6. Revisión de literatura y otras fuentes
6.2. Marco teórico
The average stem numbers, stand basal area, volume, biomass and carbon per hectare have already been presented in Table 5, in Chapter 4. The table shows the average estimates, with statistical confidence intervals (expressed as ± values) at 90% precision level. The attained precision levels of estimates (percentage confidence interval to average values of estimates) are given in brackets.
Experience shows that carbon stocks in natural forest can be estimated to precision levels within ±10% of the mean, with 95% confidence (IPCC, 2003). Against this experience, results from the forest studied indicate that the precision levels attained are relatively low (> 21%). This may be explained by the nature of the village forests as detailed in Sub-Section 5.4.1. It is not exactly certain whether the payment of carbon under REDD will be based on average or minimum estimate. As pointed out in Chapter 2, Sub-Section 2.3.1 and shown in Figure 2, it is more likely that minimum estimate (i.e. the low end value of estimates around the means) will be used and as such it is very important to have data of very high quality i.e. carbon estimates with very high precision.
With the need to improve the quality of estimates by local communities, a professional verifier was commissioned to carry out a verification study and give recommendation for improvements on the villagers’ measurements. This was done for the KSUATFR by the Tanzania Forest Research Institute (TAFORI). The commissioned verifier was asked to follow the same standard procedures for measuring forest carbon as those used by villagers.
For this forest, instead of using a plot of 5.6 m radius as done by the villagers (Sub-section 5.4.2), the TAFORI team opted to used co-centric circular sample plots in which different sized trees were measured for dbh as follows:
• within 2 m radius: all trees greater than 1 cm dbh were measured, • within 5 m radius: all trees greater than 5 cm dbh were measured, • within 10 m radius: all trees greater than 10 cm dbh were measured; and • within 15 m radius: all trees greater than 15 cm dbh were measured.
above were established randomly to cover possible variations in the entire forest. As for the villagers, stratification was also seen by the professionals to be unnecessary for this forest. Then the number of sampling units (n) required to attain a desired precision at sampling error (E) of 10% was determined to be 87 plots. It was decided to consider 89 sample plots as those used by villagers and to use the same sampling frame. Villagers who carried out forest carbon assessment for the forest were also used to assist in locating the position of plots.
Results from this verification show that, there was no significant difference between the villagers’ survey and the professional one as regards estimate of average carbon stocks (Table 12). However, the precision of TAFORI’s carbon stock estimate was as higher, at ±9% compared to ±21% attained by the villagers. This could be explained by increased sample plot sizes by TAFORI. With the same number of plots, the larger the plot size, the larger the sampling intensity and therefore the higher the precision of estimates. The sampling intensity employed by TAFORI was higher, resulting in a decrease in confidence interval and rise in the precision of the estimate. High sampling intensity has the positive effects of covering more forest variation in term of tree sizes and tree/shrub species abundance. TAFORI used the concentric plots (four different sub-plots) with maximum radius of 15 m while the villagers used 5.6 m radius plots. The concentric plots with many different sub-plots were not used by the villagers as this could have complicated the field work by them.
During this verification, it was also observed that villagers were able to accurately locate sample plots and to take tree measurements from plots correctly. The measurements by local communities therefore could have improved if the sampling intensity were increased by increasing the plot sizes; this is an important lesson which has been gained from this exercise. However, as pointed out above, the use of concentric plots would have complicated the field work, which is why this method was not selected in the first place. Since it is expected that natural forest have many small trees and few large trees, the use of concentric plots in natural forest inventory are aimed at minimizing field work, by measuring small size trees in small area plots and vice versa for large trees. The observed tree size distribution in this study (Figure 7 in Chapter 4), suggests that there are many small size trees of less than 10 cm dbh and few larger ones in all studied forests. Therefore, the instructions in the field forest inventory guide have been adapted and now call for 2 co-centric plots of 5 and 15 m radius to be used for trees of up to 10 and greater than 10 cm dbh. This will make the field work by
local communities easier and at the same increase the sampling level by increasing the plot sizes and most likely will also improve the precision of estimates.
Table 12. Stand parameters for KSUATFR by TAFORI and villagers Carbon assessment by*
Stand parameters
TAFORI Villagers
Mean
Diff. Df t-value p-value Significance Basal area (m2/ha) 9.03+0.69
(8%) 9.15+1.54 (17%) - 0.124 88 -0.160 0.8736 Not significant Volume (m3/ha) 65.54+9.18 (14%) 68.12+16.92 (25%) - 2.579 88 -0.283 0.7777 Not significant Biomass (tons/ha) 43.15+3.75 (9%) 42.19+8.65 (21%) 0.950 88 0.217 0.8286 Not significant Carbon (tons/ha) 21.14+1.84 (9%) 20.39+4.24 (21%) 0.466 88 0.217 0.8286 Not significant *The attained precision levels of estimates (percentage confidence interval to average values of estimates) are given in brackets.