improving dwm data quality 3rd of 3 part training series christopher woodall dwm national indicator...
TRANSCRIPT
Improving DWM Data Quality3rd of 3 Part Training Series
Christopher Woodall
DWM National Indicator Advisor
Outline
• QA/QC Analysis
• What Customers Want
• Measurement Errors Hot and Cold Checks
• Top Six List of Errors
• Training
QA/QC Analysis
The analysis of 2001-2004 DWM QA plots is currently ongoing. Matching algorithms are being developed for numerous measurement variables. Expect results for the 2006 P3 Training Sessions.
For more information contact: Chris Woodall @ NCFIA and Jim Westfall @ NEFIA
What Customers Want
Key
N
430°
150°
270°
30°
150°
270°
30°
150°
270°
30°
150°
270°
3
2
1
Transect Information
FWD < 0.25”& 0.26”-0.99”
FWD 1.00”-2.99”
CWD => 3.00”
6 ft. s.d.
10 ft. s.d.
24 ft. h.d.
s.d.= slope dist., h.d.=horizontal dist.
Sub-plot
Micro-plot
CWD Transect
FWD Transect
Distances between sub-plot points: 120 ft., Distance from sub-plot center and microplot center: 12 ft., Distance betweenSub-plot 1 and sub-plots 2, 3, and 4: 207.8 ft. at angles (degrees) 150, 210, and 270 respectively.
A uniform DWM sample design applied across
the entire United States producing per acre estimates of fuels, carbon, and wildlife
habitat
Measurement Errors
• Establishing Transects
• FWD Counts
• Slope versus Horizontal Distances
• CWD Diameters
• Correct Units for Duff, Litter, and Fuelbed Depths
• Microplot coverage and heights
Hot checks
Measurement Errors
• Number 1 priority is matching data and determining adherence to MQO’s
• Number 2 priority is determining cause for errors…then correcting cause
Cold/Blind Checks
Measurement Error Propagation
CWDDuff/Litter
FWD
Slash
Shrub/Herb
Transect
Database Processing Algorithms
Core TableMeasurement errors have varying magnitudes of effect
Measurement Error Simulation
Effect of 5, 10, 15% variation in down woody material measurement variables on total per acre tonnage estimates
0
1
2
3
4
5
6
7
8
5 10 15
Simulated Variation (%)
Va
ria
tio
n in
To
tal T
on
na
ge
Es
tim
ate
s
(%)FWD - 1-hr Count
Litter - Depth
CWD - Tran. Length
FWD - Tran. Length
Slash - Height1
Slash - Density
CWD - Diameter
Duff - Depth
Simulation Conclusions
Measurement variables whose errors least affect core table outputs: CWD decay, classes/transect lengths, litter depth, and FWD counts
Measurement variables whose errors most affect core table outputs: duff depths, CWD diameters, and slash pile densities
FIA’s Top Six LeastWanted DWM Errors
1. CWD Diameters2. CWD Lengths3. Duff Depths4. Litter Depths5. Slash Pile Density6. Missing Data
CWD Diameters
Crews mistakenly record CWD diameters to tenth of inch…used to P2 plots
Only measure to nearest inch!!
≠
CWD Lengths
Some log dimensions recorded in field are impossible
3 inches
15 feet
120 inches
=
Duff Depth
Duff is the heaviest down woody material per unit volume
Make sure your measurements (and units) are correct
Litter Depth
Much lighter than duff…however is usually much deeper
Don’t mistakenly enter the litter depth for duff depth
Slash Pile Density
Only neatly stacked wood can exceed 40-60% density!
