The calibration is off, which I think distracted the reviewers a bit. Possibly because we didn't emphasize the methods like #3 , but there are a few comments on this.
Reviewer 1:
The transit mode share analysis is problematic. First, the calibrated mode split does not seem correct. ActivitySim transit shares are an order of magnitude greater than WFRC. If the authors could not calibrate ActivitySim more closely, that strongly suggests a difficulty in working with the ABM and worthy of inclusion in this paper. That said, these numbers do not seem consistent with the transit numbers in Table 2 where WFRC shows much higher CR trips.
Reviewer 2:
I'm surprised how divergent the models were related to transit ridership, an order of magnitude difference in ridership (Table 1). The authors have a brief discussion in the early part about calibration, but this seems almost disqualifying or non-credible.
Reviewer 3:
calibrating the submodels, particularly activity generation and scheduling, is one of the pain points for ABM deployment. Is information on typical calibration LOE requirements available?
The calibration is off, which I think distracted the reviewers a bit. Possibly because we didn't emphasize the methods like #3 , but there are a few comments on this.
Reviewer 1:
Reviewer 2:
Reviewer 3: