Metadata for Riparian Vegetaion in Individual Classified Landsat Scenes, U.S. Forest Service Northern Region, 1992 Metadata for Riparian Vegetaion in Individual Classified Landsat Scenes, U.S. Forest Service Northern Region, 1992

Download full XML Metadata

Identification Information:

Citation:

Originator:

University of Montana
Wildlife Spatial Analysis Laboratory

Publication date: 06/12/1996

Title:

Riparian Vegetaion in Individual Classified Landsat Scenes, U.S. Forest Service Northern Region, 1992

Publication place: Missoula, Montana
Publisher: University of Montana

Larger Work Citation:

Title:: Classified Landsat Thematic Mapper Data, U.S. Forest Service Northern Region, 1992
Online linkage: http://mslapps.mt.gov/Geographic_Information/Data/DataList/datalist_Details.aspx?did={a631df49-bc51-4558-851f-6a3a1a1c02a6}

Browse graphic file name: http://docs.msl.mt.gov/geoinfo/images/metadata/wsal2.gif
Browse graphic file description: Image of data extent and data example
Browse graphic file type: gif

Abstract:

These data consist of 18 overlapping raster datasets in Esri GRID format derived from 18 Landsat Thematic Mapper scenes imaged between July 6, 1991 and September 25, 1993. The value of each cell represents a zone that the cell belongs to and there is an attribute table giving information about each zone. There are hundreds of thousands of zones in each scene. At the time the data were added to the State Library archive, Esri software seemed to be incapable of displaying the data according to the attribute values of the zones, but the identify function allowed the zone attributes of each pixel to be displayed.

The scenes were run through an unsupervised classification to reduce the approximately 56 million pixels in each scene into approximately 300,000 spectral class zones, each zone having one of from 18-37 spectral classes per scene. Only zones inside Riparian areas (as identified through analysis of hydrography and digital elevation models) are included. Each zone has attributes for the mean elevation, slope, and reflectance in the 7 Landsat Thematic Mapper spectral channels for all the pixels in the zone, and the majority aspect of all the pixels in the zone.

Additionally, each zone has attributes for which riparian landcover type the zone was subsequently classified into. These classifications are based partly on a supervised classification of the imagery with ground-truthed training sets and partly on manual classification of individual areas as urban, agricultural, or clouds. The mean Modified Normalized Difference Vegetative Index (MNDVI) of each zone is also available.

A separate data set is available with a classification of the non-riparian areas.

This metadata record documents 18 Landsat Scenes that are similar but not identical. In some cases, the values given are for a typical scene and may not exactly apply to every scene.

The lab published an executive summary and a in-depth report on the project and its result. The executive summary is at https://ftpgeoinfo.msl.mt.gov/Documents/Metadata/SILC_1_Summary.pdf and the report is at https://ftpgeoinfo.msl.mt.gov/Documents/Metadata/SILC_1_Report.pdf.

Purpose:

These data were produced for the U.S. Forest Service Northern Region to map existing vegetation and land cover in a standardized, consistent manner across Northern Idaho and Western Montana. By providing an integrated GIS database, users can query and process information contained in many layers in a cost-efficient manner and at a controlled level of accuracy. The databases can be edge-matched seamlessly to examine larger areas at once using the clipper.aml program supplied with the data. Thus, many problems associated with conventional methods of mapping large areas have been solved by the methodology developed for this project. The results provide a powerful tool for the Forest Service and other land management agencies to examine vegetation patterns across large areas of mixed ownerships and to conduct the following types of analyses and applications: broadscale assessment and planning efforts, such as the Columbia River Basin project; revision of National Forest Plans (Analysis of the Management Situation, cumulative effects and trade-off analyses, existing condition, alternative development) identification of land management projects that will achieve the ultimate goals of producing goods and services for people within the sustainable capacity of ecosystems monitoring trends in existing vegetation and evaluating whether land management activities have had desired effects.

Time period of content:

Beginning date: 07/06/1991
Ending date: 09/25/1993
Currentness reference: ground condition

Status:

Progress: Complete
Maintenance and update frequency: Unknown

Access constraints: None

Use constraints:

The Montana State Library provides this product/service for informational purposes only. The Library did not produce it for, nor is it suitable for legal, engineering, or surveying purposes. Consumers of this information should review or consult the primary data and information sources to ascertain the viability of the information for their purposes. The Library provides these data in good faith but does not represent or warrant its accuracy, adequacy, or completeness. In no event shall the Library be liable for any incorrect results or analysis; any direct, indirect, special, or consequential damages to any party; or any lost profits arising out of or in connection with the use or the inability to use the data or the services provided. The Library makes these data and services available as a convenience to the public, and for no other purpose. The Library reserves the right to change or revise published data and/or services at any time.

