速度變異在運輸研究扮演關鍵角色，例如1)研究顯示速度及速度變異是交通事故主要肇因；2)速度變異可量測旅行可靠性，因此被用於價值定價(Value Pricing)及管理車道(Managed Lane)運轉分析；3)速度變異為影響機動車輛排氣量的因子；4)速度變異可呈現交通異質性，反映量化與質化績效。然而，速度變異並非自動化交通資料蒐集設施的標準輸出項目，以致速度變異的基礎資料庫闕如，且巨觀車流研究多偏重平均速度、密度(或占有率)、流量等三項傳統車流參數，忽略速度變異的重要性以及作為相關研究之輸入變項的潛力。本計畫目的包含 1)瞭解速度變異的特性、2)連結速度變異與傳統車流參數、3)創新應用速度變異於兩項領域。本研究以近25萬筆高速公路車輛資料構建的資料庫為基礎，該資料庫含逾400組5分鐘間距之速度變異、平均速度、占有率、流量，其中速度變異採最常見的兩種指標：速度變異係數及速度標準差。在建立速度變異與傳統車流參數的關係後，將實證關係一般化，並應用於高速公路服務水準設定：x%的車流具有不高於y%的自由流旅行時間，x%反映可靠性而y%反映機動性；另一應用為美國環保署空汙模型MOVES之速度分布輸入變項，速度變異可改善既有模式過於簡化之速度分布估算。
Speed dispersion plays a key role in various transportation research. For instance, speed dispersion hazards traffic safety and studies have shown that “not only speed but speed dispersion kills.” Second, speed dispersion that measures travel reliability has been applied to value pricing studies as well as managed lane performance evaluation. Third, speed dispersion is a factor for vehicle emissions assessment. Moreover, speed dispersion reflects heterogeneity which would affect quantitative and qualitative traffic performance. However, speed dispersion is inaccessible due to absence from the standard output of automated data collection devices, and relatively sparse and incomplete studies compared to those on the fundamental traffic flow parameters (mean speed, density or occupancy, and flow). Speed dispersion is neglected and does not reach its full capacity of being a vital research input. The research objectives are to 1) identify the general characteristics of speed dispersion, 2) build connections between speed dispersion and the fundamental traffic parameters, and 3) apply speed dispersion to two fields in which traffic heterogeneity serve as a leading component. This research utilized a dataset of nearly a quarter million of freeway vehicle records, or over 400 observations of the fundamental traffic flow parameters and two speed dispersion measures－coefficient of variation of speed and standard deviation of speed. Their relationships were examined and built, supporting applications of speed dispersion that bring new concepts to the current practice. First, freeway level of service from A through E can be presented in the context of “no more x% of the vehicles with travel time up to y% greater than the free flow condition.” Such a measure not only reflects mobility (y%) and reliability (x%), but also avoids the vague descriptions associated with each service level in the current HCM. Second, the relationships can be applied to estimate speed distribution for the MOVES mobile source air emission model. The MOVES model’s approach limits the distribution in two speed bins, results in unsupported speed dispersion. A revised approach was thus proposed.