{"Bibliographic":{"Title":"An evaluation of the National Meteorological Center's experimental boundary layer model","Authors":"","Publication date":"1974","Publisher":""},"Administrative":{"Date created":"08-20-2023","Language":"English","Rights":"CC 0","Size":"0000038628"},"Pages":["A\nQC\nOF COMMUNITY\n851\nU6N5\nno.55\nical Memorandum NWS NMC-55\n*\n*\nc.l\nWHITE\nAustralia\nSTATES\nOF\nAN EVALUATION OF THE NATIONAL METEOROLOGICAL CENTER'S\nEXPERIMENTAL BOUNDARY LAYER MODEL\nPaul D. Polger\nDevelopment Division\nNational Meteorological Center\nSuitland, Md.\nDecember 1974\nREVOLUTION\nnoaa\nNATIONAL OCEANIC AND\nNational Weather\nATMOSPHERIC ADMINISTRATION\nService\n1776-1976","Nation\nries\nNational Weather Serv\n(NWS) produces weather anal-\nw\nthe\nantimal\nEnther\nThe National Meteorological Center\nyses and forecasts for the Northern e. Areal coverage 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\nNMC 34 Tropospheric Heating and Cooling for Selected Days and Locations over the United States\nTN\n22\nDuring Winter 1960 and Spring 1962. Philip F. Clapp and Francis J. Winninghoff, 1965.\n(PB-170-584)\n30 NMC 35 Saturation Thickness Tables for the Dry Adiabatic, Pseudo-adiahatic, and Standard Atmo-\nTN\nspheres. Jerrold A. LaRue and Russell J. Younkin, January 1966. (PB-169-382)\n37 NMC 36 Summary of Verification of Numerical Operational Tropical Cvclone Forecast Tracks for\nTN\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\nWBTM NMC 38 A Summary of the First-Guess Fields Used for Operational Analyses. J. F. McDonell, Feb-\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 N.1C 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. (PR-176-510)\n43 The Air Pollution Potential Forecast Program. John D. Stackpole, November 1967.\n(PR-176-\nWBTM\nNMC\n949)\nWBTM NMC 44 Northern Hemisphere Cloud Cover for Selected Late Fall Seasons Using TIROS Nephanalyses.\nPhilip F. Clapp, December 1968. (PB-186-392)\nWBTM NMC 45 On a Certain Type of Integration Error in Numerical Weather Prediction Models. llans\nOkland, September 1969. (PB-187-795)\nWBTM NMC 46 Noise Analysis of a Limited-Area Fine-Mesh Prediction Model. Joseph P. Gerrity, Jr., and\nRonald D. McPherson, February 1970. (PB-191-188)\nWBTN NMC 47 The National Air Pollution Potential Forecast Program. Fdward Gross, May 1970. (PB-192-\n324)\nNMC 48 Recent Studies of Computational Stability. Josenh P. Gerritv, Jr., and Ronald D. McPher-\nWBTM\nson, May 1970. (PB-192-979)\n(Continued on inside bacl: cover)","A\nQC\n851\nU6N5\nno.55\nC.\nNOAA Technical Memorandum NWS NMC-55\nAN EVALUATION OF THE NATIONAL METEOROLOGICAL CENTER'S\nEXPERIMENTAL BOUNDARY LAYER MODEL\nPaul D. Polger\nDevelopment Division\nNational Meteorological Center\nSuitland, Md.\nDecember 1974\nATMOSPHERIC SCIENCES\nLIBRARY\nJUL 11 1975\nN.O.A.A.\nU. S. Dept. of Commerce\nATMOSPHERIC\nAND\nNOAA\nNational Weather\nUNITED STATES\nNATIONAL OCEANIC AND\nAMOUNT\nService\nDEPARTMENT OF COMMERCE\nATMOSPHERIC ADMINISTRATION\nGeorge P. Cressman Director\nFrederick B. Dent, Secretary\nRobert M. White Administrator\nis COMPANYMENT OF COMMUNITY\n75\n2463","CONTENTS\nAbstract\n1\n1.\nIntroduction\n1\n2. Forecast Model and Analysis\n2\n3. Model Output\n3\n4. Evaluation\n4\nA.\nCeiling/Visibility and Rain Versus Snow\n4\nB.\nSevere Weather Indices\n5\nC. Time Cross Sections\n7\nD.\nBasic Statistics\n8\n5. Summary of the Evaluation\n9\nAcknow ledgements\n10\nReferences\n11","AN EVALUATION OF THE NATIONAL METEOROLOGICAL CENTER'S\nEXPERIMENTAL PLANETARY BOUNDARY LAYER MODEL\nPaul D. Polger\nDevelopment Division\nNational Meteorological Center\nNational Weather Service, NOAA\nABSTRACT. For almost a year, the National Meteorological\nCenter (NMC) experimented with a planetary boundary layer\nmodel (PBL) to determine its utility in NMC operations. The\nPBL model is similar to the operational boundary layer model\nused by the Air Force Global Weather Central (AFGWC). The PBL\nmodel has eight levels, surface to 1600 m, and a horizontal\ngrid mesh of 190.5 km over the contiguous United States. Fore-\ncasts were based on 0000 G.m.t. data and were run to 24 hours.\nThe parameters forecast by the PBL model include temperature,\nwinds, relative humidity, freezing levels, precipitation type,\nand severe weather indices. The results show that the NMC PBL\nmodel has potential for providing useful forecasts of ceiling/\nvisibility combination, rain vs. snow delineation, and areas\nof severe weather and temperature changes over the eastern two-\nthirds of the United States. The PBL model does not perform\nwell over the mountain region or during the summer months.