Mapping methods and explanation of fields and attributes for geodatabase of presettlement riverine and nearshore habitats of Puget Sound

This page includes an overview of methods used to create the geodatabase of pre-settlement land cover and landforms of Puget Sound’s river valleys, estuaries, and nearshore. The overview is followed by detail on different types of features, organized by the main fields in the geodatabase, including explanation of individual attributes within each field.

Skip ahead to:

Data fields:

General mapping methods

Records of the mid-19th century public land survey (PLS) records, consisting of field notes and plat maps, and the Coast Survey topographic sheets (T-sheets) are the two main mapping sources. Both have strengths and limitations for reconstructing historical landscapes (Table 1).

Strengths and limitations of PLS records

The PLS field notes include three main types of information. Detailed note was made along survey lines (“line notes”) of vegetation, wetlands, water bodies, landforms, and the character of the land. The survey was generally in a one-square mile grid, or 1/4 mile grids on Indian Reservations. Surveyors were to record additional information in summarizing each line (“line summaries”). “Bearing tree notes” consist of the species, diameter, and distance to trees that surveyors used to document section corners, quarter corners, and the banks of navigable streams encountered along section line surveys.

The PLS survey was made in the first few years of settlement, when land cover or landforms remained relatively unchanged (Table 1). However, the survey was restricted to a one-mile grid, excepting navigable rivers and shoreline, which were surveyed (“meandered”) along their courses. In addition, because it was a legal survey, ambiguities and biases are presented in using the records for scientific purposes.

Strengths and limitations of Coast Survey records

In contrast to the PLS, the Coast Survey was a scientific survey. The survey generated topographic sheets (T-sheets) using plane table; in the Puget Sound area most have a scale of 1:10,000, several are at a 1:20,000 scale and a few have a scale of 1:4,800. The sheets show landforms and land cover associated with the shoreline and near shore. While a reconnaissance level survey was conducted in Puget Sound earlier in the 19th century, detailed sheets were not made until a few decades into the settlement era. Some of these reflect a considerable amount of land use, primarily diking and land clearing for agriculture (Table 1). “Descriptive reports” that accompany T-sheets generally add little to the maps, leaving map symbols and drafting conventions (which were still being standardized) with a heavy burden to convey information.

Specific uses of PLS and Coast Survey records

Table 2 describes how the records of these and other 19th century field investigations can be used to map historical features. To compensate for their limitations, we cross referenced the two where possible and supplemented them with additional 19th century sources, including early maps made by other agencies (e.g. federal Bureau of Soils, Geological Survey, and Army Engineers) and text sources.

Cross referencing with 20th and 21st century sources

We supplemented 19th century sources with more recent data (Table 3), including 1930s-era aerial photographs. In addition to adding insight into areas that remained relatively unchanged from their pre-settlement condition, the photos were useful even where tidelands had been diked, or where channels had been ditched or forests cleared, because the photos record relict channels and vegetation, including features that the PLS survey would have missed in its mile-square grid.

We used high-resolution digital elevation models from lidar for several purposes: (1) Similar to aerial photographs, lidar show relict landforms. (2) By illuminating subtle topography, lidar helps in landform interpretation. (3) Lidar can refine or extrapolate boundaries of features that are partially controlled by elevation. We also made limited, local, supplemental use of modern soils and wetland data.

Transparency and “depth” of historical mapping

To make sources and assumptions transparent, we coded for each feature the sources used to map it (“source” fields, see below), and the assumptions and relative strength of evidence for a feature’s presence or absence (“certainty” field, see below).

We linked text and graphic information for individual features to a narrative (contained on this site) of the available source data, including text excerpts, maps or photographs, along with the logic we used in making use of that information in mapping the feature.

Below is more detail on how we mapped features, and explanation of attributes. At this time, we provide information for the fields we currently have complete information: “Landcover,” “chantype,” “sources,” and “certainty.” Additional fields (e.g., “landform,” summer and winter wetland inundation) will be added to a later, revised version of this document as geodatabase attributing is completed.

Land cover (“LANDCOVER”)

We developed a land cover classification (Table 4) broadly compatible with that of Anderson et al. (1976).

Channel

We categorized channels as “freshwater,” “tidal-freshwater,” or “estuarine” (Table 4). The boundary between freshwater and tidal-freshwater (generally the upstream limit of tidal influence) was from historical sources or modern sources as available. The tidal-freshwater and estuarine boundary used the estuarine and freshwater vegetation zones boundary as a surrogate (see later).

