Forecasting Summary Report: North Central Texas General Aviation Airport Activity. February 2012

Transcription

1 Forecasting Summary Report: North Central Texas General Aviation Airport Activity February 2012

2 FORECASTING SUMMARY REPORT TABLE OF CONTENTS A. GLOSSARY... 3 B. PURPOSE... 4 C. BACKGROUND... 6 D. DATA REQUIREMENTS AND FINDINGS...10 E. FORECAST MODEL STRUCTURE...29 F. RESULTS...38 G. COMPARISONS OF FORECAST DATA...39 H. CONCLUSION

3 EXHIBITS Exhibit 1 Forecast Based Aircraft Exhibit 2 North Central Texas Planning Areas Exhibit 3 System Plan Household Population Forecasts Exhibit 4 System Plan Household Population Forecasts (Annual Compound Growth Rate) Exhibit 5 System Plan Employment Forecasts Exhibit 6 System Plan Employment Forecasts (Annual Compound Growth Rate) Exhibit 7 Landing Facilities in North Texas Exhibit 8 System Plan Forecast Airports Exhibit 9 Examples of T-Hangars and Conventional/Box Hangars: Exhibit 10 Example Airport Hangar Analysis Exhibit 11 Example Hangar Space Correlation for NPIAS Airports Exhibit 12 Example Hangar Space Analysis Results Exhibit 13 Example Population Analysis (GIS) Exhibit 14 Example Population Analysis Results Exhibit 15 Regional Airport Hangar Rental Costs Exhibit 16 Regional Airport Office Space Lease Rates Exhibit 17 Regional Aviation Fuel Prices Exhibit 18 Analysis of Forecast Airport Locations and Characteristics Exhibit 19 Registered and Verified Based Aircraft Analysis Exhibit 20 Comparison of Forecast Model Methodologies Exhibit 21 Based Jet Aircraft Model Subregional Results Exhibit 22 Travel Time Distribution from Registration Address to Airports Exhibit 23 Distribution of Travel Distance from Registration Address to Airport Exhibit 24 Based Non-Jet Aircraft Model Results Exhibit 25 Based Non-Jet Aircraft Model Subregional Results Exhibit 26 Local Operations Forecast Results Exhibit 27 Subregional Local Operations Model Performance Exhibit 28 Itinerant Operations Forecast Results Exhibit 29 Subregional Itinerant Operations Model Performance Exhibit 30 Subregional Based Non-Jet Forecasts Exhibit 31 Subregional Local Operations Forecast Exhibit 32 Subregional Itinerant Operations Forecast Exhibit 33 Comparison - Total Non-Jet Based Aircraft Exhibit 34 Comparison - Total Based Jets Exhibit 35 Comparison - Total Regional Operations APPENDICES Appendix A North Central Texas System Plan Subregions Appendix B Aviation Forecast Comparisons Appendix C Demographic Forecasts Simple Growth Rates (%) 2

4 A. GLOSSARY Term Description 5010 FAA Airport Master Record ACV Airport Community Value AMP Airport Master Plan AF FAA Aerospace Forecast BA Based Aircraft BEA Bureau of Economic Analysis FAA Federal Aviation Administration GA General Aviation GAMA General Aviation Manufacturers Association GIS Geographic Information System NCTCOG North Central Texas Council of Governments REMI Regional Economic Modeling, Inc. RIMS II Regional Input-Output Modeling System RPZ Runway Protection Zone RSA Runway Safety Areas System Plan Regional General Aviation and Heliport System Plan TSZ Traffic Survey Zone TxDOT Texas Department of Transportation U.S. United States 3

5 B. PURPOSE This summary serves to document the forecasting methods and approaches researched and developed to guide the North Central Texas Council of Governments (NCTCOG) Regional General Aviation and Heliport System Plan and subsequent recommendations. Specifically, the summary will provide details of the direction the project team took to meet the FAA s System Plan grant requirements in providing a nonstandard approach to forecasting regional general aviation system plan activity. The type and level of effort needed to produce a forecast depends on the purpose for which the forecast is being made. Typically, forecasts are not a final objective in and of themselves. Short-term forecasts support operational planning and personnel requirements at an airport. Intermediate and long-term forecasts are used to plan major capital investments. 20 year and forecasts beyond assess the need for additional airports or other regional aviation facilities. In general, forecasts are usually meant to reflect the demand for aviation services so planners can look to gain the appropriate supply, in terms of infrastructure, to meet the ensuing demand. It is important to realize, however, actual activity is driven not just by demand but by the interaction between the supply and demand of aviation services. Determining these interactions is complex and historically forecasts have struggled to accurately predict outcomes driving the need for a new general aviation activity forecasting model/method predicated on the realization of the inherent difficulties in: 1) Gathering accurate input data as the foundation for the forecast. 2) The resultant variations in the forecast and the actual realized numbers over time. 3) The increasing importance of general aviation activity forecast results to the investment and development decisions made by airport sponsors and airport communities. 4) Enhancing the accuracy of general aviation forecasts in light of the aforementioned and other challenges faced in the forecasting methodologies. Some of the uses and applications of the System Plan forecasting efforts include the following: Airside Development Planning: The forecasts will indicate the future need for airside infrastructure. The length and strength of the runways and taxiways reflect the capabilities of an individual airport and strongly influence the ability to support future airport activities. Part of the dilemma in forecasting is the determination of the extent that existing facilities decide the current activity of an airport and to which future facilities and plans drive future activity and migrations to that particular airport. Navigational aids at an airport also impact the safety and ease of access into individual airports. It is postulated that there is a valid forecasting relationship between airport infrastructure and the activity the infrastructure will support and generate. Landside Development Planning: The basing and operation of aircraft at an airport drives more than just a place to land, resulting in the need to develop ramp space, fueling facilities, maintenance services, and a multitude of other support functions such as a place to eat, refresh, relax, and rent a car. Again, while activity drives facilities, facilities also influence activity which requires further examination in the development of a more accurate aviation activity forecasting model. 4

6 Financial Side Planning - Airport Business Planning: Various measures of aviation activity are directly or indirectly tied to the revenues and costs associated with operating an airport. As construction, operation and expansion of an airport requires substantial investments, it is important for an individual airport to forecast future demands for aviation services and assess the potential level of revenue and the need for future services. On a system-wide basis, accurate forecasts are needed for effective airport system planning and for the efficient provision of capacity. Bond issues and the use of public funds require an understanding and substantiation of the expected return on investments. One goal in the development of the new general aviation activity forecasting model was to better display the relative accuracy of the forecast; establish a risk factor to the validity of the future forecast. The forecasts will most likely be used to make some substantial investment decisions related to the development of an airport and ultimately, the airport community. The ability to calculate an anticipated return on an investment (and test it over time) could become a critical factor in making major airport investment decisions. NCTCOG staff and consultants created a model with region specific variables to allow for incremental monitoring of the System Plan forecasts and recommendations. Community-Side Planning: The relationship between an airport and its adjacent communities is an absolutely critical factor in the sustainability of an airport and its services to the community and the region. The ability of the surrounding communities to understand, believe in, and support the planned development of an airport is extremely important in maintaining the sustainability of the airport and in building community support. The need to protect the land around a community s airport to ensure its sustainability is unquestioned. However, the extent to which a community needs to protect the land around a planned airport expansion or airport-related development demands the evidence of facts and an endorsement of the forecasts for the future of activity at an airport to support the, many times, hard decisions that a community must make to protect its aviation asset. The forecasting model is a series of developmental steps with the goal of facilitating an understanding of the strategies and analysis therein. This summary includes the following: Background: A discussion of the factors that influence general aviation activity forecasting and additional information on how these factors are addressed in the forecasting model. Data Requirements and Findings: Findings associated with the development of general aviation activity in the region. A large amount of data was gathered and continues to be analyzed to assess relationships between various existing data points. Collection of data was on-going throughout the forecasting as new data sources become available and anomalies in the data resolved. Forecast Model Structure: Used to establish the working parameters of the new forecasting model including the following: 5

