Accurate estimation of traffic in intelligent transportation system applications, such as the advanced traveler information system and the advanced traffic management system, requires fixed location-based measurements, vehicle-based measurements, or both. Using both data sources is too expensive for most government agencies, especially in developing countries such as India, and also leads to issues related to installation and maintenance, especially on urban roads. The main drawback of vehicle-based measurements is the potential lack of participation because of privacy concerns; lack of participation would limit data collection to a sample of the population, primarily on public transport vehicles. The study aims to overcome such difficulties by using only location-based flow data for the estimation of spatial parameters, such as density and travel time. These parameters are difficult to measure or estimate on an urban arterial, especially under heterogeneous traffic conditions, because of lack of lane discipline and because of complex interactions among different vehicle types. The Lighthill-Whitham-Richards macroscopic traffic flow model discretized in both space and time was employed in the estimation scheme. The resulting partial differential equations were solved numerically with the finite difference formulation of forward-time backward-space. Both linear and exponential speed-density relationships were considered and incorporated into the macroscopic model. Linear and cubic spline interpolations of input flow values were compared. The estimated density was corroborated with the density obtained from input-output analysis. Estimated travel times were compared with manually observed travel times and travel times obtained from probe vehicles fitted with Global Positioning System devices.