\n1. INTRODUCTION\nIn September 1971, an effort was undertaken at the NMC to examine the feasi-\nbility of combining a PBL model with the NMC limited-area fine-mesh (LFM),\nprimitive equation (PE) model. In such a system, the LFM model is run first\nto provide initial and forecast values necessary to operate the PBL model.\nThe model that evolved is similar to the one that has been in operational use\nby the AFGWC since December 1969. The AFGWC boundary layer model (Hadeen\n1970) was adapted from a model for forecasting synoptic-scale low cloudiness\ndeveloped by Gerrity (1967).\nThe implementation of the PBL model at NMC required development of an\nobjective analysis code for processing surface and upper-air observations and\ntransposing the LFM model data required as boundary conditions. The actual\nforecast model was adapted from the AFGWC boundary layer forecast model and\nmodified to operate with the NMC computer system. Finally, to facilitate the\nproduction of output, the graphics were generated on microfilm. The above\nefforts have been documented by Gross et al. (1972).\nOnce the major components of the PBL model project had been completed,\nseveral test cases were carried out to obtain a preliminary evaluation of the\nmodel's performance (Gerrity et al. 1972). The results of these preliminary\ntests of the LFM-PBL model concurred, in general, with the expectations assumed\nat the onset of the study. Although some inadequacies in the analysis and","2\nprediction schemes were suggested by the results of the preliminary tests,\nfurther modification was set aside until experience could be gained from a\nmore extensive evaluation. The approach to be used in conducting this evalu-\nation was the next matter of concern.\nIn conjunction with compiling statistics on the basic forecast variables,\nit was desired that subjective and, where possible, objective evaluations\nshould be obtained by disseminating information to potential users on as near\na real-time basis as possible. To that end, the PBL model was run in an\nexperimental, semioperational framework beginning in the fall of 1972. The\nforecasts were run out to 24 hours once a day, using the 0000 G.m.t. data\ncycle. The forecasts were normally completed 9 hours after the observation\ntime, and then distributed to users outside of NMC by means of a facsimile\ntransmission. Internal distribution to the NMC Forecast Division was accom-\nplished through hard copies from the microfilm, which was processed on an\noperational basis.\n2. FORECAST MODEL AND ANALYSIS\nA detailed discussion of the PBL model, including a derivation of the model\nequations, has been given in the previously mentioned papers by Gerrity (1967)\nand Hadeen (1970). For our purposes, only a cursory examination of the model\nphysics and analysis technique will be presented in this test.\nThe PBL model forecast area is a subset of the LFM model forecast region and\nis shown by the innermost rectangle of figure 1. The horizontal area is\ndivided into a 29 X 27 grid point network with a grid interval of 190.5 km,\ntrue at 60°N. The vertical structure, figure 2, is divided into 8 levels from\nthe surface of the terrain up to 1600 m above the surface. The forecast\nvariables are the horizontal and vertical wind components, temperature,\nspecific humidity, and specific moisture. The specific moisture accounts for\nliquid water after the air becomes saturated, since the model has no mechanism\nfor precipitation.\nThe model has two regions in the vertical; a surface contact layer 50 m deep\nand a transition layer 1550 m deep. The winds within the transition layer are\ncomputed diagnostically assuming a balance of the coriolis, pressure gradient,\nand eddy viscous forces. The eddy viscosity coefficient, calculated by\napplying stability-dependent, constant-flux profile formulas within the\nsurface contact layer, is taken to be invariant with height within the tran-\nsition layer. In addition, calculations within the contact layer specify the\nheat and moisture flux at the lower boundary of the transition layer.\nThe assumption that the eddy fluxes within the contact layer are constant\nwith respect to height is the basis for the development of a similarity theory\nof the structure of the atmosphere within this layer. The results of simil-\narity theory and numerous empirical studies imply two primary turbulence\nregimes, free and forced convection, which occur in the air layer near the\nground. In the PBL model, a third regime is introduced to account for the case\nof a strong stable stratification.","3\nThe surface temperature is formulated to allow for changes principally through\nadvection and radiation. The radiation changes are calculated as a function of\nlocal time and length of day, and then scaled down by the presence of clouds\ninferred from the relative humidity forecasts of the PBL and LFM models. The\nsurface specific humidity is calculated empirically and allowed to change\nprincipally through advection.