44.6\n45.7\nB\n34.3\n33.1\n38.6\n48.2\n20 March\nA\n37.9\n41.0\nB\n30.3\n31.1\n18.4\n25.2\n30.7\n38.5\n30 March\nA\nB\n16.7\n29.4\n40.3\n49.5\n30.2\n33.4\n2 April\nA\n21.5\n23.5\n20.7\n26.2\n32.5\n33.5\nB\n16.7\n34.3\n48.0\n51.7\n4 April\nA\n20.7\n35.3\n41.9\n47.5\nB\n19.9\n32.0\n48.2\n53.2\n7 April\nA\n52.0\n19.8\n31.9\n47.3\nB\n18.3\n23.2\n32.9\n41.7\n10 April\nA\n16.5\n21.9\n35.4\n44.3\nB\n42.7\n49.2\n20.4\n31.8\nMean\nA\n32.5\n43.4\n49.4\nB\n21.1\n4\n5\n4\n4\nN\n13","Table 9. S1 scores at 500-mb for A- and B-mode forecasts on each\nforecast day. North America. N is the number of A-mode\nwins. Scores computed using the standard NMC verification\ngrid in figure 15\nForecast Period (hr)\nForecast\n36\n48\nMode\n12\n24\nDay\n33.6\n44.6\n52.6\n22.3\n11 March\nA\n46.6\n22.0\n33.2\n42.9\nB\n40.9\n22.0\n27.7\n37.5\n14 March\nA\n46.0\n54.4\n23.4\n31.9\nB\n39.7\n48.0\n20.2\n35.4\n17 March\nA\n40.9\n20.2\n34.3\n40.3\nB\n48.8\n25.6\n30.1\n41.0\n20 March\nA\n35.7\n48.4\n27.8\n29.2\nB\n37.1\n36.8\n22.7\n28.6\n30 March\nA\n47.3\n30.0\n36.0\n44.9\nB\n34.7\n23.8\n24.6\n31.0\n2 April\nA\n33.0\n37.1\n23.5\n26.8\nB\n42.2\n50.3\n23.7\n29.6\n4 April\nA\n45.1\n26.4\n33.3\n37.7\nB\n43.4\n19.2\n27.5\n39.3\n7 April\nA\n40.4\n52.6\n21.4\n29.4\nB\n36.9\n39.0\n18.8\n28.5\n10 April\nA\n35.8\n27.2\n37.7\nB\n17.6\n44.2\n29.8\n39.0\n22.1\nMean\nA\n40.0\n45.6\n23.6\n31.4\nB\n4\n5\n5\n6\nN\n14","Table 10. Surface S1 scores for each forecast day North America.\nN is the number of A-mode wins. Scores computed using\nthe standard NMC verification grid in figure 15\nForecast Period (hr)\nForecast\n48\n36\n12\n24\nMode\nDay\n68.8\n45.5\n53.9\n28.8\n11 March\nA\n65.4\n41.7\n53.3\n30.0\nB\n61.2\n48.6\n54.9\n22.8\n14 March\nA\n69.3\n58.2\n24.0\n49.3\nB\n46.7\n46.2\n50.1\n32.6\n17 March\nA\n44.1\n43.9\n49.9\n33.1\nB\n71.0\n50.4\n60.7\n38.3\n20 March\nA\n48.2\n56.6\n63.3\n37.1\nB\n52.3\n46.3\n34.3\n50.1\n30 March\nA\n53.3\n53.2\n32.2\n50.4\nB\n48.4\n26.4\n36.5\n38.5\n2 April\nA\n42.1\n50.4\n39.2\n28.5\nB\n67.6\n62.9\n51.5\n33.7\n4 April\nA\n59.6\n40.3\n54.7\n62.7\nB\n76.1\n39.0\n56.0\n23.5\n7 April\nA\n79.5\n59.2\n27.2\n38.7\nB\n67.2\n58.8\n41.0\n52.0\n10 April\nA\n64.1\n37.9\n47.0\n59.1\nB\n62.1\n46.7\n53.6\n31.3\nMean\nA\n54.9\n61.0\n45.9\n32.3\nB\n4\n6\n4\n5\nN\n15","Forecast fields examined each day include the 12- through 48-hr forecasts\nof 1000-, 500-, and 300-mb heights, 500-mb vorticities, 1000- to 500-mb\nthicknesses and 300-mb winds. Figures 16 through 24 show the differences\nbetween A- and B-mode 500-mb height forecasts on each of the 9 days.\nDifferences can be compared with differences between initial analyses shown\nin figures 6 through 14.\nA. 11 March (fig. 16)\nPositive differences in the central Pacific (fig. 6) diminish during the\nfirst 12-hr of the forecast. Differences remain less than 60-m over North\nAmerica throughout the 48-hr of the forecast. At 36-and 48-hr, S1 scores\nand RMS height errors at 500-and 1000-mb (tables 7 through 10) are slightly\nlower for B-mode than for A-mode forecasts.