ahead of the surface perturbation and will undergo little north-south   displacement,   Those   areas   with precipitation occurring will not undergo a change from one  form  to  another  since  there  is  relatively  little advection  of  warm  or  cold  air  with  a  high  zonal condition. When the upper-level wave is of large or increasing amplitude (low zonal index), it is difficult to generalize about  the  characteristics  of  the  snow  versus  rain problem  without  considering  the  surface  perturbation. Up to this point, we have discussed the snow-rain pattern in association with an active low of the classical type, The rate of precipitation accumulation here is rapid, and the transition period of freezing rain or sleet is short, usually on the order of a few hours or less. Another situation in which there is frequently a snow versus rain problem is that of a quasi-stationary front in the southern states, with a, broad west-southwest to southwest  flow  aloft,  and  a  weak  surface  low.  The precipitation area in this case tends to become elongated in   the   direction   of   the   upper-level   current.   The precipitation rate may be slow, but it occurs over a longer period. Often a broad area of sleet and freezing rain exists between belts of snow and rain, leading to a serious icing condition over an extensive region for a period  of  several  hours  or  more.  This  pattern  of precipitation  changes  either  as  an  upper  trough approaches from the west and initiates cyclogenesis on the front or as the flow aloft veers and precipitation ceases. FORECASTING TECHNIQUES AND AIDS Approaches to the snow versus rain forecasting problem  have  generally  fallen  into  three  broad categories. The first category depends on the use of observed  flow  patterns  and  parameters  to  predict  the prevalent form of precipitation for periods as much as 36 hours in advance. The second category consists of studies  relating  local  parameters  to  the  occurrence  of rain or snow at a particular station, or area. In this approach, it is assumed thermal parameters will be obtainable from prognoses. This approach tends to have its greatest accuracy for periods of 12 hours, or less, since longer periods of temperature predictions for the boundary zone between rain and snow are very difficult to  make  with  sufficient  precision.  A  third  category  used involves the use of one of the many objective techniques available.   A   number   of   stations   have   developed objective  local  techniques.  The  method  presented  here is  applicable  to  the  eastern  half  of  the  United  States. Thus, the general procedure in making a snow versus rain forecast at present is to use a synoptic method for periods up to 24 or 36 hours, and then consider the expected behavior of thermal parameters over the area to obtain more precision for periods of about 12 hours or  less. A  number  of  methods  based  on  synoptic  flow patterns applicable to the United States are described in the U.S. Department of Commerce’s publication. The Prediction of Snow vs Rain, Forecasting Guide No. 2. These methods are mostly local in application and are beyond the scope of this manual. Prognostic  charts  from  the  National  Meteorological Center  and  other  sources  should  be  used  whenever  and wherever  available,  not  only  to  determine  the occurrence  and  extent  of  precipitation,  but  for  the prediction of the applicable thermal parameters as well. Methods Employing Local Thermal Parameters The following text discusses methods of employing surface  temperature,  upper-level  temperatures,  1000-  to 700-hPa and 1000- to 500-hPa thicknesses, the height of the freezing level, and combined parameters for the prediction  of  snow  versus  rain.  All  of  these  are interdependent, and   should   be   considered simultaneously. SURFACE   TEMPERATURE.—   Surface temperature considered by itself is not an effective criterion. Its use in the snow versus rain problem has generally been used in combination with other thermal parameters. One study for the Northeastern United States found that at 35°F snow and rain occurred with equal frequency, and by using 35°F as the critical value (predict snow at 35°F and below, rain above 35°F), 85 percent  of  the  original  cases  could  be  classified, Another study based on data from stations in England suggested  a  critical  temperature  of  34.2°F,  and  found that snow rarely occurs at temperatures higher than 39°F. However, it is obvious from these studies that even though surface temperature is of some value in predicting snow versus rain, it is often inadequate. Thus,  most  forecasters  look  to  upper-level  temperatures as a further aid to the problem. UPPER-LEVEL  TEMPERATURES.—   Two studies  of  the  Northeastern  United  States  found  that temperatures at the 850-hPa level proved to be a good discriminating parameter, and that including the surface temperature  did  not  make  any  significant  contribution. The discriminating temperatures at the 850-hPa level were -2° to -4°C. Another study found that the area 4-23