A Micro-Analysis of Accessibility and Travel Behavior of a Small Sized Indian City: A Case Study of Agartala

Changing land use and lack of proper transportation planning is highly correlated with the implications of traffic congestion. Ample evidence exist that the degree of accessibility, the land use and the various opportunities available in an area, represents one of the most effective tools available for dealing with various kinds of issues related to the sustainable development of any town or city. Agartala is one of the small cities in north eastern part of India which is growing rapidly in a short span. As a result there is a need for efficient public transport infrastructure and policies which have to be adopted in order to shift the dependency on motorised private vehicles to transit facilities and other non motorised transport modes. In this study an attempt has been made to link few aspects of travel behaviour i.e. choice of motorized/non-motorized modes and trip length changes with land use, accessibility and to find the effect of travel behaviour on the same. This study quantifies the effect of accessibility of various opportunities available, in the form of cumulative and gravity indicesusing geographic information system. In both the cases it was found that accessibility to opportunities along with socioeconomic parameters; significantly affects the travel behaviour parameters.


C. Accessibility parameters
Since this study concentrates on the relationship between urban form, accessibility measures and choice of the people, the study focuses on the integral measures of accessibility derived from the residential locations of households. Handy and Niemeier (1997) state there is no best suited approach in connection with measuring accessibility [6]. A few of them relate to degree of attractiveness, size or areal extent of zones, origins and destinations, and others are function of time and cost known as the impedance factor. Out of the various measures of accessibility two trips based measures namely cumulative and gravity measures have been used in the evaluation process on the basis of certain criteria based on specification, calibration and interpretation. The study includes the comparison of these two measures of accessibility and the suitability of the same. The entire study area has been divided into transport analysis zones (TAZs) based on the municipal wards and integral measures of accessibility have been calculated. Cumulative opportunity measures and gravity based measures for all trip purposes within an arbitrary buffer zone of 1000 meters from each household location have been measured. Since most of the trips were found to be ending within a distance of 1000m, thus 1000m was considered as the cutoff region. The attractiveness portion of accessibility was taken as the entire landuse area of each buffer zone expressed as a proportion of the entire area of a buffer created around each household location. An impedance function was also incorporated with this attractiveness component.
Where, in a buffer zone j,a j is thetotal land use area expressed as a proportion of the buffer area 1, * = 0, otherwise C ij is the measure of impedance between origin i and destination j,C ij * is the threshold region of 1000m buffer.

III. RESULTS
Many of the past studies have pointed out a strong influence of accessibility of opportunities on the VKT (vehicle kilometres travelled). As a proxy to the VKT, trip length was analysed for any significant effect of the land use mix and accessibility. The accessibility indices were found to have strong influence on individual's trip length for the trips. The coefficients, with negative sign, represent that the trip length will be more if there is less accessibility to opportunities. Several logit models were also framed to check the effect of the proposed indices on the utility of each travelling mode. A multinominal logit model and a binary logit model have been used to find the most probable mode choice based on the utility of the commuters. The attributes were used to check utility maximization of different modes. BIOGEME (BIerlaire's Optimization package for GEV Models Estimation) was used to construct the modelling framework. Table3 describes the result from binary logit model on motorized and non-motorized mode choice for work trips.Having driving license, less monthly income, non-motorised vehicle ownership, increases the utility of non-motorized modes; while having motorised vehicle ownership reduces the utility of non-motorized modes. An increase in cumulative accessibility was also found to be enhancing the utility of non-motorized modes. When accessibility measure entered the model, the improvement in the model was relatively better than the model with only socioeconomic variables. The gravity measure of accessibility was found to be insignificant in explaining its behaviour in this model. The final model contains cumulative accessibility index along with socioeconomic variables. The base model was estimated by considering only individual specific and mode related variables. Increasing accessibility increases the utility of the non-motorised modes. Increasing travel cost and travel time as expected significantly reduces the utility of the non-motorised modes.  Table 4 describes the result from mode choice model based on multinomial logit model for work trips. The first model was estimated considering the socioeconomic parameters only which shows that having lesser MTW (Motorised Two Wheeler) and car ownership increases the dependency on MThW (Motorised Three Wheeler) mode. It has been found that more female commuters opt for rickshaw as their mode of travel. The utility of MTW increases with the possession of driving license and higher income. Out of the two accessibility parameters, both were found to be significant and positive for walk and MTW modes; whereas travel time and cost parameters were not found to be significant enough in the multinomial model. Thus the final model contains socioeconomic and accessibility parameters, where addition of accessibility measures showed a significant improvement in the model.
IV. CONCLUSIONS In this study an attempt has been made to find the relationship between travel behaviour and the accessibility in the context of a small unplanned city like Agartala. In order to quantify travel behaviour, few factors affecting travel behaviour such as, destination accessibility (expressed as cumulative and gravity measures), distance to destinations in terms trip lengths and the choice of motorized/non-motorized mode were used. Utilities of various modes have been formulated using the socioeconomic, land use, accessibility and mode related attributes. The following are the important conclusions drawn out of the present work on interaction of travel behaviour and land use and accessibility to opportunities: The accessibility parameters were negatively correlated with the trip length for work trips. All other socioeconomic parameters showed desired correlation with trip length following the general hypothesis.
The binary model depicted that the cumulative accessibility index was well efficient in explaining the mode choice. In this case effect of gravity measure was found to be negligible.
The coefficients of accessibility are positive for non-motorized modes of transport which implies that the trip makers residing in the areas with better accessibility to their needs prefer non-motorized modes of transport.
From the MNL model estimated on the mode choice for work trips, it can be inferred that the trip makers residing in the areas with better accessibility to their needs prefer walk and MTW modes of transport.
Thus it can be concluded from the models estimated that the socioeconomic variables have significant effect on the utility of different modes. Along with them the proposed measures of accessibility incorporating the land land use mix parameter were found to be efficient in explaining the significance of the accessibility on travel behaviour.