84 Volume 44, Number I , February 1992

A More Rational Climatic Moisture Index*

Cort J. Willmott Johannes J. Feddema
University of Delaware University of California, Los Angeles
A climatic moisture index (I,) used extensively by C. W. ThornthLvaite and others is examined, and a modified
version of the index is proposed. Arbitrary limits [- 100 5 I, 5 100 (p - l)], combined with its dimensionless
property, make the original index difficult to interpret. O u r refined version also is dimensionless, but the limits
have been recast so that - 1 5 I, 5 1. Wet climates have positive values of I, while dry climates are negative,
and the index is symmetric about zero. T h e spatial distribution of our modified annual-average moisture index
over the world's continents i 5 mapped and described. Key Words: water balance, moisture index, global climate

Introduction moisture index (Thornthwaite 1943, 1948;

limate is largely characterized by the heat
C and moisture states of the earth's surface
and near surface (Willmott 1987) and, there-
fore, concern for their (the states') spatial and
temporal variability has commanded the atten-
tion of climatologists for centuries. Moisture
has received most of the attention, owing to
its paramount role in biological processes
(Carter and Mather 1966). Climatologists often
have used simple indices to describe the mois-
ture state of the earth's surface, and one such
index-Thornthwaite's index- is the subject of
this note.
Moisture indiccs typically are functions of
the ratio of time-averaged precipitation to
time-averaged moisture demand by the atmo-
sphere. Their value lies in their ability to char-
acterize the relative wetness or dryness of
places based solely on climatic data (Mather
1974). Perhaps the moisture-index concept can
be traced to Linsser (Thornthwaite 1943), al-
though it was Thornthwaitc who first replaced
the early surrogates of moisture demand (pan
evaporation and air temperature) with a cli-
matologically meaningful moisture demand,
termed potential evapotranspiration (Mather
1990). Thornthnaite and his colleagues' many
hydroclimatological publications during the
1930s, 1940s, and 1950s served to disseminate
widely their contributions, including the
Thornthwaite and Mather 1955).
T h e evaluation and mapping of moisture
indices continue to be educational. Delworth
and Manabe (1988), for instance, use a mois-
ture index to illustrate relationships that might
occur between climate and moisture availabil-
ity a t the earth's surface. As Thornthwaite's
moisture index is well known and typical, our
purpose within this paper is to examine several
of its properties and present a more rational,
modified version of the index.
Thornthwaite's Moisture Index
Thornthwaite (1948) specified his moisture in-
dex as
I, = 100 [S -E"0.6D ] (1)
where 1, is the moisture index, S is the mois-
ture surplus, D is the moisture deficit, and E"
is the potential evapotranspiration; that is, the
evapotranspiration that would occur if the veg-
etation experiences no water stress. His mois-
ture deficit (D) is merely (E" - E), where E is
the actual evapotranspiration rate. Thornth-
waite and Mather (1955) later dropped the 0.6
scaling coefficient. Thornthwaite's surplus can
be written
> [E+(w*-w)]
= io,
r - [E+(w*-w)], r
r 5 [E+(w*-w)]

