Concepts and Contents

Contents

Overview

Features

Reaches

Encoding flow relations among transport and coastline reaches

Identifying level paths through the drainage network

Metadata and digital update units

Geographic names

Common identifier

Coordinates and related measures

Data Quality

Conterminous United States (excluding the Pacific Northwest), Hawaii, Puerto Rico, and the Virgin Islands

Lineage

Information Sources

Processes

Convert DLG data to features

Build reaches

Attribute accuracy

Logical consistency

Completeness

Horizontal positional accuracy

Vertical positional accuracy

Pacific Northwest and Alaska

Glossary

References

Appendix A. Overview of features, reaches, and related items.

Appendix B. Feature code and description field structures and definitions

Appendix C. Encoding characteristics using field names

Appendix D. Development of Reach Files and related concepts

Appendix E. Transport reach delineation rules and examples

Appendix F. Organization and examples of hydrologic units

Appendix G. Peculiarities

The National Hydrography Dataset (NHD) is a comprehensive set of digital spatial data that encodes information about naturally occurring and constructed bodies of water, paths through which water flows, and related entities. The information encoded about these features includes classification and other characteristics, delineation, geographic name, position and related measures, a "reach code" through which other information can be related to the NHD, and the direction of water flow. In addition to this geographic information, the dataset contains metadata and information that supports the exchange of future updates and improvements to the data.

The data support many applications, such as:

• Making maps. Positional and descriptive data in the NHD provide the starting point for making many different kinds of maps.
• Geocoding observations. Much like street addresses provide a way to link data to a road network, the NHD's "reach code" provides the means to link data to water features.
• Modeling the flow of water along the Nation's waterways. Information about the direction of flow, when combined with other data, can help users model the transport of materials in hydrographic networks, and other applications.
• Maintaining data. Many organizations would like to share the costs of improving and updating their collections of geographic data. Unique identifiers and other methods encoded in the NHD help to solve technical problems of cooperative data maintenance.

In 1999, coverage was made available for the conterminous United States and Hawaii. Data for Puerto Rico, the Virgin Islands, and parts of Alaska will follow. The production of data for the remainder of Alaska will be ongoing for several years. Efforts to maintain and improve the data will occur continually.

The NHD is the culmination of cooperative efforts of the U.S. Environmental Protection Agency (USEPA) and the U.S. Geological Survey (USGS). Other organizations also contributed to the effort.

This volume describes the concepts and information content of the NHD, including features, reaches, metadata, geographic names, coordinate systems and related measures, and data quality.

A feature is a defined entity and its representation. In the NHD, features include naturally occurring and constructed bodies of water, paths through which water flows, and related entities. Features are classified by type, may be described by additional characteristics, and are delineated using standard methods.

Features are classified by type. These feature types, such as "stream/river", "canal/ditch", and "lake/pond", provide the basic description of the features. Each type has a name and a definition. For example, the three most frequently encountered feature types and corresponding definitions are as follows:

Characteristics, which are traits, qualities, or properties of features, are provided for many feature types. Each characteristic has a name, a definition, and a list of values and corresponding definitions. For example, the features lake/pond and stream/river have the characteristic Hydrographic Category:

Appendix A lists the names and characteristics associated with each feature type. The "Standards for National Hydrography Dataset" (USGS, 1999) contains the names and definitions of all feature types, characteristics, and values. The document is available online through http://mapping.usgs.gov/standards/.

Encoding feature types and characteristics

A five-digit feature code encodes the feature type and combinations of characteristics and values that can be assigned to a type. The first three digits encode the feature type, and the last two digits encode a set of characteristics and values. For example, the feature type "dam/weir" has the code "343". There are five combinations of characteristics and values that can be assigned to features of this type. These combinations are assigned the values of "00" through "04". The resulting possible feature codes are listed below:

Feature codes are stored in a data element named "FCODE".

For those who prefer to use text instead of the numeric code, words also are used to encode the feature type, characteristic, and value information in the feature code:

• Feature types are encoded with a character string (for example, "Stream/River") in a data element named "FTYPE".

• Two approaches, which format the same information in different ways, encode characteristics and values using words:

1. A description field, which holds a character string that contains all characteristics and values associated with a feature code. The description field is encoded in a data element named "DESCRIPT".
2. A text field for each characteristic associated with a feature code. The name of each field is an abbreviation of the name of the characteristic.

For example, a feature classified as a "Dam/Weir" may have the characteristic of "Construction Material" with a value of "earthen", and the characteristic of "Operational Status" with a value of "operational". This information is encoded as follows:

Appendix B lists each feature code and its corresponding description. Appendix C lists the name of the field for each characteristic and the list of values for each characteristic.

A feature type and feature code are assigned to each feature. Description fields, fields for each characteristic, and feature codes are encoded in a lookup table. Associating the lookup table with features by matching the feature codes allows words denoting characteristics and values to be substituted for the numeric feature code.

The shape and extent of features are delineated using points (including nodes), lines, or areas (see figure 1).

The delineation of each feature follows three rules:

1. The delineated feature must be contiguous.
2. The delineated feature must have the same dimensionality; that is, it must be one point, one or more lines, or one or more areas.
3. The delineated feature can have only one feature type and must have the same set of characteristics and choices of values throughout its extent (including values for geographic name, surface elevation, and metadata, which are discussed in later sections).

