Appendix A1 - Reference - Fuel Loading Models.pdf

United States Department of Agriculture

Forest Service

Rocky Mountain Research Station

General Technical Report

RMRS-GTR-225

May 2009

You may order additional copies of this publication by sending your mailing information in label form through one of the following media. Please specify the publication title and number.

Publishing Services Telephone (970) 498-1392

FAX (970) 498-1122

E-mail rschneider@fs.fed.us

Web site http://www.fs.fed.us/rmrs

Mailing Address Publications Distribution Rocky Mountain Research Station 240 West Prospect Road Fort Collins, CO 80526

Sikkink, Pamela G.; Lutes, Duncan C.; Keane, Robert E. 2009. Field guide for identifying fuel loading models.

Gen. Tech. Rep. RMRS-GTR-225. Fort Collins, CO: U.S. Department of Agriculture, Forest Service, Rocky Moun-tain Research Station. 33 p.

Abstract This report details a procedure for identifying fuel loading models (FLMs) in the field. FLMs are a new classification system for predicting fire effects from on-site fuels. Each FLM class represents fuel beds that have similar fuel loadings and produce similar emissions and soil surface heating when burned using computer simulations. We describe how to estimate fuel load in the field, match the load estimates to an appropriate FLM, and use the FLMs to predict the smoke or soil heating that could result from burning those loads. The FLM names can also be used as fuel descriptors in other applications, including inputs into fire models for predicting fire effects, data layers for mapping fuel conditions, and supple-ments to vegetation data for more complete environmental descriptions to use in restoration or wildlife habitat planning.

Keywords: First-order fire effects, fuel classification, fuel loading, fuel mapping, fuel classification key

The Authors

Pamela G. Sikkink is a Research Ecologist with Systems for Environmental Management, Missoula, MT (email: pgsikkink@ fs.fed.us). Her recent research focuses on comparing the effectiveness of fuel-sampling methods used in western conif-erous forests to measure fuels and on investigating how fire intensity affects surface-fuel consumption and soil character-istics. She has been involved in the integration of FIREMON and the Fire Ecology Assessment Tool, which produced a tool to store and organize inventory and monitoring data from many different vegetation sampling methods. She has also sampled grassland and shrubland ecosystems in Yellowstone National Park and in western Montana since 2002 as a volunteer, col-lecting vegetation data for select multiple-decade monitoring programs. She received a B.S. degree in biology and geology from Bemidji State University, MN. She earned M.S. degrees in geology and forestry and a Ph.D in forestry from the University of Montana, Missoula.

Duncan C. Lutes is a Fire Ecologist with the U.S. Forest Service, Rocky Mountain Research Station at the Missoula Fire Sciences Laboratory, Missoula, MT. He has been involved in the development of the First Order Fire Effects Model, the Fire and Fuels Extension to the Forest Vegetation Simulator, the FIREMON fire effects monitoring system, Fuel Calc, and the integration of FIREMON and the Fire Ecology Assessment Tool. Duncan has B.S. and M.S. degrees from the University of Montana, Missoula.

Robert E. Keane is a Research Ecologist with the U.S. Forest Service, Rocky Mountain Research Station at the Missoula Fire Sciences Laboratory, Missoula, MT. Since 1985, he has developed various ecological computer models for the Fire Ef-fects Project for research and management applications. His most recent research includes the synthesis of a First Order Fire Effects Model; construction of mechanistic ecosystem process models that integrate fire behavior and fire effects into succession simulation; restoration of whitebark pine in the Northern Rocky Mountains; spatial simulation of successional communities on the landscape using GIS and satellite imagery;

and the mapping of fuels and fire regimes for fire behavior prediction and hazard analysis. He received a B.S. degree in forest engineering from the University of Maine, Orono; an M.S. degree in forest ecology from the University of Montana, Missoula; and a Ph.D in forest ecology from the University of Idaho, Moscow.

Acknowledgments

Partial funding for this project was provided by the U.S. Forest Service Fire and Aviation Management in conjunction with the LANDFIRE Prototype Project. We thank Wendel Hann, U.S.

Forest Service FAM, for financial support; and Roger Ottmar, U.S. Forest Service Pacific Wildfire Sciences Lab, for granting use of his photo series photographs to depict fuel loads in our appendices. We thank Matt Reeves, USFS; Laura Hudson, Na-tional Park Service; and Ben Butler of the Student Conservation Association for their suggestions to improve the manuscript.

i

Contents

Introduction What is an FLM?

How Were the FLM Classes Developed?

Forested areas Non-forested areas

How Can Managers Use FLMs?

Identification of Fuel Loading Models in the Field

Identifying an FLM in Forests Identifying an FLM in Grasslands, Shrublands, or Chaparral

Management Advice

References

Appendix A—Field Form for Recording FLM Data

Appendix B—Tables for Calculating Biomass of Duff and Litter in Forested Areas

Appendix C—Fine-Woody Debris Loadings for Forested Areas

Appendix D—Photoloads of Coarse Woody Debris Loadings for Forested Areas

Appendix E—FLM Key for Forested Areas (T acre–1)

Appendix F—FLM Key for Forested Areas (kg m–2)

Appendix G—Representative Fuel Loads in Sagebrush Areas

Appendix H—Representative Fuel Loads in Non-Sagebrush Areas

Appendix I—FLM Key for Non-Forested Areas

Appendix J—Practice Set: Using the FLM Key

1USDA Forest Service Gen. Tech. Rep. RMRS-GTR-225. 2009

Introduction ____________________

Historically, fuel classifications used to estimate fire effects have been based on the vegetative characteristics of a particular site or location (Reinhardt and others 1997;

Sandberg and others 2001). Vegetation-based classifica-tions generally use cover type, structural stage, and/or habitat type (Mueller-Dombois 1964; Pfister and Arno 1980) as surrogates for describing the type and quantity of fuels on the ground and in the forest canopy (Hawkes and others 1995; Keane and others 2006; Mark and others 1995; Shasby and others 1981). The rationale for using vegetation characteristics to classify fuels is that fuels are ultimately derived from vegetation, so knowing how much fuel a particular vegetation type produces should provide an acceptable estimate of fuel load on the ground.