\nThe PBL model analysis program generates the initial data and boundary con-\nditions for the forecast model. The initial fields are temperature, specific\nhumidity, and a parameter designed to simulate surface moisture. Also\nincluded in the analysis code are certain fixed fields such as elevation,\nroughness length, and latitude and longitude. The boundary conditions, which\nare the horizontal wind components and the cloudiness at the top of the model,\nare derived from the LFM model forecast data.\nIn the analysis of temperature and specific humidity, the fields are built\nup from the relatively data-dense surface level to the more sparse upper data\nlevels. Upper-air data are obtained from conventional radiosondes including\nsignificant level data, while surface data include both land and ship reports.\nLapse rates are analyzed at the levels above the surface and then anchored at\nthe ground to a detailed surface analysis. Analyzing lapse rates enhances the\ncontrol of vertical stability.\n3. MODEL OUTPUT\nAfter the completion of the forecast and analysis codes, the remaining task\nwas to develop an output package. This was accomplished by using the data\navailable at the multiple levels of the PBL model and adapting the data to a\nsophisticated microfilm program, which enabled the development of output\nformats that maximized the information content.\nThe forecast parameters included the temperature, relative humidity, vertical\nvelocity, horizontal wind speed and direction, multiple freezing levels, severe\nweather indices, air pollution indices, and turbulence indices. When any of\nthe above parameters required data above the PBL model levels, the information\nwas specified from the LFM model initial and forecast values. The initial and\nforecast values generated by the LFM for use by the PBL model as boundary con-\nditions or output data comprise the only interaction between the two models.\nThe output that resulted from the above data was processed at the initial,\n12-, and 24-hour times for the entire PBL forecast region. The horizontal\ndepictions included (1) 50 m vector wind and surface temperature, (2) mean\nrelative humidity 50 to 1600 m and precipitation type, (3) mean relative\nhumidity 50 to 300 m, vertical velocity, and temperatures at 300 m, (4) mean\nrelative humidity 600 to 1600 m, vertical velocity, and temperature at 1600 m,\n(6) mixing height and total wind speed, (7) relative concentration of pollutants,\n(8) best lifted index (Fujita 1970) and a modified total-total index (Miller\n1972). In addition, a vertical depiction was developed which presented forecast\nvalues at hourly intervals out to 24 hours. The variables predicted at each\nlevel of these PBL model time cross sections are the temperature, relative\nhumidity, vector wind, and freezing levels.","4\n4. EVALUATION\nThe evaluation of the PBL model was divided into four major areas: (1)\nstudies performed by the Forecast Division (FD) of the NMC which involved\nverification of ceiling/visibility forecasts and rain VS. snow forecasts; (2)\nNSSFC study of the severe weather indices generated by the PBL model; (3) study\nby several Weather Service Forecast Offices (WSFO) of the time cross sections;\n(4) study of basic statistics of temperature and relative humidity prepared by\nthe Development Division of the NMC.\nA. Ceiling/Visibility and Rain Versus Snow\ntest program for ceiling/visibility category forecasts derived from the PBL\nA\nmodel 24-hour mean relative humidity prognoses was conducted by the FD during\nthe period February 28 to May 31, 1973. The program was designed to assess the\nusefulness of the PBL model as guidance in preparing ow-level significant\nweather forecasts. The 24-hour PBL model mean relative humidity prognoses were\nconverted into ceiling/visibility category forecasts at a 60-station FD veri-\nfication network over the United States using the following criteria:\nForecasts of category 1 (ceiling 1000 feet and visibility 3 miles)\nwere assumed at all stations in the areas where the mean relative humidity\nfor the layer 50 to 300 m was forecast to be 90 percent or more.