Visitor Data Sources and Visitor

The bulk of the visitor data came from the results of overnight and day-visitor travel surveys produced by NuStats (see “Tahoe_report_final_winter[summer].pdf” for more details) These surveys were taken during both the summer and winter seasons, and attempted to capture the travel behavior of non-residents in the region. Seasonal residents, a population that is modeled along with the overnight-visitors, were surveyed during the resident surveys. It was hard to define exactly what constituted a “seasonal resident” since they come in various forms (2nd home-owners, regular visitors, monthly renters, time share owners, etc.). In the end, the surveyors asked the interviewee if he/she considered him/herself a seasonal resident and if so the data was thus marked. The last data source was external station traffic counts, which were used to help determine the flow of visitors into and out of the region during the travel day.

Visitor Model Flow

The basic unit of visitors is the travel party. This is a group of people who travel together throughout the day, each enjoying the same activities. Almost all travel decisions are made at the travel party level, and therefore for this model, travel decisions made at a more granular level (e.g. individual or joint tours from the resident model) were not allowed. As much as possible, the visitor model was kept consistent with the resident model. It is therefore a micro-simulated, activity-based travel demand model. Due to data availability limitations, the overnight and day-visitor models are more sophisticated and detailed than the thru-visitor model. Furthermore, the survey data for the overnight-visitor model was more detailed than the day-visitor, so a large part of the tour models for the day-visitors was combined with the overnight-visitors to create more robust model estimation results.

Population Stratification

The overnight-visitor population is stratified by their stay-type, or the type of accommodation used during their stay. The reason for this is partly due to the observation that people staying at campgrounds make different travel decisions than people staying in a resort hotel than do people staying in a rental cabin for example. In addition, stratifying the population provides the user more control over the overnight-visitor population in a specific scenario.

There are six stay-type stratifications for the overnight-visitor population:

• Seasonal - For seasonal residents

• Hotel/Motel - For visitors staying in a hotel or motel

• Casino - For visitors staying in a casino-based hotel

• Resort - For visitors staying in a higher-end (non-casino) resort

• House - For visitors who are not seasonal residents but who are staying in some type of an attached or detached residence

• Campground - For visitors staying at a campsite

For each of these stay-types, the synthetic population model requires the user to specify the number of units that are occupied on the model day for each zone. These vacancies are then filled by visitor parties during the synthesis procedure. A description of what form the inputs must take and how the information is used to synthesize the population is described below.

For Hotel/Motel, Casino, Resort, and Campground stay-types, the number of available units per zone is a scenario-specific input. The other required input is the occupation percentage, by zone, for each of these stay-types. By multiplying these two values, the number of occupied units is determined. (As a note, the winter Campground occupation rate is globally set to zero, as winter camping rarely, if ever, occurs in the basin).

For the Seasonal and House stay-types, the number of available units is derived from the socio-economic data used in the resident model. That data includes both the number of available housing units and the number of (full-time) occupied units. The difference between these numbers provides the number of units available for the Seasonal and House overnight-visitor stay-types. The user inputs are then the percentages of available units that are filled, and the percentage of those units that are seasonal. From this the number of filled housing units can be determined.

Sample Source

The synthetic population consists of actual travel parties sampled from the overnight-visitor and resident surveys. This is analogous to the use of the PUMS records to generate the resident synthetic population. Because there were two sources for the population, and because the samples would be merged to form the full overnight-visitor population, the data in the sample from each source had to be identical. Furthermore, the day visitor population would also be merged with this population later in the model stream, so the limited data collected during the day visitor survey determined what data could be retained/used from all the surveys. The resulting travel party characteristics were used in the models:

• travel party size

• number of children (age < 18) in the party

• presence of an adult female

• stay-type of the party

• season during which the survey was taken (summer, winter, or seasonal for seasonal residents)

For the non-seasonal resident overnight-visitors, the overnight-visitor survey records are sampled. For example, when populating campgrounds with overnight-visitors, the sample set is all surveyed visitors that stayed at a campground. However not all of the stay-types had enough survey records to provide a robust sample set and were therefore combined with other stay-types. For the same reasons, stratification by season or location could not be made. Table below summarizes the sample source for each stay-type.

For seasonal-residents, the sample source is the residential household survey. NuStats was able to survey 229 households that self-identified as seasonal.

Population Synthesis

Given the number of units to be filled by zone and stay-type, and the sample records corresponding to each stay-type, the population synthesis is a straightforward procedure. For each stay-type unit to be filled in a zone, a travel party is randomly selected from the available population sample records for that stay-type. Along with the sample data described above, the stay location (zone) is recorded and will be used as that travel party’s origin/home TAZ.

Sample Source

As described above, the day-visitor synthetic population consists of a sample of actual travel parties from the day-visitor surveys. This is analogous to the use of the PUMS records to generate the resident synthetic population. As described earlier, the day and overnight-visitor populations are merged later in the visitor model stream, and this forced the requirement that the two populations needed to contain equivalent data. As such, the following variables are included in the day-visitor sample records:

• The travel party size

• The number of children (age < 18) in the party

• The season during which the survey was taken (summer or winter)

As a note, the presence of an adult female in the travel party variable included in the overnight-visitor population was not available from the day-visitor survey. However, this variable is not used once the two populations are merged, and therefore its elimination does not affect the model. From the day-visitor survey, 597 sample records were put together to form the day-visitor sample set.