Channel codes are modified by “BAR” (gravel or sand bar) or “WOOD” (wood accumulations). While some of these features, especially wood accumulations, persisted for many years, most were transitory, and reflect conditions at the time of the early surveys, by which time substantial wood clearing had occurred in navigable channels (e.g., Collins et al. 2002).

Lake/pond (freshwater) and lagoon (marine or estuarine)

We mapped three “lakes” (Table 4): Lake Washington (9430 ha), Lake Sammamish (2080 ha), and Lake Union (350 ha). We mapped all other non-channel freshwater water bodies as “ponds.” The largest was 24 ha.

We mapped as “lagoon” water bodies having a hydrologic connection to Puget Sound, through a channel (“open lagoon”) or lacking a channel (“closed lagoon”). Symbols on T-sheets for what we mapped as “closed lagoons enclosed by estuarine emergent marsh” are indistinguishable from symbols that we mapped as “SALTPOND” (see wetlands, below). These two mapping categories are distinguished in part by size (salt ponds averaged 0.17 ha, and closed lagoons surrounded by estuarine emergent marsh averaged 1.3 ha), but more so by salt ponds being surrounded by large expanses of estuarine emergent marsh (e.g., the Puyallup River estuary), and closed lagoons being larger than, or comparable in size to, the fringing emergent marsh.

Wetland type

Our wetland classification, modified from Cowardin et al. (1979), includes: palustrine, riverine-tidal (freshwater wetlands affected by tidal influence), and estuarine wetlands (Table 4). For each, we used one of three vegetation type modifiers: forest, scrub-shrub, and emergent.

Some estuarine wetland were identified from the “saltmarsh” symbol on T-sheets. Elsewhere, where estuarine wetland had been diked and drained prior to the T-sheet, we mapped estuarine wetlands where PLS line notes or line summaries note “tide prairie.” PLS line summaries also generally note that tide prairies were subject to tidal overflow and to what depth. To extrapolate the boundary of estuarine marsh between PLS lines, we relied in some areas on the extent of tidal blind channels shown on 1930s photos or on lidar, and on elevation.

Coast Survey topographers took the outer extent of saltmarsh as the shoreline. Sometimes they mapped additional marsh, seaward of shoreline; these marshes appeared to topographers as marshy areas mostly flooded at high water (Shalowitz 1964).They used a symbol identified in published legends as “submerged marsh;” we coded these as “WT_EEM_LOW.”

We mapped riverine-tidal wetlands where the PLS notes indicated a tidal influence and there were blind-tidal channels having a freshwater source. The largest riverine-tidal wetlands are in the Skagit and Snohomish river estuaries, with smaller areas mapped in the Nooksack, Stillaguamish, Duwamish and Skokomish deltas.

Wetland vegetation type

We relied on three main types of evidence to distinguish estuarine emergent vegetation from estuarine scrub-shrub vegetation:

(1) T-sheet symbology. The hatched saltmarsh symbol on T-sheets was sometimes modified by the addition of scattered tree symbols; that this modification was intended to symbolize estuarine scrub-shrub (sometimes referred to as “spruce marsh” in historical sources, including in PLS notes) is confirmed by the correspondence of this symbol on T-sheets to the line note descriptions and bearing tree records in the PLS survey. The density of these tree symbols also appears to indicate relative tree cover density, which is also confirmed by cross-referencing to the PLS survey and to aerial photographs of remaining spruce marsh in the Snohomish and Skagit River estuaries.

(2) PLS note descriptions. Where saltmarsh had been diked prior to the T-sheet survey, we relied on the PLS survey to distinguish estuarine scrub shrub and emergent vegetation. PLS notes describe emergent vegetation with descriptors such as “grass,” “grassy prairie,” “fine grass,” “tide grass,” and “grass and tules.” Elsewhere, in areas we map as estuarine scrub-shrub, examples of PLS descriptions include “tide prairie covered with tules, flags, grass and scattering timber,” or “scattering spruce and crabapple.”

(3) PLS bearing tree records. We took the absence of any trees available to serve as bearing trees as additional evidence for the presence of emergent vegetation. In contrast, in areas we mapped as estuarine scrub-shrub, bearing trees are present, but widely spaced, as indexed by the distance measured from survey point to bearing tree.The species and diameter of bearing trees in estuarine scrub-shrub were also characteristic.

In palustrine and riverine-tidal wetlands we distinguished emergent, scrub-shrub and forested wetlands using descriptions in PLS line notes, in combination with the spacing, species, and diameter of PLS bearing trees.

Grassland, scrubland, and forest

Most polygons we mapped as “grassland” (Table 4) are nearshore features symbolized as grassland on a Coast Survey T-sheet. A few features in river valleys were mapped from PLS notes that identify “prairies;” they are small, making them likely under-represented in our mapping, because of the coarse PLS grid.

We mapped a small number of polygons as scrub-shrub (<20 polygons) and several more as “scrub-shrub/deciduous forest” (<10). These were interpreted from symbols on T-sheets, generally cross-referenced by line note descriptions in the PLS survey (e.g., in the Nooksack River delta and the Dungeness River delta).

We used three forest cover categories from the percentage basal area represented by conifers and broadleaf deciduous trees: “Deciduous” (>75% deciduous), “Conifer” (>75% coniferous), and “Mixed” (conifers and deciduous both 25-75%). This typing is based on a geodatabase (to be made available on this site in the future) of 7,995 bearing trees from river valleys in the study area. We have not completed forest typing, and forest polygon in this initial version of the geodatabase are coded as undifferentiated forest type (“FO_UN”).

Barren Land

We classified barren surfaces by elevation relative to tidal influence and by substrate (Table 4). A “BG1” zone is below the low water line, “BG2” below shoreline and above the line of low water (see later for discussion of the meaning of shoreline and low water as shown on T-sheets and in this geodatabase), “BG3” above and adjacent to shoreline, and “BG4” upland. /

In general we mapped areas above the shoreline as “barren ground” only if they were depicted by a symbol depicting “barren” land. A large number of polygons above (and adjacent to) the shoreline lacked symbols. We generally attributed these as “UNK3” to indicate their position relative to the shoreline, but with an unknown land cover. Most of these polygons likely had land cover of sand, gravel, or grass, but we generally did not make this interpretation in the absence of a corroborating 19th century source.

Substrate symbols had not been well defined or standardized at the time of the Puget Sound surveys. We developed four designations from our interpretation of T-sheet symbols: cobble (COB), gravel (GRV), sand (SND), and mud (NUD). “COB” corresponds to substrate symbols on T-sheets we appearing to correspond to “cobblestones and boulders” described in descriptive report for T-1795, surveyed by J.J. Gilbert, who surveyed 36% of the T-sheets we used. “GRV” corresponds to substrate symbols that appear to represent “gravel and rocks” as shown in a legend published in a 1905 USC&GS Annual Report of the Superintendent. “SND” corresponds to mapped symbols that appear the same as “sand” in legend published in the 1891 USC&GS Annual Report of the Superintendent.

“MUD” corresponds to a symbol on T-sheets that was not defined in a published source we could find. In all but one case, it appeared as sparse dots in a random pattern. It is dissimilar to other symbols on Coast Survey legends: "gravel and rocks" (dots are typically ordered and more densely spaced); "cobblestones and boulders" (a mix of large and small dots, densely spaced, with a spacing that varied from linear to random), and "sand" (dots spaced linearly, with density decreasing toward higher elevations). The "mud” symbol is found on sheets T-1682, T-1795, T-1952, T-1953, T-1954, T-1955, T-2194, and T-2229, all surveyed by J.J. Gilbert. The descriptive report for T-2194 includes descriptions of several areas using the symbol: “At low tide the head of [Mitchell's] bay is an extensive mud flat.” “Small boats went frequently through this entrance into Garrison Bay where at low tide they were grounded in the mud.” “Westcott Bay is quite shallow above Bell Point west of the shoreline is free from rock and the beach is gravel and mud.” J. F. Pratt (the fourth most prolific surveyor, responsible for 6% of T-sheets we used) on T-1682 used a different symbol we also mapped as “Mud”: short, faint, linear dashes, with 'Soft mud and gravel' written faintly on the sheet.

We used directly several symbols identified in Coast Survey legends: “rocky ledge,” “kelp,” “reef,” and “rock.” The symbols for “kelp” are straightforward. The scalloped symbols for ‘reef’ and ‘rocky ledge’ are similar, but represent distinct landforms:

“A reef is a rocky or coral elevation, dangerous to surface navigation, which may or may not uncover at the sounding datum. A rocky reef is always detached from shore, but a coral reef may or may not be connected with the shore. A ledge is a rocky formation connected with and fringing the shore, and generally uncovers at the sounding datum” (p. 264, Shalowitz 1964).

“Bare rocks” were digitized as polygons and attributed as “BG_RCK” if they were large enough to be digitized as a polygon at the scale of 1:1,000 to 2,000. Smaller rocks were digitized as points. The point file is available as part of the edge-mapped vector coverage of T-sheets, available on this site. /

Some T-sheets also included text descriptions of substrate. These annotations have been digitized as line features and included in a digitized vector geodatabase of edgemapped T-sheets.