\nB. 14 March (fig. 17)\nBy 36--and 48-hr, differences of more than 100-m have moved into the verifica-\ntion region from the eastern Pacific. All scores in tables 7 through 10 show\nconsistently lower errors for A-mode than for B-mode forecasts. Table 9\nindicates that the S1 score at 500-mb is lower for the 48-hr A-mode forecast\nthan for the B-mode forecast at 36-hr.\nC. 17 March (fig. 18)\nLarge differences between the two forecasts develop over the eastern Pacific,\nbut differences remain to the west of the verification region. Differences\nover North America remain less than 60-m during the entire forecast period,\nand there appears to be no meaningful difference between A- and B-mode scores\nin tables 7 through 10.\nD. 20 March (fig. 19)\nDifferences develop in the central and eastern Pacific but remain mainly to\nthe west of the verification region. Forecast scores for A- and B-modes are\nessentially the same except that 1000-mb height and height-gradient forecasts\nat 48-hr are slightly better for the B- than for the A-mode forecasts.\nE. 30 March (fig. 20)\nDifferences of more than 100-m move inland over California and Nevada.\nDifferences of more than 60-m develop over the central States. Tables 7\nthrough 10 show a clear superiority for the A-mode forecasts. S1 scores and\nroot-mean-square errors at both 500-and 1000-mb are lower for 48-hr A-mode\nforecasts than for 36-hr forecasts from the B-mode.\nF. 2 April (fig. 21)\nDifferences remain small out to 48-hr. Errors are generally slightly lower\nfor A- than for B-mode forecasts, but there is no clear indication of\nsuperiority for either mode.\n16","G. 4 April (fig. 22)\nThe large negative center to the north of Hawaii (fig. 12c) moves\nsoutheastward. Large positive differences -- exceeding 220-m -- develop in\nthe Pacific northwest. Negative differences form over Minnesota and North\nDakota. Forecasts show about equal skill through 36-hr. At 48-hr, however,\nB-mode forecasts are clearly superior. B-mode 48-hr 1000-mb and surface\nforecasts (tables 8 and 10) are more accurate than the 36-hr forecasts from\nthe A-mode.\nH. 7 April (fig. 23)\nDifferences grow rapidly to the south of Alaska but remain small over the\nNorth American continent. Root-mean-square errors in tables 7 and 8 are\napproximately the same for A- and B-mode forecasts. S1 scores at 500- and\n1000-mb at 48-hr are definitely lower for A- than for B-mode forecasts.\nI. 10 April (fig. 24)\nDifferences remain small over the verification region. Tables 7 through 10\nshow no clear superiority for either mode.\nJ. Summary\nThe sample of 9 forecast days shows a slight improvement in average forecast\nskill in 36- and 48-hr forecasts of heights, winds, and height-gradients over\nNorth America from use of VTPR data. Differences are of the order of a few\nmeters in root-mean-square height error, a few tenths of meters per second in\n300-mb wind error, and about 1-percent in S1 score. Differences are probably\nnot significant -- eigher statistically or meteorologically. The overall\naccuracy of 48-hr A-mode height forecasts is about the same as that of 45-hr\nforecasts from the parallel mode.