Abstract Wetlands provide necessary ecosystem services, such as climate regulation and contribution to biodiversity at global and local scales, and they face spatial changes due to natural and anthropogenic factors. The degradation of the characteristic structure signals potential severe threats to biodiversity. This study aimed to monitor the long-term spatial changes of the Göksu Delta, a critical Ramsar site, using remote sensing techniques. It seeks to analyze the relationship between these changes and land surface temperature (LST) and predict future land use patterns through machine learning (ML) methods. In this context, the normalized difference vegetation index, modified normalized difference water index, normalized difference bareness index, and normalized difference moisture index remote sensing spectral index analyses and LST maps were generated using Landsat 8 Operational Land Imager (OLI) satellite imagery for 1985, 2000, 2015, and 2023. Kappa accuracy assessments demonstrated a high level of agreement between the generated maps and ground truth data. Pearson correlation analysis was used to assess the consistency of the relationship between spectral index analyses and LST, revealing a statistically significant correlation at the 0.01 level. The study revealed that Lake Akgöl lost 58.85% of its water body over the 38 years of monitoring the delta. This loss was primarily attributed to increased LST and human activities. The land use land cover model for the year 2031, developed using artificial neural networks and cellular automata from ML methods, projected a 7.50% decrease in total water bodies, a 46.94% reduction in vegetated areas, and a 36.85% increase in nonvegetated areas. In conclusion, it was emphasized that the adverse land use trends within the Göksu Delta are expected to persist, degrading its ecosystem services and values. In this context, the study's findings can be utilized to identify strong strategies for protecting the delta.
Read full abstract