However, vegetation-based fuel classifications fail to recognize that (1) fuel beds, or the fuels in the surface fuel and litter/lichen/moss strata (Scott 2007), are composed of diverse fuel components (for example, a combination of downed woody debris, shrubs, and herbs as opposed to only shrubs or only litter), (2) each fuel component is highly variable in loading across space and time, (3) the fuels and the vegetation may have different disturbance histories in space and time that affect their correlation (Brown and Bevins 1986), and (4) most sampling methods are limited in their ability to capture both fuel variability and how much fuel is produced by any particular vegeta-tion type (Brown and See 1981; Lutes 1999, 2002).

Alternatives to vegetation-based classifications have been developed to classify fuels that are input into fire behavior computer models such as BEHAVE and FAR- SITE (Andrews 1986; Andrews and Bevins 1999; Finney 2004). These fuel classifications, which are also known as Fire Behavior Fuel Models (FBFMs), include only fuel bed components that are important for predicting fire behavior. They consist of a limited number of fuel beds that, in turn, have limited load values for fine fuels and live herb and woody material (Anderson 1982; Burgan 1987; Scott and Burgan 2005). They do not include fuel greater than 3 inches (7.6 cm) in diameter because this material does not substantially contribute to fire spread.

Field Guide for Identifying Fuel Loading Models

Pamela G. Sikkink, Duncan C. Lutes, and Robert E. Keane

In a sense, the fuel beds used within these FBFMs are artificial, or stylized, because individual fuel components were manually adjusted within each FBFM class to pro-duce expected fire behaviors that follow the fire spread model of Rothermel (1972). Unfortunately, classifica-tions that use artificial fuel beds, or exclude important fuel components, are inappropriate for computing fire effects like fuel consumption, smoke production, and tree mortality. For accurate simulation of these fire effects, most fire effects computer models, such as CONSUME and FOFEM (Ottmar and others 2008; Reinhardt and others 1997), require actual fuel loadings across all of the major surface fuel components.

Lutes and others (in press) recently created a new classi-fication, called Fuel Loading Models (FLMs) specifically developed to predict fire effects from on-site surface fuels.

Their FLM classification is one of the first classifications that categorize fuel beds into readily identifiable classes based on their predicted fire effects. It is unique because the FLM classes are readily identifiable in the field using on-site fuels. Over 4,000 actual fuel beds from across the United States were used to create the new classification and the individual groups within it are distinguished by two important fire effects—the amount of smoke that is produced upon combustion (specifically, the 2.5 μ particulate emissions) and the amount of soil heating.

Both of these fire effects are important indicators of the physical and chemical changes that will occur on a site when fuels are burned. Tools, such as FLMs, that aid in predicting these fire effects are critical to fire management.

Unlike the vegetation-based approaches used to clas-sify fire effects, FLMs use computer models to balance the high variability of fuel beds across a stand with the resolution needed to broadly describe unique fuel classes for the continental United States. Therefore, FLMs can be used to capture the variability of individual fuel compo-nents within a fuel bed, as well as describe differences in those fuel components across many spatial and temporal scales. FLMs are not designed to replace existing fuel classifications, such as the Fuel Characteristics Clas-sification System (Ottmar and others 2007; Sandberg and others 2001), nor are they designed to eliminate the

2 USDA Forest Service Gen. Tech. Rep. RMRS-GTR-225. 2009 need for extensive fuel inventories using planar intersect techniques (Brown 1974; Lutes and others 2006). FLMs are solely intended to be an additional tool for manag-ers to describe fuels for fire management. This report presents a quick and easy method for identifying a FLM so that its fuel information can be integrated with other applications, including computer predictions of fire effects.

What is an FLM?

Fuel Loading Models (FLMs) is a new classification system for predicting fire effects from on-site fuels. In this context, the word “model” denotes both the classi-fication itself and the specific sets of fuel loadings and fire effects that define each class within it. Fuel loadings include the quantities of duff, litter, fine-woody debris, and coarse woody debris (logs) in tons per acre (T acre–1) or kilograms per meter2 (kg m–2). Fire effects include the type and amount of surface fuels consumed, the quantity of PM2.5 emissions (smoke), and the maximum soil heating obtained during combustion at 0.8-inch (2-cm) soil depth.

Like other classification systems, such as the National Vegetation Classification System (http://biology.usgs.

gov/npsveg/nvcs.html), there is a hierarchy within the FLMs. The most basic unit, or class, is the fuel loading model. Each FLM class differs significantly from every other FLM class when its fuel load composition is com-pared statistically (p<0.05) (Lutes and others, in press).

When an FLM class is assigned to a particular location, it describes both the on-site fuels and the range of consumed fuels, particulate emissions (smoke), and maximum soil heating that may be expected from burning those fuels.

For example, FLM 71 represents distinct ranges of smoke and soil heating that result when moderate to heavy logs and light duff are consumed during computer-simulated combustion (table 1). FLM 14, however, has very dif-ferent predictions for smoke and soil heating because its main fuels consist of sagebrush (Artemisia spp.) at loadings of <12 T acre–1 (6.2 kg m–2). Because some FLM units produce emissions and soil heating effects that are similar, the classes can be grouped together at a higher hierarchical level. Groups of FLM classes that produce similar ranges of particulate emissions and soil heating are designated “Effects Groups.” The number used to name each FLM indicates (1) its Effects Group and

(2) its class within an Effect Group. For example, FLM 62 identifies the FLM as a member of Effects Group 6, but has the fuels, smoke, and soil heating characteristics of Class 2 within Effects Group 6. In general, increas-ing Effect Group numbers indicate higher particulate emissions and increasing maximum soil temperatures.

The 10 Effects Groups and their associated fuels and fire effects are described in table 1.

How Were the FLM Classes Developed?

The FLMs were developed using slightly different methods for forested and non-forested areas. In the fol-lowing sections, we provide an overview of how the clas-sification was created for forested areas, which is taken from Lutes and others (in press). We also summarize how the classification was created for non-forested areas by D. Lutes. The FLM classification for non-forested areas and its development process have not been published elsewhere. A complete description of FLM development for forested areas is provided in Lutes and others (in press).

Forested areas—We define forested areas as having greater than 10% tree cover or having a tree species name for the cover type classification within the sample data. Lutes and others (in press) developed the FLMs classification for the forested areas using the following procedures:

1. An extensive database of plot-level fuel loadings was compiled from sampled fuel beds located across the United States. Initially, data were compiled us-ing over 11,000 fuel beds, but the data used in the final classification of forested areas were ultimately reduced to 4,046 fuel beds that met selection criteria.