\nDifferences as large as 100-m between A- and B-mode 500-mb height forecasts\nover North America (figs. 16 through 24) appear on only 3 days: 14 March,\n30 March, and 4 April. And only on those days are there consistent and\napparently significant differences in the verification scores.\nOn 14 March and 30 March, A-mode forecasts are definitely better. The 48-hr\nforecast from the A-mode analysis on 4 April is clearly worse than the forecast\nfrom the parallel mode. In this case, however, the relatively large errors\nin the operational forecasts appear to be related to the creation of bogus\nreports (see previous section) that overdeveloped the trough in the eastern\nPacific (fig. 12) leading, in the forecast, to overamplification of the\ndownstream ridge and to large positive height errors in the 48-hr forecasts\nover the Pacific northwest (fig. 22 and Fawcett 1969).\n7. CONCLUS IONS\nTest results imply that the VTPR soundings, as retrieved and used at the time\nof the test produced at most a small improvement in the skill of NMC forecasts\n17","3\nThis conclusion is based upon a limited sample of 9\nover North America.\ndays and upon forecasts that extended to only 48-hr. However, the\nforecast\nclose similarities between analyses produced with and without VTPR data on\n30 test days gives rather convincing evidence that little new information\nwas provided to NMC by the satellite soundings. The problem seems not to\nbe that the data were bad or that the overall yield was too low. Instead,\nit appears to arise primarily from a bias in the locations of the VTPR\nsoundings (see figs. 9, 10, and 12). Retrievals that passed through the\nsystem were located mainly in relatively inactive regions where surface\nreports, aircraft winds, a few widely-spaced radiosonde observations, and\nforecast information may be all that we need to define the large-scale\nfeatures of the flow. It is in the meteorologically active regions where\nfields are changing rapidly and short-period forecasts are likely to be in\nerror that we most need additional reports.\nThe main problem as we see it is the very difficult one of obtaining\naccurate soundings from the areas where development is taking place -- where\nskies may be overcast or covered with multiple layers of clouds. Some\nimprovement is expected with an operational microwave sounding system (as\nplanned for TIROS N) and with continued development of infrared retrieval\ntechniques that increase the number and accuracy of soundings in areas with\nclouds.