A spatially and temporally comparable Remote Sensing Ecological Index via de-gradient national PCA for China’s resource-based cities
Year: 2026
Type: article
Abstract: Tracking the ecological quality of many cities over many years needs a Remote Sensing Ecological Index (RSEI) that stays comparable across space and time, yet standard constructions do not deliver one. A single national principal component analysis (PCA) is dominated by the between-city climate gradient, which loads heat and wetness with the wrong sign and inverts the index. We build a de-gradient national PCA (de-gradient PCA) whose weights come from the within-city component covariance under a fixed global normalization. The resulting index recovers the expected component signs where the single national PCA inverts them. It agrees with MODIS net primary productivity, an external ecological reference (city-mean r = 0.59 ), even after controlling for greenness (partial r = 0.50 ), and its weights are stable to city resampling and to leaving out any representative year. Applied to all 125 prefecture-level resource-based cities of China over 2003–2023, mean RSEI rose from 0.61 to 0.65 ( p < 1 0 − 4 ), with 83% of cities on a rising trend, faster than in 245 non-resource cities. Resource type set the ecological level more than development stage, and 13 cities in the arid northwest and coal belt stayed near 0.30. A spatial Durbin model identifies built-up land as the strongest local pressure and industrial structure as a persistent negative regional spillover; random-forest importance corroborates the built-up ranking.
Authors: Yedong Chen, Rongbing Xu, Sihao Gu, Zixuan Li
Language: English
Subfield: Global and Planetary Change
Field: Environmental Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $3,550
Award: G4536126549
OpenAlex ID: W7235703283