Point of contact:

Roland L. Redmond
Principal Investigator
Wildlife Spatial Analysis Laboratory
University of Montana
Missoula, Montana 59812
Telephone: 406-243-5372
Fax: 406-243-6064
E-Mail: red@atlas.wru.umt.edu

Data Quality Information:

Attribute accuracy report:

See the Wildlife Spatial Analysis Laboratory's Final Report, page 66 (PDF page 75), for a complete assesment of the attribute accuracy. This is a summary of that report, organized according to the item names in the grid Valute Attribute Tables.

RIPARIAN
There is no assesement of the accuracy of the Riparian classification in the Final Report. A scene by scene description of how the riparian classes were assigned is in Appendix F. The riparian classifications are based on the spectral classes of the zones. The spectral class zones produced by unsupervised classification may have been adversely affected by the inability of the software to take elevation, slope and aspect into account, by variations in the time of year and amount of haze and smoke among the images, and by the potential that some covertypes may have more variation in spectral signature within the covertype than between different covertypes. Also, the thermal band (band 6) was missing from three of the images (p43r26, p43r27, and p42r28).

ELE, SLP, and ASP
The accuracy of these attributes depends on the accuracy of the Digital Elevation Models (DEMs) they were derived from. See the metadata provided for the DEMs.

Completeness report:

Land outside of the predicted riparian region is not classified. Land inside the riparian regions that were not classified as a riparian covertype are not included.

Horizontal Positional Accuracy Report:

Claimed root-mean square error for horizontal position of the terrain-corrected Landsat TM images is 18 meters in the x direction (WNW-ESE) and 30 meters in the y direction
(NNE-SSW).

Vertical positional accuracy report:

Each zone has an mean elevation determined from Digital Elevation Models whose Root Mean Square accuracy varies from 7 meters to approximately 30 meters.

Lineage:

Source information:

Originator:: Hughes STX Corporation
EROS Data Center
Sioux Falls, SD 57198
(605) 594-6572
Publication date: 05/03/1996
Title: Terrain-corrected Landsat Thematic Mapper Images
Other citation details:: scene scene image % cloud
id name date cover
P42/R26 Lake Koocanusa/Eureka 08/14/92 2
P42/R27 Thompson Falls/Noxon 08/14/92 0
P41/R26 Glacier National Park 09/06/91 1
P41/R27 Missoula/Flathead Lake 07/20/91 1
P41/R28 Selway/Bitterroot 07/20/91 1
P41/R29 Salmon River 06/04/92 1
P40/R27 East Front/Choteau 07/31/92 2
P40/R28 Georgetown Lake 07/29/91 2
P40/R29 Beaverhead/Big Hole 07/31/92 0
P39/R27 Highwood Mts/Benton Lake 07/06/91 1
P39/R28 Canyon Ferry Res/Bozeman 07/22/91 2
P39/R29 Centennial Valley 08/09/92 1
P38/R27 Judith Mts/Winifred 07/31/91 1
P38/R28 Crazy Mountains 07/31/91 1
P38/R29 Yellowstone Park 07/15/91 5
Source scale denominator: 60000
Source contribution:: These are the Landsat TM scenes that were classified to create 15 of the 18 data sets.
Beginning date: 07/06/1991
Ending date: 08/14/1992

Source information:

Originator:: U.S. Geological Survey
EROS Data Center
Sioux Falls, SD 57198-6151
(605) 594-6151
Publication date: 07/30/1994
Title: Terrain-corrected Landsat Thematic Mapper Images
Other citation details:: scene scene image % cloud
id name date cover
P43/R26 Priest Lake 09/25/93 0
P43/R27 Coeur d'Alene/Spokane 08/08/93 0
P42/R28 Dworshak Reservoir/Kooskia 06/14/93 10
Source scale denominator: 60000
Source contribution:: These are the Landsat TM scenes that were classified to create 3 of the 18 data sets.
Beginning date: 06/14/1993
Ending date: 09/25/1993

