# Monitoring environmental impacts of a designated aquaculture area in the Karaburun Peninsula using Google Earth Engine

## TL;DR

This study uses satellite data and cloud computing to assess the environmental impact of aquaculture in the Karaburun Peninsula, finding no significant changes but highlighting the method's potential.

## Contribution

A novel framework combining DiD causal inference and MODIS satellite data in Google Earth Engine for monitoring aquaculture impacts.

## Key findings

- No significant differences in Chl-a or POC were found between aquaculture and control sites.
- Human-impacted coastal areas showed higher surface parameter concentrations.
- The method's limitations were defined, emphasizing the need for higher-resolution sensors.

## Abstract

Satellite-based monitoring of aquaculture impacts remains constrained by the absence of standardized, reproducible methodologies capable of capturing long-term environmental dynamics. This study introduces a novel framework that integrates Difference-in-Differences (DiD) causal inference with multi-decadal Moderate Resolution Imaging Spectroradiometer (MODIS) satellite data and Google Earth Engine (GEE) cloud computing to evaluate aquaculture-related changes in coastal ecosystems. Using 20 years of satellite observations (2002–2022) from the Karaburun Peninsula, İzmir, Türkiye, we compared three representative sites: an aquaculture zone, a coastal area influenced by human settlements, and an offshore reference site with minimal anthropogenic activity. The human-impacted coastal site consistently exhibited the highest concentrations of surface parameters, reflecting dominant background anthropogenic influences. However, DiD analysis revealed no statistically significant differences in chlorophyll-a (Chl-a), particulate organic carbon (POC), or other parameters between the aquaculture and control sites, indicating that potential aquaculture-related effects remained below the detection threshold of the 1 km MODIS resolution. Despite these null results, the study demonstrates the feasibility and limitations of combining causal inference and cloud-based remote sensing for aquaculture monitoring. This methodological integration provides a scalable, cost-effective, and transferable framework for detecting and interpreting environmental change across large spatial and temporal domains. By defining the sensitivity limits of satellite-based detection, this work lays a foundation for future applications that merge high-resolution sensors, in-situ validation, and process-based modeling in sustainable aquaculture management.