* Thoughtful wggestiona un a preliminary draft of this note acre made by J K \lather. li. Mink, and S. X I . Robeion. and we gratefully
acknou ledge their assistance. \Ve also thanh I<. Alatwura for his help MI drafting the map A portion of this paper is based on work
supported by S.AS 4 under grant 9AGIV-1884.
Professional Geographer, 44( I ) 1992, pages 84-87 0 Copyrighr 1992 by .Association of American Geographers
Initial submission, June 1990; revision submitsion. July IYY I;final acceptance, August 1991
 A More Rational Climatic Moisture Index
where r is the precipitation rate, w* is the
available water-holding capacity of the root
zone, and w is the actual available soil mois-
ture. Units associated with S, D, E", E and r
are typically mm/month or mndyear whereas
w* and w are depths. Computational means
of obtaining E", E, and the other terms are
discussed by Willmott et al. (1985) and, there-
fore, are not presented here.
When Thornthwaite and Mather's (1955)
expression is integrated over the "average"
year, S -+max[(r - E),O] and I, becomes
I,, = 100 - - 1
L o 1 may be used to obtain estimates of I,. T h e
Equation ( 2 ) , in other words, represents the
Thornthwaite annual moisture index that is
commonly used to characterize climates (e.g.,
Thornthwaite and Mather 1955) as well as
other aspects of the environment such as veg-
etation (Mather and Yoshioka 1968).
A difficulty in applying and interpreting the
annual moisture index, as it is represented by
equations (1) and (2), is that it is dimensionless
with arbitrary limits. It also is asymmetric
about zero. Cursory examination of equation
(2) indicates that -100 5 I, 5 100 (m - 1)
or, with the scaling deleted, -1 5 I, 5 (m
-1). Without the signpost of meaningful
physical units, such a wide range precludes
the clear-cut comparison of such values as
-0.3, 14, or 3.8 X los. Relatively wet climates
are the main problem inasmuch as there is no
upper bound on the index.
A Modified Moisture Index
A dimensionless moisture index is indeed use-
ful; however, it should be bounded meaning-
fully so that relative wetness or dryness can
be ascertained easily. It additionally ought to
be svmmetric about zero so that -0.9, for
instance, indicates that moisture supply is
equivalent to one tenth the atmospheric de-
mand while +0.9 means that the supply ex-
ceeds demand by a factor of 10. In order to
incorporate these desirable properties, our
modified annual index takes the form
I, = [ (r/E")-
1 - (E"/r),
1, r < E"
r 2 E".
(3a)
85
In the rare case that r = E" = 0, we suggest
that I, be set to zero. While I, is usually
evaluated on an annual basis, evaluation in
seasonally snow-covered environs on suban-
nual time scales requires r to be replaced by
(rl + M) where rl is liquid precipitation and
M is snowmelt (Willmott et al. 1985).
It should be noted that the demand com-
ponent of I, (E") may be evaluated in several
ways in addition to Thornthwaite's method.
Jensen et al. (1990) and Rosenberg et al. (1983)
discuss several means of evaluation. When es-
timates of net radiation are available, they too
calculation is
where L is the latent heat of vaporization and
Q is the time integral of the positive occur-
rences of net radiation. Thornthwaite and his
colleagues were among the first climatologists
to establish the strong connection between Q
and E" (Mather 1974; 1990).
Under a variety of conditions, estimates of
E" or Q can be biased representations of at-
mospheric moisture demand and, therefore, I,
should be cautiously interpreted. Biases may
occur, for example, when equation (3) is eval-
uated on short time scales (subannual) or the
area of interest is subject to nontrivial heat or
moisture advection. Other biases in E" may
arise when it is estimated from an incom-
pletely specified, empirical function of
weather-station observations (Jensen et al.
1990; Willmott 1984). It also should be men-
tioned that raingage measurements of r arc
underestimates, and this bias can be significant
in seasonally snow-covered or windy regions
(Legates and Willmott 1990a).
illustration
T h e ability of our revised annual-average in-
dex (equation 3a) to characterize the relative
wetness or dryness of climates is illustrated on
a terrestrial map of I, (Fig. 1). In this example,
annual E" was obtained according to Thornth-
waite (Wilm et al. 1944), using air-temperature
 - 1 .oo -0.75 -0.50 -0.25 0.00 0.33 1 .oo 3.00 Inf.