Delineations of linear and areal features of different types may overlap. Where they overlap, they use the same lines or areas for their delineations. For example (see figure 2) the delineation of linear features of the types canal/ditch and bridge use the same lines where they overlap. Similarly, the delineations of areal features of the types lake/pond and swamp/marsh share the same areas where they overlap.

Features delineated with lines have two additional rules: they may not have branches, and they must start and stop at decision or merge points along a network. These points exist where a path represented by a network can branch among two or more choices. For example, a decision point exists at the confluence of two stream/rivers; at the confluence, one can choose among two or more paths. Conversely, a decision point does not exist where features at different elevations cross; travel along the path of each feature is independent from that of the other. For example, a decision point does not exist where a canal/ditch passes over a stream/river.

Lines always have a direction; that is, lines trace a path between places where they start and stop. This characteristic provides a means to encode the direction of the flow of water. For features delineated with lines, for which the direction of flow is a prominent characteristic (the feature types artificial path, canal/ditch, connector, pipeline, and stream/river), and for which the direction of flow is known, the lines are oriented in the direction of the flow of water. Note that the direction of flow is not always known (for example, where source materials are ambiguous) or uniform (for example, in tidal areas), and so the lines are not always oriented in the direction of flow. In addition, along the coastline of the United States, the lines are oriented so that the water is to the right of the direction of the line.

The delineation of features stops at the borders of the United States.

The common identifier is a 10-digit integer value that uniquely identifies the occurrence of each feature. Each value occurs only once throughout the Nation. Once assigned, the value is associated permanently with its feature. When a feature is deleted, the value for its identifier is retired. The common identifier is stored in a data element named "COM_ID".

The feature types artificial path, connector, and underpass serve special functions. The artificial path and connector ensure that the hydrographic network is complete. The artificial path represents the flow of water into, through, and out of features¹ delineated using areas (that is, it serves as a centerline) and also delineates the coastline. The connector fills gaps in the delineation of other features.

An underpass and two relations ("above" and "below"), represent places where features cross at different elevations. An example is where a feature of the type canal/ditch passes over a feature of the type stream/river (see figure 3.) The canal/ditch is encoded as being above an underpass, and the stream/river is encoded as being below the (same) underpass.

Appendix A identifies the feature types that may be related to an underpass. Note that underpasses are encoded only where they can be observed from source materials.

A reach is a continuous, unbroken stretch or expanse of surface water. In the NHD, this idea has been expanded to define a reach as a significant segment of surface water that has similar hydrologic characteristics, such as a stretch of stream/river between two confluences, or a lake/pond. Reaches also are defined for unconnected (isolated) features, such as an isolated lake/pond.

Once a reach is defined for a segment of water and assigned a reach code, the reach will rarely be changed, if at all. Many activities for improving and updating the data, such as the integration of more accurate coordinate data, the replacement of linear feature delineations with areas, or the addition of smaller features, change only the alignment of existing reaches and do not require that they be redefined.

The stability of reach definition and reach code assignment makes reaches a useful foundation for geocoding observations and statistics. Changes to the surface waters (for example, the creation of a new reservoir) and corrections to erroneous delineations of reaches, of course, would change reaches and reach codes.

The NHD is the latest refinement of reaches and reach codes. Information about earlier implementations is in Appendix D

Three types of reaches are in use: transport, coastline, and waterbody reaches. A fourth type, shoreline reach, has not been implemented.

Transport reach

A transport reach represents the pathway for the movement of water through a drainage network. These reaches also are used to encode the direction in which water flows along the reach when the direction is known. They provide a basis on which locations of observations can be geocoded and linked to the drainage network.

Lines delineate transport reaches. Only lines that delineate features of the types canal/ditch, pipeline, stream/river, artificial path, and connector delineate transport reaches. For transport reaches for which the direction of flow is known, the lines are oriented in the direction of the flow of water. Note that the direction of flow is not always known (for example, where source materials are ambiguous) or uniform (for example, in tidal areas), and so some lines are not oriented in the direction of flow.

Three general rules determine the location of the ends of transport reaches: the underlying feature rule, the confluence-to-confluence rule, and the branched path rule.

The delineation of a transport reach follows that of one or more features. Where two or more features are followed, a transport reach follows delineations of :

• features of the same type (for example, exclusively stream/river) or
• features of the same type, in combination with those of types artificial path and/or connector. For an artificial path to be included, the areal feature through which the artificial path is delineated must be of the same type as the linear feature with which the artificial path makes up the reach. For example, the delineation of a reach may follow an artificial path through an areal canal/ditch and a contiguous linear canal/ditch.

A transport reach always follows the entire delineation of the underlying feature or features; the delineation of a feature is not split among reaches. Transport reaches abut and do not overlap.

In the confluence-to-confluence rule, a transport reach is a stretch of water between:

• confluences,
• a head and a confluence,
• a confluence and a mouth or other terminus, or
• a head and a mouth or other terminus.

In the application of this rule, divergences serve the same role as confluences. Reaches defined by this rule must be contiguous and may not branch (see figure 4).

Note that some confluences are not considered to be significant enough to break the delineation of a reach. Thus, although transport reaches start and end at confluences, not every confluence causes a transport reach to start or end (see Appendix E for more information).

A branched path transport reach connects reaches that enter and exit an areal feature (see figure 5). Reaches that follow this rule occur most often in large² features of type lake/pond and swamp/marsh, and they also may occur in other areal features. Artificial paths delineated within the areal feature provide the lines needed to delineate this special transport reach. The reach may branch and at times be discontiguous.