2. Each fuel bed was “burned” using computer-aided simulation. The First Order Fire Effects Model (FOFEM) (Reinhardt and others 1997) was used to simulate combustion and obtain predictions of fuel consumption, smoke emissions, and soil heating for each fuel bed.

3. The fire effects’ predictions were grouped into sta-tistically unique groups of fuel beds using cluster analysis. The unique groups were called Effects Groups.

4. Unique fuel beds were determined using classifica-tion tree analysis (Breiman and others 1984) with the Effects Groups as the independent variable.

These unique fuel beds of duff, litter, fine woody debris, and logs became the FLM classes.

5. The classification error was determined and vali-dated using two different methods.

6. A key to the FLMs was created using the classifica-tion results.

3USDA Forest Service Gen. Tech. Rep. RMRS-GTR-225. 2009

Ta b le

T he E ffe ct s G ro up s an d th ei r as so ci at ed F ue l L oa di ng M od el

F

LM

cl as se s.

E ff ec ts g ro u p n u m b er

E ff ec ts g ro u p c h ar ac te ri st ic s a

F u el lo ad c h ar ac te ri st ic s o f

F L

M cl as se s w it h in t h e E ff ec ts G ro u p

(f o re st ed s it es o n ly

A b b re vi at ed d es cr ip ti o n o f th e

E ff ec ts

G ro u p in b o ld b , c an d it s as so ci at ed F

L M c la ss es

Lo w P

M 2.

pa rt ic ul at e em is si on s

0.

T a cr e–1

°F s ur fa ce s oi l t em pe ra tu re

Li gh t t o no d uf f o r lit te r

2.

T ac re –1

F in e w oo dy d eb ris

. 2

T a cr e–1

W is p y-

C o o l-

S p ar se

F

LM

1:

L ig ht F

W D

, l ig ht to n o du ff d, e

F

LM

2:

M od er at e F

W D

, l ig ht li tte r

F

LM

3:

M od er at e F

W D

, l ig ht to m od er at e lit te r, lig ht d uf f F

LM

4:

S hr ub

-S ag eb ru sh w ith lo w to ta l l oa d

F

LM

5:

S hr ub -N on -s ag eb ru sh w ith lo w to ta l l oa d

Lo w P

M 2.

pa rt ic ul at e em is si on s

0.

T a cr e–1 to

F s ur fa ce s oi l t em pe ra tu re

2.

to

.4

T a cr e–1 du ff

.9

T a cr e–1 lo gs

W is p y-

W ar m -L ig h t F

LM

1:

L ig ht lo gs

, l ig ht d uf f

Lo w P

M 2.

pa rt ic le e m is si on s

0.

T a cr e–1 to

F s ur fa ce s oi l t em pe ra tu re

8.

T a cr e–1 ea ch o f d uf f a nd li tte r

8.

T a cr e–1 lo gs

W is p y-

H o t- M o d er at e F

LM

1:

M od er at e lit te r, lig ht d uf f, lig ht lo gs

Lo w P

M 2.

pa rt ic ul at e em is si on s(

0.

T a cr e–1 to

°F s ur fa ce s oi l te m pe ra tu re

Li gh t t o no d uf f, bu t l itt er a nd fi ne w oo dy d eb ris

2.

T a cr e–1 pr es en t

W is p y-

V er y H o t-

L ig h t

F

LM

1:

M od er at e F

W D

, l ig ht to m od er at e lit te r

Lo w P

M 2.

pa rt ic ul at e em is si on s

(0 .2 to

.4

T a cr e–1

<3

F s ur fa ce s oi l t em pe ra tu re

H ea vy c on ce nt ra tio n of d uf f (

.4 to

7.

T a cr e–1

4.

T a cr e–1 lo gs

W is p y-

C o o l-

M o d er at e

F

LM

1:

M od er at e du ff, li gh t l og s f

F

LM

3:

S hr ub

S ag eb ru sh w ith m ed iu m to ta l l oa d

F

LM

4:

S hr ub

N on

-s ag eb ru sh w ith m ed iu m to ta l l oa d

P

M 2.

pa rt ic ul at e em is si on s

0.

.8

T a cr e–1

<3

F s ur fa ce s oi l t em pe ra tu re

D uf f h ig hl y va ria bl e

Lo gs h ig hl y va ria bl e

H az y- W ar m

-M o d er at e to H ea vy

F

LM

1:

M od er at e to v er y he av y lo gs

, l ig ht d uf f

F

LM

2:

M od er at e du ff, li gh t l og s, li gh t l itt er f

F

LM

3:

M od er at e du ff, li gh t t o he av y lo gs

, l ig ht li tte r

F

LM

4:

M od er at e to h ea vy d uf f, lig ht to h ea vy lo gs

F

LM

5:

S hr ub -S ag eb ru sh w ith h ig h to ta l l oa d

F

LM

6:

S hr ub -N on -s ag eb ru sh w ith h ig h to ta l l oa d

Lo w P

M 2.

pa rt ic ul at e em is si on s

(0

.2 to

.5

T a cr e–1 to

F s ur fa ce s oi l t em pe ra tu re

4.

to

2.

T a cr e–1 du ff;

li tte r pr es en t Lo gs h ig hl y va ria bl e

W is p y to H az y-

H o t- M o d er at e F

LM

1:

M od er at e to h ea vy lo gs

, l ig ht d uf f

F

LM

2:

M od er at e du ff, li gh t t o m od er at e lo gs

, m od er at e lit te r

(c on tin ue d)

4 USDA Forest Service Gen. Tech. Rep. RMRS-GTR-225. 2009

P

M 2.

pa rt ic ul at e em is si on s

.8

T a cr e–1 to

F s ur fa ce s oi l t em pe ra tu re

U p to

6.

T a cr e–1 du ff;

lit te r va ria bl e Lo gs

.4 to

5.