\nForecasts of category 2 (ceiling 1000 to 5000 feet and visibility 3 miles)\nwere assumed at stations in the areas where the mean relative humidity for\nthe layer 600 to 1600 m was forecast to be 80 to 89 percent.\nForecasts of category 3 (ceiling > 5000 feet and visibility 3 miles) were\nassumed at stations outside the above areas.\nA utility score was determined using the FD verification matrix shown in\ntable 1, which grants more credit for correct forecasts in categories that\noccur less frequently. The evaluation was divided into three test periods.\nThe results, which were prepared independently of the FD forecast, are given in\ntable 2. Note that the differences between the utility scores decrease in the\nlast of the three test periods. Experience with objective forecast techniques\ntested by FD has shown that the objective methods tend to do better when the\nfrequency of category 3 is greater, which is the case as the season shifts from\nwinter to spring.\nTable 1.--Verification matrix employed for the computation of a utility score\nCategory forecast\n1\n2\n3\n1\n1.0\n.2\n0\n.7\n.1\n2\n.3\n3\n0\n.2\n.4","5\nTable 2. .--Utility scores for the PBL model VS. the FD, determined from the\nverification matrix in table 1. The maximum score possible is\ngiven by MAX POS. The scores are for three test periods in 1973\n2/28-4/5\n4/11-4/28\n5/1-5/31\nFD\n392\n256\n386\nPBL\n338\n241\n381\nMAX POS\n551\n317\n460\nConsidering the simplicity of the criteria used to delineate categories\nutilizing the PBL model output, the differences in the forecasts suggest that\nthe model has potential for supplying valuable guidance in forecasting ceiling/\nvisibility combinations.\nThe verification of the rain/snow line forecast was accomplished by comparing\nthe 24-hr PBL model forecasts to the FD forecasts and an objective forecast of\nthe conditional probability of frozen precipitation (POFP) developed by\nBocchieri and Glahn (1974). For the PBL model, a conditional precipitation\ntype forecast is calculated at each gridpoint. The criteria for determining\nthe conditional precipitation type were derived subjectively by analyzing\nsoundings taken during different types of precipitation occurrences. The\ncriteria were applied to a prognostic vertical temperature profile constructed\nby merging temperature forecasts from the 8 levels of the PBL model and the\ntropospheric levels of the LFM model. The forecasts discriminate between rain,\nsnow, freezing rain, sleet, and mixed rain/snow. For the purpose of verification,\nonly the delineation between rain (rain or mixed rain/snow) and snow (snow, sleet,\nor freezing rain) was taken into account. The forecasts were compared utilizing\nindependent evaluations of the PBL model vs. the FD and POFP forecasts. The\nintent was not to determine the skill of one forecast method over another, since\nrelatively few cases were considered, but rather (as with the ceiling/visibility\ntest program) to obtain a measure of the usefulness of the PBL model output. In\nthe comparison of the PBL model vs. the POFP for 29 cases, the PBL model fore-\ncasts were judged better for 56 percent and equal for 10 percent of the cases.\nIn comparison to the FD forecasts, the PBL model forecasts were judged better\nfor 50 percent and equal for 16 percent of the cases.\nB. Severe Weather Indices\nThe verification of the severe weather indices generated by the PBL model was\ncarried out at the NSSFC (Mogil 1974) The two indices evaluated were the Best\nLifted Index (BLI) and a Modified Total-Total (MTT).\nThe concept of the BLI was introduced by Fujita (1970) who noted that a lifted\nindex computed from a fixed level such as the surface might misrepresent the\nstability of the air mass. This results from the fact that the base of an up-\ndraft or unstable layer will vary from point to point in the lowest level of","6\nthe troposphere. The PBL model, with 8 levels in the lowest 1600 m of the\natmosphere, provides the resolution necessary to obtain the BLI from the model\nparameters. The BLI is the most unstable value of the lifted index computed\nfrom the 8 levels of the PBL model used in conjunction with the LFM model\ntropospheric data.\nThe MTT index is defined as the sum of the temperature and dewpoint in °C at\nthe 900-m level of the PBL model, minus twice the temperature at 500 mb from\nthe LFM model. The modification was to use 900-m values of temperature and\ndewpoint, rather than 850-mb values as originally employed by Miller (1972) to\ncompute the total-total. The result of the change is a shift in threshold\nvalues for severe weather.