Return
to Geospatial Technology Documents
Datums
And Projections: A Brief Guide
March 1999
Scientists and resource
managers often use or create data that are referenced to locations on
the earth (geospatial data). Users of geospatial data need to know the
coordinate system or projection the data are in so they can make accurate
measurements from the data or overlay different datasets for analysis.
Frequently, users or creators of geospatial data are unaware or unsure
of the projection or datum geospatial data are in, or which projection
to select when creating geospatial data (e.g., determining coordinates
of bird nest sites with a GPS receiver). This paper is written for the
scientist or resource manager who uses geospatial data occasionally
and needs a quick, non-technical guide on the use of datums and projections.
This paper defines important terms, presents a brief discussion on widely-used
projections, suggests ways to determine what datum or projection geospatial
data are in, recommends how to select the proper datum or projection
for geospatial data, and proposes methods to convert between datums
and projections.
Definitions
Ellipsoid
An ellipsoid is a mathematical figure generated by the revolution
of an ellipse about one of its axes (DMA, 1999). The earth is not a
sphere but an ellipsoid distorted by rotation about its axis, with the
globe bulging at the equator and flattened at the poles. The actual
amount of the flattening is about 21.5 km difference between the polar
and equatorial radii (Gaus, 1999). Ellipsoidal earth models are required
for accurate range and bearing calculations over long distances. For
example GPS navigation receivers use ellipsoidal earth models to compute
position and waypoint information. Ellipsoidal models define an ellipsoid
with an equatorial radius and a polar radius. The best of these models
can represent the shape of the earth over the smoothed, averaged sea-surface
to within about one-hundred meters (Dana, 1999).
Datum
A datum is a mathematical model that describes the shape of the ellipsoid.
The flattening is not uniform around the earth; the shape is different
due to the influence of the continents and there are datums for different
parts of the earth based on different measurements (Snyder, 1982).
Projection
A map or a geospatial database is a flat representation of data
located on a curved surface. A projection is a device for producing
all or part of a round body on a flat sheet. This projection cannot
be done without distortion, so the cartographer must choose which characteristic
(distance, direction, scale, area, or shape) that is to be shown accurately
at the expense of the other characteristics, or compromise on several
characteristics. (Snyder, 1982). Projections can be used with different
datums; for example the Universal Transverse Mercator (UTM) projection
could be based on the North American Datum of 1927 or North American
Datum of 1983.
Geographic System
The Geographic System is not a projection, but a representation
of a location on the earths surface using spherical coordinates
as measured in latitude and longitude the angle of a location
on the earths surface from the earths center in relation
to the equator and the Prime Meridian that intersects Greenwich, England.
The locations in the Geographic System can be converted into other projections,
but the locations represented by the Geographic System cannot be used
for determining basic information such as distance or direction between
two points. While the distance between degrees in latitude remain fairly
constant over the earth, one degree in longitude is much longer near
the equator than it is near the poles (111 km at the equator versus
19 km at 80o North (Gaus, 1999)).
Widely Used Datums
and Projections
Hundreds of different
datums have been developed since Aristotle made the first estimates
of the earths size (Dana, 1999). For many years the North American
Datum of 1927 (NAD27) was the standard in the United States. NAD27 was
based on the Clarke Ellipsoid of 1866, which was developed from ground
survey in Europe and North America in the 19th Century. During
the 1970s and 1980s satellites were able to measure the
ellipsoid flattening more accurately (the World Geodetic System ellipsoid
of 1984 or WGS84) and a new datum was developed from these measurements
called the North American Datum of 1983 (NAD83). The Global Positioning
System is based on WGS84. The earths shape is changing and measurements
are becoming more precise, and as a result new reference ellipsoids
and datums are being developed: the ITRF or International Earth Rotation
Service (IERS) Terrestrial Reference Frame is the latest ellipsoid that
has been developed.
At least 250 different
projections have been devised and described by cartographers (Gaus).
Projections fall into four major classes: cylindrical, conical, azimuthal,
and miscellaneous. Cylindrical projections result from projecting a
spherical surface onto a cylinder while conic projections result from
projecting a spherical surface onto a cone. Azimuthal projections result
from projecting a spherical surface onto a plane and miscellaneous projections
include systems such as rectangular latitude and longitude grids (Dana,
1999) that do not fall into the other three categories.
The Albers Equal-Area
Conic Projection is based on a conic representation of a portion
of the earth. It is mathematically projected on a cone conceptually
secant (a plane intersecting a curve at two points) to two standard
parallels. The Albers projection is used by the USGS for its state maps
and all U. S. maps of 1:2,500,000 scale or smaller. It is useful for
representing areas that have a predominant east-west expanse, such as
the conterminous United States. All areas on the map are proportional
to the same areas on a globe, directions are reasonably accurate in
limited regions, and distances and scale are true only along the standard
parallels.
