Mapping and estimating forest area and aboveground biomass in miombo woodlands in Tanzania using data from airborne laser scanning, TanDEM-X, RapidEye, and global forest maps: A comparison of estimated precision

Modelling above Ground Biomass in Tanzanian Miombo Woodlands Using TanDEM-X WorldDEM and Field Data

“…It was encouraging to see that the RMSE values found in this study were smaller than in Ene et al [23] who reported a RMSE of 31.16 t·ha −1 (47.4% of the mean AGB). The validity of this comparison is also confirmed by the findings by Naesset et al [29] who, in the sub-area of our AOI, found larger cross-validated RMSE than the model presented in this study both for ALS (31.8 t·ha −1 ; 62.0%) and RapidEye data (34.2 t·ha −1 ; 66.6%). Despite the consistency of these findings across the entire AOI and in the sub-area of the AOI, it was somewhat surprising to see that medium-resolution TanDEM-X WorldDEM data could yield smaller RMSE and RMSE % (30.4 t·ha −1 ; 59.2%) than ALS data.…”

“…The comparison with the study by Solberg et al [17] revealed that the use of TanDEM-X WorldDEM at medium resolution (i.e., 30 m × 30 m pixels) resulted in a smaller β 1 estimate (11.0) and in a decrease in RMSE compared to higher resolution TanDEM-X data (i.e., 10 m × 10 m pixels). Furthermore, compared to the other remotely sensed data sources compared by Naesset et al [29], the use of TanDEM-X WorldDEM always led to a reduction in the RMSE. Table 2.…”

“…The distance between consecutive plots within a cluster was 250 m. Further details on the sampling design can be found in Tomppo et al [26,28]. In this study the field dataset used by Solberg et al [17] and by Naesset et al [29] increased nearly five times, thus providing the possibility to better understand the effect of the increased sample size on the AGB models. Field plot measurements were carried out in February-June 2012 and included the measurements in concentric circles with diameters of 2, 5, 10 and 15 m from trees with diameter at breast height (DBH) of >1, 5, 10 and 20 cm, respectively.…”

“…However, because of the lack of additional higher precision measurements of the ground height, it was not possible to determine which dataset was more precise. To provide a thorough comparison between the results of the present study against previous research by Solberg et al [17] and Naesset et al [29], we used the 88 field plots from those studies located in a sub-area of the present AOI. Half of these were a subset of the total number of plots (i.e., the 513 plots previously described), while the other half were part of a new set of plots acquired by doubling the clusters' sampling intensity (see Figure 1).…”

“…The field data for the sub-area were acquired two years after (i.e., 2014) the acquisition of the field data used for the entire AOI (i.e., 2012). A further description of this additional field plot dataset can be found in Naesset et al [29]. Table 1 summarizes AGB for the entire AOI and for the sub-area of the AOI, while Figure 1 illustrates the sampling design in the entire AOI and in the sub-area used for further validation.…”