Assessing spatiotemporal eco-environmental vulnerability by Landsat data

An indicator of quantifying eco-environmental vulnerability was established by synthesizing 12 variables, mainly retrieved from satellite data with incorporation of analytical hierarchy process (AHP). Six vulnerability levels of potential, slight, light, medium, heavy, and very heavy were graded to depict changes of vulnerability over temporal and spatial scales. The proposed approach was employed to study spatiotemporal eco-environmental vulnerability with Landsat data acquired in 1989, 2003, and 2014 for the Thua Thien – Hue Province, Vietnam. Over the time periods of 1989–2003 and 2003–2014, both heavy and very heavy vulnerability levels exhibit an increasing trend in both magnitude and spatial size: The former raised from 5.9% in 1989, to 7.9% in 2003, and 15% in 2014; and the later increased from 1.2% in 1989, to 3.2% in 2003, and 7.3% in 2014. Both levels mainly appeared on urbanized area, bare land, semi-bare land, agricultural land, and sparse forests. In contrast, there was a significant decline in potential vulnerability level with 36.4% in 1989, 30.9% in 2003, and 19.2% in 2014, while the remaining vulnerability levels slight, light, and medium fluctuated slightly, increased in 2003 and decreased in 2014. Supporting reasons for such changes include: (1) deforestation, agriculture intensification, construction of three hydro-electric projects during the period 2003–2014; and (2) significant expansion of urbanized area leading to differences in thermal signatures in urban areas as compared with rural areas. The findings demonstrate that eco-environmental vulnerability is primarily exaggerated by anthropogenic activities through land use/land cover (LULC) changes and further enhanced by natural processes including disasters in the Thua Thien – Hue Province of Vietnam. The correlation between land surface temperature (LST) and Normalized Difference Built-up Index (NDBI) is found to be positively correlated with 0.87, 0.89, and 0.84 for 1989, 2003, and 2014, respectively. In contrast, LST-Normalized Difference Vegetation Index (NDVI) is found negatively correlated with respect to the spatiotemporal trend of environmental vulnerability with −0.81, −0.81, and −0.76 in 1989, 2003, and 2014, respectively. In addition, areas having potential, slight, and medium thermal environmental levels are decreased from 1989 to 2003 to 2003–2014. At the regional scale, increased anthropogenic activities through land’s modification have intensified the eco-environmental vulnerability in the study area. The currently proposed methodology is feasible for evaluating long-term eco-environmental changes processes by using remote sensing data, and valid for the other regions.