Validation of calibration strategies for macroscopic traffic flow models on synthetic data

We analyze two calibration approaches for parameter identification and traffic speed reconstruction in macroscopic traffic flow models. We consider artificially created noisy loop detector data as our field measurements. Due to the knowledge of the ground truth calibration parameter, we can give a sound assessment with respect to the performance of the considered methods. Our analysis shows that, in the proposed setting, the first order traffic flow model together with the proposed Kennedy O'Hagan approach performs better in reconstructing the speed traffic quantity than the other approaches.