Exploratory ecological-stress screening for regional planning: a quality-controlled Landsat-based mRSEI analysis of Sakarya Province, Türkiye (2000–2025)

Abstract

Regional averages can conceal localized ecological deterioration in rapidly transforming peri-urban landscapes. This study presents a quality-controlled remote-sensing workflow for screening ecological stress across Sakarya Province, Türkiye, using six benchmark-date Landsat composites (2000, 2005, 2010, 2015, 2020 and 2025). After raster alignment and quality control, the Enhanced Vegetation Index (EVI) was excluded because the 2025 layer differed substantially in spatial resolution and value distribution from the preceding series. A modified Remote Sensing Ecological Index (mRSEI) was therefore constructed from pooled, commonly normalized NDVI, NDMI, inverse NDBI and NBR layers using one fixed-loading principal component model. The first component explained 93.75% of pooled variance. Mean mRSEI increased from 0.609 in 2010 to 0.688 in 2015 but declined to 0.595 in 2025; the sensor transition and six-date design prevent this pattern from being interpreted as an unambiguous ecological trajectory. Pixel-wise Sen slope and Mann-Kendall tests indicated significant positive, significant negative and non-significant mRSEI trends over 4.83%, 1.10% and 94.08% of the common valid area, respectively, before multiple-testing correction. No trend remained significant after Benjamini-Hochberg false-discovery-rate control. A conservative Exploratory Ecological Stress Synthesis (EESS) classified 97.67% of the valid area as Low, 1.54% as Moderate, 0.57% as High and 0.22% as Very High stress. Because no independent field, cadastral or land-cover validation was available, these classes are modelled screening outputs rather than validated vulnerability zones. They can guide subsequent stratified field sampling, high-resolution image interpretation and planning review, but should not be used directly as statutory zoning boundaries.