Space to Policy: Scalable Brick Kiln Detection and Automatic Compliance Monitoring with Geospatial Data

TL;DR

The paper addresses the challenge of unorganized brick kilns contributing to air pollution by proposing a scalable ML pipeline that detects and classifies kilns across five Indo-Gangetic states using free, moderate-resolution Planet imagery. It generates a geo-located dataset of 30,638 kilns by technology (CFCBK, FCBK, Zigzag) and validates detections against state surveys, achieving strong correlations (e.g., r = 0.94 with UPPCB data) and enabling automated compliance analysis with distance-based and technology-based rules. The study couples detection with emission estimation and source apportionment, finding brick kilns contribute about 8% of PM2.5 in the Delhi airshed and exposing over 30 million people within 800 meters of kilns, highlighting substantial policy-relevant exposure. The framework relies on open data and cost-free tools, demonstrating a replicable approach to inventory, compliance validation, and policy assessment that can inform more inclusive and effective environmental regulation while considering worker livelihoods.

Abstract

Air pollution kills 7 million people annually. The brick kiln sector significantly contributes to economic development but also accounts for 8-14\% of air pollution in India. Policymakers have implemented compliance measures to regulate brick kilns. Emission inventories are critical for air quality modeling and source apportionment studies. However, the largely unorganized nature of the brick kiln sector necessitates labor-intensive survey efforts for monitoring. Recent efforts by air quality researchers have relied on manual annotation of brick kilns using satellite imagery to build emission inventories, but this approach lacks scalability. Machine-learning-based object detection methods have shown promise for detecting brick kilns; however, previous studies often rely on costly high-resolution imagery and fail to integrate with governmental policies. In this work, we developed a scalable machine-learning pipeline that detected and classified 30638 brick kilns across five states in the Indo-Gangetic Plain using free, moderate-resolution satellite imagery from Planet Labs. Our detections have a high correlation with on-ground surveys. We performed automated compliance analysis based on government policies. In the Delhi airshed, stricter policy enforcement has led to the adoption of efficient brick kiln technologies. This study highlights the need for inclusive policies that balance environmental sustainability with the livelihoods of workers.

Figures (11)

• Figure 1: Satellite view of brick kilns with bounding boxes. CFCBK is Circular Fixed Chimney Bull’s Trench Kiln, and FCBK is Fixed Chimney Bull’s Trench Kiln. We use Esri's high-resolution basemap (zoom level 17-19, 1.19-0.3 m per pixel) to create geo-referenced bounding boxes and hand-validate the predicted bounding boxes. We use Planet's moderate-resolution imagery (zoom level 15, 4.77 m per pixel) to train object detection models. Planet Imagery © 2024 Planet Labs Inc. Esri imagery © Esri, TomTom, Garmin, Foursquare, METI/NASA, USGS.
• Figure 2: Visualization of geographic regions for initial data annotation. The blue boundaries are the regions for annotation. The green boundaries denote the cities. The red dots denote the locations of brick kilns.
• Figure 3: Illustration of the labeling process. We first create a grid over the labeling region with 1 km$^2$ grid cells, as shown in (a). Blue boxes are the grids, and the red line is the boundary of the region of interest. Then, we zoom into a grid cell, put an OBB around kilns, and select the type of the kiln as shown in (b). Right menu buttons from the second button onward are used to draw, edit, move, and rotate an OBB in that order.
• Figure 4: A few samples of predicted brick kilns of each brick kiln category from our model. The first, second, and third rows show samples of CFCBK, FCBK, and Zigzag in that order. Imagery © 2024 Planet Labs Inc.
• Figure 5: Brick kiln locations across five states of the Indo-Gangetic Plain, India. CFCBKs, which use a relatively old technology, are most prevalent in Uttar Pradesh. FCBKs are present in all 5 states, but their presence is strong in eastern Uttar Pradesh. Zigzag kilns are also present in all states and are densely located in Bihar and the area close to Delhi National Capital Region (Delhi-NCR).