A Generic Global-to-Local Quantitative Transformation Model (GGTLQTM) for Modeling Socioeconomic Indicators From DMSP-OLS Nighttime Light Imagery

Abstract

To expand the application range of Defense Meteorological Satellite Program Operational Line Scanner (DMSP-OLS) nighttime light (NTL) imagery and fully utilize the value of such imagery in this letter of socioeconomic indicators on a small scale, this letter proposes a novel model, that is, the generic global-to-local quantitative transformation model (GGTLQTM), and a new comprehensive indicator of accuracy, that is, adjusted normalized cross-validation (ANCV). The consistency of the GGTLQTM, from its fundamental hypotheses to conclusions, was studied to discuss the reliability of the model. Thirteen county-level administrative regions of Wuhan, China, were selected as the research areas. It was found that the GGTLQTM can 1) obtain a relatively high accuracy by selecting an optimal linear or nonlinear model, that the quadratic polynomial is the optimal GGTLQTM assessed by ANCV; 2) be established without relying on strict hypotheses; and 3) detect universal (global) homogeneity and individual (local) heterogeneity. In letter, the proposed GGTLQTM not only expands the application range of DMSP-OLS NTL remote sensing but also proposes a new idea on the geoscience-scale transformation, which helps integrate geoscience and remote sensing.