The branched path transport reach avoids the need to define flow channels, confluences and divergences, and confluence-based transport reaches in areal features. It is used where the information needed to delineate these items reliably is not available, and at other places.

The underlying feature, confluence-to-confluence, and branched path rules govern the delineation of most transport reaches. These rules, however, have exceptions. Unusual configurations of features require modification of the rules, as do the variable condition and ambiguities of information sources. In places where unusual configurations of features or ambiguities in sources occur more often, a larger percentage of reaches delineated using modified rules will be found. Appendix E lists more examples of reach delineation rules and exceptions to the rules.

Coastline Reach

A coastline reach represents a section of coastline along the Atlantic, Pacific, or Arctic Oceans, the Great Lakes, the Gulf of Mexico, or the Caribbean Sea. These reaches provide a basis for geocoding locations of observations along the coastline.

Artificial paths that follow the coastline provide the lines used for coastline reach delineation. The delineation of a coastline reach may follow one or more artificial paths. A coastline reach always follows the entire delineation of the underlying artificial path or paths; the delineation of an artificial path is not split among reaches. Coastline reaches abut and do not overlap.

The lines are usually oriented so that the water is to the right of the line. This results in a general orientation of coastline reaches northward along the Atlantic Ocean, southward along the Pacific Ocean, eastward along the Gulf of Mexico, westward along the Arctic Ocean and the U.S. side of the Great Lakes, and counterclockwise around islands.

A coastline reach is delineated for coastal islands that have a drainage network or more than 5 miles (approximately 8.06 kilometers) of isolated drainage. Other coastal islands may or may not be delineated with a coastline reach.

The ends of coastline reaches occur where transport reaches discharge into the oceans, Gulf of Mexico, Caribbean Sea, or Great Lakes (although not every point of discharge is the end of a coastline reach). At the mouth of an areal stream/river, coastline reaches end where the artificial path used to delineate the transport reach intersects the coastline.

Waterbody reach

A waterbody is a hydrographic feature delineated using areas. Reaches assigned to waterbodies are termed waterbody reaches. These reaches provide a means to geocode observations for areas of water. (In contrast, transport reaches represent the path of a flow of water and provide a means of geocoding observations along the path.) Areal delineations of features provide the areas used to delineate waterbody reaches³.

Shoreline reach

The shoreline reach has not been implemented and is discussed for information purposes only. A shoreline reach would represent all or part of the shoreline of an inland waterbody. Analogous to coastline reaches, they would provide a basis for geocoding locations of observations along the shoreline. Lines would delineate shoreline reaches. These reaches would abut and not overlap. The types of features that would provide these lines are being investigated, as are questions of the direction, relative to the water, in which these reaches should be ordered, means of handling islands, and rules for deciding where these reaches would start and stop.

A reach code is a numeric code that uniquely labels each reach. This 14-digit code has 2 parts: the first 8 digits are the hydrologic unit⁴ code for the subbasin in which the reach exists; the last 6 digits are assigned in sequential order, and arbitrarily among the reaches.

Each reach code occurs only once throughout the Nation. Once assigned, a reach code is associated with its reach permanently. If a reach is deleted, its reach code is retired. A reach code should not be altered.

Reach codes can serve to geocode an observation to a reach or a position along a reach. Observations can be geocoded to an entire reach by associating the reach code with the observation data, or to sections of a transport, coastline, or (the planned) shoreline reach by using the reach and reach code as the basis of a linear referencing system.

Reach codes are stored in data elements named "RCH_CODE". In addition to the reach code, the date on which the code was assigned in the NHD is encoded. The date is stored in data elements named "RCH_DATE".

In addition to identifying each feature, the common identifier uniquely identifies each reach. Each 10-digit integer value occurs only once throughout the Nation. Once assigned, the value is associated permanently with its reach. When a reach is deleted, the value for its identifier is retired. The common identifier is stored in a data element named "COM_ID".

Table 1 summarizes the types, related items, delineation, and underlying features of reaches.

Flow relations among transport and coastline reaches encode drainage network connectivity among reaches independently from their delineations. Relations among transport reaches define a connected hydrographic network and encode the direction of water flow among reaches. This connectivity enables hydrologic sequencing of reaches (what is upstream and downstream of a given point in the hydrographic network) and navigating the network in an upstream or downstream direction. Similarly, relations among coastline reaches allow the traversal of the coastline using reaches. Relations among transport and coastline reaches connect hydrographic networks to the coastline and enable the ordering of the mouths of these networks along the coastline. Flow relations are especially useful in applications that require information about connectivity among, but not about, the underlying delineations and coordinate positions of reaches.

Flow relations are encoded by identifying a pair of reaches that touch and describing the flow of water between the reaches. Note that the direction of flow is not always known (for example, where source materials are ambiguous); in these cases, flow relations are not created.