Day-visitor Rates

To synthesize the day-visitor population, the number of day-visitors originating at each external station must be known. To determine this, external station counts (in each direction) for the base scenario and model results from the residential and external worker models are used. Subtracting the residential and external worker trips which use the external stations from the external station counts gives the amount of flow that the day-visitor and thru visitor trips must make up. During validation, the split between the day-visitor and thru visitor trips is determined and, given this, the specific number of day-visitors per external station is known for the base scenario. To make the day-visitors vary when running forecast scenarios, two assumptions are made:

1. That the day-visitors rise and fall at the same rate that overnight-visitors do; that is, that the number of day-visitors is a function of the number of overnight-visitors

2. That the distribution of day-visitors between the external stations is constant across scenarios

Assumption 2 is probably more controversial and less realistic than 1, but it is required as there is little to no information as to what drives the day-visitor external station distributions

Given the above assumptions, the number of day-visitors for each external station is calculated using a simple linear formula:

D_e = O\(\sigma\)_e

where D_e is the number of day-visitors coming in through external station e
O is the number of overnight-visitors
\(\sigma\)_e is the day-visitor rate factor for external station e

The values for \(\sigma\)_e are given in Tables below for the two seasons. Summing all of the rate factors shows that the number of day-visitors is roughly equal to 78% of the number of overnight-visitors for the summer, and 58% for the winter.

Population Synthesis

Given the number of day-visitor travel parties per external station, and the day-visitor survey sample records, generating the day-visitor synthetic population is a straightforward procedure. Essentially, for each external station, a travel party is randomly selected from the available day-visitor sample records until the required number of parties is achieved. Along with the sample data described above, the external station is recorded and will be used as that travel party’s origin/home TAZ.

Daily Activity Pattern Alternatives

For both the day-visitor and overnight-visitor daily activity pattern models, a set of possible daily activity patterns has been specified. These were selected from the patterns found in the overnight and day-visitor surveys. Each pattern begins and ends at the home or external station location (H) and can have at most one stop going to and one stop coming from the primary destination.

There are 4 main tour types - recreation, gaming, shopping, and other. Each pattern specifies how many tours (H x H x H = 2 tours), the primary purpose and the number of stops that occur in the party’s daily travel. For example, the pattern “HOHTGH” says that the travel party left home to participate in an other activity and went back home (HOH). Later the same day, the same travel party made a quick stop (T) on the way to a gaming activity (G) before returning directly home (H). Because day-visitors are not staying in the region, their daily activity patterns consist of only one tour. The abbreviations used in the patterns for each of the destinations are described in table below.

The daily activity pattern models are multinomial logit models where each pattern is an alternative. Information specific to both the travel parties and the patterns are considered during model prediction.

Destination Choice Sub-model

The destination choice model is a multinomial logit model in which each potential destination zone is an alternative. Each zone’s attractiveness is calculated from a utility function, where the utility consists of variables such as distance, stay-type, and area type.
To provide a measure of a zone’s attractiveness based on tour-specific characteristics, a size term is included in the utility expression. The size terms are stratified by tour type and are calculated as the natural logarithm (ln) of a sum of variables. Also included in the utility expression is the logsum from the mode choice model, which provides an index of accessibilities for a destination zone.

Time-of-Day Choice Sub-model

The time-of-day sub-model is a multinomial logit model in which start/stop hour pairs make up the alternatives. The earliest allowed start/stop time is 5:00 am (corresponding to the 5:00-6:00 hour), and the latest allowed is midnight (corresponding to the 12:00am-1:00am hour). As far as skim periods are concerned, the following definitions are used:

Mode Choice Sub-model

The mode choice model is a multinomial logit model in which each mode is an alternative. For overnight tours, the following alternatives are available:

• Drive

• Shuttle

• Walk to transit

• Drive to transit

• Non-motorized

The shuttle mode represents tour buses and commercial shuttles. Day-visitors are assumed to use the drive mode. The primary component of the mode choice model is travel time, which uses the same coefficient across all modes. For the modes that have costs associated with them (transit has fares, auto modes have operating costs), a value of time factor was estimated; this factor converts dollar costs into time, for which a utility can be calculated using the travel time coefficient.

Population Synthesis

The population synthesis for the thru-visitor population is very simple and closely tied to the day-visitor population synthesis. Essentially, the number of thru-visitors originating at each external station is indexed to the number of overnight-visitors in the region in the scenario. This implies that there is a linear relationship between the number of overnight-visitors and the number of thru-visitors on any given day. The formula for calculating the number of thru-visitors is:

T_e = O\(\lambda\)_e

where T_e is the number of thru-visitors coming in through external station e
O is the number of overnight-visitors \(\lambda\)_e is the thru-visitor rate factor for external station e

The rate factors were initially set such that the number of thru-visitors equals 2% of overnight-visitors. These factors changed slightly as the model was validated against external counts.

Each travel party only contains one identifying characteristic: its origin external station.

Thru-Visitor Destination Choice Model

The thru-visitor destination choice model is a very simple multinomial logit model where every external station is an alternative. Because so little information concerning the thru-visitors is known, only alternative specific constants were specified for the model. Distance was not included as a variable because there is no indication that a thru-visitor travel party is more inclined to prefer shorter (or longer) trips through the region.

Thru-Visitor Time-of-Day Choice Model

The thru-visitor time-of-day (TOD) choice model is a simple multinomial logit model where each skim period is an alternative. Since a thru-trip has no stops, the TOD choice sets the beginning and end time for the tour. Because so little information concerning the thru-visitors is known, only alternative specific constants were specified for the model.