Smithsonian miscellaneous collections . amount, the change which he wishes toknow in advance. I regard the results of this test of forecasting Washington tempera-ture as so promising that I have ventured to synthesize the expectedWashington temperatures from 1952 to 1959. This forecast is givenin table 8. These forecasted 8 years of Washington temperatures Ihave reduced from the status of departures from normal to actualtemperatures Fahrenheit. In making the forecast I have assumed thatthe lag between synthesis and event will be reduced to zero, and that 14 SMITHSONIAN MISCELLANEOUS COLLECTION
Smithsonian miscellaneous collections . amount, the change which he wishes toknow in advance. I regard the results of this test of forecasting Washington tempera-ture as so promising that I have ventured to synthesize the expectedWashington temperatures from 1952 to 1959. This forecast is givenin table 8. These forecasted 8 years of Washington temperatures Ihave reduced from the status of departures from normal to actualtemperatures Fahrenheit. In making the forecast I have assumed thatthe lag between synthesis and event will be reduced to zero, and that 14 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 122 the scale of mean temperatures will remain 2 degrees above synthesis,as now prevailing. The comparison of forecast is to be with WeatherBureau Records, means between averages of monthly maxima andmonthly minima, at the main Weather Bureau Office, 26th and MStreets, NW., Washington, D. C. Obviously, to check the accuracyof the forecast, the observed temperatures of future years must firstbe smoothed by 5-month running Fig. 5.—Washington temperature departures, 1950 to 1952, predicted (lightcurve) and observed (heavy curve). Correlation, ± percent. Tempera-tures, degrees Fahrenheit. All temperatures smoothed by 5-month runningmeans before used. To fix upon the probable scale difference and lag, I prepared figure 5,in which departures from normal in the synthesis are plotted from theupper zero line and the right-hand scale of ordinates. The departuresobserved are plotted from the lower zero line and the left-hand scaleof ordinates. The plot begins with 1950 and extends through lag of one to two months is seen, as stated above, in the years 1950and 1951, but seems to vanish in 1952. As for the scale, the syntheticvalues seem to run about 20 Fahrenheit below the observed values inthese three years. So I have assumed that the same scale differenceand zero lag will continue till 1959, as stated above. In view of unpredictable changes of scale and
Size: 2323px × 1075px
Photo credit: © The Reading Room / Alamy / Afripics
License: Licensed
Model Released: No
Keywords: ., bookcentury1800, bookdecade1860, bookidsmiths, booksubjectscience