The common identifiers for a pair of reaches and a description of the direction of water flow between the reaches are used to encode flow relations. Five values describe the direction of water flow between transport reaches; a sixth value describes connections between pairs of touching coastline reaches or between touching transport and coastline reaches:

1. in - the first reach flows into the second reach at the top of, or at an interior point along, the second reach. (See Figure 6, Figure 7, Figure 8, and Figure 9.) Most flow relations between transport reaches are of this type.
2. out - the first reach flows from the second reach from an interior point somewhere along the second reach. (See Figure 7.)
3. bidirectional - the first and second reaches touch end-to-end, or the first reach touches an interior point along the second reach, and flow can occur in either direction. This value has not been implemented.
4. network start - the second reach starts at a head of the drainage network. (The field for the first reach is null.) (See Figure 6.)
5. network end - the first reach ends at a terminus, other than at the coastline, of a drainage network. (The field for the second reach is null.) (See Figure 6.)
6. non-flowing connection -

1. between coastline reaches. Coastline reaches are directed so that water is to their right. The flow relation between coastline reaches is ordered so that the end of a coastline reach (the first reach in the relation) connects to the beginning of, or to some point along, the next coastline reach (the second reach in the relation) that continues the traversal of the coastline, with water to the right of the traversal. Flow relations between coastline reaches allow the coastline to be traversed. (See the flow relation between coastline reaches 1 and 4 in Figure 9.)
2. between a transport and coastline reach. The transport reach connects to the coastline reach that continues the traversal of the coastline from the place where they touch. The flow relation is ordered so that the end of the transport reach (the first reach in the relation) connects to the beginning of, or at some point along, the coastline reach (the second reach in the relation) that continues the traversal of the coastline. Relations between coastline and transport reaches allow the order in which transport reaches intersect the coastline to be computed. (See the flow relation between transport reach 3 and coastline reach 4, and transport reach 6 and coastline reach 1, in Figure 9.)

(A common identifier value of "0" represents a null entry.)

Figure 7. Flow relations illustrating out and in directions.

Figure 8. Flow relations for a branched path reach.

Figure 9. Flow relations illustrating non-flowing connections.

(A common identifier value of "0" represents a null entry.)

Flow relations are stored as four data elements. The common identifier for the first reach is stored in a data element named "COM_ID_1". The common identifier for the second reach is stored in a data element named "COM_ID_2". The direction description is stored in two data elements: as a character string in a data element named "DIR_TEXT" and as a numeric alias⁵ in a data element named "DIRECTION".

Sequence numbers order flow relations of transport reaches that flow into or out of the interior of another transport reach. The reach that flows in or out is always the first reach in the flow relation; the reach along whose interior the inflowing and outflowing reach is being sequenced is the second reach in the flow relation. The flow relations are numbered sequentially, starting with 1, from upstream to downstream along the interior of the second reach.

Sequence numbers also order flow relations of transport reaches that have a non-flowing connection to the interior of a coastline reach. The non-flowing connection flow relations of intersecting transport reaches are numbered sequentially, starting with 1, from start to end along the interior of the coastline reach.

A sequence number of 0 means that the first reach touches the second reach at its end and not along its interior. The sequence number is stored with the flow relations in a data element named "SEQUENCE".

Figure 10 illustrates the sequencing of flow relations.

(A common identifier value of "0" represents a null entry.)

Transport reaches and flow relations provide the components of the drainage network. The level path builds on these elements to create a sequence of transport reaches that trace the main stem for a given flow of water. The level path traces the flow between a head and the next largest flow of water or between a head and the ocean. Geographic names often follow these paths.

Level paths are encoded by associating stream levels to transport reaches and by associating a delta level with flow relations.

The stream level is a numeric code that identifies each main path of water flow through a drainage network. Stream level is assigned by identifying the terminus of a drainage network (see Figure 11). The lowest value⁶ for stream level is assigned to a transport reach at the end of a flow and to upstream transport reaches that trace the main path of flow back to the head. The stream level value is incremented by one and is assigned to all transport reaches that terminate at this path (that is, all tributaries to the path) and to all transport reaches that trace the main path of the flow along each tributary back to its head. The stream level value is incremented again and is assigned to transport reaches that trace the main path of the tributaries to to their heads. This process is continued until all transport reaches for which flow is encoded are assigned a stream level.

For example (see Figure 12), the Mississippi River terminates at the Gulf of Mexico. The transport reaches that trace the main flow of the river, from the head to the mouth on the Gulf, are assigned a stream level of 1. The transport reaches that trace the main flow of each tributary to the Mississippi River (such as the Ohio/Monongahela Rivers) from their heads to their termini on the Mississippi River, are assigned a stream level of 2. The transport reaches that trace the main flow of each tributary to the level 2 tributaries (such as the Tennessee River, which is a tributary to the Ohio/Monongahela Rivers), from each head to each mouth on their level 2 tributary, are assigned a stream level of 3.

Ideally, each main path would trace the flow of the largest volume of water. Stream levels encoded in two predecessors⁷ to the NHD were based on flow volume data, and these paths were retained wherever possible. Data about flow volumes were not available for other reaches. Instead, the path with the same geographic name was used to determine the level path. Where a geographic name was not available or known, the longest and straightest path was used to determine a main path. Where this rule did not adequately discriminate among the choices, the rightmost (looking upstream) transport reach was assumed to be the continuation of a main path.

Stream level is stored in a data element named "LEVEL". The special value "-9998" means that a value can be applied to the transport reach but has not been specified. This value usually occurs where flow relations cannot be determined or have not been encoded. Without this information, main paths cannot be identified and stream level cannot be assigned. This value is also assigned to coastline reaches.

To help identify main paths in the flow relations table, the difference, or delta level, between the stream levels of inflowing and outflowing reaches is encoded. To calculate this value, the stream level of the second reach in the relation is subtracted from the stream level of the first reach in the relation (see Figure 13).

Delta levels of "0" mean that a relation links two reaches that are on the same main path. In most other cases, the delta level will have a value of "1". Areas of complex divergence and braided stream configurations can yield other values. A special value of "-9999" means that a value for delta level is not applicable to the relation, and this occurs when the direction is bidirectional, network start, network end, or non-flowing connection. A special value of "-9998" means that a value could not be specified because the stream level was not specified for one or both reaches that are associated by the flow relation.