7 o o o o System Inventory The airports to be included in the forecasting effort are discussed in this section along with the rationale for their levels of inclusion through categorization. Data Accuracy, Availability, and Accountability Issues associated with data collection and maintenance associated with improving the accuracy and relevancy of the forecasting efforts are discussed. Flexibilities Flexibility built into the forecasting will allow the study teams to more easily address known components of change which may occur within the system over time. Model Construct This section offers a delineation of the model components. Several of the model components represent sub-models and are described in separate papers. Examples include the concept of Airport Community Value (ACV) and Airport Categorization. The heliport forecasting model and methodology is presented in the Vertical Flight Report and subsequent resources previously distributed to the Air Transportation Technical Advisory Committee. Results: Staff and consultants initially researched the possibility of creating a renewable forecasting model. This model was designed to integrate flexibility with recommendations, for supporting capital investment decisions; however, given time constraints combined with inherent modeling and data limitations, this was not possible. The forecasts stemming from the findings related to the nonstandard forecasting and data analyses used throughout the planning process is detailed in this section. Comparisons of Forecasts Data: FAA acceptance of an aviation forecast, whether at an airport or regional level, is subject to a variety of factors. Accurate data representation and proof of methodologies may need to be provided for further review if they lie outside of the parameters of the FAA s documented forecasting processes. Conclusion: The conclusion recaps the goals and work performed throughout the System Plan forecast development. Appendices: The appendices contain aviation data tables and data analyses collected in the model development process. C. BACKGROUND There are several accepted methodologies for preparing general aviation activity forecasts. Some of these methodologies are discussed below: Market Share Identifies the relationship between activities at an airport as a percentage of regional activity. Important assumptions to consider that may or may not hold true are that the presumed relationships between an individual airport and the region remain constant over time by default or that strategic planning may indicate a proactive change to that the relationship. Econometric This modeling method uses explanatory variables that influence or drive changes in demand or supply of aviation activity. These variables can often include macroeconomic and demographic factors, regulatory environment, infrastructure constraints or improvements, and technological changes. This method of modeling is 6

8 potentially very sound and powerful but lack of statistical validations and data inaccuracies can cause problems. Regression In regression analysis, the dependent variable may be related to numerous other variables. The independent, or explanatory, variables explain the estimated value(s). An example of a regression equation is to estimate operations related to based aircraft. The relationship is estimated using historic data for the independent and dependent variables. Forecasts of the independent variables are used in the regression equation to calculate forecast values for the dependent variable. Times Series This approach involves extrapolating existing data into the future and is usually based on current and past trends. The general philosophy is that past trends will identify future trends. Simulation This method of modeling provides forecasts based on the relationships between activity patterns within a region or system. Simulation modeling employs rules for flow and the effect of decisions and then develops results. This can then be used to assess the needs of each airport and cumulatively for the system. Combinations Most accepted forecasts will usually develop a methodology that combines several of the above methods and uses a weighting method to produce the final forecasts of activity. The traditional variables used in general aviation activity forecasting are regionally segregated population and employment numbers. These numbers are used by the regional planning agency to help guide capital improvement investment decisions on transportation. Other agencies and local governments also use this demographic information to aid decisions related to housing and commercial and industrial development. These decisions can also be supported with locally-adopted zoning ordinances and subdivision regulations that provide direction and timing for these investments. The forecasting of general aviation activity cannot be validated until the passage of time reveals the accuracy of the forecast and ultimately the accuracy and validity of the original numbers and assumptions used to generate a forecast. The 1991 Regional General Aviation System Plan provided forecasts that guided development of aviation and its communities over the last 16 years. A review of previous system plan forecasts data, see example in Exhibit 1, and their realizations are a part of the development process for the new forecasting model. Understanding the developments from 1990 to the present will highlight new (and likely stronger) interrelationships between the activities and the developments that have occurred. 7

9 Exhibit 1: Forecast Based Aircraft % Subregion Forecast Actual Difference Central 2,464 1, % East % North 978 1,183-17% South % West % Regional Total 4,332 3, % Source: NCTCOG 1991 Regional General Aviation System Plan and Current FAA 5010 The new general aviation and heliport forecasting model is designed to develop correlations among several factors known to have an influence on general aviation (GA) activity. These factors include: Economics Sustained growth in national, regional, and local economies increases demand for business-related activity. The FAA Aerospace Forecast (AF) projects economic growth for 2011 at 2.3 percent; future economic growth may be subject to downturns due to rising oil prices, credit markets, and international currency valuations. Regional economic forecasts are utilized as a foundation for regional GA forecasts, which are then modified with specific subregional influencing factors. General Aviation as a Business Tool Increasingly, GA aircraft are seen as business and productivity tools, particularly in light of limitations and delays of commercial air travel as well as the time value of the corporate/business traveler. According to the FAA AF, the hours flown in expensive and sophisticated turbine-powered aircraft (including rotorcraft) and turbine jets is projected to grow at an annual rate of 3 percent and 4.2 percent, respectively, over the forecast. Additionally, the 2010 General Aviation Manufacturers Association (GAMA) GA statistical handbook reports $4.9 billion in new airplane export revenue. Development of the new model attempted to draw a correlation between business aircraft and regional demographic data; household population, employment, and income. Airport Locations and Access The location and accessibility of an airport, in relation to regional growth and development, contributes to the identification of the level of activity that can be expected at an individual airport. Other factors include airspace and ease of access to infrastructure such as: o Controlled airspace o Approach instrumentation o Environment surrounding the airport o Nearby land-use compatibility It is important to note that varying travel times to regional airports is an indicator of the impact that ground access has on activity (current and forecasted) at regional GA airports. As the region s population increases, surface transportation congestion may increase which can affect GA airports success. 8

10 Airport Characteristics The availability and utility of airport facilities are other key factors in the allocation of activity to particular locations within the region. Factors like runway strength and length, approach procedures and minima, fuel service, and aircraft storage contribute to the level of activity likely to occur at an airport. The new forecasting model and its metrics are founded on the provision of these facilities and services as attractants for aviation activity occurring at one airport rather than another. Service Area Population and Employment The local review of population, employment, and income data along with related forecasts within an airport s service area offers indications of an airport s current and potential future roles. Fuel Prices Fuel has a global influence on aviation activity. Associated fuel costs have been seen to affect aviation development decisions on a regional level. As a part of the system planning process, the project team developed a survey to research the effects of fuel prices on regional aviation activity. Based upon the mixed responses from the survey, no significant correlation was found linking the availability of fuel and airports based aircraft and operations. Regional Pilot Population Pilots often select an airport based on proximity to their residence or place of business. The number of U.S. student and private pilots dropped below 600,000 in 2006 for the first time in several decades. However, forecasts by the Texas Workforce Commission, as seen in the North Texas Aviation Education Initiative, display a need for 440 new pilots annually from The new forecasting model collects and analyzes pilot registrations and based aircraft locations as an indicator of concentrations of aviation activity Risk Factors and Thresholds An important part of the forecasting process is the assessment of uncertainty. Forecasting inherently includes a degree of uncertainty that decision makers should consider when planning different strategies based on the ranges of assumptions built into individual forecasts. The new GA activity forecasting model offers an analysis and modification method to assess the relative risk inherent in the use of new forecasting results. This risk factor is currently being defined to address the impacts of forecast aviation activity information on planned regional airport developments and on anticipated return on investment decisions. The new model takes into consideration that the risk of any forecast grows exponentially with time. When decisions are made utilizing forecast information that is increasingly aged, the risks increase. The new model provides a method to minimize risks by decreasing the time between forecasts. The development of the new model also explores the opportunities to create and validate the concept of thresholds in the forecasting methodology so that threshold factors can reveal strategic points for further airport investments based on anticipated returns. The new forecasting tool will allow periodic updates and offer a method to incorporate the updated information into the airport capital improvement processes. Regional Variations North Central Texas includes characteristics that are both similar and different from other metropolitan regions around the country. The regional variations that may offer some influence in GA forecasting include weather, geography, state and local legislation, and commercial and residential development patterns. Such factors were assessed as part of this System Plan. 9