Shoreline

With few exceptions, we took the shoreline from T-sheets. The Coast Survey’s shoreline was intended to represent the Mean High Water line (MHW). By convention the outer limit of marsh was take as the shoreline rather than attempting to define a high water line. We included the shoreline as far up-river as it appears on T-sheets.

Where cultural features such as docks obscured the shoreline, we interpolated a line (attributed “SHORELINE_INTRP”) (Table 4). In cases where the T-sheet was poorly legible, we estimated the shoreline location (attributed as “SHORELINE_APPROX”). On a few T-sheets (see later discussion on precision; also see meta-data for vector, edge-mapped geodatabase of T-sheets for Puget Sound), we shifted the shoreline to compensate for what appeared to be survey error on the T-sheet. In those areas, we attributed the shoreline as “SHORELINE_SHIFT.”

The “submerged marsh” symbol on T-sheets (see “wetlands” section, above) was not bounded by lines. We digitized an approximate seaward boundary of these marshes and attributed the line “EEM_LOW_OUTER_EDGE.”

Low water

We digitized the line shown on T-sheets as the low water line [“LOW_WATER” (Table 4)]. This line is not believed to represent an actual, measured mean low water line, but instead the topographer’s estimate of the mean low water line (Shalowitz 1964). For this reason we use the generic designation “low water” rather than the more specific “mean low water.” Unlike the shoreline, the low water line was not shown continuously or on all T-sheets. We attributed the line marking the outer boundary of a reef or rocky ledge polygon as “LOW_WATER_IMPLIED” (<Table 4). Where cultural features obscured the low water line, we interpolated a line and attributed it as “LOW_WATER_INTRP.”

Channel type (“CHANTYPE”)

We attributed channels having bankfull widths >25 m as “mainstems” (“MA”), and <25 m as “tributaries” (“TR”). The distinction is on size only. We subdivided into five bankfull width classes: “MA5” (>50m), “MA4” (25-50 m), “TR3” (10-25 m), “TR2” (5-10 m) and “TR1” (<5 m) (Table 5). We attributed as “secondary” channels (“SC”) channels that branch from mainstems or tributaries and either rejoin the main channel downstream, or dead-end on the floodplain, and “distributaries” (“DT”) those branching from mainstems or tributaries, typically on river deltas, and flowing to saltwater. We categorized as “blind tidal” (“BT”) channels that are primarily created by and drain tide or flood introduced water (Simenstad, 1983) and rapidly narrow with increasing distance from the tidal source.

Sources (“SOURCE_1” “SOURCE_2” “SOURCE_3” “SOURCE_4”)

Source materials used in the four source fields (“SOURCE_1,” “SOURCE_2,” etc.) are listed in the order of the importance with which they were considered in interpreting a particular feature (Table 6).

Certainty (“CERTAINTY”)

We categorized the level of evidence for the presence or absence of each of four major land cover types: large channels (channels mapped as polygons), small channels (channels mapped as lines), ponds and lakes, and wetlands. We developed these ratings by grouping all of the mapping situations we encountered—all combinations of sources and relative levels of evidence—into 33 categories. We then grouped these mapping situations into four categories based on the nature and amount of evidence bearing on the situation.

The first category (“category I;” Table 7) includes feature that had been mapped by a 19th century field survey, and the nature of the survey methods make the feature’s presence certain. For example, in the PLS survey, the presence of a meandered channel is certain. This assumes the absence of survey fraud. Fraud did occur in Washington (White 1983, Cazier 1976), but none has been documented in the Puget Sound area. Most situations fall within this category.

A second category (“Category II;” Table 7) includes features mapped by a 19th century field survey; while the survey methods make the feature’s presence uncertain, other sources can corroborate it. This category applies to small channels (lines) on T-sheets, which do not appear to have been mapped with the same attention as larger (polygon) channels. Some surveyors in other regions of the country mapped small tidal channels in detail (e.g., Grossinger 1995). In the Puget Sound region, different surveyors either mapped, sketched, or entirely made up the location of small channels. The two situations include features we mapped when we could corroborate their presence as relict channels on 1930s photographs or lidar in diked areas, or using active channels using the same sources. In situations where we could not confirm the presence of small channels on T-sheets, we generally did not map them; cases where we did represent a situation in a third category (Table 7). “Category III” differs from “Category II” in there being no corroborating source available.

The final “Category IV” includes features not mapped by a 19th century field survey, but survey methods make it reasonable to assume that the survey would have missed the feature, and more recent sources are available to corroborate the feature (Table 7). This encompasses those features which the PLS survey could have missed because they are located within the interior of land sections and thus would not have been encountered by surveyors.

