Technical
Notes
Stratified Sampling for Determining
Duff
Moisture
in Mountainous
Terrain
DonaldF. Potts,KevinC. Ryan,andHansR. Zuuring
is visually stratifiedinto proportions
of 0.3, 0.6, and 0.1 for dry, mesic,and
dures, yet accurateestimatesof duff
wet, respectively.
The midslopeor mesicstratumwill
moisturearenecessary
for prescribing
a givenlevelof duff reductionthrough exhibit the greatestvariationin duff
burningtBrownet al. 1985).Duff moisture.Ridgesand drawstend to be
moisture•.anvary considerably
across moreuniformlydry and wet. Because
of the large expectedvariancein the
a given site, particularlyin steep,irmesic stratum, we recommend a minregularterrain.Adequatesamplingto
There are no publishedguidelines
on duff moisture sampling proce-
obtain duff moisture estimatescan be
imum
difficultfind time-consuming.
We
slopepositions.In both the wet and
of ten observations from mid-
40 or more observationsto obtain de-
recommended
havefouodthat sirr•ple
randomsam- dry strata,a minimumof threeobserplingof c•uffmoistureoftenrequires vations should be made. Thus, the
sampling
ratio
is 3:10:3.In largeareas
siredprecision.Stratifiedrandom (dry:mesic:wet)
samplingcarlprovidemorereliable
with multipleaspects,it maybe desir-
estimates
•th"an
simple
random
sam-
pling witti• lesstimeand effort.The
able to divide the arealnto
two or
more subunits based on dominant car-
key is to'separatethe area into nonoverlappingsitesor strata,so that the
dinal aspect. Each subunit should
then be sampledindependently.
variation
in duff
moisture
within
strata i•:1!essthan in the area as a
whole.',:i•
Duffm•6tsture
contentis closely
re-
lated to exposureand surfaceand
subsurface
drainagein steep,.t•egular
COLLECTING
DUFF
Once the three strata have been de-
fined,duffcollection
ma•,begin.The
amount and composition';of forest
floor
material
varies
within
and be-
tweensitesdependingon climate,the
type of vegetationpresent,and stand
history. The litter layer consistsof
freshly fallen needles, twigs, and
other debris.Litter is looselypacked
and hasundergoneonlyslightdecomposition.It freely exchangesmoisture
with the air and is a majorsourceof
fuel in a spreading
fire. Thefermentation layer, or upper duff, is belowthe
litter. Although darker in color and
more denselypacked than litter, organicmaterialhasnot decomposed
so
much
that
individual
needles
and
twigscannotbe identified.The humus
layer, or lower duff, is underneaththe
fermentationlayer. Humus is in an
advanced state of decomposition.
Dark brown or black in color, indi-
vidual pieces are no longer distinguishable.
Technically, duff observations
should be made at random locations
within eachstratum.In practice,however•'it is considerably
easierto estab-
terrain(Potts•etal. 1983).Iri short,
topographicposition may identify
strata
to be used
in stratified
sam-
pling. Basedon thispremise,we suggestthe followingsamplingprocedure
forestimating
duffmoisture
in mountainousterrdinto providethebestin-
formation
With minimum
time and ef-
fort.
SAMPLING
PROCEDURE
The firststepin stratifiedduff moisture samplingis to estimatethe areal
•$800
proportionsof the site occupiedby
dry, mesic, or wet strata. Convex
areas, such as ridges, are typically
dry. Concave areas, particularly
draws,are•omparatively
wet. Slopes
betweenthesetopographicpositions
generally have an intermediate, or
mesic,moistureregime. Stratification
can be done easily with the aid of a
large-scaletopographicmap of the
site, and by observationin the field.
The efficiencyof the procedureresults
from groupingareas of similar duff
moisture,not in the absoluteaccuracy
of the estimatedarealproportions.For
example,the 34-acclearcutin Figure 1
• Donald F. Pottsand Hans R. Zuuring are
associateprofessors,Schoolof Forestry,
University of Montana, Missoula, MT.
KevinC. Ryanis a research
forester,USDA
Forest Service Intermountain
Laboratory,Missoula,MT.
Fire Sciences
Fig. 1. Stratification
of a clearcutinto dry, mesic,and wet areasand exampleduff moisturesampling.Transects
with samplepoints(x) at fixedintervalsbeginningwith random
starting
points•').
(40ft contours).
WJAF1(1)1986 29
hsh a transect across each stratum
make observations
and
at fixed intervals
startingfrom a randompoint (see
Figure1). Avoid collectingduff where
disturbances such as logging or
windthrowhaverecentlyalteredduff
structure.Removefreshlyfallen, uncompactedlitter from the surface.The
entiredepth of duff maybe collected,
or the profilemay be separatedinto
upper and lower duff. The decision
whether to treat the two duff layers
separatelyor as one dependson how
the information
will
be used.
duff moisture estimates, while Brown
et al. (1985)have provisionsfor using
either lower duff moisture or total duff
moistureestimates.In practice, it is
difficultto separatethe duff into two
layerswhen it is lessthan 1-in deep.
with mineral soil
as this will lead to erroneouslylow
duff moisture estimates.