T a cr e–1

H az y- W ar m t o

H o t- H ea vy

F

LM

1:

V er y he av y lo gs

, l ig ht d uf f

F

LM

2:

M od er at e du ff, li gh t t o he av y lo gs

, m od er at e lit te r

F

LM

3:

H ea vy to v er y he av y lo gs

, m od er at e du ff

P

M 2.

pa rt ic ul at e em is si on s

.2

T a cr e–1

°F s ur fa ce s oi l t em pe ra tu re

D uf f, lit te r, lo gs a nd

F W

D b io m as s va ria bl e bu t r el at iv el y he av y bi o-m as s fo r ea ch

S m o ke y- C o o l t o

H o t- V er y

H ea vy F

LM

1:

H ea vy d uf f, lig ht to h ea vy lo gs F

LM

2:

V er y he av y lo gs F

LM

3:

H ea vy d uf f, he av y to v er y he av y lo gs

P

M 2.

pa rt ic ul at e em is si on s

2.

T a cr e–1

M os t <

F s oi l t em p

V er y he av y du ff

O R

Li gh t d uf f b ut h ea vy li tte r an d lo gs

8.

T a cr e–1

S m o ke y- C o o l- V er y

H ea vy F

LM

: V er y he av y du ff F

LM

: H ea vy li tte r, lig ht d uf f, lig ht lo gs a To c on ve rt

E ffe ct s

G ro up s ch ar ac te ris tic s to m et ric

F or m et ric e m is si on s:

M ul tip ly

T a cr e–1 b y

0.

to g et

M g km –2 m et ric te m pe ra tu re

: M ul tip ly

°F

2) b y (5

/9 to g et

C m et ric fu el lo ad : M ul tip ly T a cr e–1 b y 0.

to g et k g m –2 b E ffe ct s G ro up d es cr ip tio ns a re a rr an ge d as

E m is si o n s- S o il

S u rf ac e Te m p er at u re

-T o ta l

F u el L o ad s (e

.g

S m ok ey

-C oo l-V er y H ea vy

W ith in e ac h

E ffe ct s

G ro up in di vi du al F

LM

c la ss es a re d is tin gu is he d by d iff er en ce s in p ro po rt io ns o f d uf f, lit te r, fin e w oo dy d eb ris , a nd lo gs c R an ge s fo r th e ab br ev ia te d E ffe ct s G ro up d es cr ip tio ns

P

2.

E m is si o n s (s m o ke p ar ti cl es

2.

m µ i n d ia m et er

W is p y

0.

T a cr e

–1

M g km –2

H az y

0.

t o

0.

T a cr e–1 t o

M g km

S m o ke y

0.

T a cr e–1

M g km

Te m p er at u re

C o o l <

°F

°C

W ar m to

F to

C

H o t to

F to

C

V er y H o t

°F

°C

To ta l

F u el L o ad s

S p ar se

V er y lit tle d uf f or l itt er us ua lly l es s th an

.7

T a cr e–1

0.

k g m

–2 of t ot al f ue l lo ad

L ig h t

M os tly d uf f;

to ta l lo ad i s le ss t ha n

T a cr e–1

3.

kg m

–2 ev en i f co ar se w oo dy d eb ris i s pr es en t;

M o d er at e D uf f, lit te r an d co ar se w oo dy d eb ris p re se nt to ta l lo ad r ar el y ex ce ed s

T a cr e–1

6.

kg m

–2

H ea vy A bu nd an t b io m as s bo th in d uf f a nd lo gs ; t ot al lo ad c an v ar y be tw ee n

T a cr e–1 a nd

T a cr e–1

7.

to

3.

kg m –2

V er y

H ea vy A bu nd an t d uf f a nd v er y ab un da nt b io m as s fo r lo gs ; t ot al lo ad u su al ly e xc ee ds

T a cr e–1 k g m –2 d

R an ge s fo r fu el lo ad c om po ne nt s in a bb re vi at ed fu el lo ad d es cr ip tio ns o f F

LM

c la ss es fo re st ed s ite s on ly

D ff

L ig h t

.9 T a cr e–1

2.

kg m –2

M o d er at e

.9 t o

.6

T a cr e–1

2.

to

.9 k g m

H ea vy

5.

to

T a cr e–1

8.

to

3.

kg m

V er y h ea vy

T a cr e–1 k g m it te r:

L ig h t

≤2 .7

T a cr e–1

≤0

.6 k g m

–2

; M o d er at e

2.

to

.9

T a cr e–1

0.

to

.0 k g m –2

; H ea vy

8.

T a cr e–1

2.

kg m in e w o o d y d eb ri s

(F W

D

L ig h t

2.

T a cr e–1

0.

kg m -2

M o d er at e ≥2

.2 T a cr e–1

≥0 .5 k g m g s:

L ig h t <

8.

T a cr e–1 kg m

–2

M o d er at e

.9 to

2.

T a cr e–1

(2 .0 to

.9 k g m

–2

H ea vy

3.

to

6.

T a cr e–1

(3 .0 to

.0 k g m

–2

V er y h ea vy

.8 T a kg m

N te

R an ge s fo r th e fu el c om po ne nt s de sc rib ed a bo ve a re b as ed o n th e F

LM

d ev el op m en t r ul es o f L ut es a nd o th er s

(in p re ss

S om e ad ju st m en ts to th os e ru le s ha ve b ee n m ad e so th at th ey c or re sp on d to b re ak s in th ei r di ch ot om ou s ke y an d th e F

LM

k ey d ev el op ed fo r th is p ap er e U nl es s ot he rw is e no te d, d es cr ip tio ns fo r

F

LM

s ar e fo r fo re st ed s ite s.

f F

LM

c la ss es an d h av e id en tic al d es cr ip tio ns a cc or di ng to th e cr ite ria li st ed a bo ve (t ha t i s, M od er at e du ff, li gh t l og s, li gh t l itt er

F

LM

ha s ap pr ox im at el y

2.

