Effects of Extreme Climate Shifts to Pavement Infrastructure in Tennessee

Details

• Creators:
  Onyango, Mbakisya ; Fomunung, Ignatius ; Bathi, Jejal Reddy ; Owino, Joseph ; Liang, Yu ; Otieno, Maxine ; Vanga, Maneesha
• Corporate Creators:
  University of Tennessee at Chattanooga
• Corporate Contributors:
  Tennessee. Department of Transportation ; United States. Department of Transportation. Federal Highway Administration
• Subject/TRT Terms:
  AASHTOWare PMED Alternatives Analysis Climate Change Data Collection Forecasting Infrastructure Machine Learning Maintenance Pavement Performance Temperature Transportation Departments

  More +
• Publication/ Report Number:
  RES 2023-11
• Resource Type:
  Tech Report
• Geographical Coverage:
  United States ; Tennessee
• Edition:
  Final Report: September 2025
• Corporate Publisher:
  Tennessee. Department of Transportation
• Abstract:
  Resilient pavement infrastructure is a dream of every transportation agency. Many factors affect the health of pavement infrastructure to include extreme climate shifts. This study evaluated the effects of climate shifts to pavement infrastructure in Tennessee. The study used AASHTOWare PMED software to predict distresses on selected pavement sections in Tennessee using historical MERRA climate data (or baseline) and machine learning projected climate scenarios. Results showed that pavement sections of the baseline scenario performed at an acceptable level to the end of design period. Climate data projections used machine learning models to predict five climate inputs: temperature, wind speed, percent sunshine, precipitation, and humidity from 2024 to 2044. NeuralProphet and LSTM models were selected for the study. However, the models did not capture peak temperatures accurately. A hybrid model, based on a Variational Autoencoder (VAE) with LSTM layers, gave more accurate results capturing peak temperatures. The VAE compressed the large climate dataset into a smaller representation to capture seasonal and temporal patterns, and the LSTM encoder learned the sequential climate data and reconstructed the values while keeping these critical weather patterns. Comparative analysis between historical and projected climate data files indicated that the projected climate data predicted distresses that were higher than historical climate data; however, the difference was not statistically significant and did not exceed the distress threshold at the end of design period. This indicates that there is no immediate need to change design and maintenance parameters for Tennessee, but close monitoring of weather events and improving severely distressed sections may be required. Moreover, historical climate data can still be successfully used to design pavements. Tennessee maintenance data showed a state average PSI of 3.35 and strong negative correlation of 68% between PSI and IRI, meaning PSI decreases with increase in IRI.
• Format:
  PDF
• Funding:
  RES 2023-11
• Collection(s):
  Federal Aviation Administration
• Main Document Checksum:
  urn:sha-512:bd5b16a5d50962db57fc25a0c05d146fc7da87c86f1d1a30160f8a0d58df925fe95a08e8132b5abb7d0aa4c2f2bb60cf8c6eb311432953913fa3d92ade49a97a
• Download URL:
  https://rosap.ntl.bts.gov/view/dot/89743/dot_89743_DS1.pdf
• File Type:
  [PDF - 20.63 MB ]