Accuracy and Output.md 2.1 KB
The reliability of the traffic‑volume figures you obtain from the “watershed” (or flow‑network) approach is strongly tied to how accurately the road‑condition classes are defined and applied.
Pavement condition as a key input – In most traffic‑flow models the surface quality is entered as a categorical variable (e.g., 0 = good, 1 = poor). The data you quoted from source 8 show exactly how such classifications are coded for pavement and speed‑limit levels, and these codes feed directly into the capacity or impedance functions that determine the “flow” through each link. If the pavement class is mis‑assigned (e.g., a deteriorated segment labelled as “good”), the model will underestimate travel resistance and consequently overestimate traffic volume on downstream links.
Other geometric and environmental factors – Source 5 lists additional surrogate variables commonly used in road‑traffic crash‑prediction models (RTCPMs), such as lane width, shoulder width, gradient, weather, lighting, etc. When these attributes are also grouped into discrete classes, any inconsistency or coarse categorisation propagates through the network model and reduces the precision of the volume estimates.
Propagation of classification errors – Because the watershed analogy treats traffic like water flowing from sources (intersections with main roads) to sinks (other main‑road intersections), an error in one segment’s condition class alters the “resistance” for all downstream paths. The cumulative effect can be substantial, especially in dense rural networks where many local links converge on a few major junctions.
Bottom line:
The more granular and objectively measured your road‑condition classes (pavement state, geometry, weather exposure, etc.), the tighter the correspondence between modeled impedance and real‑world travel behaviour, leading to higher accuracy of the derived traffic volumes at the intersections between local and national roads. If you rely on coarse or outdated classifications, expect larger deviations from observed counts.