T a

(0

.5 k g m

–2 m or e du ff lo ad th an F

LM

2, w hi ch p ro du ce s un iq ue fi re e ffe ct s du rin g co m pu te r si m ul at io ns b ut d oe s no t p ro du ce s ho rt , u ni qu e fu el d es cr ip tio ns u si ng th e cr ite ria a bo ve

. W e el im in at ed th e lig ht li tte r de sc rip to r fr om F

LM

to e ns ur e al l F

LM

d es cr ip tio ns w er e un iq ue

E ff ec ts g ro u p n u m b er

E ff ec ts g ro u p c h ar ac te ri st ic s a

F u el lo ad c h ar ac te ri st ic s o f

F cl as se s w it h in t h e E ff ec ts G ro p

(f re st ed s it es o n ly

A b b re vi at ed d es cr ip ti o n o f th e

E ff ec ts

G ro u p in b o ld b , c an d it s as so ci at ed F

L M c la ss es

Ta b le

C on tin ue d)

5USDA Forest Service Gen. Tech. Rep. RMRS-GTR-225. 2009

Data were selected for statistical analysis by Lutes and others (in press) based on the completeness of the data for surface fuel components and their spatial distribution across the United States. Each sample had to include load estimates of fine-woody fuels (i.e., the 1 hr, 10 hr, and 100 hr fuel-moisture classes), coarse-woody fuels (logs >3 inches or 7.62 cm in diameter), and depths of the litter and duff. If any one of these six components was missing, the sample was eliminated from further consideration for the FLM study. Plot-level data quality was maintained by using only datasets that (1) were collected on plots not greater than 0.25 acres (0.1 ha); (2) did not include subjective assessments of loading; and (3) included all six fuel components needed for the FOFEM simulations (Lutes and others, in press). All of the selected datasets contained estimates of downed woody debris loads derived from using the planar intersect method (Brown 1971;

van Wagner 1968; Warren and Olsen 1964). Duff and litter load was (a) estimated by averaging multiple depth measurements and multiplying by a predetermined bulk density or (b) calculated from dried samples. To insure that the classification was pertinent to many regions of the United States, the data were also selected based on regional distribution. Most data came from recent research and large inventory or monitoring projects conducted by the Department of Defense, Bureau of Land Management, Bureau of Indian Affairs, U.S. Forest Service, and the Student Conservation Association. Their projects were spread throughout the contiguous United States.

FOFEM was used to simulate fuel bed burning. FO- FEM provided outputs for many fire effects; however, only two outputs were used to create the FLMs. These included 1) the smoke estimates measured in lb acre–1 or kg m–2 for the 2.5 mμ particulate (PM2.5) emissions and 2) the maximum temperature measured at the soil surface. These two estimates represented important ef-fects resulting from a real burn. They were also poorly correlated, which made them good variables to include in the cluster analysis used to develop the Effects Groups.

Plots were grouped by cluster analysis based on the particulate and soil heating effects using agglomerative hierarchical clustering (Lance and Williams 1967). In this type of clustering, a plot is located in two-dimensional space based on the soil heating (x-axis) and emissions (y-axis). Each plot starts out as its own group; however, during an iterative process, plots are added to groups or groups are recombined until all of the plots are members of one cluster. At each iteration of the clustering process, there will be between 1 and n (for FLMs, n=4,046) clusters, and the plots are grouped in a way that minimizes the increases in the overall sum of the squared within-cluster differences. In the FLM study, the final number of clusters was set at 10 because it was found, through a number of exploratory analyses, that classification rules applied during the FLMs process could not uniquely identify differences between clusters when more than 10 clusters were used. The cluster analysis, and a complementary classification tree procedure used to verify the groupings, produced 21 forest-type FLMs in 10 Effects Groups. The accuracy of the FLM key developed from this process was tested with cross-validation and contingency table analyses, which estimated the misclassification error as 34% and <30% respectively (Lutes and others, in press).

Each FLM had a range of loading values for each fuel component that, when consumed, produced a respective smoke and heating effect. The median values for each of these fuel components are summarized for forest FLMs in table 2.

Non-forested areas—Non-forested areas have <10% tree cover and fuels that originate primarily from grass, herbs, or shrubs. The main difference between the process used to create the FLM classification for forested areas and the process used for the non-forested areas concerned how fuel loading was used. In forested areas, loadings of the six individual fuel components were entered directly into FOFEM and emissions and soil heating were calculated automatically. In shrub and grassland areas, fuel data was often collected as TOTAL biomass without distinction as to how much fuel was in each of the six fuel components.

Therefore, fire effects had to be estimated using the total biomass for these areas. To use total biomass, a correla-tion had to be established between total fuel load and the amount of emissions that might result from burning that load so that fire effects were comparable for both forest and non-forest areas. The correlations between load and emissions were established using published emission factors for sagebrush and chaparral (DeBano and others 2005; Fahnestock and Agee 1983; Frandsen 1987; Ottmar and others 1996; Sandberg and others 2002; Taylor and Sherman 1996).

Several assumptions were made to assign maximum soil temperatures as a fire effect in non-forested areas. In general, the heat pulse was considered short in these ecosystems and burn severities were considered minimal (Molina and Llinares 2001; Ryan 2002), thus maximum soil tempera-tures were also assumed to be low. Soil-temperatures in the non-forested areas were considered to be equivalent to, or less than, the lowest temperatures obtained by burning the forest fuel beds, which were obtained in Effects Groups 1, 5, and 6. None of the temperatures for these Effects Groups exceeded 400 °F (200 °C).

6 USDA Forest Service Gen. Tech. Rep. RMRS-GTR-225. 2009

Six FLM classes were created for non-forest areas using the following procedure:

1. We used the PM2.5 emissions at the upper and lower boundaries of each Effects Group (from the forest classification) as upper and lower emission limits in the non-forest classification.

2. We calculated the amount of total plot fuel load needed to produce the emission values at the upper and lower boundaries of the selected Effects Group using published emission factors. The formula used for calculating total fuel load from emissions at the Effects Group boundaries was:

, where

L is the total plot fuel load in T acre–1 (kg m–2), Epm2.5 is the PM2.5 emissions in lb ac–1 (MG km–2), and EF is either (a) the PM2.5 emissions factor for sagebrush (26.7 lb ton–1 or 13.35 kg per metric megagram [MG–1]) OR (b) the PM2.5 emissions factor for chaparral or herbaceous (17.3 lb ton–1 or 8.65 kg MG–1), depending on whether you are calculating sagebrush load or non-sagebrush and using English or metric measures.

(Note: The emissions factors were taken from the Smoke Management Guide [National Wild-fire Coordinating Group 2001]. The equation assumes 90 percent consumption of the shrub and herbaceous fuel beds. Fuel consumption in herb and shrub systems is highly variable but 90 percent consumption was used to limit complexity and to represent a typical “worst case” scenario for emissions production. Ac-cording to Green [1970], even heavily loaded fuel beds such as chaparral can approach consumption levels of 90%. Emissions factors for grassland dominated fuels are similar or slightly lower than in shrub dominated systems [National Wildfire Coordinating Group 1985] so we estimated grassland emissions using the chaparral emissions factors).

