PubAg

Main content area

Using multi-indices approach to quantify mangrove changes over the Western Arabian Gulf along Saudi Arabia coast

Author:
Li, Wenzhao, El-Askary, Hesham, Qurban, Mohamed A., Li, Jingjing, ManiKandan, K.P., Piechota, Thomas
Source:
Ecological indicators 2019 v.102 pp. 734-745
ISSN:
1470-160X
Subject:
Internet, Landsat, United States Geological Survey, algae, anthropogenic activities, artificial intelligence, biomass, breeding, climate, coasts, databases, decision support systems, fisheries, forest products, habitats, mangrove forests, normalized difference vegetation index, remote sensing, salt marshes, spectral analysis, surveys, tides, Persian Gulf, Saudi Arabia
Abstract:
Mangroves habitat present an important resource for large coastal communities benefiting from activities such as fisheries, forest products and clean water as well as protection against coastal erosion and climate related extreme events. Yet they are increasingly threatened by natural pressure and anthropogenic activities. We observed an inaccurate distribution of mangroves over the Western Arabian Gulf (WAG) which is a vital habitat and resource for the local ecosystem, according to the United Stated Geological Survey (USGS) mangrove database through spectral analysis. Change detection analysis is conducted on mangrove forests along the Saudi Arabian coast of the WAG for the years 2000, 2010 and 2018 using Landsat 7 & 8 data. Three supervised classification methodologies are employed for mangrove mapping, including Supported Vector Machine (SVM), Decision Tree (DT), referred to as Classification and Regression Trees (CART) and Random Forest (RF). CART’s accuracy was recorded to be >95% while other classifiers were >90%. The CART supervised learning classifier, mapping mangroves’ distribution and biomass using Google Earth Engine (GEE) online platform, indicates an overall increase in the northern Tarut Bay and Tarut Island, by 0.21 km2 from 2000 to 2010 and by 1.4 km2 from 2010 to 2018. The increase might be due to mitigation strategies such as mangrove breeding and plantation. It can be challenging to detect changes in certain regions due to the inadequate resolution of Landsat where submerged mangroves can be confused with salt marshes and macro algae. We employed a new method to identify and analyze submerged mangrove forests distribution via a submerged mangrove recognition index (SMRI) and Normalized Difference Vegetation Index (NDVI) in Abu Ali Island. Our results show the robustness of SMRI as an effective indicator to detect submerged mangroves in both high and medium spatial resolution satellite images. NDVI values differentiated submerged mangroves from tidal flats between Landsat 7 & 8 as well as during conditions of low and high tides. High resolution WorldView-2 image showed agreement of mangroves distribution with the SMRI and NDVI results.
Agid:
6352482