The values are stored with the flow relations in a data element named "DELTA_LVL".

Metadata, or data about data, are data that describe the content, quality, condition, and other characteristics of data. Metadata answer questions such as "How current are these data?"; "How accurate are they?"; "Are there any restrictions on their use?"; "What is their coordinate system?"; and many others. Metadata help organizations manage data, advertise and share data, and make informed use of data.

Metadata for the NHD use data elements from the "Content Standards for Digital Geospatial Metadata" (Federal Geographic Data Committee, 1994). The standard allows the identity, quality, spatial data organization and reference, entity and attribute definitions, distribution sources and forms, and metadata of the data to be documented. The metadata are provided as text files.

For the NHD, a general set of metadata accompanies each set of data. This NHD (for National Hydrography Dataset) metadata provides general information that applies to all data.

Within the NHD, values differ for metadata elements such as currentness and quality. For example, features originate from more than 1,800 datasets of different vintages, and reaches from more than 2,100 datasets. In the future, many organizations will collaborate to maintain and improve the data. These new data will differ in currentness, source, accuracy, and other characteristics. A single document that provides metadata for these dynamic and varied data would be either very general or very unwieldy.

Digital update units associate specific metadata information with selected features and reaches. A digital update unit is a collection of one or more features and (or) reaches to which a set of metadata values applies. These values include only those needed to describe unique aspects of the associated features or reaches. A feature or reach may be a member of one or more digital update units.

Metadata associated with digital update units supplement information provided in the NHD metadata. These digital update unit metadata amplify, and in some cases replace, the more general information provided in the NHD metadata.

In the initial release of the NHD, there are two types of digital update units; subbasin and quadrangle digital update units (see Figure 14):

• subbasin digital update units associate reaches with metadata about sources of information and processing steps used to construct reaches and analysts' notes about unique situations encountered in the process.
• quadrangle digital update units associate features with metadata about the sources of information (including the currentness of the data) and processing steps used to construct features.

As development and improvement of the data continue, additional digital update units will be created. These may have collections of features, reaches, and shapes different from the digital update units in the initial release. These new digital update units also may have metadata elements different from those included in the digital update units in the initial release.

Digital update units are assigned unique identifiers and stored in data elements named "DUU_ID".

A geographic name is "the proper name, specific term, or expression by which a particular geographical entity is, or was, known" (Orth and Payne, 1997, p. 43). Geographic names designated as being official for Federal use are encoded for many reaches and features. These names were taken from the National Geographic Names Database of the Geographic Names Information System⁸ (USGS, 1995), the Federal Government's primary source for identifying official names.

Reaches and features can carry geographic names (see Figure 15 and Table 2). Reaches carry geographic names most often. If there is no reach to carry a name, then the feature carries the name. Appendix A lists the types of features that may carry names.

Geographic names are stored in data elements named "NAME". In addition, the eight-digit identifier for a name in the Geographic Names Information System is stored in data elements named "GNIS_ID".

Note that geographic names have not been applied to all reaches and features. Names are applied to transport and coastline reaches that were named in the USEPA's Reach File version 3 (see Appendix D). Names are also applied to waterbody reaches and features delineated using points or areas in cases where batch processes can reliably associate the name with the reach or feature.

(Notes for table: (1) Type: The table lists all reach and feature types defined for the NHD. Note that not all types have been collected in the initial release. (2) Geographic name: An "X" in the column indicates that a geographic name may be associated with reaches or features of this type. Note that geographic names have not been encoded for many reaches and features. (3) An asterisk (^*) means that the feature or names are not in the initial release of the NHD. (4) Coastline reaches carry the name of the water feature that they bound. (5) For the State of Washington, waterbody reaches carry the names of the features types ice mass, lake/pond, reservoir, and swamp/marsh. Elsewhere, waterbody reaches carry the names of the features type lake/pond only.)

(Adapted from "Principles, policies, and procedures: domestic geographic names" (Orth and Payne, 1997, p. 3); "The national gazetteer of the United States of America: concise: (USGS, 1990, p. xi); and "Geographic Names Information System" (USGS, 1995, p. 7-8).)

Geographic names normally originate in and are influenced by spoken language. It is important to remember this fact because many people are concerned with written forms of names, including matters of spelling, capitalization, word form, and writing marks, that may have little to do with the way names are spoken. Geographic names in the United States most often reflect English, French, and Spanish naming traditions.

Most geographic names are binomial in that they have two parts, denoting specific and generic information. The generic part tells the kind of place, feature, or area to which the name refers and usually is a single topographic term, such as brook, creek, lake, spring, or river. The specific part uniquely identifies the particular place, feature, or area, and it may consist of one or more words. For example:

The binomial (two-part) form of specific and generic information is strong and in written usage often leads to combining words in the specific part of the name, such as Threemile Run and Redhill Gulch. The names of some features can be long, especially if the specific part is a prepositional phrase: Cliffs of the Seven Double Pillars, Foot of the Mountain Run, and Cañon del Rajadero de los Negros.

Some geographic names have a "false" generic that does not describe the feature correctly. For features of the type stream/river, common false generics are terms for depressions in the Earth's surface (such as "draw", "gully" or "gulch") that usually are not defined as applying to water flowing through the depression.