Collected duff materials
should be
immediatelysealedin air-tight containers,clearlylabeled,protectedfrom
temperatureextremes,and processed
as quicklyas possible.Moisture content may be determined gravimetrically or from electrical or chemical
properties--personal preference or
equipment availability dictate the
choice.
Once moisturecontent(percentby
weight) hasbeen determinedfor each
observation, calculate the average
moisturecontent(y-)in each stratum.
Mean
duff moisture
content
on the
site(•st)isthenestimated
by:
•st = N• •) + N2•2) + N3
where N•, N2, and N3 are the proportions of the total site in dry, mesic,
•0 WJAF
1(1)1986
content would
be:
Yst = 0.3 (54%) + 0.6 (76%)
+ 0.1 (121%) = 74%
Duff
consumptionmodels have differing
duff moisturedata requirements.For
example, Norum's (1977) and Sandberg's (1980) models require lower
Avoid contamination
and wet strata,respectively(N1 + N2
+ N•) = 1 and yl, y2, and •3 are the
averageduff moisture contentsobserved in the dry, roesic, and wet
strata,respectively.
For example,if the averagemoisture contentsobservedfor the dry,
mesic,and wet stratain Figure1 were
54, 76, and 121%respectively,the estimate of the site'saveragemoisture
INTERPRETING
AND
APPLYING
RESULTS
Before using the stratified mean
duff moisturecontentto predict duff
consumptionon a site, the withinstratum
estimates
should
be in-
spected. It is possible that burning
with
the observed
moisture
contents
may resultin unacceptable
fire effects
in one or more strata. If duff is too
wet, duff consumption may be too
low to accomplish
the objectivesof the
burn. The oppositemay occurif the
duff is too dry. In the examplegiven
above,Brownet al. (1985,equation9)
predict69,58, and 34%duff reduction
in the dry, mesic,and wet strata, respectively.The managerhas the ultimate choiceof burning for acceptable
resultsin all strata,or perhapsthe best
results in one stratum.
The recommended total of 16 obser-
vations made for the stratified random
sample in a 3:10:3 (dry:mesic:wet)
ratio shouldprovide acceptableprecisionfor mostburningsituationsin the
Northwest.
We have found
that site-
average duff moisture content can
usuallybe estimatedwith 90% confidence
and
-+20%
allowable
error
using the procedure.Managersconcernedwith the accuracyof their esti-
mates should compute confidence
levels based on their data. For further
discussion of stratified random sam-
pling, samplesizedetermination,precision, and accuracy, interested
readers
are advised
to consult
stan-
dardstatistics
texts(e.g.Snedecorand
Cochran 1967).
As stated at the beginning of this
discussion,there are no published
guidelines or procedures for duff
moisture sampling. Some sampling
techniques are more efficient than
others. Stratified sampling for estimatingduff moisturehas two distinct
advantagesover simplerandom sampling. We have found that strattfied
duff moisturesamplinggenerallyrequires half the observations for a
given level of confidenceand allowable sampling error. Second, and
perhaps more important, stratified
samplingprovidesestimatesof duff
moisture means and variances in the
differentstrata. This givesa clearer
pictureof how duff consumption
and
associated
fire effectscanbe expected
to vary acrossa site.
LITERATURE
CITED
BROWN,J. K., M. A. MARSDEN,
K. C. R¾,• and
E. D. REINHARDT.
1985. Predictingduff and
woodyfuel consumed
in the northernRocky
Mountains.USDA ForestServiceRes. Pap
INT-337.23 p.
NORUM,
R. A. 1977. Preliminaryguidelinesfor
prescribed
burningunderstandingtimberin
westernlatch/Douglas-fir
forests.USDA Forest
ServiceRes.Note INT-229.15p.
Ports, D. F., H. ZWR•C,
and M. I. I-hkU•OUSE
1983. Spatial analysisof duff moisture and
structurevariability.In Seventhconference
fire
and forest meteorology,April 25-28, Ft
Collins,CO, American
Meteorology
Society,
Boston, MA. P. 18-21.
SANDBER½,
D. V. 1980.Duff reductionby prescribedunderburningin Douglas-fir.USDA
ForestServiceRes.Pap.PNW-272.18p.
SNEDECOR,
G. W. and W. G. COCHP•N. 1967 Sta-
tisticalmethods.IowaStateUniversityPress,
Ames,IA. 593p.