23.12.2012 Views

Thematic Accuracy Assessment Procedures. Version 2 - USGS

Thematic Accuracy Assessment Procedures. Version 2 - USGS

Thematic Accuracy Assessment Procedures. Version 2 - USGS

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Method 3 (Method 4 with a post hoc review of limited floristic data collection) is a less desirable<br />

methodology for the NPS Vegetation Inventory, but may be employed if the following<br />

conditions are met:<br />

-Field data collected for post hoc review are limited to a list of dominant and/or frequent species<br />

in each stratum (e.g., not species cover estimates, photographs, etc.). This measure is intended to<br />

save field time and also direct limited field time toward carefully assessing criteria needed to<br />

follow the field key. It will limit the ability to “overrule” reference data values made in the field<br />

to the more significant errors. “Reasonable but wrong” calls will often stand. Exhibit I is an<br />

example of a field data assessment form that meets these criteria.<br />

-Neither the identity of both the map class (sample data value) nor the reference call that has<br />

been made by the field observers can be revealed to the reviewing expert prior to review.<br />

Revealing the map class identity is a clear violation of independence in a hypothesis test.<br />

Revealing the observers’ reference data values will likely bias the reviewer’s judgment. The<br />

reviewer knows that he/she is reviewing only reference data values that are mismatches with the<br />

sample data values and is likely to avoid the same reference data when there is ambiguity. The<br />

reviewer must be given an unbiased opportunity to agree with, as well as disagree with the field<br />

observers’ calls. If the reviewer lacks sufficient information to make a judgment about the best<br />

reference data call, then no alternative value need be offered, and the field observers’ reference<br />

data values will be regarded as correct.<br />

-Data used in mapping (e.g., properties of a vegetation image and other remotely sensed data)<br />

may not be used to influence either reference data value.<br />

If these procedures are followed, then only field data for sites with mismatches between the<br />

sample data value and reference data value need be reviewed and the expert reviewer’s<br />

assessment from the [limited] field data may be regarded as the most correct reference data<br />

value, when there is disagreement between the field observer and the reviewer.<br />

3.2 Field Methodology<br />

3.2.1 Field Observer Skill Level Considerations<br />

While the NPS Vegetation Inventory accepts Method 4 as a best practice, a minimal amount of<br />

preparation and oversight is necessary to assure that observer skill and that the accuracy gap<br />

between use of Method 4 and Methods 1, 2, and 3 is minimized, within a reasonable amount of<br />

project control.<br />

Field observers should be hired and/or vetted and trained, as needed to represent an adequate<br />

skill level to:<br />

Identify all species named in the field key, as well as likely look-alikes. For example, if northern<br />

red oak (Quercus rubra) is named in the key, then observers should be able to recognize this<br />

species whenever it is encountered, to avoid errors of omission of northern red oak in the key. If<br />

the similar-appearing scarlet oak (Quercus coccinea) is also present in the study area, observers<br />

should be able to recognize it and differentiate it from northern red oak, regardless of whether or<br />

not scarlet oak is named in the field key (in order to avoid errors of commission in situations in<br />

the key that call for recognizing northern red oak). Familiarity with most of the flora is<br />

33

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!