\nConclusions about the current utility of VTPR data should be tempered by\nthe fact that numerous changes have been made in the systems for processing\nand analyzing these data. NESS has replaced the Minimum Information Technique\nby a statistical regression scheme that is relatively independent of the NMC\nforecasts. NMC has replaced the analysis scheme described with a three-\ndimensional spectral method (Flattery 1971) and introduced final analysis\ncycles at both 00 and 12 GMT. We do not expect that improvements in forecast\nskill are automatic with the introduction of new types of data. Better\nforecasts are likely to be achieved only from a combined effort involving\nimprovements in the observing systems and methods for processing or retrieving\nthe data, development of better methods for assimilating asynoptic reports,\nand formulation of better numerical prediction models.\nResults of this study have been widely disseminated through informal\npublication and at national and international meetings. Publication at this\ntime is intended primarily to provide a reference for use in comparing these\nearly results with those from a series of tests planned in 1976.\nACKNOWLEDGMENTS\nThis test was conducted under NASA Contract S-70252-AG. Carl Amorose and\nGene Brown participated in the parallel mode operation. Doris Gordon of\nAutomation Division, NMC, wrote the programs used in daily comparisons between\n3Forecast verifications over Europe, and eastern Atlantic and eastern Pacific\n(not discussed) show no notable improvement from use of VTPR data. A more\nrecent study conducted by the British Meteorological Office (Atkins and Jones\n1975) indicated a small beneficial effect from use of VTPR data in forecasts\nover Europe.\n18","A- and B-mode analyses. Dr. Ken Bergman performed many of the tabulations\nand calculations involved in the analysis comparisons. We appreciate the\nexcellent cooperation received from Automation Division, NMC, and from Dr.\nHalem and his staff at GISS. We would also like to thank Dr. C. Hayden of\nNESS for his careful reivew of this paper.\nREFERENCES\nAtkins, M. J. and M. V. Jones, 1975: An experiment to determine the value\nof satellite infrared spectrometer (SIRS) data in numerical forecasting.\nMeteorological Magazine, 104, 125-142.\nBergthorssen P. and B. Doos, 1955: Numerical weather map analysis. Tellus,\n7, 329-340.\nBrownlee, K., 1960: Statistical Theory and Methodology in Science and\nEngineering. John Wiley and Sons, Inc., New York, N. Y., 570 pp.\nCressman, G. P., 1959: An operational objective analysis system. Monthly\nWeather Review, 87, 367-374.\nDesmarais, A., 1972: Updating asynoptic data for use in objective analyses.\nNOAA Technical Memorandum, NWS NMC-51, National Weather Service, National\nOceanic and Atmospheric Administration, U.S. Department of Commerce, Silver\nSpring, Maryland, 19 pp.\nFawcett, E. B., , 1969: Systematic errors in operational baroclinic prognoses\nat the National Meteorological Center. Monthly Weather Review, 97, 670-\n682.\nFlattery, T. W., 1971: Spectral models for global analysis and forecasting.\nProceedings Sixth AWS Technical Exchange Conference, U.S. Naval Academy.\nAir Weather Service Technical Report 242, 42-54.\nHubert, L. F. and L. F. Whitney, Jr., 1971: Wind estimation from geostrationary\nsatellite pictures. Monthly Weather Review, 99, 665-672.