Methodological Comparison of Google Earth Engine and ArcGIS for ESRI 10 m Land Cover Analysis in Baghdad, Iraq

This study presents a comparison between the Google Earth Engine platform and the ArcGIS Pro software, two distinct methodological approaches within a rigorous comparative framework for ESRI 10-meter resolution land cover datasets derived from the Sentinel-2 satellite for Baghdad Governorate, Iraq, from 2017 to 2023. The study area is located between 33°00′ and 33°45′ north latitude and 44°00′ and 44°45′ east longitude, with a total area of approximately 5,132 km². The ESRI 10 m Land Cover product uses supervised deep learning classification (U-Net). GEE indicated that built-up areas increased by 40.7% from 1,096.99 km² in 2017 to 1,543.45 km² in 2023, while cropland area declined from 2,621.95 km² to 2,587.11 km² (-1.3%). The Pearson correlation coefficient value of 0.9987 indicates a very strong correlation; the accuracy of the data used (ESRI 10 m Land Cover) exceeds 85% globally in terms of classification. The main methodological differences between these two platforms are cloud-based processing versus local desktop processing (manual), procedures for handling scales, and methods for resampling pixels. Statistically, both datasets returned suitable, reliable, and statistically supportable data, and therefore the selection of one of these solutions is based on analytical requirements, performance, and resource usage.