Abstract:
Mangroves provide numerous ecological and biophysical services in the tropics and subtropics that
support flood regulation, carbon sequestration, and reducing erosion from storm surges. Remote
sensing satellite imagery provides valuable information for mangrove mapping and monitoring. The
objective of this study is to detect the spatio-temporal changes in mangroves in Ampara District from
2004 to 2019 based on Landsat data. A semi-automated image classification technique was used to
delineate and detect changes of mangrove vegetation in the Ampara District from 2004, 2009 and
2019 using Landsat 5 and 8 images. The multi-index approach was constructed using: (i) water
masking using Normalized Difference Water Index (NDWI), (ii) mangrove detection using red and
shortwave infrared (SWIR), SWIR and near-infrared (NIR) band ratios, Normalized Difference
Vegetation Index (NDVI), (iii) mangrove classification using Principle Components Analysis (PCA)
and an unsupervised classification. The historic Google Earth imagery was used to validate the
classified mangrove habitats. The results estimated that the total mangroves in Ampara District were
424 ha in 2004, 355 ha in 2009, and 569 ha in 2019. The total mangrove habitat which was estimated
through available land-use/cover maps was 770 ha. In addition, habitat suitability of mangroves for
current and future (year 2050) climate change scenarios was mapped using a maximum entropy
(MaxEnt) model and bioclimatic variables. The current MaxEnt was resulted in 11% area in high
habitat suitability (H) and a moderately suitable (M) class in each. While the suitable habitat
projection for the year 2050 was 11% (H) and 16% (M). In conclusion, a loss of mangrove was
observed five years later in tsunami, and a gain of mangrove was occurred after 15 years resulting in
best land management practices.