11 Relationships of Activity between Individual Airports and Regional Activity There are relationships between based aircraft and operations at individual airports and between regional airports (particularly those without control towers) which can be calculated. This will be challenged with the upcoming public introduction of reported based aircraft counts at individual airports, soon to be released by the FAA from its recent survey of based aircraft counts by tail number. The actual counts significantly differ from 5010 data that has been historically reported as the FAA s Master Record for airports. The information, maintained in FAA databases is used for record keeping purposes and in airspace studies. Some of the information on the 5010 is critical to aviation safety and is published in flight information handbooks and charts for pilot use. The new forecasting model provides a method to address new 5010 data to include changes in based aircraft numbers. Next Generation Air Transportation System (NextGen) NextGen was endorsed in 2003 by the U.S. Congress in the FAA Reauthorization Bill Vision 100 and was designed to look seriously into the development of a replacement to the existing airspace management system utilizing technologies that are available today. NextGen will provide the GA industry with the ability to operate more freely. The potential forecasting impacts of NextGen and its effects on airport capacity and future needs remain speculative at this point. The new forecasting model offers the ability to quickly adapt to changes as their impacts become reality. Fuel Sales and Activity The project team designed and distributed a survey with one of the concluding results showing that no significant correlation was identified between the availability of fuel at an airport and the number of operations and based aircraft. Regional Aviation Data Points The new forecasting model evaluates and compares regional data as factors in airport choice and GA activity. This includes, but is not limited to, travel times to local airports, proximity to major employers, pilot concentrations by zip code, airport amenities, and employment by classification/type. Ground Access to Regional Aviation Infrastructure Congestion on the ground via roadway networks can significantly impact aviation operations and efficiency. The System Plan Ground Access report covers details on how surface access, as a component of transportation, can impact the region s air passenger and cargo facilities. D. DATA REQUIREMENTS AND FINDINGS As the use of forecasted activity represents the consensus view of future demand, it is important to gain a foundational understanding of what should be projected as forecasting variables. For the System Plan forecasts, these requirements will be examined in terms of meeting accepted uses of forecast information as well as consideration for new uses and values. The FAA Advisory Circular on Airport Master Plans, AC 150/5070-6B, recommends specific forecast elements to be included. These are: Annual Operations Itinerant and local operations categorized as air carrier, air taxi and regional commuter, general aviation and military Annual Passenger Enplanements Air carrier and commuter passengers; further categorized as originating and connecting 10

12 Aircraft Types Categorized as based aircraft and aircraft mix (jet, multi-engine, piston, etc.); includes identification of the design or critical aircraft. Other Elements Domestic and international operations, annual instrument approaches, IFR and VFR operations, air cargo aircraft operations, helicopter operations, training (touch and go) operations, average passenger load factors, fuel volumes, general aviation passengers, and other items such as number of student pilots and hours flown or average seats per aircraft The development of the System Plan s GA forecasting model focuses on the collection of key pieces of data to better understand the relationships among the many factors influencing activity in the region. For System Plan purposes, the project team focused the forecast efforts on jet and non-jet based aircraft and local and itinerant operations for the forecast airports at a subregional level. The details of the aforementioned are outlined in the following pages but in summary it is assumed subregional GA local operations and non-jet based aircraft are closely associated, similarly to jet aircraft and itinerant operations. The uses and applications of the forecast data can be multifaceted. For a metropolitan area such as North Central Texas, forecast demographic data assists in the evaluation and potential need for NPIAS, public, and private airports in the region. It will also present data to aid in evaluating the service level of existing airports and their current FAA classification as a GA or Reliever Airport with associated additional amenities. The Metropolitan Planning Area (MPA), included in the NCTCOG travel demand modeling and demographics forecasts, contains all of Collin, Dallas, Denton, Rockwall, and Tarrant counties as well as Ellis, Hood, Hunt, Kaufman, Johnson, Parker and Wise counties. See Exhibit 2. 11

13 Exhibit 2: North Central Texas Planning Areas During the last two decades, the population in North Central Texas has been rapidly increasing and by 2035 North Central Texas is forecast to be home to more than 9.8 million people. Forecasts for household populations and employment, collected by county, have been compiled and analyzed to better understand the current projections for regional growth. See exhibits 3-6 as it applies to aviation growth. The source of the data that has been analyzed in this section is NCTCOG s official forecasts adopted in 2011, with the preparation of Mobility The most current NCTCOG forecast information has been aggregated by county to facilitate the development of the new GA forecasting model. The new forecasting model uses boundaries aligned as a result of the System Plan Subregion Analysis, described later in this report. NCTCOG s Mobility 2035 Plan (Mobility 2035) is a comprehensive, long-range, multi-modal plan to meet the mobility needs of the Dallas-Fort Worth area given existing fiscal constraints. NCTCOG s Executive Board approved the demographic data sets used in the development of Mobility 2035 and the Plan was approved by the Regional Transportation Council (RTC). It is important to note a few assumptions concerning the demographic data for the System Plan forecasting model. The forecast methodology called for many non-standard analyses that included demographic data inputs. The county level Household Population (HHPOP) and Employment (EMP) forecasts used in the model required defined service areas around the 12

14 proximity of each airport. These areas often times include counties that are not in the NCTCOG demographic forecast model. To fill these gaps in the model, the project team used data from the Real Estate Center at the Texas Transportation Institute (TTI) For those counties not in the NCTCOG model, the average annual population and employment growth rates for 2007, 2008, and 2009 were calculated based on the data available at the TTI Center for Real Estate Research. These rates from TTI were used to make assumptions on what a reasonable growth rate would be for these counties and the project team assumed these average growth rates will be experienced from 2007 until Note: Industry norms utilize a 1.5 percent annual growth rate that is considered normal, 3 percent is considered above average, and any growth rate above 3 percent is considered high. Therefore, many of the growth rates are set at 1.5 percent, assuming normal growth, with few at 2 percent. The forecasts for counties not in the NCTCOG model suggest the annual HHPOP and EMP growth rate will reduce from 2 percent in 2010 to 1.5 percent in 2040 (1.5/2 = 0.75). Therefore, growth rates between 2035 and 2040, for counties outside the NCTCOG modeling area, will be 75 percent of their growth, reported by TTI, between 2007 and Taking this into consideration, growth rates for every 5 years between 2010 and 2040 were interpolated. From this point the project team could develop HHPOP and EMP forecasts for the exterior counties based on the estimated growth rates for each 5 year period. Next the project team estimated what percentage of each of the counties is within the service area of forecasts airports and included any additional POP and EMP data from this analysis to the numbers in NCTCOG s demographic model forecasts. The results yielded forecast data in 5 year increments, utilizing assumptions in a practical and reasonable range for counties outside NCTCOG s modeling region. Overall the final forecast data for counties not included in NCTCOG s demographic model have a rather small affect on the final aviation forecast numbers as they contributed little to the overall total POP and EMP in the service area of each forecast airport. Appendix C demonstrates the relationship of forecast employment and population and simple annual growth rates for counties not included in the NCTCOG model. The income analysis was less straightforward as NCTCOG does not have an income forecasting model. For the purpose of the forecast, the assumption is that the incomes grow by the average national inflation rate with the average income from 2007 as a baseline for the methodology. Unlike the HHPOP and EMP calculations largely based at the county level, the income data inputs were calculated for each year in the service area of each airport. 13

15 Exhibit 3: System Plan Household Population Forecasts COUNTIES COLLIN 761, ,969 1,007,638 1,139,019 1,264,006 1,404,147 DALLAS 2,397,572 2,511,578 2,666,344 2,825,629 2,976,809 3,125,281 DENTON 625, , , , ,223 1,053,892 ELLIS 152, , , , , ,772 HILL 44,992 52,733 54,611 56,306 62,817 69,165 HOOD 64,427 72,569 74,016 75,477 86,717 97,800 HUNT 90, , , , , ,452 JOHNSON 163, , , , , ,060 KAUFMAN 95, , , , , ,511 PARKER 116, , , , , ,736 ROCKWALL 79,234 96, , , , ,560 TARRANT 1,785,206 1,944,211 2,151,696 2,378,257 2,604,119 2,823,526 WISE 66,908 75,078 80,726 85,720 90,853 95,621 NCTCOG MODEL TOTALS 6,444,454 7,083,282 7,755,830 8,455,364 9,175,782 9,902,523 EXTERIOR COUNTIES COOKE 40,151 43,094 46,130 49,249 52,439 55,687 GRAYSON 123, , , , , ,299 PALO PINTO 28,574 30,669 32,830 35,049 37,320 39,631 ERATH 37,079 39,796 42,600 45,481 48,426 51,426 SOMERVELL 8,164 8,962 9,804 10,688 11,611 12,569 NAVARRO 51,015 54,757 58,612 49,249 52,439 55,687 EXTERIOR COUNTY TOTALS REGIONAL TOTALS 288, , , , , ,299 6,732,947 7,393,122 8,087,708 8,796,576 9,539,326 10,288,822 Source: NCTCOG Demographic Forecast, TTI Real Estate Center Exhibit 3 illustrates the following key points relative to population forecasts for the region: The populations of Collin, Kaufman, and Rockwall counties are expected to double by the year The 12 county population is expected to increase by nearly 3.5 million people from

