EFFECT OF LEAF PHENOLOGY, TOPOGRAPHY AND WIND SPEED ON FOREST CANOPY HEIGHT AND ABOVE GROUND BIOMASS ESTIMATION USING OPTICAL UAV DATA IN MALAWI’S MIOMBO WOODLANDS

Abstract

The study investigated the effect of tree leaf phenology, topography and wind speed on estimating forest canopy height (FCH) and above ground biomass (AGB) using optical drone data in Miombo Woodlands of Malawi. Data were acquired in 2019 from thirteen plots of Ntchisi Forest Reserve, in two-fold; tree measurements from 0.1ha in-situ plots and airborne RGB ortho-images overlaid on the terrestrial plots from 50 ha boxes. The imagery were acquired by DJI Phantom-4 Unmanned Aerial Vehicle (UAV) at ~7cm pixel resolution. These were processed using Structure-from-Motion photogrammetry algorithm in Agisoft Metashape Professional software. Malawi’s local allometry  was used to estimate AGB while the Analysis of Variance was employed to test accuracy. Results reveal that leaf phenology, topography and wind speed significantly (P<0.05) affected estimation of FCH and AGB using the airborne observations. Wind speed contributed highly (53.95% and 55.52%) to the total variation of FCH and AGB, respectively. Furthermore, wind speed upsurge >5m.s-1 (-666.90±27.62%; -64.78±8.47%), topography >15% slope (-114.53±2.92%; -17.07±2.31%) and leaf-off condition (-188.45±25.62%; -27.80±7.98 %) bore higher errors for FCH and AGB, respectively. A significant variation of relationship (R2=0.71; P<0.05) was observed between ground-based height and UAV canopy height. Nevertheless, the variation of the relationship strongly increased by 10% to R2=0.81 when two plots highly affected by the wind speed were excluded from the analysis. Similarly, a significant variation of relationship (R2=0.81; P<0.05) was registered between the ground-based and UAV-based AGB. Correspondingly, a very strong variation of relationship (R2= 0.95; an increase by 14%) was yielded when the two plots highly affected by wind speed were excluded from the analysis. These findings imply that seasonal leaf-fall, slope and wind speed significantly influence the accuracy of FCH and AGB estimations in Miombo Woodlands. Hence, for accurate results, it is a pre-requisite to consider leaf-phenology, topography and wind speed prior to undertaking ground and airborne inventory in the Woodlands.