108 



METEOROLOGY— FORECASTING 



8.1.3 Army Analysis and Forecasts 



The program of the x\rmy forecasters included the 

 forecasting of transmission and radar ranges ; the ap- 

 proach to this problem was empirical. The basis of the 

 program was again the analysis of the first 6 weeks' 

 data, this time inchiding the signal strength meas- 

 urements which have been described. Signal strengtlis 

 were divided into four ranges qualitatively described 

 as low, standard, high, and very high, corre.sponding 

 to M curves of the types substandard, standard, super- 

 standard, and trapping. This analysis did not con- 

 sider variations in the M curve over the path but 

 rather related signal to the prevailing type of curve. On 

 this basis, then, a transmission forecast for a 24-hour 

 period involved the forecasting of prevailing M curves 

 over the transmission path for appropriate time in- 

 tervals. The length of these time intervals was deter- 

 mined by the rapidity with which the weather factors 

 aifecting the M distribution were changing. Specific- 

 ally, a 24-hour transmission forecast involved two 

 ilf-curve forecasts plus forecasts of temperature and 

 dew point trends. These forecasts were supplemented 

 frequently with i¥-curve forecasts for times of mini- 

 mum or maximum propagation conditions. These 

 meteorological data could then be translated quali- 

 tatively into values and trends of signal strength. This 

 information was presented in the form of a graph of 

 signal strength versus time. 



^■^■* How the Forecast Is Made 



The forecast in general involves two determinations : 

 one, of the initial conditions of the air before it leaves 

 land; and two, the modifications of the air by the 

 water surface. A study of the synoptic situation and 

 the low-level circulation reveals the location of the 

 point where the air in question leaves the land. The 

 synoptic situation shows the general flow pattern ; 

 local winds from the surface to 3,000 ft indicate the 

 specific pattern over the area under consideration. 



The initial temperature and moisture distributions 

 are determined by studying the local hourly teletype 

 sequences and radio.sonde observations. The modifica- 

 tion of the air is determined by considering the relation 

 of the surface water temperature to the representa- 

 tive air temperature and dew point, the over-water 

 travel, and the rate of modification. 



Time forecasts were also made by the Army fore- 

 casters. They involved straight meteorological fore- 

 casts of the initial conditions to which were applied 

 the space forecast technique just described. 



Example. Tliis is a forecast made by the Weather 

 Bureau. The synoptic weather map on the morning 

 of July 26 indicated a rather weak flow of modified 

 continental polar air moving in an easterly direction 

 from the mainland of eastern Massachusetts out over 

 the waters of Massachusetts Bay. The temperature of 

 tliis air was potentially more than 21 C and under 

 sunshine was developing surface temperatures near 

 the shore line of more than 21 C by 0800. The fore- 

 cast was for 1000 about .5 miles southeast of Eastern 

 Point, Massachusetts. The temperature over land 

 about a half-hour before this was exjjected to be about 

 24 C, and the air flow as indicated by winds aloft was 

 such as to allow the air warmed to about this figure 

 to be out over this position within a half-hour. The 

 lapse rate over land would be approaching the dry 

 adiabatic by this time; so, as a guide, a lapse rate 

 amounting to about 3 C per 1,000 ft was projected 

 to 1,000 ft starting from 24 C at the surface. A value 

 for the sea water temperature of 17 C was predicted 

 from recent observations made in the Bay. Using past 

 experience, one then assumed a water modification vip 

 to about 300 ft, and the T curve was constructed 

 starting from the surface value of 17 C, showing a 

 sharp inversion at first and a gradual inversion until 

 it met the guiding line representing the air from the 

 land. The radio observation made at MIT about mid- 

 night, July 25 to 26, was considered to be a fairly 

 good check of the properties of the air mass involved. 

 A surface temperature of between 21 and 22 C was 

 indicated. 



In forecasting the moisture curve, a value at the 

 surface corresponding to the water temperature was 

 made the base of the curve. Over-land dew points were 

 initially predicted to be about 13.5 C, whicli would 

 give a vapor pressure value of between 15 and 16 mb 

 at the top of the water modification zone. An examina- 

 tion of the raobs, both MIT and Portland, show mix- 

 ing ratios of about 10.5 g per kg between 500 and 

 1,000 ft, which corresponds to 15 or 16 mb. This makes 

 a good check on the prevailing initial dew points. The 

 raob at Albany indicated that air which was a little 

 drier was moving in from the west so that a slight 

 decrease in the vapor pressure was forecast between 500 

 and 1,000 ft. (This part of the forecast did not prove 

 to be correct, since, as the verification of the forecast 

 in the figure shows, the moisture value remained 

 fairly uniform from 400 up to 1,000 ft.) Another 

 curve was drawn similar to the T curve to connect 

 the surface vapor pressure value with that value at 