Source information:

Originator: U.S. Geological Survey
Publication date: 1996
Title: 7.5 Minute Digital Elevation Models
Other citation details:: See http://edcwww.cr.usgs.gov/nsdi/html/dem75/dem75 for a list of sources used by the USGS to create these files.
Source scale denominator: 60000
Source contribution:: This is the source of about 97 percent of the elevation data covering the study area. The mean elevation of each zone was a factor in the final supervised classification of some of the land cover types.
Beginning date: 07/01/1979
Ending date: 1996

Source information:

Originator: U.S. Geological Survey
Publication date: unknown
Title: 3 arc-second DEM Data
Other citation details:: See http://nsdi.usgs.gov/nsdi/wais/maps/dem1deg.html for a complete description of how the 3 arc-second DEMs were created.
Source scale denominator: 250000
Source contribution:: This is the source of about 2 percent of the elevation data in the study area. The mean elevation of each zone was a factor in the final supervised classification of some of the land cover types.
Calendar date: Unknown

Source information:

Originator:: Hughes STX Corporation
EROS Data Center
Sioux Falls, SD 57198
(605) 594-6572
Publication date: unknown
Title: 3 arc-second DEM Data, Resampled to 30-meter resolution
Other citation details:: See http://nsdi.usgs.gov/nsdi/wais/maps/dem1deg.html for a complete description of how the 3 arc-second DEMs were created. The process by which Hughes STX resampled the data to 30-meter resolution is proprietary.
Source scale denominator: 250000
Source contribution:: This is the source of about 1 percent of the elevation data in the study area. The mean elevation of each zone was a factor in the final supervised classification of some of the land cover types.
Calendar date: Unknown

Source information:

Originator: U.S. Forest Service Northern Region Ranger Districts
Publication date: unpublished material
Title: Ground-Truth Field Data
Other citation details:: 671 7.5-minute quadrangles were selected for field sampling. A spectral class map was created for each quadrangle, and Forest Service field crews attempted to visit as many different spectral zones as possible in each quadrangle and classify them. The locations where the classifications were made were recorded by GPS. In all, 22,125 locations were sampled.
Source contribution:: 80 percent of the Ground Truth points were entered as training sets for the supervised classification. The remaining 20 percent of the points were used to assess the accuracy of the classification.
Beginning date: 04/30/1994
Ending date: 10/03/1995

Source information:

Originator: U.S. Forest Service Northern Region Ranger Districts
Publication date: unpublished material
Title: Ground-Truth Existing Databases
Other citation details:: These are 9244 field samples from existing Forest Service databases collected before 1994.
Source contribution:: 80 percent of the Ground Truth points were entered as training sets for the supervised classification. The remaining 20 percent of the points were used to assess the accuracy of the classification.
Calendar date: unknown

Source information:

Originator: U.S. Geological Survey
Publication date: 1993
Title: 1:100,000 Scale Digital Line Graphs
Other citation details:: A list of the quadrangles used and their source dates is in the Final Report on page 5, Appendix C (PDF page 113). For more information on 1:100,000 scale Digital Line Graphs, see: http://edcwww.cr.usgs.gov/nsdi/html/dlg100/dlg100
Source contribution: This data was used to help define the riparian regions.
Beginning date: 1975
Ending date: 1993

Process step:

Perform unsupervised classification of TM scenes. This step was performed with ERDAS and VIMAP software. See pages 14 (PDF page 23) and G.1 (PDF page 900) of the Final Report for more detailed information.


Process date: 04/24/1994

Process step:

Regroup the classes produced in the unsupervised classification to produce fewer classes. This typically reduced the number of classes per image from 60-80 down to 30-40. See pages 15 (PDF page 24) and G.9 (PDF page 908) of the Final Report for more detailed information.


Process date: 04/25/1994

Process step:

Run majority filter to eliminate "salt & pepper" pixels. See pages 15 (PDF page 24) and G.16 (PDF page 915) of the Final Report for more detailed information.


Process date: 04/25/1994

Process step:

Merge zones less than the Minumum Mapping Unit size of 2 hectares (about 22 pixels) with their most similar neighbors. See pages 16 (PDF page 25) and G.11 (PDF page 910) of the Final Report for more detailed information. Data was converted from ERDAS format to Arc/Info Grid format after this step.