\nMcMillin, L. M. et al., 1973: Satellite infrared soundings from NOAA\nspacecraft. NOAA Technical Report, NESS 65, National Environmental\nSatellite Service, National Oceanic and Atmospheric Administration,\nU.S. Department of Commerce, Washington, D.C. 112 pp.\nShuman, F. G. and J. B. Hovermale, 1968: An operational six-layer primitive\nequation forecast model. Journal of Applied Meteorology, 7, 525-547.\nTeweles S. and H. B. Wobus, 1954: Verification of prognostic charts.\nBulletin American Meteorological Society, 35, 455-463.\n19","NMC ANALYSIS/FORECAST CYCLE\n12 GMT\n00 GMT\n00 GMT\n12 GMT\n00 GMT\nt o +96\nFCST\nFG\nFG\nFCST\nto +48\nOA\nO\nFCST\nF A\nFG\nFOST\nto +96\nOA\nFG\nOA= OPERATIONAL ANALYSIS\nFA= FINAL ANALYSIS\nFG= FIRST GUESS\nCIRCLE DENOTES DATA CUT-OFF TIME\nFCST\nt o +48\n0A\nO\nFCST\nFA\n12\n00 GMT\n12 GMT\n00 GMT\nGMT\n00 GMT\nFigure 1. NMC analysis/forecast cycle. March and April 1973.\n20","AREA 4\nAREA\n3\nAREA5\nAREA\nAREA\nFigure NMC analysis/forecast grid. Map projection 60°N. is\n2. stereographic. Grid spacing is 381-km at in\nAreas polar 1, 2 and 5 are used in analysis comparisons\nSection 5.\n21","NMC DATA BASE\nSATELLITE\nIMAGERY\nMODIFIED NMC\nDATA BASE.\nVTPR AND A-\nMODE BOGUS\nDELETED\nB-MODE BOGUS\nB-MODE DATA BASE\nMAPPED B-MODE\nGUESS FIELDS\nB-MODE ANALYSIS\nNESS-GISS\nB-MODE\nAT 1000 AND\nCDC 6600\nDATA LINK\nINITIALIZATION\n300 MB\nIBM 360/95\nANALYSES ON TAPE\nB-MODE\nFOR COMPARISON\nFORECAST\nAND DISPLAY\nIBM 360/195\nPROCESSED\nFORECAST\nFIRST-GUESS\nDIVERGENCE\nGISS-NESS\nFIELDS FOR\nFOR NEXT\nDATA LINK\nNEXT B-MODE\nB-MODE\nANALYSIS\nINITIALIZATION\nFigure 3. B-mode operations.\n22","RMS DIFFERENCES, A-MODE vs.B-MODE\nFOR 500MB GEOPOTENTIAL HEIGHTS\nCYCLE 1\n50\nNORTH AMERICAN\nATLANTIC\nMEAN\nSTD. DEV.\nPACIFIC\nNO. AMER.\n4.47\n1.66\nATL.\n14.77\n4.33\nPAC.\n25.34\n6.14\n40\n30\n20\n10\n0\nMAR\n9\n10\n11\n12\n13\n14\n15\n16\n17\n18\n19\n20\n21\n1973\n61\n40\n16\n29\n36\n36\n55\n6\n43\n40\nNO. OF\n52\n37\n60\n34\n48\n5\n60\n36\n35\n40\n43\n38\n44\nM\n27\nVTPR RPTS.\n40\n38\n?\n28\n39\n34\n44\n3\n40\n41\n4)\n22\n39\n36\n40\n0\n38\n33\n30\n35\n36\n40\n39\nM\n28\nFigure 4.\nRoot-mean-square differences between\nA- and B-mode 500-mb height analyses. Areas\n1, 2, and 5 of figure 2. Period 1.\n23","RMS DIFFERENCES, A-MODE vs. B-MODE FOR 500MB GEOPOTENTIAL HEIGHTS\nCYCLE 2\nas\nMEAN\nSTD DEV\nNORTH AMERICA\nNO. AMER.\n4.29\n+ 2.01\nATLANTIC\nATL.\n14.76\n+ 4.70\nPACIFIC\nPAC.\n27.17\n+ 10.17\n50\n40\n30\n20\n10\n0\n6\n7\n8\n9\n10\n11\n12\n13\n3\n4\n5\n31\n1\n2\n27\n28\n29\n30\nAPRIL\nMARCH 1973\n38\n71\n60\n59\n64\n37\n53\n42\n62\n43\n59\n54\n32\n39\n47\nM\n42\n57\n58\nNO. OF\n39\n46\n60\n50\n64\n49\n37\n32\n31\nM\n67\n42\n44\n34\n35\n44 56\n4,1\n37\n34\n45\n44\n36\n40\nVTPR RPTS.\n38\n28\nM\n58\n28\n32\n51\nSame as figure 4, except Period 2.\nFigure 5. .\n24","5880\n5880\n5880\n5880 5880\n5880\n5850\n5880\n5880\n5880\n5880\n5880\n5880\nyou\n5880\n5880\n5880\n5880\n5880\n500 MB HTS B MODE VALID 00 HOURS AFTER\n500 MB HEIGHT A MODE VALID 00 HOURS AFTER\n002 11 MAR 73\n00Z 11 MAR 73\n20\n20\na\n20\n889\n:-\n500 MB HT DIFF A-B MODES VALID 00 HOURS AFTER\n002 11 MAR 73\nVTPR REPORTS\n00Z 11 MAR, 1973\nFigure 6. A- and B-mode analyses of 500-mb - heights\nfor 00 GMT 11 March.