16 Exhibit 4: System Plan Population Forecast Growth Rates COUNTIES (Annual Compound Growth Rate) COLLIN 2.80% 2.89% 2.48% 2.10% 2.13% 13.02% DALLAS 0.93% 1.20% 1.17% 1.05% 0.98% 5.44% DENTON 2.67% 1.98% 2.13% 1.92% 1.84% 10.99% ELLIS 2.94% 1.88% 1.99% 1.83% 1.52% 10.58% HILL 3.23% 0.70% 0.61% 2.21% 1.94% 8.98% HOOD 2.41% 0.40% 0.39% 2.82% 2.43% 8.71% HUNT 4.00% 1.10% 0.95% 2.05% 1.84% 10.30% JOHNSON 2.88% 2.09% 1.66% 1.76% 1.87% 10.69% KAUFMAN 7.14% 1.86% 1.31% 2.32% 1.80% 15.16% PARKER 2.57% 2.35% 1.91% 1.88% 1.65% 10.78% ROCKWALL 4.07% 3.98% 2.94% 2.65% 2.19% 16.84% TARRANT 1.72% 2.05% 2.02% 1.83% 1.63% 9.60% WISE 2.33% 1.46% 1.21% 1.17% 1.03% 7.40% EXTERIOR COUNTY GROWTH RATES COOKE 1.42% 1.37% 1.32% 1.26% 1.21% 6.76% GRAYSON 1.42% 1.37% 1.32% 1.26% 1.21% 6.76% PALO PINTO 1.43% 1.37% 1.32% 1.26% 1.21% 6.76% ERATH 1.42% 1.37% 1.32% 1.26% 1.21% 6.76% SOMERVELL 1.88% 1.81% 1.74% 1.67% 1.60% 9.01% NAVARRO 1.43% 1.37% -3.42% 1.26% 1.21% 1.77% Source: NCTCOG Demographic Forecast, TTI Real Estate Center Exhibit 4 offers the following observations for demographic activity within the region over the next 25 years: Average growth rates for Collin, Denton, Ellis, Hunt, Johnson, Kaufman, Parker, and Rockwall counties is forecast to be over 10%. Regional growth rates appear to be concentrated outside the central core counties, extending outward from the traditional urban areas in Dallas and Tarrant counties. 15

17 Exhibit 5: System Plan Employment Forecasts COUNTIES COLLIN 361, , , , , ,348 DALLAS 2,080,698 2,245,960 2,407,600 2,560,446 2,710,842 2,854,287 DENTON 220, , , , , ,102 ELLIS 59,292 69,229 80,030 91, , ,146 HILL 11,948 13,607 15,410 17,398 19,181 20,997 HOOD 19,683 23,068 26,673 30,064 33,458 37,037 HUNT 45,881 51,541 57,913 65,361 71,716 78,162 JOHNSON 65,623 77,254 89, , , ,918 KAUFMAN 39,989 46,602 54,536 64,003 72,624 81,646 PARKER 46,265 54,002 62,412 72,546 81,892 91,661 ROCKWALL 24,009 28,800 34,247 40,553 47,019 53,934 TARRANT 1,053,934 1,166,574 1,282,220 1,401,548 1,522,314 1,644,457 WISE 28,789 32,891 37,375 42,807 47,538 52,312 NCTCOG MODEL TOTALS 4,057,670 4,470,571 4,895,603 5,328,141 5,762,602 6,198,007 EXTERIOR COUNTIES COOKE 21,778 23,375 25,002 26,713 28,444 30,205 GRAYSON 54,979 59,009 63,166 67,437 71,805 76,252 PALO PINTO 13,934 14,955 16,009 17,091 18,198 19,325 ERATH 18,276 20,062 21,947 23,925 25,992 28,138 SOMERVELL 4,043 4,636 5,290 6,005 6,783 7,623 NAVARRO 20,475 21,976 23,524 25,115 26,742 28,398 EXTERIOR COUNTY TOTALS REGIONAL TOTALS 133, , , , , ,941 4,191,155 4,614,584 5,050,541 5,494,427 5,940,566 6,387,948 Source: NCTCOG Demographic Forecast, TTI Real Estate Center Exhibit 5 illustrates the following key points relative to the geographical distribution of employment growth among the 12 counties: Almost two million new jobs are expected over the next 25 years. Employment is expected to come close to or actually double in Collin, Denton, Ellis, Hood, Johnson, Kaufman, Parker, Rockwall and Wise counties during this period. Regional growth will move outward from the employment centers of Dallas and Fort Worth. 16

18 Exhibit 6: System Plan Employment Forecast Growth Rates COUNTIES (Annual Compound Growth Rate) COLLIN 2.49% 2.38% 2.19% 2.14% 2.00% 11.70% DALLAS 1.54% 1.40% 1.24% 1.15% 1.04% 6.53% DENTON 2.77% 2.65% 2.53% 2.30% 2.14% 13.02% ELLIS 3.15% 2.94% 2.80% 2.47% 2.27% 14.39% HILL 2.63% 2.52% 2.46% 1.97% 1.83% 11.94% HOOD 3.22% 2.95% 2.42% 2.16% 2.05% 13.48% HUNT 2.35% 2.36% 2.45% 1.87% 1.74% 11.24% JOHNSON 3.32% 3.10% 2.86% 2.62% 2.42% 15.16% KAUFMAN 3.11% 3.19% 3.25% 2.56% 2.37% 15.35% PARKER 3.14% 2.94% 3.06% 2.45% 2.28% 14.65% ROCKWALL 3.71% 3.53% 3.44% 3.00% 2.78% 17.57% TARRANT 2.05% 1.91% 1.80% 1.67% 1.56% 9.31% WISE 2.70% 2.59% 2.75% 2.12% 1.93% 12.69% EXTERIOR COUNTY GROWTH RATES COOKE 1.43% 1.35% 1.33% 1.26% 1.21% 6.76% GRAYSON 1.42% 1.37% 1.32% 1.26% 1.21% 6.76% PALO PINTO 1.42% 1.37% 1.32% 1.26% 1.21% 6.76% ERATH 1.88% 1.81% 1.74% 1.67% 1.60% 9.01% SOMERVELL 2.78% 2.67% 2.57% 2.47% 2.36% 13.52% NAVARRO 1.42% 1.37% 1.32% 1.26% 1.21% 6.76% Source: NCTCOG Demographic Forecast, TTI Real Estate Center Key observations from Exhibit 6 and the annual compounded growth of employment reveal the following: Employment growth rates for Collin and Denton counties will continue to increase for the next five years, and then growth rates, while still substantial, will decline. All counties, less Dallas and Tarrant, are forecast to experience over 10% growth from Comparing the forecasts for population growth with employment forecasts suggests that employment opportunities follow population growth. 17

19 Additional key information required for development of the forecasting model included the following: Reassess the forecasting baseline numbers and assumptions used in the previous regional general aviation system plan, while evaluating the effect of the proximity of population and employment forecast made in 1990 with actual data from 2000 and Identify available demographic and economic data for those portions of North Central Texas that can provide a basis for comparable analysis with previous studies. Prepare a comprehensive history of major general aviation airport improvements. It is anticipated that the following airport improvements will be documented over the last years: runway extensions, taxiway additions, ramp expansions, major changes in NAVAIDS, and hangar developments that have occurred in North Central Texas since Develop a list of planned major transportation improvements over the next 10 years for identifying potential factors that may influence aviation forecasts. Detailed analysis of the FAA s National Plan of Integrated Airport Systems (NPIAS) for individual airports in North Central Texas. Collect and evaluate existing forecasts for airports in the region from the FAA s TAF and master plans for years 2010, 2015, 2020, 2030 and beyond as available. Identify the structure of the National Airspace System (NAS) Architecture and understand its influence on North Central Texas. Consider developing subregional airport systems for use in forecasting and in identifying specific areas of activity and influences. Evaluate the individual and cumulative economic impacts of North Central Texas regional airports. Investigate and evaluate based aircraft and operation activity for North Central Texas s airports since Develop description criteria for categorizing airports in the system plan and identify the effect on forecasting activity, allocation of capacity and funding implications. Evaluate current master plans for North Central Texas airports concerning their assessments of existing and proposed capacity. Identify current and potential initiation of legislative considerations and their impacts on basing aircraft and level of operations at particular airports. There are innate inaccuracies built into existing general aviation forecasting models. These were examined in the forecast s developmental process and include considerations such as: 18