3. We selected all non-forested plots from the data set. Most of these plots were from the grasslands and shrublands of the western United States.

4. We compared the total fuel load of each individual plot to the upper and lower fuel load limits that were calculated for each Effects Group in step 2.

Table 2—Median loadings for each forested FLM by fuel component in tons per acre (T acre–1) and kilograms per meter squared (kg m–2).

Litter Duff 1-hour 10-hour 100-hour Logs Effects T T T T T T FLM group acre–1 kg m–2 acre–1 kg m–2 acre–1 kg m–2 acre–1 kg m–2 acre–1 kg m–2 acre–1 kg m–2

011 01 0.18 0.04 0.00 0.00 0.07 0.01 0.08 0.02 0.04 0.01 0.00 0.00 012 01 0.27 0.06 0.00 0.00 0.27 0.06 1.56 0.35 2.68 0.60 2.59 0.58 013 01 2.50 0.56 1.20 0.27 0.23 0.05 1.52 0.34 2.05 0.46 2.23 0.50 021 02 1.16 0.26 3.30 0.74 0.20 0.04 0.64 0.14 0.67 0.15 0.94 0.21 031 03 1.87 0.42 7.31 1.64 0.27 0.06 0.89 0.20 1.07 0.24 1.52 0.34 041 04 2.41 0.54 0.00 0.00 0.27 0.06 1.67 0.37 2.57 0.58 2.59 0.58 051 05 1.52 0.34 15.83 3.55 0.16 0.04 1.10 0.25 1.42 0.32 1.43 0.32 061 06 0.89 0.20 3.61 0.81 0.29 0.06 1.31 0.29 2.46 0.55 16.73 3.75 062 06 1.34 0.30 20.56 4.61 0.14 0.03 0.93 0.21 1.25 0.28 1.61 0.36 063 06 1.52 0.34 17.04 3.82 0.20 0.04 1.39 0.31 2.38 0.53 7.76 1.74 064 06 2.90 0.65 25.87 5.80 0.31 0.07 1.11 0.25 1.66 0.37 3.35 0.75 071 07 2.19 0.49 9.37 2.10 0.45 0.10 1.42 0.32 2.19 0.49 11.51 2.58 072 07 3.79 0.85 16.77 3.76 0.38 0.09 1.01 0.23 1.33 0.30 2.85 0.64 081 08 0.89 0.20 3.97 0.89 0.33 0.07 1.16 0.26 2.65 0.60 36.35 8.15 082 08 3.79 0.85 17.71 3.97 0.53 0.12 1.58 0.35 3.16 0.71 12.04 2.70 083 08 2.50 0.56 11.91 2.67 0.50 0.11 1.58 0.36 2.87 0.64 22.43 5.03 091 09 1.16 0.26 45.94 10.30 0.31 0.07 1.42 0.32 1.66 0.37 2.90 0.65 092 09 3.03 0.68 14.85 3.33 0.55 0.12 1.60 0.36 3.32 0.74 46.12 10.34 093 09 6.20 1.39 32.83 7.36 0.57 0.13 1.40 0.31 2.40 0.54 21.50 4.82 101 10 9.99 2.24 99.99 22.42 0.39 0.09 0.94 0.21 1.08 0.24 1.61 0.36 102 10 17.97 4.03 10.48 2.35 0.14 0.03 0.49 0.11 1.05 0.24 4.59 1.03

7USDA Forest Service Gen. Tech. Rep. RMRS-GTR-225. 2009

We then assigned plots to the Effects Group with the appropriate range of total load and calculated the median emissions value for each Effects Group.

5. We calculated the total on-site fuel load required to obtain those emissions for each of the three Ef-fects Groups using the median emissions value and the equation above (see table 3). We established the range of loadings for each Effects Group by plotting each Effects Group’s loading distribution (Appendix I).

The non-forest FLMs should not be used for tall (>6 ft) shrub communities (for example, California’s tall chaparral communities). Tall shrub communities have more biomass than the shrub communities analyzed for this classification, which results in higher fuel loads.

They also have a more open structure than short shrub communities, which affects fuel bulk density. With the total fuel load higher, vegetation more volatile, and bulk densities structurally optimal for fuel consumption, the fuels in these systems can burn more completely and at higher intensities. Intense burning can lead to sub-stantially higher emissions and soil heating compared to the minimal-fuel, low-severity fires typical of the non-forested areas classified for this study. Developing FLMs that are appropriate for these shrub communities will require further research.

How Can Managers Use FLMs?

Managers can use FLMs to quickly estimate the fuel loads of six fuel components while in the field. FLMs can be an economical alternative for fuels sampling because sampling can be done quickly or without visit-ing an area. FLMs can also be easily integrated with other types of plot-level data, such as stand structure or vegetation cover, to create a more comprehensive description of a plot’s characteristics with little additional field sampling. Because FLMs can be consistently and accurately identified in the field, they can also be used as map units of fuel loadings. The map units can be used to quantify fire effects and plan, prioritize, and implement fuel treatments. Map accuracy can be easily assessed because the map units can be checked in the field using the FLM identification key presented in this report. The FLM classification also allows users to quickly enter fuel loading data into simulation models to compute fire effects. For example, fuel descriptions can be input into FOFEM using a FLM number instead of inputting detailed information on six separate fuel components.

Many ecosystems were not represented in the FLM fuel bed data so we do not recommend that this classification be applied in some rare ecosystems, such as pocosin bogs, boreal forests, deserts, and some hardwood forest types.

More targeted sampling and additional analysis will be required to extend this FLM classification system to these special systems.

Identification of Fuel Loading Models in the Field ______________

FLMs are identified using the tools provided in this field guide, including (1) a field form that outlines which fuel components must be sampled and which FLM key to use (Appendix A); (2) FLM keys to forested and non-forested vegetation types (Appendices E, F, and I); and

(3) photographic examples of threshold load values that are specified in the FLM keys (Appendices C, D, G, and H). The field form has space to record all the fuel load information collected in the field and to assign the FLM that summarizes the plot’s fuel beds.

During the identification of FLMs, users must make some coarse measurements or visual estimates of fuel load components within their sample area. For users who lack experience visually estimating fuel load, we provide photographs of known fuel loads in this field guide to compare with plot conditions (Appendices C, D, G, and H). Users only need to decide whether the

Table 3—Median of plot loadings for non-forested FLMs.