Process date: 04/27/1994

Process step:

Analyze all the U.S.G.S. quadrangles in the study area to determine their spectral diversity according to the merged unsupervised classification grids.


Process date: 04/30/1994

Process step:

Select the U.S.G.S. quadrangles where training set data will be collected. Select approximately 40 quads in each of the 18 TM scenes such that the spectral diversity of the selected set is maximized, while ensuring that rare spectral classes are represented. See page D.1 (PDF page 118) of the Final Report for more detailed information.


Process date: 05/30/1994

Process step:

Gather the ground truth database. Performed in the field by Forest Service personell according to the ECODATA method (Keane, R. E., M. E. Jensen, and W. J. Hann. 1990. ECODATA and ECOPAC: analytical tools for integrated resource management. The Compiler, 8(3):24-29, 31, 33-37.)


Process date: 10/03/1995

Process step:

Process the ground truth databases: Verify that the location attributes of the plots in the databases agree with the attributes of which quad, TM scene, and spectral class the plots were supposed to be in. Plots which could not be resolved to their correct locations and spectral values were discarded. See the page 27 of the Final Report (PDF page 36) for more information.


Process date: 11/30/1995

Process step:

Build a Digital Elevation Model (DEM) for each TM scene: Collect 30-meter DEMs from the USGS and Hughes STX, and choose the best data where duplication existed. Create patches from 250K DEM data where no 30-meter data exists. Manually correct any errors in the 30-meter data, and merge the individual 7.5 minute quadrangles into 60x30 minute quadrangles. Fill in any null values between 7.5 minute quads with averages of adjacent values. See Appendix B of the Final Report (PDF page 103) for additional details.


Process date: 01/20/1996

Process step:

Build a Value Attribute Table for each TM scene: For each spectral zone, determine the mean elevation and slope and the majority aspect using the DEM data. Determine the mean MNDVI and mean value of each TM spectral channel using the original scene data. Determine the area in hectares, perimeter, and x and y center coordinate.


Process date: 02/21/1996

Process step:

Label agricultural, urban, and cloud areas: Manually determine which spectal zones are agriculture, urban, or cloud and assign the covertype of these zones.


Process date: 02/29/1996

Process step:

Label cover type, size class, life form, and canopy: Perform supervised classification of the spectral zones based on the training data. This was an iterative process that included automatic assignment of a cover type to each spectral zone based on the training data, evaluation of the results, modification of the training data and addition of new training sets based on the Forest Service's knowledge of ground truth, re-doing the automatic classification, and manual or automatic modification of the results based on combinations of the covertype assigned by the automatic classification with elevation and slope. See page 30 (PDF page 39) and Appendix F (PDF pages 131-899) in the Final Report for more details.


Process date: 02/29/1996

Process step:

Build scene hydrography: Modify U.S. Geological Survey 1:100,000 scale Digital Line Graph hydrography by digitizing streams that appear on U.S. Geological Survey 1:100,000 scale printed maps and correcting any errors in topology and attribute coding. See Appendix C (PDF page 109) of the Final Report for more information.


Process date: 01/30/1996

Process step:

Predict riparian regions: Determine what lands adjacent to hydrography are less than 5 meters in elevation higher than the hydrography. See page 48 (PDF page 57) of the Final Report for more information.


Process date: 02/29/1996

Process step:

Label landcover in riparian zones: Determine relationship between ground truth plots and spectral types and assign a riparian covertype to each spectral class that was determined to represent riparian vegetation. Determine which zones represent agricultural land by a partly manual, partly automatic comparison of the riparian covertypes with the agricultural zones from the 2-hectare covertype classification. See page 50 (PDF page 59) of the Final Report for more information.


Process date: 03/13/1996

Process step:

Set dominance data for each spectral zone: Determine which spectral zones should be retained when mapping areas covered more than one TM scene. See page 36 (PDF page 46) of the Final Report for more details.


Process date: 03/11/1996

Process step:

Standardize final products and validate attributes. See page 47 (PDF page 56) of the Final Report for more details.