Missing/Mismatched Data
AKA: Excruciating Headaches for Analysts
Missing/Mismatched Data
Example: DWM plot sheet indicates CWD transects on a condition class 2…however, only one condition class recorded in P2 record
Example: CWD piece is decay class 2, but is missing small and large end diameters
Might be your fault, might be data management’s fault, might be computer’s fault…no matter…do what you can to minimize mismatch errors
Training
Problem Areas
Problem: Field crews disturb the CWD too much trying to determine decay class or if segmented
Correction: Although field crews must disturb CWD pieces in order to acquire measurements, try to keep disturbance to a minimum
Problem Areas Cont’d
Problem: Field crews mistakenly enter extra digit for CWD diameter (40 instead of 4 inches)
Correction: Unless PDR’s catch them, be sure of very large CWD diameters
Problem Areas Cont’d
If CWD piece ends in water, treat as if underground,
measure piece to water edge
For FWD, if transect under water try to enter “0” values and
indicate in plot notes
Problem Areas Cont’d
Problem: Crews dig through litter hunting down pieces of FWDCorrection: Crews should only tally obvious FWD pieces, namely those on litter surface
Problem: Crews aren’t tallying FWD pieces hung up in slash/saplings Correction: Crews should tally all FWD pieces from forest floor up to 6 feet above ground
Problem Areas Cont’d
Problem: Crews either include too much of the litter layer or upper soil mineral horizons in estimation of duff depth
Correction: Crews should be absolutely sure of what is duff, litter, and mineral horizons. Be absolutely sure of duff measurements!!
Problem Areas Cont’d
Duff Depths:
1) Identify duff from mineral soil2) Don’t include moss or litter
material3) What to do with deep duff4) Anything over 1 foot be
absolutely sure
Problem Areas Cont’d
Problem: Crews can’t decide on the fuelbed height measurement
Correction: Crews should only take 15-seconds to determine height of dead, down woody material, don’t over analyze, use local knowledge and reasonable definition of fuel ladders
Problem Areas Cont’d
1) Measure from top of duff to top of fuel complex
2) Fuel complex composed of dead FWD, CWD, shrubs, and litter
3) Gaps allowed in fuel complex where one would reasonably expect flame lengths to connect
4) Plum-bob not required, ocular estimate around sample point
5) 15-second rule…Don’t over analyze height of fuelbed…Use your experience and logic
Fuelbed Depths
Problem Areas Cont’d
Problem: Condition class boundary runs through microplot
Correction: Use entire forested condition of microplot to estimate coverage and heights
=100% cover of litter for forested conditions (don’t include asphalt or other non forested conditions in cover assessment)
Problem Areas Cont’d
1) Train with idea of imaginary 6.8 foot radius cylinder
2) Make sure crews know what herbs and shrubs include
3) Gaps allowed in fuel complex as long as reasonable
4) Branches from shrubs rooted outside microplot allowed
5) Train about vines and canopy herbaceous plants
Microplot Heights
Problem Areas Cont’d
Only include epiphytes or hanging moss up to
6 feet in height
Include vines that are within microplot
Problem Areas Cont’d
Only estimate density of CWD within pileDensity should rarely exceed 40%
70% 20% 01%
Slash Pile Densities
Organizing Training Sessions
1) Part 1: Introduction to DWM2) Part 2: Field Methods3) Analyst Example (optional)4) Part 3: Improving DWM Data Quality5) Certification Test
1) Stations (test optional)2) Go over one subplot together as group3) Trainees do at least one subplot on their
own – hot audit and/or compare results
Classroom
Field
Bringing it all Together
1) Pick training location where many conditions classes and sampling scenarios exist (see word file)
2) Use powerpoint files to sculpt training session so trainees have understanding of why we need quality DWM data, what we use it for, the theory behind the sampling design, field methods, and problem areas
3) Setting up a quality station course can reduce questions during actual field season – may conduct test
Sample Design Changes
The DWM Indicator must be responsive to customer needs and improving science/techniques…
Don’t assume your ideas are insignificant, you collect the data, assume you know best and pass ideas upwards…
Submit your suggestions [email protected]
End of Part 3 of 3
http://www.ncrs.fs.fed.us/4801/national-programs/indicators/dwm/