Median plot load

FLM Effects group System T acre–1 kg m–2

014 01 Sagebrush 0.75 0.17

053 05 Sagebrush 21.0 4.70

065 06 Sagebrush 40.9 9.20

015 01 Chaparral and herbaceous 1.15 0.26

054 05 Chaparral and herbaceous 32.5 7.30

066 06 Chaparral and herbaceous 63.2 14.2

8 USDA Forest Service Gen. Tech. Rep. RMRS-GTR-225. 2009 load for a specific fuel component on their plot appears greater or less than the load shown in the photographs.

We also provide three problem sets to practice identify-ing FLMs from known fuel loads (Appendix J). Each example in the problem set has a photograph of on-site fuels, gives the estimated on-site fuel components, and provides step-by-step instructions to key the data to a specific FLM class.

To key FLMs on your plot, first determine if you are in a forested or non-forested area, then follow the process detailed in the appropriate section below or the steps outlined in the plot form presented in Appendix A. To be considered a forested area, canopy cover of trees must be greater than 10% or the habitat type should key to a forest type. Non-forested areas can consist of sagebrush canopy cover, which have at least 10% canopy cover of Artemisia spp., or non-sagebrush canopy cover, whose fuels are derived from other shrubs or grasses that are <6 ft (2 m) tall.

Identifying an FLM in Forests ______

Step 1: Estimate duff and litter depths. Do this at several points within the sampling unit or plot and take an average of your measurements. Methods for defining and measuring duff and litter can be found in Lutes and others (2006). Record the aver-age depth in the appropriate place on the plot form (Appendix A).

Step 2: Pick an appropriate duff and litter bulk density value. Consult the scientific literature or lo-cal experts before you go into the field so that you choose the most appropriate bulk density value for your area. Typical bulk density values are provided in Appendix B, but they may not be realistic for all parts of the United States. You can interpolate between these bulk density values to get values that are more appropriate for your area.

Step 3: Calculate the duff and litter biomass. Bio-mass is calculated as thickness (depth) multiplied by bulk density (T acre–1 in–1 or g cm–3). Examples of duff and litter biomasses that have been calculated using common bulk densities are provided in Ap-pendix B. Record the biomasses in the appropriate places in Appendix A.

Step 4: Estimate the fine-woody debris loading.

Use Appendix C to determine if the fine-woody debris load is less than 2.4 T acre–1 (0.5 kg m–2). Record the load as greater or less than 2.4 T acre–1 (0.5 kg m–2) on the field form. If you need to record more precise estimates of the fine-woody component for project ob-jectives, measure the 1 hr, 10 hr, and 100 hr fuels using standard fuel sampling procedures or estimate loads using the methods described in Keane and Dickinson (2007). Record values on the plot form in Appendix A.

Step 5: Estimate log loadings (1000 hr fuels).

Precision is important for this fuel component, so we have provided photographs of known log loads (photoloads) to help users estimate this component more accurately in the field. Use the photoloads (Ap-pendix D) to decide whether you have the minimum fuel loads required in the key. Record the value on the plot form in Appendix A.

Step 6: Key to the appropriate FLM. Using the values that you recorded on the plot form, follow the directions through the FLM key for forested areas.

Use Appendix E if you have measured the loads in tons per acre and Appendix F if you have measured loads in kilograms per meter squared. Stop at the FLM that best fits your estimates. Watch the greater than or equal to (≥) and less than (<) signs in the key to be sure you are making the correct decisions for the pathways.

Step 7: Record the FLM number on the field form. If you need fuel load values for individual load components to input into a software application, the median loads for each FLM are summarized in table 2.

Identifying an FLM in Grasslands, Shrublands, or Chaparral

Non-forested areas can be keyed to an FLM in the field using several methods, including (a) making visual estimates of the total fuel load using photo series guides,

(b) calculating total fuel load from counts or measures of separate fuel components, or (c) clipping and weigh-ing all on-site fuels. The FLM key for the non-forested areas requires an estimate or measurement of total site fuel load that includes FWD, coarse woody debris, duff, litter, and shrub and herbaceous cover. To help users visualize total plot loads, we provide examples of mea-sured site loads using photographs from several photo series guides (Ottmar and Vihnanek 2000; Ottmar and others 2000; Wright and others 2002). The photo series examples contain total load values for each picture plus a list of the individual fuel components that comprise the total load. Any photo series guide can be used to estimate site load. The photographs and data used in this guide are from the Digital Photo Series developed by the U.S.

9USDA Forest Service Gen. Tech. Rep. RMRS-GTR-225. 2009

Forest Service, Pacific Northwest Research Station, Fire and Environmental Research Applications Team (http:// depts.washington.edu/nwfire/dps/). However, users may want to use a photo series guide that is specific to their ecosystem or landscape. As long as the photo series has measures of total site load and can be compared with the values required in the non-forest FLM key, it can be used to estimate loads and identify FLMs. Users should visually compare the total loads on their plot with the photo series photographs and decide if their plot looks like it has more or less load than the photos. When total fuel load is determined, the non-forest FLMs are easily identified. The identification process is as follows:

Step 1: Determine the cover type. In non-forested environments, the FLM key is based on whether the fuel bed is created by sagebrush or non-sagebrush vegetation. Sagebrush plots must have at least 10% canopy cover of Artemisia spp. Remember, the FLMs were not developed for areas with shrubs greater than 6 ft tall.

Step 2: Estimate the total fuel load on site. Compare your site conditions with the photographs of known fuel loads in Appendix G (sagebrush sites) or Ap-pendix H (non-sagebrush sites). Note that each photo series already includes duff and litter in its total load value so you need only to match the picture with your plot conditions to assign an approximate load. In the appropriate boxes on the field form (Appendix A), record the number and total load of the photo series that most closely matches load on your plot.

Step 3: Key to the appropriate FLM. Use the total load determined in step 2 and match it with the load ranges in the non-forested area key (Appendix I) to pick the FLM that best describes the cover type and load estimate for your plot.

Step 4: Record the FLM number on the field form. If you need fuel-load values for individual load components to input into a software application, the median loads for each FLM are summarized in table 3.

Management Advice _____________

As users begin to apply this classification, we would like to stress several points about its use and applicabil-ity. First, make this classification as convenient for your work situation as possible. You can use this classifica-tion without using the field form provided in this paper.