\nSevere weather forecasts were verified on a digitized radar (DR) data grid\nover much of the eastern two-thirds of the United States for grid squares\nroughly 95 km on a side. The forecasts were verified using both DR data and\nthe SELS Severe Weather Log. The results of the verifications for the 1973\nspring period are given in table 3. The verification period was composed of\n39 forecast days during which an average of 24.6 severe weather reports were\nrecorded daily. Of the 39 days, 31 were considered important severe weather\ndays during which 10 or more severe weather reports were recorded in the SELS\nlog.\nTable 3. -- -Percentage frequency (f (%)) of severe weather associated with\nthe MTT and BLI. The number of squares covered by a given index\nvalue or range of values is given by N.\nf (%)\nf (%)\nBLI\nN\nMTT\nN\n.24\n.37\n-1\n6371\n41\n2709\n-2\n2392\n1.09\n42-44\n2748\n.73\n1.07\n1.04\n-3\n3094\n45-47\n4538\n3533\n1.57\n48-50\n6541\n1.77\n-4\n-5\n3173\n1.76\n51-53\n5543\n3.23\n3.78\n-6\n2698\n54-56\n3096\n5.52\n1943\n5.04\n7.60\n-7\n57\n1622\n-8\n1326\n5.67\n9\n2267\n9.00\nThe values shown in table 3 indicate the direct relations between the values\nof the BLI and MTT, and severe weather frequency. For the BLI and MTT, 80 per-\ncent of the occurrences of severe weather were recorded for values less than\nminus 6 and greater than 50, respectively. It should be noted that both\nindices had a bias toward overforecasting the anticipated area of severe\nweather, particularly in the Gulf States.","7\nC. Time Cross Sections\nThe time cross sections were designed to depict the vertical structure of\nthe atmosphere through the 8 levels of the PBL model and indicate changes in\nthis structure with time. An example of a PBL model time cross section (TCS)\nis shown in figure 3. The time runs from right to left, zero to 24 hours.\nThe solid contours are temperature in degrees Celsius, the dashed lines are\nrelative humidity in percent, and the vector winds are in knots. The surface\ntemperature in °F is shown at the bottom of the chart, with the freezing\nlevels--including those from the LFM model troposphere forecasts, shown at the\ntop of the TCS. The elevation in feet above sea level for the particular grid\npoint location of the TCS is given at the right margin. It is important to\nnote that because of the horizontal resolution of the PBL model grid network,\ngrid point values were not interpolated to the actual location of Weather\nService Forecast cities.\nThe results for two of the WSFOs, which participated in the evaluation and\nverified the TCS over an extended period, will be discussed in this section.\nA subjective evaluation was performed for the grid point closest to New York\nCity, and an objective study was conducted for the grid point closest to Sioux\nFalls, S. Dak.\nTo evaluate the TCS at New York City, weather elements were delineated into\nseveral categories which included ceiling 5000 ft, onset or ending of ceiling\n5000 ft, onset or ending of precipitation, and temperature changes 25°F in\n12 hours. These elements were then evaluated relative to how well the observed\nelement corresponded to the PBL model predictors which included relative\nhumidity, temperature stratification, change of wind direction, trend of\nrelative humidity, change of wind speed, and freezing levels. The verification\nwas carried out for 103 cases between December 1972 and May 1973. of a total\nof 1108 responses relating the PBL model predictors to the observed weather, 42\npercent were recorded as well related, 23 percent were moderately related, and\n35 percent were poorly related. In individual categories, the relative humidity\nfactors proved to be the best related--while the wind factors were the PBL model\npredictors most poorly related to the observed weather.\nThe relative humidity forecasts from the PBL model previously discussed in\nconnection with ceiling/visibility category forecasts at NMC were also investi-\ngated at the grid point nearest Sioux Falls, S. Dak. For 100 cases during the\nperiod November 30, 1972, and April 30, 1973, the PBL mean relative humidity\nforecasts for the layer from the surface to approximately 2500 ft above the\nground level were compared to the occurrence of ceilings at or below 2500 ft\nand those below 1000 ft. The frequency of ceilings for both the 1000-ft level\nand the 2500-ft level, which includes the values of the lower level, are shown\nin figure 4. The frequency of ceilings is directly proportional to the mean\nrelative humidity of the PBL model. During the same test period, a comparison\nof the PBL model 24-hour surface temperature forecast to the observed value at\nSioux Falls resulted in a root-mean-square (rms) error of 4.11°c.