Estimating Carbon Emissions
from Construction Sites
Construction activities contribute significantly to carbon emissions through heavy machinery usage, transportation, material processing, and land-use changes. This study integrates in-situ data collection, satellite-based observations, machine learning techniques, and geospatial analysis to estimate carbon emissions and evaluate their environmental impact. The approach aims to provide actionable insights for emission reduction strategies and promote sustainable construction practices.
Methodology
Data Collection
- In-Situ Measurements: Data is collected from construction sites to capture critical parameters such as:
- Carbon monoxide (CO)
- Carbon dioxide (CO₂)
- Particulate matter (PM2.5 and PM10)
- Satellite Data:
- Sentinel-2: Used for mapping land use and vegetation changes.
- Sentinel-5P: Provides atmospheric pollutant concentrations.
- High-Resolution Imagery: Identifies construction zones and associated activities.
- Supplementary Data:
- Air quality data from pollution control boards.
- Meteorological variables, including temperature, wind speed, and humidity, to account for environmental factors influencing emissions.
Machine Learning Techniques
Machine learning models are employed to analyze and predict emissions across various construction scenarios:
Random Forest Regression (RFR): Identifies relationships between variables such as site activity levels, vegetation loss, and pollutant levels, providing accurate and interpretable predictions.
- Gradient Descent Regression (GDR): Optimizes regression models by minimizing error and improving performance, especially for nonlinear interactions.
- Feedforward Neural Networks (FNNs): Captures intricate relationships between multi-source data inputs, ideal for analyzing complex datasets with diverse variables.
- Support Vector Regression (SVR): Adds robustness by effectively handling nonlinearity and noisy datasets, ensuring precise emission estimates.
- K-Means Clustering: Groups construction sites with similar attributes during preprocessing, enabling tailored modeling for specific site conditions.
Geospatial Analysis
- Spatial and temporal analysis of emissions is conducted to produce high-resolution emission maps.
- Temporal trends are assessed across different construction phases to identify periods of peak emissions.
Outputs and Insights
The study provides detailed spatial and temporal estimates of carbon emissions, including:
- Emission Hotspot Maps: Highlight areas with the highest emissions, enabling targeted interventions.
- Carbon Footprint Analysis: Accounts for direct emissions from machinery and indirect impacts such as vegetation loss and soil disturbance.
- Temporal Emission Trends: Offers insights into emission patterns across various construction phases, helping identify periods of highest impact.
Policy and Practical Implications
The findings enable policymakers, construction managers, and environmental agencies to:
- Identify High-Emission Activities: Pinpoint construction activities contributing the most to emissions.
- Implement Targeted Mitigation Strategies: Develop interventions to reduce emissions effectively.
- Promote Sustainable Practices: Support the adoption of low-emission technologies and strategies in construction.
Conclusion
By integrating field data, remote sensing, and advanced machine learning techniques, this study provides a scalable and accurate framework for estimating carbon emissions from construction sites. The approach not only identifies emission patterns but also supports the development of sustainable construction practices, aligning industry operations with climate mitigation goals.
Get in touch
Registered Office:
32/16 Cassia Road Shipra Sun City
Indirapuram, GZB, UP 201014
Email: info@skyxtend.com; 
+91 820-0577352
© 2025, SkyXtend. All rights reserved.
Made by ShilpCurrie