- 1 .oo -0.75 -0.50 -0.25 0.00 0.25 0.50 0.75 1 .o

Figure 1: Moisture regions of the terrestrial world as depicted by the modified Thornthwaite moisture index (Im). Values of the new index
appear under the gray scale and, for comparative purposes, corresponding values of the original index (scaling by 100 deleted) appear atop
the gray scale. The isoline interval is 0.25 with four levels of constant gray-scale shading from - 1.00 to -0.50, -0.50 to 0.00, 0.00 to
0.50, and 0.50 to 1.00. Our index was evaluated at a 0.5" of latitude by 0.5" of longitude resolution using the monthly average precipitation
and temperature data sets described by Legates and Willmott (199Oa, 1990b) and the climatic water-balance procedure discussed by
Willmott et al. (1985).
 A More Rational Climatic Moisture Index
data compiled by Legates and Willmott
(1990b), while annual r was taken from the
raingage-corrected precipitation data set of Le-
gates and Willmott (1990a). Soil available
water-holding capacity (w") is held constant at
150 mm. Moist climates are represented by
positivc valucs of I,,,, while dry climatcs are
negative. Supply (r) exceeds demand (E") over
much of the tropics, in eastern North Amer-
ica, throughout much of Europe, in Southeast
Asia, and in several other regions. Demand
exceeds supply over the world's great deserts
as well as in adjacent semi-arid zones.
While these results are expected, the new
index is superior in that it varies only between
-1.0 and +1.0, and it is symmetric about
zero. This allows for a more rational (linear)
progression of isolines for the wet climates,
relative to the original index (Fig. 1). It also
makes possible the use of a single, constant
contour interval for both wet and dry climates.
This, in turn, facilitates cross-comparisons
among places. Within much of the forested
area of western and central Brazil ( I , = 0.5),
for instance, only about half the rainfall can
be evapotranspired back into the atmosphere
whereas, in the relatively dry Caatinga region
of eastern Brazil (I, = -0.5), only about half
the atmospheric moisture demand can be met
by the supply (Fig. 1). It should he noted that
our 0.25 isoline interval was selected solely to
describe the spatial variability of I,. Our iso-
lines, therefore, do not represent boundaries
associated with newly defined climatic types.
Summary
A modified version of Thornthwaite's annual
moisture index has been presented. O u r ver-
sion of the index is symmetric about zero and
bounded by -1.0 and 1.0; as a result, it re-
moves some of the difficulties associated with
the original index (e.g., its asymmetric limits
and variable contour intervals). T h e theoreti-
cal advantages of the new index have been
discussed and the practical advantages illus-
trated by mapping the index for the terrestrial
world. It is hoped that our refinement will be
of use to those interested in describing and
mapping the relative wetness and dryness of
climates. W
87
Literature Cited
Carter, D. B. and J. R. Mather. 1966. Climatic
classification for environmental biology. Puhlica-
tiom in Climatology 19(1):305-95.
Delworth, T.L. and S. Manabe. 1988. The influ-
ence of potential evaporation on the variabilities
of simulated soil wetness and climate. Journal of
Climate 1( 5 ) : 5 23-47.
Jensen, M. E., R. D. Burman and R. G. Allen,
eds. 1990. Evapotranspiration and Irrigation Watet-
Requirements. New York: American Society of
Civil Engineers.
Legates, D. R. and C . J. Willmott. 1990a. hlean
seasonal and spatial variability in gauge-corrected,
global precipitation. International Journal of Cli-
matology lO:ll1-27.
Legates, D. R. and C. J. Willrnott. 1990b. Mean
seasonal and spatial variability in global surface
air temperature. Theoretical and Applied Climatology
41:11-21.
Mather, J. R. 1974. Climatology: Fundamentals and
Applications. New York: McGraw-Hill.
Mather, J . R. 1990. Personal communication. De-
partment of Geography, University of Delaware,
Newark, DE.
Mather, J. K. and G. A. Yoshioka. 1968. The role
of climate in the distribution of vegetation. Annals
of the Association of American Geographers 58(1):ZY-
41.
Rosenberg, N. J , , B. L. Blad and S. B. Verma.
1983. Microclimate: The Biological Environment.
New York: John Wiley and Sons.
Thornthwaire, C. W. 1943. Problems in the classi-
fication of climates. Geographical Review 3 3(2):2 3 3-
55.
Thornthwaite, C. W. 1948.A n approach toward a
rational classification of climate. Geographical Re-
view 38(1):55-94.
Thornthwaite, C.W. and J. R. Mather. 1955.The
water balance. Publications in Climatology 8(:)I 1-
104.
Willmott, C.J. 1984. On the evaluation of model
performance in physical geography. In Spatial Sta-
tistics and Models, eds. G. I,. Gaile and C . J. \W-
mott, 443-60. Boston: D. Keidel.
Willmott, C. J. 1987. Models, Climatic. In 73e En-
cyclopedia of Climatology, eds. J . E. Oliver and K.
W. Fairbridge, 584-90. New York: Van Nostrand
Reinhold.
Willmott, C. J., C. M. Rowe and Y. Mintz. 1985.
Climatology of the terrestrial seasonal \rater
cycle. Journal of Climatology 5:589-606.
Wilm, H.G.,C . W. Thornthwaite, E. A. Colman,
N. W. Cummings, A. R. Croft, 14. T. Gisbornc,
S. T. Harding, A . H. Hendrickson, M. D. fioo-
ver, I. E. Houk, J. Kittridge, C. H. Lee, C.-G.
Rossby, T . Saville and C. A. Taylor. 1944. Re-
88 Volume 44, Number I , February 1992