\nA-mode analysis\na.\nb.\nB-mode analysis\nDifference (A-B) between A- and B-mode analyses\nC.\nd.\nLocations of VTPR reports available to the A-mode.\nDifference contours with absolute values of 60-m\nor more are transcribed on map d.\n25","5880\n5880 5880\n5580\n7310\n5880\n5880\n5880\n5900\n5880\n5100\n5100\n5820\n5820\n5820\ne\n5820\n5880\n5220\n5880\n5880\n5880\n5820\n5880\n5880\n5880\n5880\n00Z 14 MAR 73\n500 MB HTS B MODE VALID 00 HOURS AFTER\n500 M8 HEIGHT A MODE VALID 00 HOURS AFTER\n00Z 14 MAR 73\nO\n002 14 MAR 73\nVIPR REPORTS\neaz 14 MAR. 1973\n500 MB HT DIFF A-B MODES VALID 00 HOURS AFTER\nSame as figure 6, except 00 GMT 14 March.\nFigure 7. .\n26","5880\n5830\n5880\n5880\n5880\n5880\n5880\n5880\n5700\n5700\n5640\n5640\n5500\n5580\n5820\n5880\n5880\n5880\n5880\n5880\n5880\n5880\n5'80\n00Z 17 MAR 73\n500 MB HEIGHT A MODE VALID 00 HOURS AFTER\n500 MB HTS B MODE VALID 00 HOURS AFTER\n002 17 MAR 73\n20\nO\n20\n20\n20\n20\nwe\nVTPR REPORTS\n00Z 17 MAR. 1973\n500 MB HT DIFF A-8 MODES VALID 00 HOURS AFTER\n002 17 MAR 73\nSame as figure 6, , except 00 GMT 17 March.\nFigure 8. .\n27","5880\n5880 5880\n5880\n5940\n5880\n5880\n5880\n5880\n$6\n5880\n5880\n5880\n5820\n5880\n5880\n5880\n5829\nwe\n5880\n5880\n5880\n5880\n5080 5880\n5880\n5880\n500 MB HEIGHT A MODE VALID 00 HOURS AFTER\n500 MB HTS B MODE VALID 00 HOURS AFTER\n002 20 MAR 73\n00Z 20 MAR 73\n20\nwell\nVTPR REPORTS\n500 MB HT DIFF A-B MODES VALID 00 HOURS AFTER\n00Z 20 MAR 73\n00Z 20 MAR. 1973\nFigure 9. .\nSame as figure 6, , except 00 GMT 20 March.\n28","5850\n5880 5880\n5880\n5880\n5880\n5850\n5880\n5880\n5820\n5820\n5850\n5880\n5880\n500 MB HTS B-MODE VALID 00 HRS AFTER\n500 MB HEIGHT A MODE VALID 00 HOURS AFTER\n00Z 30 MAR 73\n00Z 30 MAR 73\n:\n0\n002 30 MAR 73\n500MB HT DIFF A-B MODE VALID 00 HRS AFTER\nVTPR REPORTS\n002 30 MAR, 1973\nFigure 10. .\nSame as figure 6, except 00 GMT 30 March.\n29","5880\n5880\n5880\n5880\nthe\n&\nSRAO\n5880\n5880\n5880\n5880\n5880\n5880\n5820\nis\n5880\n5880\n5880\n5880\n5880\n5760\nare\n5880\n5880\n5380\n5880\n002 02 APR 73\n500 MB HTS B-MODE VALID 00 HRS AFTER\n500 MB HEIGHT A MODE VALID 00 HOURS AFTER\n00Z 02 APR 73\nQ\n8\n-\n500MB HT DIFF A-8 MODE VALID 00 HRS AFTER\n002 02 APR 73\nVTPR REPORTS\n00Z 2 APR. 1973\nSame as figure 6, , except 00 GMT 2 April.\nFigure 11.\n30","5840\n5880\n5880\n5880\n5880\n5880\n5880\n5880\n5658\n5880\n5880\n5880\n5800\n5880\n5860\n5880\n5880\n5880\n5820\n5940\n5940\n5880\n5880\n5820\n5820\n5820\n5880\n5820\n5820\n5830\n5880\n5880\ninco\n5880\n5880\n5880\n5880\n500 MB HEIGHT A MODE VALID 00 HOURS AFTER\n500 MB HTS B MODE VALID 00 HOURS AFTER\n002 04 APR 73\n00Z 04 APR 73\n-\n20\n20\n-20\n-20\nO\nwas\n-\n500 MB HT DIFF A-B MODES VALID 00 HOURS AFTER\n002 04 APR 73\nVTPR REPORTS\nsoz,\n4\nAPR. 1973\nSame as figure 6, except 00 GMT 4 April.\nFigure 12.