20 Operational counts at towered and non-towered airports - Towered airports provide relatively accurate accounting of aviation activity at that airport. There can be a small margin of missed activity depending on the hours of tower operation, as most towers at general aviation airports do not operate 24 hours per day. Reporting of activity at nontowered airports relies on the varied reporting of airport management who sometimes use previous years estimates as a basis for the current or coming year. There is currently not a sound method for validating the operational activity estimates other than the installation of an activity counter at the end of each runway for a period of time. While counters can provide more accuracy, the numbers are still just estimates with varying degrees of precision depending on the data collection methods. Accuracy and availability of historical aviation data - The accuracy and availability of historical general aviation data is also suspicious primarily due to the previously mentioned method of estimating that has been used in the recording of historical activity. Historical aviation data consists of operations and based aircraft. The operations data is then grouped by activity and aircraft type. During the system planning process, the FAA started an initiative to document aircraft tail numbers by airport to greatly assist in the validation of based aircraft data. Other benefits of this effort are yet to be seen but may include future confirmation of other activity numbers and forecasts. The development of the new general aviation forecasting model studied the use of this new data and its application toward improved general aviation activity forecasting. As previously discussed, forecasts are used to plan for potential future facilities and infrastructure. The forecasts assist in determining the quantity and timing of investment decisions, hence the importance of ensuring the validity, accuracy, and applicability of the numbers and the factors generated in the forecasting process. Considering forecasts are established from a baseline set of numbers and assumptions, the accuracy of these figures and relationships is critical to the outcome and validity of a forecast. Understanding and testing these baseline figures was a substantial element in the development of a new model. The development of the forecasting model will assess these relationships and the applicability to general aviation activity forecasting. Before moving forward with the nonstandard approach to the forecasts, staff needed an inventory of the aviation system in North Central Texas. This process is summed up in the System Plan s Inventory Report but the simplified results of the inventorying process can be seen in Exhibit 7. 19

21 Exhibit 7: Landing Facilities in North Central Texas Source: NCTCOG System Plan Inventory Report Due to the dense number of aviation facilities and infrastructure present in North Central Texas, staff narrowed the focus of forecast airports down to 41 total facilities, shown in Exhibit 8, created from: Towered and non-towered airports NPIAS and non-npias facilities Public and Private Airports (Ten or more based aircraft) Exhibit 8: System Plan Forecast Airports Addison Dallas Love Majors Aero Country Decatur Mesquite Metro Airpark East Denton Mid-Way Regional Arlington Ennis Mineral Wells Bourland Field Fort Worth Alliance Northwest Regional Bridgeport Fort Worth Meacham Parker County Caddo Mills Fort Worth Spinks Possum Kingdom Clark Field Gainesville Propwash Cleburne Granbury Rockwall Collin County Grand Prairie Sycamore Strip Commerce Grayson County Terrell Corsicana Dallas Airpark Dallas Executive Dallas/Fort Worth Hicks Airfield Hillsboro Lakeview Lancaster 20

22 Research into various non-standard approaches to acquire forecast-related variables included obtaining unique airport data through the following analyses: Surrogate based aircraft analysis Public use airport survey Subregional analysis Surrogate Based Aircraft Analysis Forecasting for the NCTCOG System Plan includes airports that are part of the National Plan of Integrated Airport Systems (NPIAS) and some airports that are not included in the NPIAS (non-npias). This forecast is based on the aviation activity data from the Airport Master Record (5010) for airport operations and based aircraft. In the fall of 2008, Staff conducted a series of surrogate analyses for airports to ensure that input data is as accurate as possible, to produce reliable forecasts for our regional aviation facilities. Additionally, in an effort to improve the accuracy of 5010 data, the FAA surveyed all NPIAS airports for based aircraft tail numbers. Although response rates varied initially, NCTCOG, with help from the FAA, the Texas Department of Transportation (TxDOT) Aviation Division, GCR Associates (GCR), and local airports, was able to improve the response rate to an acceptable level in the spring of However, non-npias airports that will be forecasted have not had the same level of data accuracy because they did not receive an FAA survey. Thus, their 5010 data has not had the same requirements for tail-number reporting. As such, acquiring accurate aviation activity data for non-npias airports is essential to ensure aviation demand forecasts are realistic. Due to the data discrepancy in reporting requirements, it was suggested that an alternate method for determining airport activity at non-npias airports was needed. This process would create surrogate based aircraft numbers for inclusion in the System Plan forecasting model in an effort to create a more accurate standard for non-npias airports. In lieu of previous 5010 Airport Master Record database deficiencies, identified throughout the aviation industry, and the varying levels of response accuracy to the additional FAA based aircraft surveys, NCTCOG Staff identified the following goals for the surrogate analysis: 1. Determine steps to create proposed based aircraft numbers for a given airport based on unique characteristics. 2. Find a group of similar NPIAS airports to use as a control group 3. Compare a control group to similar non-npias airports. 4. Apply surrogate methodology analysis to non-npias airports to create surrogate based aircraft numbers. 5. Identify airports that fall out of the norm for accurate data thresholds. 6. Compare analyses independently to determine cause of disparity. Hangar Space Income Population 7. Utilize an alternate method for reporting based aircraft if needed. 21

23 The surrogate analysis methodology is intended to identify correlative NPIAS airports with the following criteria: Non-Reliever Five or less based jets Runway length of approximately 5000 feet or less Verified based aircraft from the 5010 data The NPIAS based aircraft would then be plotted against NPIAS airport hangar space to correlate an average ratio of based aircraft to hangar space (in sq. ft.) for these airports. Applying non-npias airports based aircraft and hangar ratio data to the benchmark defined by NPIAS airports generated a surrogate based aircraft count. The primary assumptions of the analysis suggested that hangar space is the limiting factor for an airport s based aircraft (supported by regional weather characteristics and survey information) and the fact that airport hangar space is reasonably accurate as calculated in GIS. Additional assumptions for the surrogate analysis included the following: Helicopters are not included in based aircraft counts. An airports tie down space is utilized largely by itinerant aircraft, not based aircraft. The population data was derived from the 2030 forecast by Traffic Survey Zones (TSZ) (2005 HHPOP) and income data from the 2000 Census median income. Analysis was included for any TSZ touching a 5 mile airport buffer. A five-mile buffer was considered because there was minimal overlap; travel times could not be utilized due to airports outside the MPA. The steps used in the 3 analysis referenced above; hangar space, population, and income are outlined in the following pages: Hangar Space Analysis and Results: Using GIS, staff calculated total hangar space for NPIAS and non-npias airports (see Exhibit 9). T-hangar space and conventional hangar space were isolated at select NPIAS and non- NPIAS airports based upon a mean industry standard for T-hangars space of 1,600 square feet and conventional hangar space at 2,500 square feet. Exhibit 9: Examples of T-Hangars and Conventional/Box Hangars: 22

24 Exhibit 10: Example Airport Hangar Analysis NCTCOG staff input the total hangar space data into Microsoft Excel 2003 and performed a hangar space correlation for T-hangars and conventional/box hangars. Exhibit 11 displays examples for results found from the correlations for NPIAS airports. Exhibit 11: Example Hangar Space Correlation for NPIAS Airports Once the hangar space correlation analysis was completed for both kinds of airports, ann X/Y Scatter Chart was created to calculate a linear trend line for NPIAS airports based aircraft (BA), and hangar space data leading to development of a standard deviation. Further analysis of the non-npias airport hangar space correlation to the standard deviation, illustrated over 73% of non-npias airports within the NPIAS standard deviation indicating a fairly strong correlation between based aircraft and hangar space supported by the data s R 2 statistics as seen in Exhibit

