bounded by the 0°C isotherm at 850-hPa level and the 32°F  isotherm  at  the  surface,  when  superimposed  upon the   precipitation   area,   separated   the   snow-rain precipitation shield in a high percentages of cases. A range of -2° to -4°C at the 850-hPa level should be used along coastal areas, and also behind deep cold lows, At mountain stations a higher level would have to be used. A technique that uses temperatures at mandatory levels (surface, 1000-, 850-, 700-, and 500-hPa, etc.) is advantageous because of the availability of charts at these levels. There is, however, the occasional problem where temperature inversions are located near the 850- or 700-hPa levels, so that the temperature of one level may not be indicative of the layer above or below. This difficulty can be overcome by using thickness, which is a measure of the mean temperature of the layer. THICKNESS.— The National Weather Service has examined both the 1,000- to 700-hPa and 1,000- to 500-hPa thickness limits for the eastern half of the United States. A generalized study of 1,000- to 500-hPa thickness as a predictor of precipitation forms in the United States was made by A. J. Wagner, More complete details on this study can be found in The Prediction of Snow vs Rain, Forecasting Guide No. 2. Wagner’s  data  was  taken  from  a  study  of  40 locations in the United States for the colder months of a 2-year   period. Cases   were   limited   to   surface temperatures  between  10°F  and  50°F.  The  form  of precipitation in each case was considered as belonging in one of two categories-frozen which includes snow, sleet, granular snow, and snow crystals; and unfrozen, which includes rain, rain and snow mixed, drizzle, and freezing  rain  and  drizzle. Equal probability, or critical thickness values, were obtained from the data at each location. From this study it was clear that the critical thickness values increase with  increasing  altitude.  This  altitude  relationship  is attributable to the fact that a sizable portion of the thickness  stratum  is  nonexistent  for  high-altitude stations,  and  obviously  does  not  participate  in  the melting  process. To compensate for this, the equal probability  thickness  values  must  increase  with  station elevation.  For  higher  altitude  stations,  thickness  values between  the  850-  to  500-hPa  or  700-  to  500-hPa stratums, as appropriate, should prove to be better dated  to  precipitation  form. The Wagner equal probability chart is reproduced in figure 4-21. Wagner’s  study  also  indicates  that  the  form  of precipitation can be specified with a certainty of 75 percent at plus or minus 30 meters from the equal probability   value,   increasing   to   90   percent certainty at plus or minus 90 meters from this value. Stability  is  the  parameter  that  accounts  for  the variability of precipitation for a given thickness at a given  point.  This  fact  is  taken  into  account  in  the following reamer: if the forecast precipitation is due to a warm front that is more stable than usual, the line  separating  rain  from  frozen  precipitation  is shifted  toward  higher  thickness  values.  Over  the  Great Lakes,  where  snow  occurs  in  unstable,  or  stable conditions, the equal probability thickness is lower than that shown in figure 4-21 for snow showers, and higher than that shown in figure 4-21 for warm frontal snow. HEIGHT OF THE FREEZING LEVEL.— The height of the freezing level is one of the most critical thermal parameters in determining whether snow can reach the ground. It was pointed out earlier that theoretical  and  observational  evidence  indicates  that a freezing level averaging 1,200 feet or more above the  surface  is  usually  required  to  ensure  that  most  of the snow will melt before reaching the surface. This figure  of  1,200  feet  can  thus  be  considered  as  a critical  or  equal  probability  value  of  the  freezing level. COMBINED   THERMAL   PARAMETERS.— From  the  foregoing  discussion,  you  can  conclude that  no  one  method,  when  used  alone,  is  a  good discriminator  in  the  snow  versus  rain  forecasting problem. Therefore, you should use a combination of the  surface  temperature,  height  of  the  freezing  level, 850-hPa temperature, and the 1,000-to 700-hPa and /or 1,000- to 500-hPa thicknesses to arrive at the forecast, There  is  generally  a  high  correlation  between  the 850-hPa  temperature,  and  the  1,000-  to  700-hPa thickness and between the 700-hPa temperature and the 1,000- to 500-hPa thickness. Certainly an accurate temperature forecast for these two levels would yield an approximate  thickness  value  for  discriminating purposes. Additional Snow versus Rain Techniques The determining factor in the form of precipitation in  this  study  was  found  to  be  the  distribution  of temperature  and  moisture  between  the  surface  and  the 700-hPa  level  at  the  time  of  the  beginning  of precipitation.  The  median  level  of  850-hPa  was  studied in  conjunction  with  the  precipitation  area  and  the  32°F 4-24