\n31","5880\n5880\n5880\n5880\n5880\n5880\ns\n5880\n5880\n5880\n5820\n5820\n5880\n5880\n5880\n5820\n5820\n5880\n5880\n5850\n5880\nlaws\n5880 5880\n5880 5850\n500 MB HTS 8 MODE VALID 00 HOURS AFTER\n002 07 APR 73\n002 07 APR 73\n500 MB HEIGHT A MODE VALID 00 HOURS AFTER\n-20\n20\n20\n20\n38\n-20\n00Z 07 APR 73\n00Z 7 APR. 1973\n500 NO HT DIFF A-8 MODES VALID 00 HOURS AFTER\nVTPR REPORTS\nSame as figure 6, except 00 GMT 7 April.\nFigure 13.\n32","3880\n5880\n5880\n5880\n5700\n5580\n5580\n2460\n2650\n5340\n5340\n5880\n5880\n10\n5520\n5880\n5520\n5880\no\n5100\n5100\n$400\n5850\n9340\n5340\n5880\n5160\nSTE\n5220\n5220\n$288\nRE\n5880\n5880\n5880\n5880\niii\n5880\n5760\n5940\n500 MB HEIGHT A MODE VALID 00 HOURS AFTER\n500 MB HTS B MODE VALID 00 HOURS AFTER\n00Z 10 APR 73\n00Z 10 APR 73\n-20\n-20\n-20\n20\nQ\n-20\n-\n-20\n-20\n500 MB HT DIFF A-B MODES VALID 00 HOURS AFTER\n002 10 APR 73\nVTPR REPORTS\n00% 10 APR. 1973\nFigure 14. Same as figure 6, , except 00 GMT 10 April.\n33","so\n5 g\n$6.8\n100\n140\n120\n70\nN\nx\n60\nto\n50\n140\n130\n40\n30\n100\n20\n140\n110\n130\nFigure 15. . Verification region and grid used in\ncalculations of S1 scores.\n34","20\n20\n20\n20\n20\n20\n20\n20\n20\n20\na\n$20\n20\n20\n500 MB HT DIFF A-B MODES VALID 12 HOURS AFTER\n500 MB HT DIFF A-B MODES VALID 24 HOURS AFTER\n002 11 MAR 73\n002 11 MAR 73\n20\n20\n20\n20\n28\n20\n20\n20\n20\n20\n20\n20\n20\n:-\n20\n20\n500 MB HT DIFF A-B MODES VALID 36 HOURS AFTER\n00Z 11 MAR 73\n500 MB HT DIFF A- 8 MODES VALID 48 HOURS AFTER\n002 11 MAR 73\nFigure 16. Differences (A-B) between 12-, 24-,\n36-, and 48-hr A- and B-mode forecasts of\n500-mb height from initial analyses at 00 GMT\n11 March.\n35","20\n20\n20\n.\n20\n20\n00Z 14 MAP. 73\n500 MB HT DIFF A-B MODES VALID 12 HOURS AFTER\n500 MB HT DIFF A-B MODES VALID 24 HOURS AFTER\n00Z 14 MAR 73\n2020\n2020\n20\n20\n20\nOctober\n20\n-20\n-20\n20\n20\n20\n20\n20\n20 20 20 20 20\n20\n,\n002 14 MAR 73\n500 MB HT DIFF A-B MODES VALID 48 HOURS AFTER\n002 14 MAR 73\n.\n500 MB HT DIFF A-8 MODES VALID 36 HOURS AFTER\nFigure 17. . Same as figure 16, except 14 March.\n36","20\n20\n20\n20\n20\n20\n:-\n500 MB HT DIFF A-B MODES VALID 12 HOURS AFTER\n500 MB HT DIFF A-B MODES VALID 12 HOURS AFTER\n00Z 17 MAR 73\n002 17 MAR 73\n20 20\n20\n20\n20\n20\nLY\n20\n20\n-20\n-20\n20\n20\n20\n20\n500 NB HT DIFF A-B NODES VAL ID 48 HOURS AFTER\n002 17 MAR 73\n500 MB HT DIFF A-B MODES VALID 36 HOURS AFTER\n00Z 17 MAR 73\nSame as figure 16, except 17 March.\nFigure 18.\n37","20\n20\n20\n20\n20\n20\n20\n20\n500 MB HT DIFF A-B MODES VALID 24 HOURS AFTER\n00Z 20 MAR 73\n500 MB HT DIFF A-B MODES VALID 12 HOURS AFTER\n002 20 MAR 73\n20 20\n20\n20\n20\n20\n20\n20\n50\n20\n20\n20\n20\n20\n20\nse\n2\n20\n20 20 20\n20 20\n00Z 20 MAR 73\n500 MB HT DIFF A-B MODES VALID 48 HOURS AFTER\n002 20 MAR 73\n500 MB HT DIFF A-B MODES VALID 36 HOURS AFTER\nSame as figure 16, except 20 March.\nFigure 19. .