\nAdditional statistics were compiled from upper air observations taken at Huron,\nS. Dak. The error and bias were determined for temperature, relative humidity,","8\nwind direction, and wind speed. For the purpose of verification, the data from\nthe Huron upper air sounding was always within 50 m of the appropriate level of\nthe TCS. The results are shown in table 4. It will be seen later that the rms\nerrors of temperature and humidity are comparable to those calculated for the\nregion over the eastern two-thirds of the United States. The freezing level at\nHuron was also evaluated and found to have an rms error of 930 ft with a positive\nbias of 514 ft.\nTable 4.--PBL time cross section for Sioux Falls, S. Dak., verified at 24\nhours against Huron, S. Dak. sounding. The rms errors are given\nby the top number, with the bias in parenthesis below.\nWind\nRelative\nWind\nhumidity\ndirection\nspeed\nTemperature\n(°C)\n(%)\n(deg)\n(kt)\n14.9\n64.3\n5.75\nSurface\n3.24\n(.61)\n(-3.88)\n(2.80)\n(-4.14)\n300 m\n3.44\n15.0\n80.3\n9.75\n(.10)\n(1.15)\n(-6.50)\n(41.5)\n10.2\n900 m\n3.76\n18.0\n87.6\n(.85)\n(-7.7)\n(27.2)\n(-.35)\n1600 m\n4.42\n17.5\n61.0\n9.43\n(2.68)\n(-8.75)\n(26.6)\n(-1.02)\nD. Basic Statistics\nThe statistics compiled on the PBL model temperature and relative humidity\nforecasts are summarized in table 5. They are based on 24-hour forecasts valid\nat 0000 G.m.t. and are verified against the 0000 G.m.t. PBL model analysis,\nwhich is independent of the forecast. The verification area was divided into\ntwo regions, which included the eastern two-thirds of the United States as one\nregion and the Western Mountain States as the other. The forecasts were veri-\nfied at the surface and 300, 600, and 1600 m for the forecast temperature\nchange, forecast relative humidity changes, and the corresponding observed\nchanges. The values shown in table 5 are mean values computed for 168 cases\nover a 10-month period from November 1972 through August 1973. The statistics\nare presented in terms of rms error, bias, and the percent difference between\nthe rms error of forecast changes and the rms of observed changes.\nThe mountain region yields consistently poor results for both forecast param-\neters as indicated by the bottom row of table 5. For the region over the\neastern two-thirds of the United States, the temperature change forecasts have\nan average error of approximately 3°c, except at 1600 m where there is a large\nbias in the forecasts. The results of the relative humidity change forecasts\nare mixed, but on the average there is an improvement over the observed rms\nchange--which is essentially the error of persistence. The bias of the observed\nchanges, not shown, is small as would be anticipated when averaged over the 10-","9\nmonth verification period. The biases of forecast values were relatively small\nexcept for the 1600-m temperature, as noted above, and for the surface relative\nhumidity.\nTable 5.--Statistics of PBL model temperature and relative humidity change\nforecasts based on 168 cases between November 1972 and August 1973.\nThe statistics given for the surface and 300, 600, and 1600 m are\ncompiled only over the eastern two-thirds of the United States.\nThe statistics for the mountain region are averaged over all the\naforementioned levels.\nTemperature\nRelative humidity\nfcst\nrmse rms\nfcst\nrmse\nrms\nfcst\nobs\nchg\npct\nfcst\nobs\npct\nchg\nchg\nchg\ndiff\nbias\nchg\nchg\ndiff\nbias\n1600 m\n3.74\n3.48\n-7.5\n2.90\n21.2\n25.9\n18.1\n-3.9\n600 m\n2.69\n3.27\n17.7\n-.18\n18.5\n18.0\n-2.8\n3.3\n300 m\n2.87\n3.16\n9.2\n.22\n15.6\n15.8\n1.3\n-2.0\nSurface\n3.04\n3.19\n4.7\n.39\n15.1\n13.8\n-9.4\n-5.4\nMtn. reg.\n4.32\n3.37\n-28.2\n.65\n20.9\n17.6\n-18.7\n.9\nThe characteristics of monthly variations in performance are presented in\nfigure 5, which shows the percent increase or decrease of the rms error of\nforecast temperature and relative humidity changes vs. the rms of the observed\nchanges. The computation which includes the 300- and 600-m levels over the\nnonmountain region indicates that the PBL model does not skillfully forecast\nchanges of temperature and relative humidity during the summer months due,\nin part, to the persistence of that season. Similar statistics were not\ncompiled for the wind components because the initial values are determined\ndiagnostically rather than being analyzed from observations.