Simply add fields on your normal field form if it makes recording the FLM data more convenient. By using the key regularly, we are confident that you will soon be able to recognize fuel load thresholds and identify FLMs in your area very quickly so you should make recording the data as convenient as possible. Second, there may be times when the fuels on your site may not seem to fit into this classification. In these situations, first ensure that you are in an appropriate vegetation type for the FLM key. As stated previously in this paper, do not use this key in rare ecosystems such as pocosin swamps or on sites with tall shrub vegetation (>6 ft tall). If you are in an applicable vegetation type and are unable to make the key fit, make sure that you are reading the greater than or equal to (≥) and less than (<) symbols correctly, as they can be confusing when you start out. If one col-umn does not seem to fit your fuel load data, then step back a row and try again, paying close attention to these symbols. This key is designed to help you make estimates accurately and quickly. Estimating fuels takes practice, so recheck your field estimates to make sure that you have accurately portrayed the on-site fuels. Some of the FLMs in this classification have very small differences in fuel components, such as duff thickness, which may require measurement instead of estimation of their values.

When you begin your identification of FLMs, pull out the ruler and measure these critical thicknesses, if necessary.

If you still have trouble fitting a site’s characteristics to the FLM key after checking all of these items, remember that this key misclassified plots 34% of the time during its development and this could be what is affecting your ability to correctly identify an FLM (Lutes and others, in press). New classification systems invariably need adjust-ments as they are applied to new locations and used by more people and programs. As more plots and data are collected and analyzed across ecosystems, the extent of the FLM misclassification problem can be investigated and rectified to improve the FLM classification. Finally, and perhaps most importantly, we want to stress that this classification does have limitations. Do not attempt to extrapolate fire effects for FLMs to effects caused by canopy fuel consumption. The FLM key is designed to predict fire effects only from the surface fuels.

FLMs constitute an important advance in fuel clas-sification because they relate actual on-site fuels to the smoke and soil heating that may result from burning those fuels. As such, the FLMs can be an important tool in many fire studies and management decisions. There may be fire effects in addition to smoke and soil heating that could be incorporated into the FLM classes in the future that would make them more pertinent to some wildlife, vegetation, and microbe studies. However, the current FLM classes are a positive step in the process

10 USDA Forest Service Gen. Tech. Rep. RMRS-GTR-225. 2009 to create a fuels classification that directly relates cause to effect in fuels consumption, and they should be an improvement over earlier fuel classification methods for many applications.

References _____________________ Anderson, Hal E. 1982. Aids to determining fuel models for esti-mating fire behavior. Gen. Tech. Rep. INT-122. Ogden, UT: U.S.

Department Agriculture, Forest Service, Intermountain Forest and Range Research Station. 22 p.

Andrews, P. L.; Bevins, C. D. 1999. BEHAVE Fire Modeling System:

redesign and expansion. Fire Management Notes. 59(2):16-19.

Andrews, P.L. 1986. BEHAVE: Fire behavior prediction and fuel modeling system. Gen. Tech. Rep. INT-194. Ogden, UT: U.S.

Department of Agriculture, Forest Service, Intermountain Research Station. 130 p.

Breiman, L.; Friedman, J.H.; Ohshen, R.A.; Stone, C.J. 1984. Clas-sification and regression trees. New York: Chapman & Hall.

Brown, James K. 1971. A planar intersect method for sampling fuel volume and surface area. Forest Science. 17(1):96-102.

Brown, James K. 1974. Handbook for inventorying downed woody material. Gen. Tech. Rep. INT-16. Ogden, UT: U.S. Depart-ment Agriculture, Forest Service, Intermountain Forest and Range Experiment Station. 24 p.

Brown, James K. 1981. Bulk densities of nonuniform surface fuels and their application to fire modeling. Forest Science. 27:667-683.

Brown, James K.; See, Thomas E. 1981. Downed dead woody fuel and biomass in the northern Rocky Mountains. Gen. Tech. Rep.

INT-117. Ogden, UT: U.S. Department of Agriculture, Forest Service, Intermountain Forest and Range Experiment Station. 48 p.

Brown, James K.; Bevins, Collin D. 1986. Surface fuel loadings and predicting fire behavior for vegetation types in the Northern Rocky Mountains. Res. Note INT-358. Ogden, UT: U.S. Department of Agriculture, Forest Service, Intermountain Forest and Range Experiment Station. 9 p.

Burgan, Robert E. 1987. Concepts and interpreted examples in advanced fuel modeling. Gen. Tech. Rep. INT-238. Ogden, UT:

U.S. Department of Agriculture, Forest Service, Intermountain Research Station. 40 p.

DeBano, L.F.; Neary, D.G.; Folliott, P.F. 2005. Volume 4 Report Title:

Chapter 2—Effects on soil physical properties. Gen. Tech. Rep.

RMRS-GTR-42. U.S. Department of Agriculture Forest Service Rocky Mountain Research Station:29-51.

Fahnestock, G.R.; Agee, James K. 1983. Biomass consumption and smoke production by prehistoric and modern forest fires in western Washington. Journal of Forestry. 81(10):653-657.

Finney, Mark A. 2004. FARSITE: Fire Area Simulator: model de-velopment and evaluation. Res. Paper RMRS-RP-4. Ogden, UT:

U.S. Department of Agriculture, Forest Service, Rocky Mountain Research Station. 47 p.

Frandsen, William H. 1987. The influence of moisture and mineral soil on the combustion limits of smoldering forest duff. Canadian Journal of Forest Research. 17:1540-1544.

Green, Lisle R. 1970. An experimental prescribed burn to reduce fuel hazard in chaparral. Res. Note PSW-216. Berkeley, CA: U.S.

Department of Agriculture Forest Service, Pacific Southwest Forest and Range Experiment Station. 6 p.

Hawkes, B.; Niemann, O.; Goodenough, D.; Lawson, B; Thomson, A.; Sahle, W.; Fuglem, P.; Beck, J.; Bell, B.; Symingto, P. 1995.

Forest fire fuel type mapping using GIS and remote sensing in British Columbia. In: Heit, M.; Parker, H.D; Shortreid, A., comps.

Proceedings: GIS applications in natural resources 2 symposium- the 9th symposium on Geographic Information Systems, Vancouver, British Columbia, Canada:290-299.

Keane, R.E.; Frescino, T.L.; Reeves, M.C.; Long,…