25 Exhibit 12: Example Hangar Space Analysis Results Population Analysis and Results: Another component of the surrogate analysis was utilizing 2030 TSZ forecast data (most current available at the time) to perform an analysis of population within 5 miles of airports in GIS as seen in Exhibit 13. Similar to the hangar space analysis, NPIAS airports based aircraft was plotted against population data inputs to develop its unique trend line and a standard deviation. Exhibit 13: Example Population Analysis (GIS) Exhibit 14 displays sample results of the population analysis which illustrates a poor correlation between BA and population with a R 2 statistic of less than 50%, shown by numerous facilities plotted outside the standard deviation. 24

26 Exhibit 14: Example Population Analysis Results Income Analysis and Results: Using 2000 Census data (most current available at the time), NCTCOG Staff performed a median household income analysis within 5 miles of airports with the use of GIS. Much like the preceding analyses, NPIAS airports based aircraft was plotted against median income population data inputs to identify a trend line and develop a standard deviation. Results of the income analysis indicated a weak correlation between BA and median income supported by the data s R 2 statistic of 43%. Conclusions of Surrogate Based Aircraft Analysis The surrogate analysis data results indicated that the strongest correlation for airports based aircraft was in the hangar space analysis with no significant correlation resulting from the income or population analysis. Further review of the hangar analysis showed four airports needing further review with ultimately only one facility requiring the use of a surrogate number. Public Use Airport Survey The public use airport survey was conducted in an effort to acquire a sampling of numerous factors, covering a wide array of airport related activates including: Airport Governance Airport Planning Airport Tenants Airport Services Airport Finances Airport Ground Access Airport Infrastructure Airport Safety 25

27 The focused design of the survey allowed NCTCOG staff to collect airport specific data in order to identify airport characteristics related to attracting airport tenants and activity. Many variables were derived from the survey responses and provided data for inclusion into the new forecasting methodology. The survey enabled NCTCOG staff to increase the level of detail for the forecast model development upon which System Plan recommendations are based. The following exhibits illustrate the recorded costs at regional airports for hangar use, fuel, and leasing office space. Exhibit 15: Regional Airport Hangar Rental Costs Source: NCTCOG Public Use Airport Survey Exhibit 16: Regional Airport Office Space Lease Rates Source: NCTCOG Public Use Airport Survey 26

28 Exhibit 17: Regional Aviation Fuel Prices Source: NCTCOG Public Use Airport Survey Subregional Analysis Another component of the nonstandard System Plan forecast model development was the Subregion Analysis. The project team conducted the analysis to divide the 19-county System Plan study area into subregions. Identifying subregions can increase the accuracy of future aviation infrastructure and demand forecasts by considering growth demographics and other characteristics unique to airports and their geographic area. The detailed technical analysis performed to determine subregional boundaries can be viewed in the Subregion Report, but a summary of the process is outlined below: Step 1: Analysis of County and City Employment and Population Analysis of the Metropolitan Planning Area s (MPA) 2030 Forecast Employment (EMPGROWTH) and Population (HHPOPGROWTH) Growth by traffic survey zones (TSZ) and 2030 forecast Total Employment (EMP30) and Population (HHPOP30) by city, identified similar trends for growth in the region. Step 2: Analysis of Naturally Occurring Breaking Points Analysis of naturally occurring breaking points such as the geographic position of lakes and 2030 planned roadway networks in the region proved useful in selecting specific census group blocks acting as subregional boundaries. Step 3: Analysis of Airport 2030 Travel Time Contours Analysis of reliever airports 2030 ten minute travel time contours isolate specific TSZs around an airport for inclusion in a given subregion. Additionally, airports ten minute interval contours provide a precise level of detail to prove the selection of census blocks acting as boundaries for subregions. 27

29 Step 4: Analysis of Forecast Airport Locations and Characteristics Analysis of the geographic distribution of forecast airports, shown in Exhibit 18, and their unique characteristics determine demographics that a forecast airport s facility might attract. Exhibit 18: Analysis of Forecast Airport Locations and Characteristics Step 5: Analysis of Registered and Verified Based Aircraft Data To confirm results of the subregion analysis tail numbers and addresses in the registered aircraft database, verified 28

30 aircraft tail numbers reported to the FAA by airport managers and directors were compared. This data verified correlations between the location of an aircraft owner s residence to the subregion and forecast airport their aircraft is based. See Exhibit 19. Exhibit 19: Registered and Verified Based Aircraft Analysis Region Based Aircraft In Region (%) South 52% East 61% West 64% North 64% Central 85% Note: Verified based aircraft data is available for select airports only. The combination of these analyses resulted in the development of five subregions that are based on census blocks, increasing the accuracy of forecast airports future demand and capacity. The technical analyses identified subregional boundaries accounting for: Demographic growth of employment and population Naturally occurring breaking points Surface access capabilities Airport characteristics and geographic distribution of airports in the region Distance correlations for aircraft owners residences to their aircrafts based airport The different segments of airspace in the region were also considered in the subregion analysis, with no significant correlation linking airspace and forecast airports growth. Professional judgment was also exercised to aid in identifying the 5 subregions. E. FORECAST MODEL STRUCTURE Forecast Model Development The System Plan forecast update includes the items listed below as they relate to the general aviation activities in the 19-county region: Based aircraft (jet, non-jet) Operations (local, itinerant) Specifically, the based non-jet aircraft model is an input to the local operations. The local operations and the based-jet aircraft models are an input to the itinerant operations. The two airports that are excluded from this forecast update are DFW International and Dallas Love Field Airports due to their operations being primarily commercial in nature. The airparks were excluded from the model estimation data set due to their different growth and operational characteristics. However, the final model was applied to them as well. As previously discussed, there are several established methodologies for forecasting aviation activities; market share, econometric, time series, etc. As staff ventured further into 29

31 development of the new forecasting methodology, three of the most important factors in the forecasting processes were isolated: The availability of accurate, comprehensive, and consistent data for both the current conditions and the historical trends; The identification of a comprehensive set of airport-specific and demographic attributes that could define the forecasting variable in the study area; and The knowledge of the region s aviation activities and trends and the main forces that affect them. This methodology used for the System Plan update uses a combination of market share and single variable regression analyses for forecasting the based jet aircraft and a multinomial linear regression for the non-jet based aircraft and their number of operations. All of these approaches attempt to find the relationship between a dependent variable (forecast aviation variable) and a set of independent variables (airport-specific or demographic attributes). None of the above methodologies directly addresses the risks that could potentially affect the aviation activities in the future. However, this can be achieved through a probabilistic approach where the risk factors are defined and the effect of each factor on aviation activities is identified. This approach would require the involvement of a team of experts (e.g., pilots, stakeholders, airport operators, consultants) in the modeling process as well. The probabilistic approach is rather new in the aviation field but has been used in the construction cost estimating process for the last decade. A comparison of the above methodologies is shown in Exhibit 20: Exhibit 20: Comparison of Forecast Model Methodologies Methodology Market Share, Single Variable Regression Multinomial Linear Regression Attributes Macroscopic modeling process Straight forward selection of independent variables Range of possible forecasts can be broad Approach considers multiple independent variables requiring more detailed analysis of inputs Comprehension of regional aviation dynamics critical It should be noted, throughout the forecasting process the project team accounted for the general aviation operations and based aircraft at Dallas Fort Worth International Airport and Dallas Love Field. Although due to these airports conducting primarily commercial operations, the forecast data developed for these facilities was not included in development of the general aviation system s future recommendations including the reallocation process of forecast based aircraft and operations amongst forecast airports. Based Jet Forecast Initially staff expected to use a multinomial linear regression model for all the forecasting variables, including the based jet-aircraft. However, after further analysis and review of the FAA s aircraft registration data for the owners of jet aircraft, and a random sampling of the type of businesses in which they are involved, the following conclusions were evident: Some private and corporately-owned based jets at the forecast airports are registered to an out-of-state address, increasing the difficulty of access to their income information; 30

