Like crude petroleum, natural gas is also an OOS commodity because NAICS 211 was not included in the CFS sampling frame.  Information obtained from EIA publications, including the Natural Gas Annual and the International Energy Annual, served as the principal data sources used to generate tonnage and value estimates for shipments in FAF4.  In addition, data collected by the Federal Energy Regulatory Commission (FERC) and published annually in Pipeline Economics by the Oil & Gas Journal was used as supplemental information.  

10.1 DATA SOURCES

10.1.1 Domestic Natural Gas

Domestic movements of natural gas flows are separated into two groups, interstate and intrastate.  Data sources, used in estimating volumes of shipments for the two groups, are slightly different.

Interstate Domestic Flows

The baseline state-level data for interstate movements of natural gas is obtained from EIA’s published Table 12 of the 2012 Natural Gas Annual entitled “Interstate movements and movements across U.S. borders of natural gas by state.”  For domestic shipments, only the interstate shares from this table are utilized.  Data on movements across U.S. borders is covered in the imports and exports sections discussed later.  To disaggregate interstate movements into FAF regions, a series of auxiliary data are also used.  This includes the following:

• Natural gas receipt/delivery points database
  According to the EIA, the U.S. natural gas pipeline network consists of over 11,000 delivery points (transport to end-use customer), 5,000 receipt points (involved in “gathering” natural gas), and 1,400 interconnection points that transfer natural gas throughout the U.S. in 2008³⁹ .  The FAF4 uses an updated dataset where the total number of natural gas receipt/delivery points reached nearly 17,800 according to the 2011/2012 database obtained by the FAF team.
  
  In the context of FAF4, these receipt/delivery points (usually at the beginning of a natural gas transport route) are treated as natural gas shipment starting locations (production).  The sum of “scheduled capacity” from all receipt/delivery points located within a given FAF zone is used in calculating shares, and then applied to disaggregate interstate movements from state-level to FAF regions. 
   

• Natural gas consumption by end use
  This statistic is obtained from the EIA’s Natural Gas data website⁴⁰ .  This annual data series contains total “volumes delivered to consumers” by state and end-use sectors of residential, commercial, industrial, vehicle, and electric power.  Additional databases used in disaggregating state-level natural gas consumptions to FAF regions included: population data, CBP, vehicle population data, as well as data for electric generating units.

Intrastate Domestic Flows

Information on dry production and withdrawals from underground storage⁴¹ by individual states is gathered from the EIA’s 2012 Natural Gas Annual⁴² and used to create control totals for intrastate natural gas movements.  The same auxiliary data used in the interstate case is also used to disaggregate intrastate movements.

10.1.2 Imported Natural Gas

U.S. Natural Gas Imports by Point of Entry

The major data source used in estimating imported natural gas flows is the U.S. Natural Gas Imports by Point of Entry from 2012 Natural Gas Annual published by the EIA.  Specifically, Table 9,entitled “Summary of U.S. natural gas imports by point of entry, 2008 – 2012” of that report is used.  The table provides volume (in million cubic feet), as well as price for natural gas transported by pipeline and liquefied natural gas (LNG).  Except for natural gas transported by pipeline (from Canada or Mexico), mode of transportation on movements of LNG is not specified.  It is expected that LNG imported from countries other than Canada/Mexico is transported to the U.S. by vessel.  

U.S. Natural Gas Imports by State

In addition to imports by point of entry, the EIA also reports data on natural gas imports by state⁴³, which provides total imported volumes (in million cubic feet) and prices (dollars per thousand cubic feet) of natural gas at the state level.  Total volumes by state from both tables (one by point of entry and the other by state) are examined and the results assumed that an imported natural gas shipment is typically delivered to U.S. destinations within the same state where its point of entry is located.  According to the EIA, natural gas is mostly transported by pipeline domestically. 

A processing plant is the place where pipeline-quality natural gas is produced.  Therefore, under FAF4 processing, it is assumed that the U.S. destination of imported natural gas would be a FAF region with processing plants in it, and within the same state as the port of entry of the given imported shipment.  A database that contains natural gas processing plant locations is used to determine destinations of natural gas shipments.

10.1.3 Exported Natural Gas

Data sources used in the imported natural gas estimation process typically also contain information for exported natural gas shipments.  Instead of point of entry for imports, exported shipments involve point of exit.  Therefore, the list of data sources is fundamentally the same as those used for imports.  Unlike imports, LNG exported to both Canada and Mexico can be transported by land modes, while exports to other countries are shipped only by vessel.

Similarly, an examination on the two sets of data on natural gas exports (i.e., one by point of entry and the other by state) suggested that an exported natural gas shipment typically originated in the same state as its point of exit.  Based on the natural gas transportation process and flow, the natural gas enters the transmission system through a pipeline gathering system once it leaves the producing wells.  Thus, the locations where gathering system connects to the transmission pipeline can be treated as the domestic origins for export flows.

Again, the natural gas receipt/delivery points database, briefly discussed above for imports, contains geographic information representing locations on interstate natural gas pipelines where natural gas gathering systems connect to the pipeline, or where natural gas local distribution systems and other end-users connect to the pipeline, within North America.  These receipt points are used to identify originating domestic FAF zones for exported flows.