The Lambert Conformal-Conic
Projection has many of the same characteristics as the Albers Equal-Area
Conic Projection. It is mathematically projected on a cone conceptually
secant at two standard parallels. It is useful for representing countries
or regions that have a predominant east-west expanse. It is used in
the State Plane Coordinate System for North Carolina and other states.
All areas on the map are proportional to the same areas on a globe,
directions are reasonably accurate, and distances and scale are true
only along the standard parallels.
The Lambert Azimuthal
Equal-Area Projection was derived for the specific purpose of maintaining
equal area. It is mathematically projected on a plane tangent (a plane
touching a curve at one point) to any point on the globe. It is useful
for representing areas that extend equally in all directions from center
points, such as Asia, and the Atlantic Ocean. Areas are in true proportion
to the globe, while direction is true only from the center point, and
scale decreases gradually from the center point.
The Universal
Transverse Mercator Projection is based on the Transverse Mercator
projection. It is mathematically projected on a cylinder tangent to
a meridian (which is transverse or crosswise to the equator). It was
adopted by the military in 1947 for designating coordinates on large-scale
military maps for the entire world. The world is divided into 60 east-west
zones each 6o wide in longitude. It is used by many 1:24,000
scale to 1:250,000 scale USGS quads and is useful for mapping large
areas that are oriented north-south. The projection is conformal so
shapes and angles within any small area (such as a USGS 7.5 quadrangle)
are essentially true.
The State Plane
Coordinate System (SPCS) is not a projection, but a coordinate system
that divides all 50 states into 120 zones, which are represented by
different projections, based on the shape of the zone. Three projections
are used: the Lambert Conic Conformal for east-west oriented zones (e.g.,
zones in Tennessee), the Universe Transverse Mercator for north-south
oriented zoned (e.g., zones in Alabama), and the Oblique Mercator for
the panhandle of Alaska.
Determining the
Datum or Projection for Existing Data
Often scientists
use geospatial data created by someone else, and frequently must overlay
datasets that are in different or unknown datums/projections. How is
the datum/projection determined? There are four methods to determine
the datum/projection: 1) refer to the metadata accompanying the dataset,
2) query the dataset using your geospatial software package, 3) refer
to the header file for the dataset, or 4) visually compare the dataset
in question to a dataset of known datum/projection.
Metadata,
or data about data, should contain information on the datum and projection
of a geospatial dataset. If metadata files follows FGDC requirements,
they will contain this information. Often, however, metadata are incomplete
or missing altogether; in that case another method must be used.
The software
used to view and manipulate the geospatial data will often allow
queries to determine which datum/projection a dataset is in. For example,
in ARC/INFO, the command "describe" will provide this information,
if the datum/projection has been defined by the originator of that dataset.
ArcView, on the other hand, will not allow the user to query the dataset
to determine the projection or datum. Other software packages such as
ERDAS Imagine, PCI, ENVI, or GRASS also allow the user to query the
dataset to determine the datum/projection.
The header of
a dataset may contain the information, if the dataset has a separate
header. Headers are common for raster datasets. For example, the header
of a USGS DOQQ contains the information on datum/projection, along with
other information. A text editor can read the header.
When other methods
fail, visually compare the dataset with a dataset of known datum/projection
that has similar features in the software program. For example, a landcover
vector dataset could be overlain on a DOQQ to allow comparison of features
such as road intersections or permanent shorelines common to both layers.
A GPS receiver that is set to a known reference system, can collect
coordinates at well-defined locations, such as road intersections and
those data can be overlain on the dataset in question. If these points
do not align, the datasets are in different datums/projections. If the
locations are approximately 200 meters apart in the north-south direction
and slightly offset in the east-west direction, the difference is commonly
due to the southern dataset being in NAD27 and the northern dataset
in NAD83. If the datasets are hundreds or thousands of miles apart,
the projections are different.
In any event, it
is a good idea to visually compare a new dataset with a known dataset,
even if datum/projection information exists to ensure that the information
is correct.
Selecting Datums/Projections
In the conterminous
Unites States NAD27 and NAD83 are the two common choices for datums.
NAD27 was the standard for many years because the USGS published its
topographic maps in this datum, principally the 1:24,000 scale 7.5-minute
quadrangle series, which served as a reference map for many datasets.
When the USGS began publishing digital data, particularly the digital
ortho quarter quads (DOQQs), the datum used was NAD83, which provided
a more accurate representation of the earths shape and a more
accurate depiction of the location of objects on the earth. This created
the problem of datasets that were generated in NAD27 that were not registered
correctly to the new, more accurate, higher resolution DOQQs.
