2.2 Data acquisition and preparation
This study considers both spatial and non– spatial data and
they have been acquired from a variety of sources (Table 1). Satellite
data includes Sentinel 2A, RapidEye, World View– 2,
multi– date UAV and a digital elevation model from Shuttle
Radar Topographic Missing (SRTM). Sentinel images represent winter
season and pre– and
post– influx of Rohingya situations. Vector data includes
administrative boundary and location of camps, obtained from Survey of
Bangladesh (SoB), BFD and UNHCR. In addition, human trails are derived
from the UAVs and Google Earth imageries and population data of each
camp was obtained from FCN– UNHCR and NPM– IOM. In
addition, a field survey was conducted in February 2018 to understand
state of the environmental condition. Observational technique along with
photographic method was accepted, during the field works, to support
satellite– based evaluation of forest function loss in the
study area.
Sentinel 2A imageries are first geometrically corrected and a root mean
square error (RMSE) of <1 pixel is accepted. Atmospheric
correction is then carried out with SEN2COR toolkit to convert Top of
Atmosphere (TOA) value to surface reflectance (Cleverset et al., 2017;
Quintano et al., 2018). A Universal Transverse Mercator (UTM) system
with 46N is used to project spatial datasets and then clipped to the
study area boundary.