10.2 ESTIMATION PROCEDURES FOR NATURAL GAS FLOWS

10.2.1 Domestic Flows

Overview of Methods

The estimation methodology for domestic natural gas movements is more complex than for imports and exports.  This is because the most detailed data available for domestic natural gas shipments is at state level, thus modeling approaches are needed in order to disaggregate movements into FAF zones.  This disaggregation process involved two modeling efforts, the development of a spatial interaction model (e.g., gravity model) and an application of the IPF process.  The spatial interaction model is applied to estimate the initial flow matrix of the natural gas movements.  Once the initial natural gas flow matrix is produced, the IPF procedure is then applied to generate the final natural gas flow matrix for FAF.

The IPF is a procedure for adjusting a matrix (of data cells) so that they would add up to available control totals for both the columns and rows of the given matrix.  The general process of IPF can be described as:

1. Each row of initial cells is proportionally adjusted to equal the marginal row totals;
2. Each column of (already row-adjusted) cells is proportionally adjusted to match the marginal column totals.  This is the end of the first ‘Iteration’; and
3. The above steps are repeated multiple times until a pre-determined level of convergence is reached.

Estimation Framework for Interstate Flows of Natural Gas

The general estimation process for interstate movements of natural gas is described in detail in the following steps.  The state-to-state movements that came from state-level control totals are extracted from Table 12 of the EIA-published 2012 Natural Gas Annual.  The production estimations are calculated by aggregating scheduled capacity of receipt points into FAF regions based on geospatial joining of receipt point locations.  Then, the attraction estimation for the flows is performed.

The attraction estimation process relied on available data from the EIA, which is the consumption by state and end-use sector.  For each given end-use sector, the consumption data is disaggregated to county level and then added up to FAF zones by applying one of the following methods depending on the consumption type.

• Residential:  Assuming the consumption is proportional to population, consumptions by state was disaggregated to county level.
• Industrial & Commercial:  Assuming the consumption by industrial and commercial sector is highly correlated to payrolls in each industry, CBP data was used to disaggregate data into county level.
• Vehicle:  Vehicle natural gas use is assumed as proportional to number of natural gas vehicles.  Natural gas vehicle population data from Polk was used to disaggregate vehicle consumption data into county level.
• Electric Power:  Natural gas cost information for electric generating units was used to disaggregate natural gas consumption by electric power to the county level.

The total consumption by FAF zone is obtained by combining estimates from all end-use sectors for counties within each FAF region.  Note that a rank matrix of distance is used in this process instead of using distance matrix directly.  Specifically, the initial assignment uses a gravity model by utilizing the production and attraction estimates applied to the spatial interaction procedure.  Then, with the marginal control totals from EIA data for the given state-to-state pair, the assignment and IPF approach are repeated for all state-to-state records until the pre-determined level of convergence is met.  

Intrastate Movements

The only difference for the intrastate movement estimation from the interstate process is in the preparation for the state-level data.  Because this information is not directly available from the EIA, it was derived using statistics published in Tables S1–S52 in the 2012 Natural Gas Annual.  All subsequent steps are the same as those conducted for disaggregating interstate movement to FAF regional level.

10.2.2 Imported Natural Gas

Several processing steps are required to estimate flows of imported natural gas shipments. First, with the aid of a GIS tool, locations of entry points for the EIA-published Natural Gas Imports by Point of Entry data are assigned to FAF trade zones.  Then, each processing plant is flagged in each FAF zone by conducting a spatial join between FAF-zone layer and the geographical locations of “natural gas processing plants.”  For each import record, a domestic destination area is assigned by pinpointing the FAF zone that has processing plant, which is also closest to the entry point of imported natural gas.  The foreign mode (mode used to reach the U.S.) is then adjusted and added to the domestic mode.  Mode assignment for the foreign mode from Canada is assumed truck, while all domestic modes are assumed pipelines.  Finally, the data is aggregated to the FAF zones and natural gas volumes are converted to tons, and values of shipments are estimated.

10.2.3 Exported Natural Gas

Similar to the imports, data from the EIA -published Natural Gas Exports by Point of Exit data is used with the GIS tool to create spatial layers of exit points for the foreign countries involved.  A spatial join is also performed to “connect” them with FAF4 zones.  Then, each processing plant is flagged in each of the FAF zones by the spatial join between the FAF–zone layer and the geographical locations of “natural gas processing plants.”

For each export record, a domestic destination area is assigned by pinpointing the FAF zone that has receipt points which is also closest to the exit point of exported natural gas.  The foreign mode (mode used to exit the U.S.) is then adjusted and added to the domestic mode.  Mode assignment for the foreign mode to Canada was assumed truck, while all domestic modes are assumed as pipeline.  Finally, the data is aggregated to the FAF zones and natural gas volumes are converted to tons, and values of shipments are estimated.

³⁸ http://www.eia.gov/naturalgas/annual/archive/2012/pdf/table_012.pdf.

³⁹“About U.S. Natural Gas Pipelines – Transporting Natural Gas”, Nature Gas, EIA website at:  http://www.eia.gov/pub/oil_gas/natural_gas/analysis_publications/ngpipeline/index.html

⁴⁰ http://www.eia.gov/dnav/ng/ng_cons_sum_dcu_SAL_a.htm.

⁴¹ EIA table “Natural Gas annual Supply & Disposition by State”, Natural Gas data, available at: http://www.eia.gov/dnav/ng/ng_sum_snd_a_EPG0_FPD_Mmcf_a.htm. 

⁴²Natural Gas Annual 2012, Office of Oil, Gas, and Coal Supply Statistics, EIA, U.S. Department of Energy, available at: http://www.eia.gov/naturalgas/annual/pdf/nga12.pdf.

⁴³ EIA statistics are available at http://www.eia.gov/dnav/ng/ng_move_state_a_EPG0_IM0_Mmcf_a.htm.