Extent and canopy height maps of Trees outside Forest (ToF) for Bangladesh
Abstract
Trees in Bangladesh are disproportionately important. An estimated 11% of Bangladesh is forested and it has one of the lowest forested land per capita (0.009 ha person) estimates globally (FAO., 2015). Approximately 40% of the Bangladesh natural forest area has been lost since 1930 (Reddy et al., 2016) due to demands for wood products and agricultural land expansion. Trees outside forest (ToF), defined as discrete trees and small groups of trees in non-forest settings, therefore provide important ecosystem services and are an essential socioeconomic and ecological asset. Previous assessments of ToF have often been limited by the unavailability of sufficient high-enough resolution remotely sensed imagery and have primarily utilized optical data only. We have generated a new extent and canopy height map for all trees within Bangladesh. We used a combination of ESA Sentinel-1 and Sentinel-2 and DLR Tandem-X imagery within the Google Earth Engine environment. Thresholds on Sentinel-1 Synthetic Aperture Radar (SAR) HV and Sentinel-2 NDVI were used to derive tree extent, at 10 m resolution (mmu >1/8 ha). Tree canopy height at 30 m resolution was generated using an InSar Tandem-X digital surface model (DSM) and associated interpolated Tandem-X digital terrain model (DTM) (RMSE 1.26 m). We mapped 4.5 million ha of trees, increasing existing estimates of tree cover extent, as a consequence of the use of higher resolution multi-modal imagery. We derived country wide canopy height (excluding mountainous terrain), with an average of 9.5 m for trees outside of continuous forests (i.e., forested hillsides, mangroves). Further work is oriented towards improving canopy height estimates over sloped terrain and deriving biomass estimations from field collected allometry. We successfully generated a new tree extent and canopy height map that will serve as baseline products for the monitoring of terrestrial carbon within a heavily fragmented yet important forest resource. This work contributes to a USAID SilvaCarbon initiative to facilitate the monitoring of forest extent and terrestrial carbon by the Bangladesh Forest service.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2018
- Bibcode:
- 2018AGUFM.B31I2599T
- Keywords:
-
- 1632 Land cover change;
- GLOBAL CHANGEDE: 1640 Remote sensing;
- GLOBAL CHANGEDE: 1855 Remote sensing;
- HYDROLOGYDE: 1942 Machine learning;
- INFORMATICS