|Author's First Name||Harun|
|Author's Last Name||Makandi|
|Author's Contacts||Cell: +255 689 89 2205; E-mail: email@example.com|
|Thesis Title||Evaluation of Remotely-sensed Thermal Fluxes in Monitoring Woodland Carbon: the Case of Liwale and Kilwa in Lindi, Tanzania|
|Supervisor's full name||Philip K. Mwanukuzi, Ph.D.|
|Supervisor's Contacts||Cell: +255 754 018 648; E-mail: firstname.lastname@example.org|
A functional, cost-effective, and comprehensive system for repetitive measurement, reporting and verification (MRV) of forest carbon is important for sustainable forest management. Optical remote sensing datasets are critical for the development of such a system because they are free, and have a wall-to-wall and repetitive coverage. However, their accuracy in estimating woodland above-ground biomass and carbon (AGB and C) using mainstream methods is limited. One such method is using the magnitude of woodland greenness quantified using the normalised difference vegetation index (NDVI) drawn from the imagery. NDVI saturates with increasing AGB and C, thereby limiting the range estimations. Also, the greenness fluctuates seasonally in tropical woodlands and evaluates the variable canopy moisture than the stable AGB and C. Cloud contamination on the datasets is another limitation. These limitations justify estimations using radar and Lidar as alternatives. These have their own limitations, including limited repetitive imaging and geographical coverage.
To enhance the accuracy of the estimations while leveraging the strengths of optical satellite data, a Forest Biomass Index (FoBI) was developed to model the magnitude of the latent and sensible thermal fluxes prevalent in woodland conditions. The satellite-derived surface temperature (Ts) and NDVI were combined in the modelling using the index. The magnitude of the fluxes correlates better with woodland AGC and is less prone to seasonal fluctuations than does that of the commonly used woodland greenness. The resulting FoBI maps were regressed with plot-based AGC measurements to estimate the AGC in Liwale and Kilwa districts in 2014 and 2018 and its change between the two years.
The regression of the FoBI maps of 2014 and 2018 with plot-based AGC returned R2 of 0.52 and 0.58 respectively. This compared favourably to R2 of 0.44 from pairing the annual NDVI map of 2014 with plot estimates. Also, the range of estimation of FoBI map was from 0 to 266 t ha-1 C, which was over twice that of NDVI. NDVI’s range plateaued at its peak, indicating saturation. FoBI’s extended range indicates the elimination of the saturation problem at least in estimating AGC in miombo woodlands. Cloud cover was also eliminated in the extensive processing that included maximum value compositing (MVC) multiple Ts and NDVI layers. Using the regressed FoBI maps of 2014 and 2018, the mean carbon stock density in the study area was estimated to be 44 t ha-1 at 95% confidence level in both years. The total AGC was about 220 Mt in 2014 and 213 in 2018. Change analysis shows a decline of 6.6 Mt (ca. 3%) of total AGC between the two years, which indicates general stability of the AGC pools in Liwale and Kilwa.
The developed FoBI enhances the accuracy of comprehensive and repetitive estimations of woodland AGC using free and widely available optical satellite datasets by eliminating the main cited problems with using them. Using FoBI, monitoring and reporting woodland carbon stocking that meets the standards of UNFCCC reporting can be done.
|Entry Date||May 19, 2019|