Process date: 03/11/1996

Spatial Data Organization Information:

Raster object information:

Raster object type: grid cell
Row count: 7330
Column count: 7720

Spatial Reference Information:

Horizontal coordinate system definition:

Map projection:

Map projection name: Albers Conical Equal Area
Albers conical equal area:

Planar distance units: meters

Geodetic model:

Horizontal datum name: North American Datum of 1927
Ellipsoid name: Clarke 1866
Semi-major axis: 6378206.4
Denominator of flattening ratio: 294.98

Altitude system definition:

Altitude datum name: National Geodetic Vertical Datum of 1929
Altitude resolution: 1.0
Altitude distance units: meters
Altitude encoding method: Attribute values

Entity and Attribute Information:

Entity type label: (scene-id).VAT

Entity type definition: Grid cell value attribute table

Attribute label: VALUE

Attribute definition:: Unique zone identification number, assigned by Arc/Info software when the final spectral zones were imported from ERDAS.
Range domain minimum: 1
Range domain maximum: 366581

Attribute label: COUNT

Attribute definition: Number of 30mx30m pixels in the spectral zone
Range domain minimum: 0
Range domain maximum: 1165795

Attribute label: RIPARIAN

Attribute definition: Riparian Land cover type assigned to the zone.

Attribute label: ELE

Attribute definition: Mean elevation of zone
Range domain minimum: 0
Range domain maximum: 5000
Attribute units of measure: meters
Attribute measurement resolution: 1

Attribute label: SLP

Attribute definition: Mean slope of zone
Range domain minimum: 0
Range domain maximum: 90
Attribute units of measure: degrees
Attribute measurement resolution: 1

Attribute label: ASP

Attribute definition: Majority aspect of zone

Attribute label: MNDVI

Attribute definition:: Modified normalized difference vegetation index; calculated according to the following equation: (TM4 - TM3) / (TM4 + TM3 + 1) * (256 / (TM5 + 1)) * 100
Range domain minimum: -24576
Range domain maximum: 22400

Attribute label: TM1

Attribute definition: Mean spectral value of zone in Thematic Mapper channel 1
Range domain minimum: 0
Range domain maximum: 255

Attribute label: TM2

Attribute definition: Mean spectral value of zone in Thematic Mapper channel 2
Range domain minimum: 0
Range domain maximum: 255

Attribute label: TM3

Attribute definition: Mean spectral value of zone in Thematic Mapper channel 3
Range domain minimum: 0
Range domain maximum: 255

Attribute label: TM4

Attribute definition: Mean spectral value of zone in Thematic Mapper channel 4
Range domain minimum: 0
Range domain maximum: 255

Attribute label: TM5

Attribute definition: Mean spectral value of zone in Thematic Mapper channel 5
Range domain minimum: 0
Range domain maximum: 255

Attribute label: TM6

Attribute definition: Mean spectral value of zone in Thematic Mapper channel 6
Range domain minimum: 0
Range domain maximum: 255

Attribute label: TM7

Attribute definition: Mean spectral value of zone in Thematic Mapper channel 7
Range domain minimum: 0
Range domain maximum: 255

Attribute label: HECTARES

Attribute definition: Area of the zone in hectares (COUNT * 0.09)
Range domain minimum: 0
Range domain maximum: 104921
Attribute units of measure: hectares
Attribute measurement resolution: .09

Attribute label: X-COORD

Attribute definition: X coordinate of a point within the zone.
Range domain minimum: -97890
Range domain maximum: 712395
Attribute units of measure: meters
Attribute measurement resolution: 1

Attribute label: Y-COORD

Attribute definition: Y coordinate of a point within the zone.
Range domain minimum: -66418
Range domain maximum: 644794
Attribute units of measure: meters
Attribute measurement resolution: 1

Attribute label: SCENEPOLY_ID

Attribute definition:: Unique zone identifier for entire project area. Generated by multiplying the scene number (1-18 below) by 1 million and adding the VALUE item. Required for clipping, and should not be modified.