port of thc Committee on Transpiration and tionships between land-surface processes and cli-
F,vaporation, 1943-44. Transactions, American Geo- mate, and the statistical analysis of large-scale cli-
physical Union 2 5 1638-93. mate fields.

J O H A N N E S FEDDEMA is Assistant Professor of

C O R T WILLhlOTT is Chair of the Ikpartment
of Geography and Professor of Geography and hta-
rine Studies at the University of Delaware. IIe also
is a member of the Universityk Center for Climatic
Research. His research interests include the rela-
Geography at the University of California, I n s An-
geles. H e was formerly a Ph.11. student at the Uni-
versity of Delaware. His research interests include
water balance climatology and climate/teleconnec-
tion processes.

On the Issues of Scale, Resolution, and Fractal Analysis in the

Mapping Sciences*
Nina Siu-Ngan Lam Dale A. Quattrochi
Louisiana Stute C'niversity hTASAScience and Technology Laboratory,
Stennis Space Center
Scale and resolution have long been key issues in geography. The rapid development of analytical cartography,
CIS, and remote sensing (the mapping sciences) in the last decade has forced the issues of scale and resolution to
be treated formally and better defined. This paper addresses the problem of scale and resolution in geographical
studies, with special reference to the mapping sciences. l h e fractal concept is introduced, and its use in identifying
the scale and resolution problem is discussed. The implications of the scale and resolution problem on studies of
global change and modeling are also explored. Key words: scale, resolution, fractals, mapping sciences.

Scale, Resolution, and Geography
he concept of scale is central to geography
T H arvey 1969; Meentemeyer 1980; Wat-
son 1978; Woodcock and Strahler 1987). It is
one of the main characteristics that portrays
geographic data and provides a unique percep-
tion of spatial attributes as they relate to form,
process, and dimension. Geographers often
deal with spatial phenomena of various scales.
For example, geomorphology encompasses
studies ranging from patterns of river net-
works, river basins, and coastline changes to
potholes, cave, and gully formation based on
international, national, regional, or local
scales. Climatologists study upper air circula-
tion around the globe as well as effects of local
climate on agricultural production and health.
Urban geography includes studies ranging
from analyzing the urban systems in an inter-
national, national, or regional context to as-
sessing the impact of facility location on local
* W e thank (;regor> Carter. Lce De C:ola, the nnon!.mous re\ icuers, and the editor for improving this paper; Clifford Duplechin and Xhrv
Lee Eggart, and the LSU graduate students fnr preparing the graphic?. and the N;\SA, John C. Stennis Space Cmter, Director's Discre-
rionar! Fund for supporting in part the de\rlopmcnt of this paper.
areas. Some geographers rely on data obtained
from satellites, yet others depend on data re-
garding pollen counts and soil particles ob-
tained through electronic microscopes. Diver-
sity within the discipline results in the necd
to address spatial problems from multiple
scales and resolutions.
This variation in scales can be regarded both
as a strength and weakness of the discipline.
hmalyzing geographical phenomena using a
range of scales offers a special view and meth-
odology that other disciplines seldom employ,
enhancing geography's strength. To the con-
trary, the massive amount of data needed for
analysis of spatial phenomena at various scales,
coupled with the possibility of applying an
inappropriate methodology, often leads to a
meaningless study. This invites criticism and
confusion from within the discipline and from
other related disciplines.
This paper does not attempt to solve issues
of scale and resolution, but rather brings to-

Gwgrdpher, +&(I) 19Y?, p a p 88-YX 0 Cop! right 1992 b! .issociation of .-\nicrican (;eogrdpherS
Pn~fess~onal
. \ p i 1 1991; final acceptancr. .J.uguyt 19')I
Initial subi ni ~ ~ i i i n,