Some names have rare generic forms. Examples include colorful American names such as Bowl of Tears (lake), Butlers Toothpick (pinnacle rock), Titans Piazza (hill), and Devils Racepath (ridge). Among variations of the binomial form are one-word names that require a capitalized article: The Canal, The Lagoon, The Lakes, and La Laguna.

The entries for geographic names include uppercase and lowercase alphabetic, numeric, and punctuation and other special characters. Most names are entered in the way they are commonly written; for example "Adams Creek" and "Green Lake". Exceptions include the following:

• Diacritical mark. A diacritical mark is a mark near or through a character or combination of characters. The mark indicates a phonetic value (a spoken sound) different from that given the unmarked or otherwise marked character. For example, the acute accent ("´") is the diacritical mark in río, and the tilde ("~") is the diacritical mark in cañada and cañoncito. If the entry for a name has a diacritical mark in the Geographic Names Information System, the mark is retained in the NHD⁹.
• Generic part entered last. Sometimes the generic part of the name may be in the first position, followed by the specific part. For example, the name Rio Grande del Rancho begins with the generic "rio" (from the Spanish río for river). In most of these cases, the geographic name is entered with the position of the generic within the name reversed. For example, Rio Grande del Rancho becomes "Grande del Rancho, Rio". The generic, however, is not reversed if the specific part of the name contains an Arabic number.

• Initial definite articles entered last. Some geographic names start with a definite article (for example, "a" and "the" in English, and "el", "la", "los", and "las" in Spanish). Examples include El Capitan Wash and The Cat Hole. In many of these cases, the geographic name is entered so that the initial definite article is provided last. For example, El Capitan Wash becomes "Capitan Wash, El" and The Cat Hole becomes "Cat Hole, The".

The common identifier is a 10-digit integer value that uniquely identifies each feature or reach. Each value occurs only once throughout the Nation. Each feature and reach is assigned an identifier, stored in the data element named "COM_ID". Once assigned, the identifier is associated permanently with its feature or reach and should not be altered. When a feature or reach is deleted, its identifier is retired.

Common identifiers serve several purposes. They are the basis for relating features and reaches for several purposes, including the following:

• underpasses: the data elements "ABOVE_ID" and "BELOW_ID" store the common identifiers of the features above and below an underpass respectively.
• reach flow relations: the data elements "COM_ID_1" and "COM_ID_2" store the common identifiers of the reaches that participate in a flow relation.
• linear features and their transport and coastline reaches: the data element "RCH_COM_ID" stores the common identifier of the transport or coastline reach that underlies a linear feature.
• waterbodies and their waterbody reaches: the data element "RCH_COM_ID" stores the common identifier of the waterbody reach that underlies a waterbody.
• waterbodies and linear features they contain: the data element "WB_COM_ID" stores the common identifier of the waterbody that contains a linear feature.
• features and reaches, and their digital update units: the data element "COM_ID", when associated with the data element "DUU_ID", associates features and reaches with their digital update units.

Common identifiers are also the basis for tracking and sharing deletions, additions, and modifications of features, reaches, and relations.

Positions are encoded using a common coordinate referencing system. Other measurements, including lengths, areas, and selected surface elevations, also are encoded.

The locations of points, lines, and the boundaries of areas are encoded using geographic (longitude-latitude) coordinates. The coordinates are encoded in decimal degrees, with west longitude and south latitude represented by negative values. The horizontal datum is the North American Datum of 1983 (NAD83).

Lengths of linear features and areas of areal features are supplied for convenience in applications that require these measures. The coordinate data were projected from longitude-latitude coordinates to the Albers Equal Area projection, and lengths and areas were calculated and saved. The parameters for the projected coordinate systems are shown below:

The North American Datum of 1983 was specified for the projected coordinate system. The distance units of the projected system were meters.

Lengths are computed in meters and stored in data elements named "METERS". Areas are computed in square kilometers and stored in data elements named "SQ_KM".

The elevation of water surfaces where water pools is encoded for a few features. Water surface elevations may be encoded for features of the type area to be submerged, canal/ditch, inundation area, lake/pond, reservoir, and stream/river. Note that these elevations have not been entered for most features. The vertical datum for the conterminous United States and Alaska is the National Geodetic Vertical Datum of 1929. For Puerto Rico and the Virgin Islands, the datum is Local Mean Sea Level.

Elevations are recorded in meters and stored in data elements named "ELEV". Two encoded values have special meanings. A value of -9999 means that an elevation value will not be applied to the feature. A value of -9998 means that a value can be applied to the feature but has not been specified.

In addition to the elevation, the prevailing condition (for example, "average water elevation") to which the elevation applies is provided. The condition is stored as a text string in data elements named "STAGE". Where the "ELEV" field has a value of -9999 or -9998, the "STAGE" field is blank.

The NHD resulted from a multiyear effort to process and integrate datasets containing delineations and classification of features, reaches, hydrologic units, and geographic names. This section provides an overview of the production processes, sources of information used in these processes, and other statements related to data quality.

This section describes what is "generally true" about the initial release of the data for the conterminous United States, Hawaii, Puerto Rico, and the Virgin Islands. The NHD, even in its initial release, has variations in currentness, processes, and other characteristics. It is important to note that, even before the initial release of the data, collaborative efforts are under way to correct and improve the data. Completion of these efforts will contradict even these "generally true" statements for places where the improvements are made. It is important to review the metadata to understand the condition of data.

The discussion below applies to the conterminous United States (excluding the Pacific Northwest¹⁰), Hawaii, Puerto Rico, and the Virgin Islands.

Lineage

Figure 16 illustrates the information sources and processes used to develop the data.