1: P38/R27 2: P38/R28 3: P38/R29 4: P39/R27 5: P39/R28 6: P39/R29
7: P40/R27 8: P40/R28 9: P40/R29 10: P41/R26 11: P41/R27 12: P41/R28
13: P41/R29 14: P42/R26 15: P42/R27 16: P42/R28 17: P43/R26 18: P43/R27
Range domain minimum: 1000001
Range domain maximum: 18287082

Attribute label: DOM

Attribute definition:: 4-character value indicating specific dominance relationships between the scene in which this zone falls, and the surrounding four scenes. The four characters correspond to N, E, S, and W (in that order); see items below. Values of 1 indicate that, after the edge-matching process, this zone will be kept when certain combinations of scene grids are merged and clipped in subsequent processes; the position(s) of 1 values indicate scene combinations for which this holds true. In all, there are sixteen possible combinations: 1000 means that the zone will be kept when merging with the grid to the north; 1010 means that the zone would be kept in merging with the grid to the north and/or the south, but not to the east and west. Required for clipping, and should not be modified. Similar to KEEP (see below), but stores information on more complex relationships. Populated for all zones, regardless of whether or not they fall within the project area boundary.
Range domain minimum: 0000
Range domain maximum: 1111

Attribute label: DOM_N

Attribute definition:: Single-character value for this zone indicating dominance relationship between the scene in which this zone is located and the scene to the north.

Attribute label: DOM_W

Attribute definition:: Single-character value for this zone indicating dominance relationship between the scene in which this zone is located and the scene to the west.

Attribute label: DOM_S

Attribute definition:: Single-character value for this zone indicating dominance relationship between the scene in which this zone is located and the scene to the south.

Attribute label: DOM_E

Attribute definition:: Single-character value for this zone indicating dominance relationship between the scene in which this zone is located and the scene to the east.

Attribute label: KEEP

Attribute definition:: Single-character value for this zone indicating whether or not the zone will be kept when adjacent files are merged according to the currently defined edgematching scheme for the project area. A value of 1 indicates that the zone will be kept, at least in part (zones along edges may be subdivided when files are merged); 0 indicates that the zone will be entirely dropped. As with COVERTYPE and similar attributes, the KEEP item has only been populated for zones that fall at least partly within the project area boundary. If working with areas outside the project area boundary, DOM must be used.

Note that if a new grid is created based solely on the KEEP item (i.e., OUTPUT = P43R27ZP.KEEP), there may appear to be holes (areas where KEEP = 0 entirely surrounded by KEEP = 1) well within the edge-matched boundary on edges where the scene is subordinate (Fig. I-1). These holes are an artifact of the manner in which KEEP was created; if even a part of a zone will be kept in the edge-matching process, the entire zone will be coded KEEP = 1.

Where large polygons from the dominant scene extend into the subordinate scene, some configurations of subordinate polygons will be such that a subordinate polygon (S1) closer to the edge spans some portion of a dominant polygon (D). S1 will be coded KEEP = 1; when displayed, S1 will essentially cut off any subordinate polygon farther from the edge (S2) where KEEP = 0 because it is entirely within D. If completely surrounded by zones like S1, S2 will appear to be a hole. When the two scenes are merged into one, S2 will be entirely overwritten by D, as will the extra portion of S1. Thus, KEEP offers an appropriate shortcut for the entire edge-matching process (clipper.aml), as long as the dominance order of scenes is maintained when merging scenes together (see below). It will, however, provide some extra information along scene borders (e.g., the part of S1 that will later be overwritten) and, when mapped out, will not present an exact geographic match with the specific area to be kept per scene. Furthermore, KEEP is not intended for single-scene operations.

SCENE DOMINANCE ORDER: To be used when merging adjacent scenes together based on KEEP in conjunction with other attributes.
(i.e., OUTPUT = MERGE(CON(P41R27ZP.KEEP == 1, P41R27ZP.COVERTYPE), CON(P41R26ZP.KEEP ==
1, P41R26ZP.COVERTYPE).
1: P41/R27 2: P39/R27 3: P39/R28 4: P38/R29 5: P38/R28 6: P38/R27
7: P41/R28 8: P40/R28 9: P40/R29 10: P39/R29 11: P40/R27 12: P43/R27
13: P42/R27 14: P42/R26 15: P41/R26 16: P43/R26 17: P42/R28 18: P41/R29

Distribution Information:

Distributor:

Montana State Library
PO Box 201800
Helena, MT 59620-1800
Telephone: 406-444-5354
E-Mail: geoinfo@mt.gov

Resource description: Offline Data

Distribution liability:

Metadata Reference Information:

Metadata date: 03/21/2016

Metadata review date: 09/29/2004

Metadata review date: 08/15/1996

Metadata contact:

GIS Implementation Specialist
Montana State Library
PO Box 201800
Helena, Montana 59620-1800
Telephone: 406-444-5354
E-Mail: geoinfo@mt.gov