32 A random sampling of business jet owners revealed that their privately-owned jets are commonly linked to a business with multiple corporate jets; The type of industries utilizing business jets are advertising, distribution, farming, legal, and sales; and The available income data is not comprehensive enough to identify the actual income level of the private-jet owners or the respective corporations. Therefore, it was decided to follow a more traditional market share and single variable regression analysis to forecast based jet aircraft at airports. The following steps detail the process involved in forecasting based jet aircraft for the System Plan and the results can be seen in Exhibit 21: Step 1: To forecast the future number of aircrafts, a single variable regression model is built based on the historical relationship between registered jet aircraft in the region and the following variables: County population County employment Per capita personal income (PCPI) Step 2: To calculate the number of based jet aircraft at each airport, the existing share of jet aircraft is applied to the totals from Step 1 to calculate the number of based jet aircraft at every airport. Step 3: This process also included a simple trend analysis which looks at the total regional based jet aircrafts in the region in the past years and extrapolates them into the future. Exhibit 21: Based Jet Aircraft Model Subregional Results Subregion Central East North South West Regional Total The different forecast scenarios developed for based jet aircraft are listed below: 1. Constant (Market Share): This projection of Registered Aircraft represents a constant share of the FAA s Registered Aircraft projected through the forecast period. The most recent year of actual data is used to generate the market share percentage. 2. Dynamic (Market Share): This projection of registered aircraft examines historical market shares and develops a linear trend of those numbers to yield a projection of demand. 3. Population (Single Variable Regression): This projection of Registered Aircraft is generated via a regression between county population and county registered aircraft. The R squared value is shown for this regression. 31

33 4. Employment (Single Variable Regression): This projection of Registered Aircraft is generated via regression between county employment and county registered aircraft. 5. PCPI (Single Variable Regression): This projection of Registered Aircraft is generated via regression between county per capita personal income and county registered aircraft. 6. Exponential Smoothing (Trend): This projection of Registered Aircraft is generated using the double exponential smoothing method applied to registered aircraft. 7. Linear Trend: This projection of Registered Aircraft is generated using a simple regression of registered aircraft against time. 8. High/Low Average: This projection of registered aircraft is produced by averaging the high and low reasonable projections from the first seven projections of demand. 9. Multi-Average: This projection takes an average of all reasonable projections. The multiaverage projection reflects the values of the majority of projections and is not easily skewed by unusually high or low values in one of the projections. 10. Constant Market Share: This projection uses a constant market share average of the regional based jet history. The selected forecast can be any one of the above projections. However, it is generally either the high/low average or the multi-average projection. These projections take into account the upper and lower ranges of demand and produce mid-range estimates of future activity. These results should be reviewed by experienced aviation personnel who then select the best forecast. This step was formulated in Microsoft EXCEL through a set of macros that utilizes the application s available regression functions. Based Non-Jet Aircraft and Operations Forecasts The non-jet based aircrafts and the local and itinerant operations are forecast through a multinomial linear regression model. Unlike the based jet forecast model, this is a bottom-up approach where the forecasts aggregated to the subregional level. Therefore, the modeler needs to have a more detailed understanding of the characteristics and the dynamics of the aviation activities in the region. The steps in this process are as follows: 1. Define the dependent variable; 2. Define a set of possible independent variables; 3. Define the set of main variables that have to enter the regression model; 4. Other variables are entered in to the regression model in a stepwise process based on their statistical significance (T-Test); 5. The multipliers, calculated for each independent variable in each step, are evaluated for correctness of its sign, magnitude, and physical significance; 6. Evaluate the results in each step and find new independent variables to improve the forecast numbers at the airports; and 7. Repeat steps 4 through 6 until the model reaches acceptable performance levels. The statistical significance of the independent variables, not already entered the model, is reported in each step. However, this does not mean that insignificant variables should not enter the model or that all significant variables must enter the regression model. The Statistical Package for the Social Sciences (SPSS) software was utilized in this step. 32

34 Trips Cumulative Percentage (%) Based Non-Jet Forecasts This model is based on the demographic attributes within the service area of each airport. The service area was defined as a 25-mile radius around each airport based on mapping the registration addresses and their based airport for about 2,300 records (out of a total of 4,245 non-jet based aircrafts at the study airports). The results are shown in Exhibits 22 and 23. Exhibit 22: Travel Time Distribution from Registration Address to Airports Airport Trip Length Airport Distribution Trip Length based Distribution 2007 Expanded Network (Travel Time) 1200 Trips Total Trips: 2312 Maximal Travel Time: minutes Cumulative Percent (0,15] (15,30] (30,45] (45,60] (60,75] (75,90] (90,105] (105,120] (120,135] Travel Time (Minutes)

35 (0,5] (5,10] (10,15] (15,20] (20,25] (25,30] (30,35] (35,40] (40,45] (45,50] (50,55] (55,60] (60,65] (65,70] (70,75] (75,80] (80,85] (85,90] (90,95] (95,100] (100,105] >105 Trips Cumulative Percentage (%) Exhibit 23: Distribution of Travel Distance from Registration Address to Airport Airport Trip Length Airport Distribution Trip Length based Distribution 2007 Expanded Network (Travel Distance) Trips Cumulative Percent Total Trips: 2312 Maximal Travel Distance: miles Travel Distance (Miles) These figures show that approximately 80% of the non-jet aircraft owners travel less than 25 miles and about 35 minutes to reach their base airport. The model results, with a R 2 statistic at 97%, show that the number of non-jet based aircrafts at every airport, with the exception of a correction added for a few outliers, is: Highly correlated with the median income in the service area; Higher correlation with the population compared to the employment of the service area; and Is highly affected by the existence of airports in its vicinity that do not serve the non-jet general aviation demand. Demographic information used in the forecasts was extracted from NCTCOG s travel demand model forecasts and TTI for areas located outside NCTCOG s modeling boundary. The regression results are shown in Exhibit 24 and summarized at a subregional level in Exhibit

36 Model Based-Aircrafts Exhibit 24: Based Non-Jet Aircraft Model Results Actual Based-Aircrafts R^2 =.969 Exhibit 25: Based Non-Jet Aircraft Model Subregional Results Total Total Model Total Actual Subregion Difference Based Aircraft Based Aircraft North -5 1,913 1,918 South East West Central -12 1,331 1,343 The next long-term step in improving this forecast model was to build all, or any combination of, the models listed below: Aircraft ownership model to forecast the geographic location of the aircraft owners; Airport choice-set model to define what the plausible based-airport choices are for each of the owners; and Airport selection model to define how owners rank the airports that are within their choice set. However, the process outlined above would require performing a very detailed survey of the aircraft owners and the airport operators. This data is currently not available. Local Operations Forecasts The most time-consuming part of building the forecasting model for the local operations was identifying a set of variables that could potentially define operations at the airport level and explain differences between the forecast airports. 35

37 Model Operations Review of the local operations forecast displayed an over-estimation at Spinks airport, in all scenarios, determined primarily as a result of an unpaved runway commonly used for training purposes; an airport-specific constant was added for this airport. The regression results displayed a high correlation, as seen with a R 2 statistic at 93%, are shown in Exhibit 26 and summarized at a subregional level in Exhibit ,000 Exhibit 26: Local Operations Forecast Results 100,000 80,000 60,000 40,000 20, ,000 40,000 60,000 80, , ,000 Actual Operations R^2 =.937 Exhibit 27: Subregional Local Operations Model Performance Total Total Model Annual Total Actual Annual Subregion Difference Operations Operations North 3, , ,297 South 5, , ,640 East -6,842 86,446 93,288 West -1,399 58,401 59,800 Central 3, , ,256 Itinerant Operations Forecasts The itinerant operations forecasts are generated both by an airport s based aircraft and aircraft based elsewhere. However, the breakdown of based and non-based generated itinerant operations is not readily available. Therefore, practically the model can only include variables related to the airport itself. The regression results and the subregional level summaries are shown in Exhibits 28 and