Sources of information used to construct the initial release of the NHD include:

• Digital line graph 3 (DLG-3) data. These data, captured from USGS topographic maps and unpublished source materials, provide the delineations and classification of features (except for artificial paths in areal features and connectors). The data were organized in the DLG optional format, were tiled in quadrangles, were edge-matched, were from the DLG "hydrography" data category, and were in the Universal Transverse Mercator coordinate system, North American Datum of 1927 (NAD27). The scales of the map source materials used for the initial release of the data are shown below:

• Reach File Version 3 (RF3) data. These data, developed by the USEPA, provide the starting point for reach delineation, reach codes, direction of water flow information, and positions of geographic names for the feature type stream/river. These data were developed on 1:100,000-scale DLG data for the conterminous United States and Hawaii as part of a previous project.
• Subbasin boundaries. These data, developed by different agencies, are used to assign features and reaches to subbasins.

• Geographic names.
• Final version. Extracted in March 1999 from the National Geographic Names Database of the USGS's Geographic Names Information System, these data were the final source of geographic name entries.
• Earlier versions. Two earlier extracts from the Geographic Names Information System were used to assign names to reaches (December 1996 extract) and waterbodies (February 1995 extract).
• Features. These quadrangle-tiled data, containing features and characteristics, were created from the DLG-3 data.
• California and Arizona updates. These data from California and Arizona State agencies contain additional name assignments and minor spatial and attribute corrections to Reach File Version 3 data.

This process converted DLG data to features and associated characteristics and converted the coordinate system to geographic (longitude-latitude) coordinates in the NAD83 in five steps:

1. The USGS's "Batch DLG-3 to DLG-F Conversion System" converted DLG-3 nodes, lines, areas, and associated attribute codes to temporary features and associated characteristics. Known conditions for which conversions could not be reliably made were flagged for later inspection. Only known conversion problems were flagged, and no additional steps were taken to detect or repair discrepancies in the original DLG-3 or the converted NHD.
2. A default value of a characteristic was added in cases where the description was incomplete.
3. All instances in which data were flagged were inspected and resolved interactively.
4. Feature delineation rules were applied to the temporary features in a batch process to create the final version of features.
5. Coordinate values were converted to geographic coordinates and to the NAD83 using the NADCON software version 2.1 (National Geodetic Survey, n.d.).

This process generated the "features" data.

The basic steps for building reaches are as follows:

1. Convert RF3 to RF3" (RF3 double prime). This batch operation processed Reach File Version 3 to delete duplicate reaches, reassign reaches to the correct subbasin, validate geographic names assigned to reaches against data from the Geographic Names Information System (December 1996 extract), apply updates supplied by the States of California and Arizona, redelineate reaches on the basis of standards used for the NHD, and identify inflow/outflow points where transport reaches entered and exited waterbodies.
2. Create artificial paths. Using waterbodies from the feature data and inflow/outflow points extracted from RF3", this process automatically generated the centerlines used to delineate artificial paths within waterbodies by using subroutines within the ARC/INFO ®¹¹ GRID module.
3. Blind pass. This batch step conflated features and RF3" reaches and transferred reach information (reach code, reach date, name, stream level, and flow relationships) to the features. It also integrated the artificial paths generated in the previous step with the other features, built reaches on the artificial paths, and assigned geographic names (February 1995 extract) to waterbodies.
4. Quadrangle-based visual pass. During this interactive step, analysts ensured that the data developed in the previous batch processes conformed to reach delineation rules and that reaches were assigned to the appropriate subbasin. Batch procedures identified and developed a list of possible errors. (Errors not detected by the software may continue in the data.) Using the list, software presented each case to analysts to make appropriate edits to the data. Analysts recorded notes about repairs that could not be made and about other errors in the data. (These notes are encoded in the subbasin digital update units.)
5. Build superquads. After the quadrangle-based visual pass was complete, all quadrangles that cover all or part of each subbasin were paneled into a superquad. In this batch process, reaches that cross quad boundaries were corrected to conform to reach delineation rules.
6. Subbasin-based visual pass. As they did with the quadrangle-based visual pass, analysts ensured that reaches conformed to reach delineation rules. Batch procedures identified and developed a list of possible errors. (Errors not detected by the software may continue in the data.) Analysts examined each error and corrected the data. Analysts recorded notes about repairs that could not be made and about other errors in the data. (These notes are encoded in the subbasin digital update units.)
7. Central quality assurance/quality control. This step (1) confirmed that integrity checks were performed successfully during the visual pass activity, and (2) assessed statistics gathered during the earlier processes to determine if additional review of data was needed. A check of data from the subbasin-based visual pass was run in batch; any data that did not pass the procedure were reviewed interactively. If substantive changes were required, the data were reprocessed using procedures (as required) described in previous steps. The edited data then were rechecked using the central quality assurance/quality control process.
8. Data preparation and database load. This batch procedure performed final processing to the data emerging from the quality assurance/quality control step. Some of the activities included assigning the final reach codes, building waterbody reaches, adding final artificial paths in waterbodies, and implementing any recent changes in standards for the NHD. The spelling of geographic names was replaced using the March 1999 data extract from the Geographic Names Information System. After this, reaches, features, characteristics, geographic names, and relations were loaded into the database that holds the NHD.
9. Flow relation correction and validation. The flow relations were checked for consistency through a batch procedure, which generated a list of possible errors. Software presented possible errors to analysts, who corrected flow relations and, occasionally, the delineation of reaches. Changes were posted to the database.
10. Extract distribution copies of data. Data for a subbasin were extracted from the database and converted into an ARC/INFO ® workspace containing coverages and other files. Data available in the Spatial Data Transfer Standard format were developed from the workspaces. The workspaces and the Spatial Data Transfer Standard-formatted files were made available to the public.