\n5. SUMMARY OF THE EVALUATION\nThe development of a planetary boundary layer forecast capability at NMC was\napproached from the viewpoint of testing an existing boundary layer model to\ndetermine its usefulness as a forecast tool within the NMC operational frame-\nwork. The NMC PBL model was adapted from the AFGWC boundary layer model, which\nevolved from a model conceived for the prediction of synoptic-scale low cloudi-\nness (Gerrity 1967). Previous evaluations by Gerrity (op cit) and Diercks\n(1970) of these boundary layer models, which are similar to the NMC PBL model,\nsuggested several problem areas with regard to model performance. However, the\nintention set forth at the onset of the feasibility study precluded extensive\ndevelopment of the basic model. Hence, only minor adjustments were introduced\nduring the coding of the PBL model for the NMC computer system. For example,\nmodification to the radiational component of the temperature change altered\na","10\nthe original formulation which yielded a minimum temperature at 3 a.m. local\nstandard time (1.s.t.) and a maximum temperature at 3 p.m. l.s.t., without\nregard to latitude or season. The modified formulation yields a minimum\ntemperature at sunrise and a maximum temperature 2 hours before sunset.\nModifications, such as the one just described, are simple in nature and did\nnot resolve basic model deficiencies.\nThe experience gained from operating the PBL model for an extended test\nperiod suggests that it should be possible to improve the forecasts without\ndeveloping a new model substantially different in character from the present\nversion. The statistics indicate an effort should be directed at the elimina-\ntion of important biases, such as those of the temperature at 1600 m and the\nsurface relative humidity. Inherent in this effort would be an improved formu-\nlation of the surface temperature and parameterization of the cloud cover. It\nshould also be possible to refine the procedure for diagnostically determining\nthe wind field which the limited evaluation has shown to possess a degree of\ninaccuracy that, in turn, adversely affects advection-dependent parameters.\nThis study of the PBL model and the earlier study by Diercks (1970) on the\nAFGWC boundary layer model show the deficiency of the model's capability to\nmake forecasts for the mountain region.\nThe results of the evaluation show that the NMC PBL model has the potential\nof providing useful forecast information and that improved forecasts are likely\nthrough improved analysis and modeling techniques. In particular, the verifi-\ncation indicates that the PBL model has skill in forecasting ceiling/visibility\ncombinations, rain vs. snow delineation, areas of severe weather, and temperature\nchanges over the eastern two-thirds of the United States. The time cross\nsection, presently limited by the lack of local detail, is a useful forecast\ntool if one considers the forecast trends rather than a special spatial or\ntemporal prognostication. The PBL model does not perform well over the mountain\nregion or during the summer months.\nACKNOWLEDGMENTS\nI wish to express gratitude to Dr. Joseph Gerrity for his help in analyzing\nthe results, and to Mrs. Mary Daigle for typing the manuscript. In addition,\nthanks are due to those in the National Weather Service whose participation in\nthe evaluations made this study possible.","11\nREFERENCES\nBocchieri, J. R., and Glahn, H. R., \"Predicting the conditional probability of\nfrozen precipitation,\" NOAA Technical Memorandum NWS TDL-51, NOAA,\nWashington, D.C., March 1974.\nDiercks, J. W., \"Preliminary verification of AFGWC boundary-layer and macro-\nscale cloud-forecasting models,\" USAF ETAC TN 71-5, Air Force Global\nWeather Central, Offutt AFB, Neb., June 1971.\nFujita, T. T., Bradbury, D. L., and Van Thullenar, C. F., \"Palm Sunday tornadoes\nof April 11, 1965,\" Monthly Weather Review, Vol. 98, No. 1, January 1970,\npp. 29-69.\nGerrity, J. P., \"A physical-numerical model for the prediction of synoptic-\nscale low cloudiness,\" Monthly Weather Review, Vol. 95, No. 5, 1967,\npp. 261-282.\nGerrity, J. P., , Gross, E. M., and McPherson, R. D., \"On the feasibility of\nintegrating a combined LFM-PBL model,\" NMC Office Note 77, NOAA, 1972.\nGross, E. M., , Jones, R., and McPherson, R. D., \"A description of the NMC\nplanetary boundary layer model,\" NMC Office Note 75, NOAA, 1972.\nHadeen, K. D. \"AFGWC boundary layer model,\" AFGWC Technical Memo 70-5,\nAir Force Global Weather Central, Offutt AFB, Neb., April 1970.\nMiller, R. C. , \"Notes of analysis and severe-storm forecasting procedures of\nthe Air Force Global Weather Central,\" AFGWC Technical Report 200 (rev.),\nChapter 3, Air Force Global Weather Central, Offutt AFB, Neb., 1972.