Tracking the industrial growth of modern China with high-resolution panchromatic imagery: A sequential convolutional approach
Ethan Brewer, Zhonghui Lv, Dan Runfola
TL;DR
This study addresses the challenge of quantifying industrial-site development in data-sparse regions by applying deep learning to high-resolution panchromatic satellite imagery of 419 sites in China (2002–2021). It compares two primary approaches—Mask R-CNN to estimate total structural footprint and CNN-LSTM to predict site-level development over time—with VIIRS nighttime-light radiance as a baseline proxy. The MR-CNN–CNN-LSTM pipeline best tracks ground-truth trends, estimating site growth around $+4{,}084~\mathrm{m^2}$ on average, while single-band panchromatic input and VIIRS radiance show notable limitations for sub-km, site-scale changes (e.g., $R^2$ near 0 for NTL-based approaches). The work demonstrates that panchromatic imagery can yield quantitative site-level development estimates ($\approx 0.021~\mathrm{km^2}$ for area and $\approx 10~\mathrm{nW/cm^2 sr}$ for radiance) but highlights the need for multispectral data and more labeled sites to improve robustness and generalization for industrial-growth monitoring. Future directions include extending to multispectral datasets and expanding labeled site counts to enhance trend analyses and regional applicability.
Abstract
Due to insufficient or difficult to obtain data on development in inaccessible regions, remote sensing data is an important tool for interested stakeholders to collect information on economic growth. To date, no studies have utilized deep learning to estimate industrial growth at the level of individual sites. In this study, we harness high-resolution panchromatic imagery to estimate development over time at 419 industrial sites in the People's Republic of China using a multi-tier computer vision framework. We present two methods for approximating development: (1) structural area coverage estimated through a Mask R-CNN segmentation algorithm, and (2) imputing development directly with visible & infrared radiance from the Visible Infrared Imaging Radiometer Suite (VIIRS). Labels generated from these methods are comparatively evaluated and tested. On a dataset of 2,078 50 cm resolution images spanning 19 years, the results indicate that two dimensions of industrial development can be estimated using high-resolution daytime imagery, including (a) the total square meters of industrial development (average error of 0.021 $\textrm{km}^2$), and (b) the radiance of lights (average error of 9.8 $\mathrm{\frac{nW}{cm^{2}sr}}$). Trend analysis of the techniques reveal estimates from a Mask R-CNN-labeled CNN-LSTM track ground truth measurements most closely. The Mask R-CNN estimates positive growth at every site from the oldest image to the most recent, with an average change of 4,084 $\textrm{m}^2$.