\n38","-20\n20\n-20\n20\n20\n20\n20\n20\n20\n20\n20\n126\n20\n20\n-20\n-\n20\nwas\n20 20 -20\n-20\n-20\n-20\n500 MB HT DIFF A-B MODES VALID 12 HOURS AFTER\n002 30 MAR 73\n002 30 MAR 73\n500 MB HT DIFF A-8 MODES VALID 24 HOURS AFTER\n20 20\n-26\n20\n20\n20\n20\nA\n20\n28\n-20\n500 MB HT DIFF A-B MODES VALID 48 HOURS AFTER\n002 30 MAR 73\n500 MB HT DIFF A-8 MODES VALID 36 HOURS AFTER\n00Z 30 MAR 73\nFigure 20.\nSame as figure 16, except 30 March.\n39","20\n20\n-20\n20\n20\nwas\nsee\n500 MB HT DIFF A-B MODES VALID 24 HOURS AFTER\n500 MB HT DIFF A-B MODES VALID 12 HOURS AFTER\n00Z\n2 APR 73\n002 02 APR 73\n20\n-20\n20\n20\n20\n8\n20\n20\n8\n-,\nwe\n20\n20\n20 20\n20 20\n500 MB HT DIFF A-B MODES VALID 48 HOURS AFTER\n00Z 02 APR 73\n500 MB HT DIFF A-B MODES VALID 36 HOURS AFTER\n002 2 APR 73\nSame as figure 16, , except 2 April.\nFigure 21.\n40","-20\n-20\n20\n20\n20\n20\n20 20 20 20\n500 NB HT DIFF A-B MODES VALID 12 HOURS AFTER\n00Z 04 APR 73\n500 MB HT DIFF A-8 MODES VALID 24 HOURS AFTER\n002 04 APR 73\n-20\n20\n20\na\nwas\n20\n2020\n20\n20 2020\n20\nCOMMENT\n002 04 AP\n500 MB HT DIFF A-B MODES VALID 48 HOURS AFTER\n002 04 APR 73\n500 MB HT DIFF A-B MODES VALID 36 HOURS AFTER\nSame as figure 16, 9 except 4 April.\nFigure 22.\n41","20\n20\n20\n:-\nto\nto\n,\n500 MB HT DIFF A-B MODES VALID 24 HOURS AFTER\n002 7 APR 73\n500 MB HT DIFF A-B MODES VALID 12 HOURS AFTER\n00Z 07 APR 73\n20\n20\n20\n20\n20\n-,\n20 20\n-20\n20\n20\n20\n20\n2020 20\n500 MB HT DIFF A-8 MODES VALID 48 HOURS AFTER\n500 MB HT DIFF A-B MODES VALID 36 HOURS AFTER\n.\n002 07 APR 73\n00Z 7 APR 73\nSame as figure 16, except 7 April.\nFigure 23.\n42","7\na\n20\n500 MB HT DIFF A-B MODES VALID 12 HOURS AFTER\n500 MB HT DIFF A-B MODES VALID 24 HOURS AFTER\n002 10 APR 73\n00Z 10 APR 73\n7\n9/0\n0\n:-\n20 20 20\n20\n500 MB HT DIFF A-B MODES VALID 48 HOURS AFTER\n500 MB HT DIFF A-B MODES VALID 36 HOURS AFTER\n002 10 APR 73\n002 10 APR 73\nSame as figure 16, except 10 April.\nFigure 24.\n43","(Continued from inside front cover)\nNOAA Technical Memoranda\nNWS NMC 49\nA Study of Non-Linear Computational Instability for a Two-Dimensional Model. Paul\nD.\nPolger, February 1971. (COM-71-00246)\nNWS NMC 50\nRecent Research in Numerical Methods at the National Meteorological Center. Ronald D.\nMcPherson, April 1971.\nNWS NMC 51\nUpdating Asynoptic Data for Use in Objective Analysis. Armand J. Desmarais, December\n1972. (COM-73-10078)\nNWS NMC 52\nToward Developing a Quality Control System for Rawinsonde Reports. Frederick G. Finger\nand Arthur R. Thomas, February 1973. (COM-73-10673)\nNWS NMC 53\nA Semi-Implicit Version of the Shuman-Hovermale Model. Joseph P. Gerrity, Jr., Ronald D.\nMcPherson, and Stephen Scolnik. July 1973. (COM-73-11323)\nNWS NMC 54\nStatus Report on a Semi-Implicit Version of the Shuman-Hovermale Model. Kenneth Campana,\nMarch 1974. (COM-74-11096/AS)\nNWS NMC 55\nAn Evaluation of the National Meteorological Center's Experimental Boundary Layer model.\nPaul D. Polger, December 1974. (COM-75-10267/AS)\nNWS NMC 56\nTheoretical and Experimental Comparison of Selected Time Integration Methods Applied to\nFour-Dimensional Data Assimilation. Ronald D. McPherson and Robert E. Kistler, April\n1975. (COM-75-10882/AS)"]}