\nMogil, M. H., \"Evaluation of severe weather and thunderstorm forecasts using\nmanually digitized radar data and the SELS severe weather log,\" Preprints,\nFifth Conference on Forecasting and Analysis, Saint Louis, Mo., March 4-7,\n1974, pp. 270-275.","S\na\n100°\nLFM rectangle). rectangle) and\nFigure\nthe\n12","Figure 2. --Vertical depiction of the PBL model. The forecast variables are the horizontal (U,V) and\nvertical (N) wind components, temperature (T), , specific humidity (2), and specific moisture (R).\nVERTICAL DEPICTION OF NMC PLANETARY BOUNDARY LAYER MODEL\n(LFM-U AND V WIND COMPONENTS AND RELATIVE HUMIDITY)\n1600 METERS\nFREE ATMOSPHERE\n50\n1200\n900\n600\n300\n150\nMODEL TERRAIN\nLFM\nU,V,W,T,Q,R\n700\n850\n500\nMB","(METERS\nABOVE\nSFC)\n900\n600\n300\n150\n50\n0\n1600\n1200\nFigure 3.--A forecast time cross section. The solid contours are temperature (°C),\nlevels including those from the LFM model tropospheric forecasts - at the top of\nThe surface temperature ( is shown at the bottom of the chart, with freezing\n30\n40\n-1\n0\n.11\n3.22\nthe dashed lines are relative humidity (%), and the vector winds are in knots.\nTEAP 43.2 44.0 44.9 45.5 45.4 44.6 43.2 41.2 38.5 35.4 31.8 28.0 24.1 22.6 23.5 24.3 25.2 265 28.0 29.4 30.8 32.3 33.8 35.3 36.8\n-.2\n-1.9\n18\n22\n44\n5,1\n5,4\n5,4\n0\n.3\n1.5\n2.3\n1.8\n-1.1\n4,1\n5.4\n5.0\n5,4\n52\n60-2\n50\n39\n19\n24\n23\n1\nEXPERIMENTAL\n0\n1.5\n2.4\n55\n38\n40\n62\n52\n2.\n5.5\n5,3\n2,1\n24\n25\n2\n2\n3\n5\n2\n1.7\n2\n2.1\n55\n37\n24\n40\n5,1\n5.2\n4\n5h\n26\n28\n3\n700\n0\n-.4\n1.6\n9\n-.2\n-10\n32\n26\n4\nBF\n29\n42\n50\n52\n66\n30\n-1\nis\n1.6\n-.2\n-.3\n28\n3\nB,F\n32\n43\n50\n5.2\n5,9\n35\n5\n-.3\n1.3\n5,3\n29\n8\nBF\n16\n39\n45\n5,1\n6,5\n6\n00Z\n1.2\n-1.1\n-.6\n-.2\n72\n29\n8\n40 -1\n8F\n46\n5,4\n64\n19\n42\n90\n9\n-1.1\n-.6\n0\n3.2\n55\n31\n10\nBF\n41\n8\n74/ 2128\n.1\n2.7\n-.5\n-1.1\n80\n33\n10\nBF\n43\n45\n48\n9\n-1.0\n48\n93\n10\n35\n11\nBF\n44\n46\n.0\n2\n-.9\n-.3\nsh\n4.8\n11\n37\n11\nBF\n44\n-.8\n5\n1.1\n48\n50\n8.1\n12\n40\n11\nBF\n45\nNYC\n1.3\n42\n-1.7\n-.5\n.2\n8\n48\n77\n13\n45\n11\nBF\n44\n46\n1.6\n1.7\n0\n.7\n1.3\n43\n55\n48\n14\n11\n45\n46\nB,F\n23\n1.3\n1.9\n2/1\nTIME X-SECTION\n62\n62\n15\n59\n1\n43\n44\n46\n40\n1.4\n1.9\n2.4\n2.3\n6\n6,4\n16\n63\n4.5\n39\n1\n42\n504\n1.6\n2.2\n2.7\n2.6\n2.0\n3.5\n5\n64\n3.9\n4.1\n4.1\n5.1\n5.8\n20\n17\n14\n5.1\n1.6\n2.3\n2.9\n2.3\n3.2\nP.O\n6/2\n5.3\n4.\n46\n48\n18\n64\n40\n4.1\n42\n5.7\n40\n1.7\n2.5\n2\n4.0\n5.\n6.\n4,3\n46\n19\n64\n40\n4.1\n60\n3.\n2)\n1.9\n2.7\n3.3\n2.2\n4.4\n5.9\n42\n65\n6,5\n42\n45\n20\n4.1\n4.1\n60\nPBL MODEL\nNO\n3.4\n242\n6.1\n15\n2.1\n6.\n49\n44\n66\n4.1\n43\n6.1\n45\n21\n3\n2.0\n3.0\n3.2\n2.5\n5.1\n6.0\n.\n2\n22\n65\n5,5\n43\n42\n46\nthe ICS.\n2\n2.4\n4.9\n6.0\nV\n1.8\n2.7\n3.0\n40\n64\n44\n57\n42\n45\n23\n48\n46\n1.6\n2.6\n2.8\n2.3\n6.1\n42\n50\n24\n63\n48\n45\n46\n57\n2\n50\n4\n5\n6\nFREEZING\nLEVEL($)\n(OF)\nSFC\nABOVE\nSL)\n(FEE)\n5490\n4178\n3194\n2210\n1226\n734\n606\n242","Ceiling VI 1000'\nCeiling VI 2500'\n60\n50\n40\n30\n20\n10\n0\n10-49\n50-59\n60-69\n70-79\n80-89\n90-100\nPBL MEAN RELATIVE HUMIDITY %\nFigure 4. -- The frequency of ceilings < 2,500 feet related to the mean rel-\native humidity from the PBL model time cross section for Sioux Falls, S.\nDak. Data are averaged for 100 cases during the period November 30, 1972,\nand April 30, 1973.\n15","Nov. Dec. Jan. Feb. Mar. Apr. May Jun. Jul. Aug.\nFigure 5. - - The characteristics of monthly variations in performance\nas shown by the percentage increase or decrease of the rms error\nof forecast temperature and relative humidity vs. the rms of the\nTemperature\nRelative Humidity\nobserved changes.\n40\n30\n20\n10\n0\n-10\n-20\n-30\n-40","(Continued from inside front cover)\nNOAA Technical Memoranda\nA Study of Non-Linear Computational Instability for a Two-Dimensional Model. Paul D.\nNWS NMC 49\nPolger, February 1971. (COM-71-00246)\nRecent Research in Numerical Methods at the National Meteorological Center. Ronald D.\nNWS NMC 50\nMcPherson, April 1971.\nUpdating Asynoptic Data for Use in Objective Analysis. Armand J. Desmarais, December\nNWS NMC 51\n1972. (COM-73-10078)\nToward Developing a Quality Control System for Rawinsonde Reports. Frederick G. Finger\nNWS NMC 52\nand Arthur R. Thomas, February 1973. (COM-73-10673)\nA Semi-Implicit Version of the Shuman-Hovermale Model. Joseph P. Gerrity, Jr., Ronald D.\nNWS NMC 53\nMcPherson, and Stephen Scolnik. July 1973. (COM-73-11323)\nStatus Report on a Semi-Implicit Version of the Shuman-Hovermale Model. Kenneth Campana,\nNWS NMC 54\nMarch 1974."]}