As the DOQQs begin to cover significant portions of the United
States, this issue has become important for many scientists and data
centers who use the DOQQs as a base layer in their geospatial
databases. In addition, datasets in NAD27 do not join together well
across large regions (such as a state or a large ecoregion), due to
the inaccuracies of that datum (Brown, 1999).
When creating a
new geospatial dataset, it is critical to select the proper projection.
Selection considerations are 1) the extent of the project area: (e.g.,
the world, a continent, or an ecoregion), 2) the location of the project
area (e.g., polar, mid-latitude, or equatorial), and 3) the predominant
extent of the project area (e.g., circular, primarily east-west axis,
north-south axis, or oblique axis). If a base layer, such as DOQQs
or USGS 1:100,000 scale maps, is used, the base layers projection
may determine the projection. For small areas, such as a 7.5-minute
quad, or a DOQQ, the UTM projection is nearly correct in every respect
(USGS, no date). For continental or smaller regions in temperate zones
that 1) have an east-west axis select the Lambert Conformal projection
for conformal accuracy and the Albers Equal Area for areal accuracy
2) have a north-south axis select UTM for conformal accuracy, 3) have
an oblique axis select the Oblique Mercator projection for conformal
accuracy, and 4) have equal extent in all directions select the Polar
or Stereographic Projections for conformal accuracy and the Lambert
Azimuthal Projection for areal accuracy.
Conversions between
Datums/Projections
The best way to
convert between datums/projections is to use geospatial softwares
conversion or projection commands. The algorithms used by geospatial
software vary in quality, so the user must be aware of the algorithms
the software uses. The best conversion programs are based on the National
Geodetic Surveys NADCON (North American Datum Conversion). NADCON is
a reasonably accurate (errors of less than 0.5 meters) way of converting
between NAD27 and NAD83 (although there is no exact transformation between
NAD27 and NAD83 because NAD27 has different errors in different parts
of North America). Programs such as ARC/INFO have sophisticated commands
to convert between projections and datums, while programs such as ArcView
do not support conversion in their basic modules (although the Projector
extension can convert coverages and shapefiles to shapefiles in different
projections). Image processing software such as ERDAS Imagine and PCI
also have conversion capabilities. The U.S. Army Corps of Engineers
has developed Corpscon, a MS-Windows-based program which allows the
user to convert coordinates between Geographic, State Plane and Universal
Transverse Mercator (UTM) systems on NAD 27, NAD 83 and High Accuracy
Reference Networks (HARNs). Corpscon uses NADCON to convert between
NAD 27, NAD 83 and HARNs (http://crunch.tec.army.mil/software/corpscon/corpscon.html).
Another method to
convert data from one projection to another, if projection information
does not exist for the dataset in question, is to "rubber-sheet"
the dataset to a base dataset. This is accomplished by visually connecting
similar points in the two datasets and allowing the software to adjust
the dataset. This is not a perfect method, but is useful if the data
do not have projection information, or are not projected initially (such
as a scanned aerial photograph).
Conclusions
Common errors when
using or creating geospatial data are 1) to overlay datasets that are
not in the same datum/projection, 2) collect field data with a GPS receiver
that are in a different projection than the base data layers or to collect
the data in different datums/projections on different days due to neglect
or change of the settings, 3) digitize new data in an incorrect datum/projection,
4) use an inappropriate datum/projection for a project, or 5) convert
datums in GPS receivers that have a poor conversion algorithm. Again,
geospatial data that are in the same projection may be in different
datums. It is critical to carefully check spatial datasets to ensure
that they are the proper datum/projection. Useful Internet sites for
more detailed information are listed in the references.
Karl Brown
Geospatial Technology Specialist
USGS Center for Biological Informatics
karl_brown@usgs.gov
303-202-4259
References
Brown, K.E. 1999.
Personal communication.
Dana, P.H. 1999.
Map Projection Overview. http://www.colorado.edu/geography/gcraft/notes/mapproj/mapproj_f.html.
Defense Mapping
Agency. 1999. DMA Technical Manual 8358.1 Datums, Ellipsoids, Grids,
and Grid Reference Systems. http://164.214.2.59/GandG/tm83581/tr83581a.htm.
Gaus, M.P. 1999.
CIE 303 Geodesy, Surveying, GIS and GPS Course http://overlord.eng.buffalo.edu/ClasshomePages/cie303/index.htm.
Snyder, John P.
1982. Map Projections Used by the U.S. Geological Survey. 2nd
Edition. United States Government Printing Office, Washington, D.C.
313 pages.
U.S. Geological
Survey. No date. Map Projections. Poster. U.S. Geological Survey National
Mapping Division. Reston, VA.
U. S. Geological
Survey. 1996. National Mapping Program Technical Instructions, Part
2, Specifications, Standards for Digital Orthophotos. U.S. Geological
Survey National Mapping Division. Reston, VA, 37 pages.
|