38 Model Itinerant Operations Exhibit 28: Itinerant Operations Forecast Results 100,000 90,000 80,000 70,000 60,000 50,000 40,000 30,000 20,000 10, ,000 40,000 60,000 80, ,000 Actual Itinerant Operations R^2 =.977 Exhibit 29: Subregional Itinerant Operations Model Performance Subregion Total Difference Total Model Annual Operations Total Actual Annual Operations North 18, , ,707 South 2,057 89,656 87,599 East 19,299 54,499 35,200 West 6,235 36,135 29,900 Central -24, , ,086 Application of Residuals A regression model attempts to build a model that minimizes the errors between observed and actual data. Therefore, it will not necessarily be able to reproduce the exact observations in the base year. The model could produce a better match if all the possible factors that define the dependent variable were known. The difference between the model generated number and the actual observation is called the residual, which accounts for the differences caused by unknown factors. In this forecasting process the actual value of the residual in the base year was applied to all forecast years. 37

39 F. RESULTS Subregional Forecasts Results The model forecast results at the subregional level are summarized in Exhibits 30, 31, and 32. Subregion Exhibit 30: Subregional Based Non-Jet Forecasts Current Forecast Year North 1,865 2,051 2,216 2,381 2,544 2,710 South East West Central 1,314 1,436 1,567 1,700 1,832 1,963 SUM 4,245 4,796 5,287 5,779 6,266 6,756 Subregion Exhibit 31: Subregional Local Operations Forecast Current Forecast Year North 407, , , , , ,067 South 143, , , , , ,673 East 93, , , , , ,488 West 59,800 66,302 71,528 76,901 82,410 87,987 Central 264, , , , , ,678 SUM 968,279 1,026,569 1,100,388 1,177,867 1,257,740 1,340,893 Subregion Exhibit 32: Subregional Itinerant Operations Forecast Current Forecast Year North 241, , , , , ,643 South 92,621 96, , , , ,749 East 55,904 59,729 63,126 67,006 70,691 74,840 West 36,135 39,282 42,153 45,096 48,120 51,192 Central 290, , , , , ,781 SUM 716, , , , ,728 1,011,206 Initial results have shown significant growth throughout the region which is strongly related to demographic and economic projections. Regional operations are anticipated to grow steadily at an annual rate of approximately 1.5% and a total of 40% over 2010 baseline counts. Based nonjet aircraft mirror that with a slightly higher 2% annual growth rate and nearly 60% over the life of the forecast. Most promising is the growth of based jets from around 378 to nearly 800 by the end of forecast year 2035 (see Exhibit 21). 38

40 Through refinement of the forecasting model, NCTCOG staff identified several important factors affecting the forecasting processes: Accurate, complete, and consistent data Comprehensive set(s) of airport specific and demographic attributes in the study area Knowledge of the main forces affecting the region s aviation activities and trends Ultimately, the forecasting results are consolidated by each of five subregions identified through the system planning process. These results will feed into a needs assessment model that will identify capacity shortfalls as a result of aviation growth in the future. Once these capacity needs are identified, they will be utilized in a series of scenario-based planning recommendations that should yield a suggested plan for future investment and improvements. G. COMPARISONS OF FORECAST DATA A major FAA requirement affecting the ultimate acceptance of proposed forecasts is that baseline forecasts of operations and based aircraft must be compared with the FAA s Terminal Area Forecasts (TAF). The current FAA policy is to review and approve short-term and intermediate-term forecasts. As a matter of policy, short-term forecasts that differ from the FAA TAF by more than 10%, or intermediate term forecasts that differ more than 15%, must include clearly justified methodology and input data. System planning forecasts that are used to support AIP projects are also subject to the 10% and 15% variation review from the TAF. Generally, forecasts at GA airports that have fewer than 100,000 annual operations are not subject to percentage conformity to the TAF unless there are specific issues regarding special programs (e.g. establishment/discontinuance of federal contract tower facilities). To confirm the regional System Plan forecasts were conforming to the FAA policy, NCTCOG gathered 2 sets of forecast data TAF and AMPs for comparison to NCTCOG s updated forecast methodology. Exhibits display the comparison of FAA TAF, Airport Master Plans, and NCTCOG regional forecasts for total: Based Jets Based Non-Jets Regional Operations Note: Due to the dated material in many AMPs, a comprehensive set of forecast data was unavailable beyond the year Exhibit 33: Comparison - Total Non-Jet Based Aircraft Airport Master Plan Forecasts 3,346 3,541 3,776 NCTCOG Forecasts 2,797 3,173 3,507 Terminal Area Forecasts 2,928 3,078 3,217 As seen in Exhibit 33, both AMP and NCTCOG s forecasts closely reflect the TAF with a smaller difference margin realized between the TAF and NCTCOG forecasts. Overall, regional non-jet growth is expected to increase throughout the region through 2020 in this comparison. 39

41 Exhibit 34: Comparison - Total Based Jets Airport Master Plan Forecasts NCTCOG Forecasts Terminal Area Forecasts When comparing the System Plan based jet forecasts to AMPs and the TAF the NCTCOG regional forecast more accurately reflects the FAA s outlook concerning the growth of this type of aircraft in North Central Texas. See Exhibit 34. Exhibit 35: Comparison - Total Regional Operations (in thousands) Airport Master Plan Forecasts 1,707 1,873 2,113 NCTCOG Forecasts 1,365 1,430 1,523 Terminal Area Forecasts 1,501 1,568 1,675 As previously mentioned, because many AMPs have not been updated for some time, and in the past decade monumental changes impacting aviation activity and its businesses have been seen, many AMP forecasts have not been able to account for these differences which is reflected in their forecasts. See Exhibit 35. See Appendix B for more comparisons of historical TAF data and the System Plan forecasts. H. CONCLUSION As the foundation of airport system planning, NCTCOG developed an innovative model to forecast aviation activity in the region. These forecasts are the main product to be utilized in recommendation planning for infrastructure improvements from current year to horizon year In previous system plans, critical areas for additional capacity were identified, which in turn resulted in municipal and industry site selection of Fort Worth Alliance and Spinks Airports. NCTCOG s forecasting process uses a combination of market share and single variable regression analyses for forecasting the based jet aircraft and a multinomial linear regression for the non-jet based aircraft and the number of airport operations. It is unique to traditional methodologies such as market share, econometric, time series and simulation. As noted, during the forecasting process data accuracy and consistency was an issue. The current data reporting methods for NPIAS airports provides them with a window of time for updating their data. Therefore, the available airport level data is not necessary collected at the same time for all airports. This issue becomes more complicated when considering non-npias airports. Therefore, the modeler and the end-users need to have a good understanding of the possible sources of errors in the input data. The new forecasting methods and approaches incorporated by NCTCOG for the System Plan reflect the expressed desire by the FAA and NCTCOG to explore alternative forecasting 40

42 methodologies for general aviation and vertical flight activity. Evolving interactions between the supply and demand today for aviation services reflect a powerful need for gathering more accurate baseline data to be used as the foundation for the forecasts. With the goal to compare variations in forecasts over time, to help focus investment and development decisions by airport sponsors and by airport communities, the accuracy of these forecasting efforts and ultimately the accuracy and validity of the original numbers and assumptions used to generate The forecasts. Ultimately NCTCOG staff and consultants anticipate the new forecasting methodology generated will serve to complement tools such as the Airport Community Value Model to aid in future return on investment assessments to help guide future airport capital improvement investment decisions. Exploration of the relationships between based aircraft and operations at individual airports, coupled with their regional airport counterparts, tied with fuels sales, and actual aircraft movements yield new validated points of departure for general aviation activity forecasting. Basic return on an investment and testing this new valid data are critical factors in making major airport planning and investment decisions. To ensure its sustainability, the need for advanced community planning to protect the land around airports is vital. A few of the new data requirements identified will understandably pivot away from baseline numbers and assumptions used in the previous regional general aviation system plan. Demographic comparisons for portions of the region may be found inconsistent and improvements since 1990 and planned major improvements over the next ten years will certainly influence the forecasting paths moving forward. Side by side comparisons between master plans and TAF s need to be identified, analyzed, and cataloged. Structural changes in the National Airspace System (NAS) architecture have developed a system of subregional airport and heliport systems when compared to other regions and the State as a whole. Growth and decline in reported based aircraft and operations for the region s airports examined with individual airport master plans may reflect or imply future changes in regional capacity. In some circumstances, legislative considerations for potential impacts on basing and operations at particular airports may also become obvious when viewed over the next 20 years and beyond. 41

43 APPENDIX A North Central Texas System Plan Subregions 42