Attribute accuracy

The accuracy of the attributes of the DLG data is estimated to be 98.5 percent. One or more of the following methods was used to test attribute accuracy:

• the source was manually compared with hardcopy plots.
• a symbolized display was done of the DLG on an interactive computer graphics system.

• selected attributes that could not be visually verified on plots or on screen were interactively queried and verified.

In addition, software was used to validate feature types and characteristics against a master set of types and characteristics, to check that combinations of types and characteristics were valid, and to check that types and characteristics were valid for the delineation of the feature. Feature types, characteristics, and other attributes conform to the "Standards for National Hydrography Dataset" (USGS, 1999) as of the date they were loaded into the database.

The entry and identifier for geographic names match those in the Geographic Names Information System as of March 1999. The association of each name to reaches has not been methodically checked, and so a name may be applied to the wrong reaches. Anecdotal reviews indicate that 80 percent or more of the named reaches have the correct name.

Reaches were delineated with a batch procedure and were checked extensively during the "visual pass" steps of processing. According to automated quality assurance/quality control checks performed at various intervals during the processing, it is estimated that approximately 99 percent of the reaches are delineated according to standards.

Logical consistency

Points, nodes, lines, and areas conform to topological rules. Lines intersect only at nodes, and all nodes anchor the ends of lines. Lines do not "overshoot" or "undershoot" other lines where they are supposed to meet. There are no duplicate lines within a dataset. Lines bound areas, and lines identify the areas to the left and right of the lines. Gaps and overlaps among areas do not exist. All areas close.

Completeness

The completeness of the data reflects the content of the sources, which in the initial release of the NHD, are most often USGS topographic maps. Features found on the ground may have been eliminated or generalized on the source graphic because of scale and legibility constraints. In general, streams longer than 1mile (approximately 1.6 kilometers) were collected. Most streams that flow from a lake were collected regardless of their length. Only definite channels were collected, so not all swamp/marsh features have stream/rivers delineated through them. Lake/ponds having an area greater than 6 acres (approximately 2.4 hectares) were collected.

Note, however, that these general rules were applied unevenly among maps during compilation. Some map quadrangles have a much sparser pattern of hydrography than do adjoining maps, and these differences continue in the digital rendition of these features. Rectification of these differences is a priority for maintenance of the NHD.

Transport reaches are defined on almost all of the features of types stream/river, canal/ditch, pipeline, artificial path, and connector. Waterbody reaches are defined on the subset of lake/pond features that were identified as waterbodies during the development of Reach File Version 3.

Most attention in applying geographic names was given to transport reaches that follow stream/rivers and to waterbody reaches.

Near the international boundaries with Canada and Mexico, only the parts of features within the United States are delineated.

Detailed capture conditions are provided for every feature type in the "Standards for National Hydrography Dataset" (USGS, 1999), available online through http://mapping.usgs.gov/standards/.

Horizontal positional accuracy

Statements of horizontal positional accuracy are based on accuracy statements made for USGS topographic quadrangle maps. These maps were compiled to meet National Map Accuracy Standards. For horizontal accuracy, this standard is met if at least 90 percent of points tested are within 0.02 inch (at map scale) of their true positions¹². Additional offsets to positions may have been introduced where there are many features to improve the legibility of map symbols. In addition, the digitizing of maps is estimated to contain a horizontal positional error of less than or equal to 0.003-inch standard error (at map scale) in the two component directions relative to the source maps. Visual comparison between the map graphic (including digital scans of the graphic) and plots or digital displays of points, lines, and areas is used to assess the positional accuracy of digital data.

Linear features of the same type along the adjoining edges of datasets are aligned if they are within a

0.02-inch tolerance (at map scale). To align the features, the midpoint between the end of the corresponding features is computed, and the ends of features are moved to this point. Features outside the tolerance are not moved; instead, a feature of the type connector was added to join the features.

Vertical positional accuracy

Statements of vertical positional accuracy for elevation of water surfaces are based on accuracy statements made for USGS topographic quadrangle maps. These maps were compiled to meet National Map Accuracy Standards. For vertical accuracy, this standard is met if at least 90 percent of well-defined points tested are within one-half contour interval of the correct value. Elevations of water surface printed on the published map meet this standard; the contour intervals of the maps vary. These elevations were transcribed into the digital data; the accuracy of this transcription was checked by visual comparison between the data and the map.

Similar approaches are being used to develop data for the Pacific Northwest and Alaska. Specific statements will be provided when these data are released.

decimal degree = degrees + (minutes/60) + (seconds/3600). See latitude, longitude.

Notes on table: (1) Feature type: The table lists all feature types and characteristics defined for the NHD. Note that not all feature types have been collected. (2) Geo(graphic) name: An X in the column indicates that a geographic name may be associated with features of this type. Note that geographic names have not been encoded for many features. (3) For the State of Washington, waterbody reaches carry the geographic names of lake/pond, reservoir, ice mass, and swamp/marsh. (4) Surface elevation: An X in the column indicates that the elevation of the surface of the water may be associated with the features of this feature type. Note that the surface elevations have not been encoded for most features. (5) An asterisk (^*) means that the item is not in the initial release of the NHD.