Appendix 3

CONSISTENCY CHECK

Methods

The consistency of the field team’s measurements was assessed by the repeat measurement of two survey units randomly selected out of the first 50 units. The field team had operated under considerable supervision throughout the second half of the felling block, where the exercise had started with on-the-ground training in the field method. For this reason units 51 to 100 were excluded form the sampling frame. The two units selected were numbers 3 and 14. These were re-measured by the same staff, with the exception that the booker was a different person in unit 3. Re-location was straightforward as the survey stakes remained in place along the internal transect cut-lines. Table 1 gives the dates of assessment.

Table 1. Dates of assessment

Survey unit	Date of measurement	Date of re-measurement
3	13 May 2000	22 May 2000
14	17 May 2000	22 May 2000

Results and Discussion

Table 2 summarizes the results of the consistency check for the five attributes that were recorded: the number of trees, the species identification, the stem diameter measurement, the merchantability score and the logging damage score. Overall, 96 per cent of the trees were measured twice. Two out of the three trees missed at re-measurement and all three trees gained were trees at or close to the minimum diameter limit of 30-cm dbh. This result shows that care is required when checking small trees and that such checks should be made with the diameter tape rather than solely by eye. Also, the re-measurement exercise failed to record one 62-cm dbh Morabukea tree, suggesting that even with a high standard of field recording gross errors will occur. The field method developed for this programme, particularly the narrow, demarcated enumeration sweeps, is designed to keep this type of error to a negligible level.

The field team benefited from the expertise of Mr. Sam Roberts, the chief tree spotter, who was well supported by Mr. Ignatius Stoll. The 100 per cent match in tree names between the two assessments demonstrates their skill in species identification. Fifteen tree species (13 of which are considered commercial) were measured within the two survey units. The GFC can therefore place considerable confidence on this important part of the survey results.

Stem diameters are difficult to measure accurately due to the irregular stem shape of most tree species in tropical moist forest. Exceptions, such as Baromalli and Locust occur with near cylindrical stems, but they constitute a small number of the total trees measured. For this survey an acceptable consistency level for the stem diameter measurement was set at +/- 2 cm, implying that most trees would not cross a 10 cm dbh size-class assuming they were within this range. This range should also take into account any discrepancy between the two measurements caused by the point of measurement not being marked on the tree. A minor error level was set at 3 to 9 cm implying a possible size-class change; and a gross error level of 10 cm difference or greater that would place the tree in another 10 cm size-class. Eighty-five per cent of stem diameter measurements fell within the acceptable range, with minor errors accounting for another 13 percent. Only two percent of the re-measured trees were classified in the gross error category. It can therefore be concluded that considerable confidence can be placed on the resultant stand tables. Eight out of the Twenty Two (22) trees that were classified as being in error were Morabukea, a species well known for its high buttresses (and hence difficult to measure). The data were also checked for bias, which might have occurred due to the stretching of the measurement tapes during field-work. The near balance of under and over-measurements suggest an absence of such bias.

Table 2. The results of the consistency check

Attribute		Unit 3	Unit 14	Total	Per cent
Number of trees	Match	85	60	145	96.0
	Loss at re-measurement	3	0	3	2.0
	Gain at re-measurement	1	2	3	2.0
Species identification	Match	85	60	145	100
	non-match	0	0	0	0
Stem diameter	-10 + cm	1	0	1	0.7
	-3 to -9 cm	7	1	8	5.5
	-1 to -2 cm	18	11	29	20.0
	Match	44	38	82	56.5
	1 to 2 cm	7	5	12	8.3
	3 to 9 cm	7	4	11	7.6
	10 + cm	1	1	2	1.4
Merchantability score	Match	74	54	128	88.3
	non-match	11	6	17	11.7
Logging damage score	Match	85	58	143	98.6
	non-match	0	2	2	1.4
Number of tree stumps		0	2	2

 

The merchantability score is an assessment of the tree stem with regards its future utilization value. It had been decided to keep this subjective score as simple as possible, judged on the presence or absence of defect (both visible and internal) and stem form. Eighty-eight per cent of all trees were assessed consistently using this criterion.

The assessment of logging damage was consistent between the two measurements, with an almost 99 per cent match in this score. Overall, logging had occurred at a low level within this felling block and this was reflected in the consistency check samples, with only two trees having been felled in unit 14. Five trees in the same unit showed signs of felling damage, although two of these had gone unnoticed at the time of the first assessment.

Conclusions and recommendations

The consistency check is an important component of the silvicultural survey, providing a quantitative measure of the quality of field observations. The results described above all point to a high standard of data collection, which is to the credit of the GFC field team.

With regard to the two subjectively scored attributes, some strengthening of the field method is recommended. The scoring system for stem merchantability had been discussed earlier with the company’s foresters elsewhere in the felling block, who had agreed with the assessments made on a small number of trees. This important element of the survey would be further enhanced by inviting company staff to join the field team during future consistency checks to validate the measurement of this particular attribute. The assessment of felling damage suggests the results of this method are rather imprecise at present and therefore the procedure warrants further development.