PROPOSAL FOR A MODIFICATION OF KOEPPEN'S DEFINITIONS OF THE DRY CLIMATES Harry P. Bailey University of California, Los Angeles W. Koppen's classification of climates sets forth five major climatic regions, intended to correspond to the distribution of the major plant associations described by A. de Candolle. Boundaries of the climatic regions are defined quantitatively by monthly and annual means of temperature and precipitation. Koppen used these quantities in such a way as to take account of the effectiveness of precipitation, on the general principle that rate of evaporation from the ground and of transpiration from plants increases with temperature. That is to say, more precipitation is required in warm than in cool seasons or climates, if the water requirements of plants are to be met with equal effectiveness. This relation can readily be seen from Köppen's definitions of the dry climates [1], which, when mean annual precipitation, P, is expressed in inches as a function of mean annual temperature, T, in degrees Fahrenheit, are as follows: Mean annual precipitation at boundary between Desert and steppeSteppe and humid climates Precipitation chiefly in winter (1) P = 0.22 (T — 32.0) (4) P = 0.44 (T — 32.0) Seasonal contrast negligible(2) P = 0.22 (G — 19.4) (5) P = 0.44 (T — 19.4) Precipitation chiefly in summer (3) P = 0.22 (T — 6.8) (6) P = 0.44 (G — 6.8) Koppen divided the dry climates into two grades of intensity, steppe and desert, in such a manner that, with the same seasonal distribution, precipitation at the boundary between desert and steppe is only hala that at the humid boundary of the steppe. Precipitation has a twofold relation to temperature: to mean annual temperature, and to the season in which most of the precipitation falls. Thus, along the boundaries of the dry climates, precipitation is heaviest where average annual temperature is highest and where most of the year's precipitation falls in the warm season. Both of these relation* involve the prenciple of effectiveness of precipitation stated above. Unfortunately, Koppen did not define the dry climates in such a way as to take account of changes in the seasonal concentration of precipitation along their boundaries. In their present form, therefore, the definitions of desert and steppe climates do not provide continuous boundaries of the dry climates. Each boundary is segmented, displaying offsets at the points where the precipitation regime changes [2]. The magnitude of these offsets is not inconsiderable; they amount to as much as 5.5 inches of precipitation along the boundary between the steppe and humid climates, and to half as much along the boundary between desert and steppe. Of Köppen's major climatic regions, only the dry climates show this objectionable lack of continuity. In the interest of logic, it would therefore seem desirable to modify the definitions of the dry climates so as to obtain continuous boundaries without arbitrary smoothing. Formulation of the New Definitions A start in this direction may be made by writing a general formula for the boundaries of the dry climates: (7)P= k (Ttx), in which P and T have the same meanings as in equations 1 to 6, & is a coefficient of proportionality having a value of 0.22 at the boundary between desert and steppe, and ? is a variable whose value depends on the seasonal concentration of precipitation. An 34Yearbook of the AssociationVol. 10 appropriate definition of ? based on the seasonal concentration of precipitation should complete equation 7 so that it will yield a reasonable and continuous delineation of, the dry climates. Instead of trying to establish a definition of ? by trial and error, I have chosen to evaluate the effects of seasonality of precipitation by the use of monthly data, and to apply the results thus achieved in the form of a term suitable for use in the Koppen classification. Such a procedure can be conveniently carried out by reference to Thornthwaite 's work on effectiveness of precipitation [3]. Thornthwaite found that a suitable index of "precipitation effectiveness" for a single month can be determined from the expression ( P ? 10/9 (8)/ = 115 \ ----------------- \ L T — 10 J in which I is the index of effectiveness...