We applied these certainty categories to features in the valleys and estuaries of major rivers, and to a few nearshore features for which we were able to use multiple sources, but most nearshore features were mapped directly from the T-sheet without the benefit of additional sources. We indicated these situations in the “certainty” field by the code “T” when the feature was mapped from a T-sheet, and “H” when the feature was mapped from a hydrographic sheet (“H-sheet”).

Accuracy and precision of sources

Because multiple sources were used to map most features, there is not an exact correspondence between the mapping situations in Table 7 and the accuracy or precision with which the feature’s location is known. However, general estimates of typical precision and accuracy associated with different sources (Table 8) can be used to assess individual map features.

Coast Survey T-sheets

The process we used to register T-sheetsintroduced an error of between 2 to 20 m (Table 8). In an error assessment, we found the horizontal difference between still-existing benchmarks and the same benchmarks as shown on the registered T-sheets ranged from 2 to 8 m on 1:10,000 scale sheets, and from 6 to 20 m on 1:20,000 scale sheets.

The accuracy of some T-sheets are compromised in several situations. A few were inaccurate in their up-river portions (e.g. T-1755, Stillaguamish River, and T-1681, Snohomish River). In these situations we gave precedence to other sources (e.g. PLS survey) or other sources were used to adjust the location of T-sheet mapping (e.g. 1930s aerial photographs). We also identified several shoreline segments in the southern study area that appear to have been shifted during the original survey. For mapping that appear to have been in error by more than 20 m, we manually shifted features to correspond with modern reference points, and coded in geodatabase for those features the distance and direction of shift (see metadata for digitized vector geodatabase of T-sheets).

1930s-era Aerial Photographs

We orthorectified photographs except those for the Skagit River and Nisqually River upstream of the river’s delta. We tested our orthorectified 1936 photographs in the Cedar River watershed using National Standard for Spatial Data Accuracy (NSSDA) methods and found that the NSSDA 95% statistic averaged between 5 and 10 m. A portion of the Nisqually River photos (upstream of the river delta) could not be orthorectified because too few identifiable reference points were present on the photos, and the Skagit photos have not been orthorectified for budgetary reasons. An estimated error for these georeferenced photos is 10-50 m.

1900-era US Geological Survey Topographic Maps

The small scale of US Geological Survey topographic sheets (mostly 1:125,000) introduces substantial error, in addition to an unquantified error introduced by georeferencing. We generally restricted use of these sources to confirm the presence or absence of a feature or its general location.

PLS Survey Records

The spatial precision of information on PLS maps and field notes varies with type of feature. Channels were meandered and later drafted using survey segments that are long relative to the curvature of rivers, giving meandered channels (and meandered shorelines) a geometric shape that introduces some imprecision in channel shape and location. Some inaccuracy is also introduced to maps of some river channels. We did not examine this systematically to determine the relative importance of survey errors or drafting errors. However, local inaccuracy is evident by the frequency with which PLS-meandered channels are plotted in locations that are topographically impossible. Error was also introduced by the drafting of meandered channels; we examined this error by comparing widths on plat maps to widths in field notes along all section lines in the Nooksack, Skagit, Stillaguamish, and Snohomish river valleys. We found that while individual surveyors depicted channel widths with varying accuracy, on average the width shown on plat maps and as measured and recorded in the field notes agreed within a few percent. The greatest potential for inaccuracy on PLS plat maps is associated with non-meandered features which are sketched within the interior of land sections. The location of channels, for example, can potentially be in error by up to 1000 m, making it necessary to rely on additional sources to locate the channel, as discussed above.

Limitations

Mapping certain features, or feature characteristics, suffer from inherent limitations for which using multiple source can’t compensate. These limitations stem from the original survey methodology, the consistency with which they were conducted, the survey’s timing relative to land uses, or the dynamic nature of some geomorphic processes.

Dynamic geomorphic processes

Secondary channels in our mapping are underrepresented in river valleys occupied by a river that rapidly migrates and avulses over a large percentage of its valley. The original PLS survey did not map secondary channels except where crossed by section lines. Subsequent to the PLS survey, the river would either have actively migrated for some period of time, eradicating the topographic trace of the channels that would be evident on lidar, or the valley was subsequently converted to land uses that also eradicated topographic traces of channels. Consequently, the mapping of secondary channels in such river valleys has widely varying detail, depending on the valley’s subsequent history. While this limits the direct use of the data for some purposes, such as, for example, quantitative estimates of aquatic habitat, it is possible to extrapolate data from undisturbed valleys to valleys in similar geomorphic settings.

Effects of land uses

The historic record very likely grossly underestimates the importance of beaver ponds in river valleys. In part this is because of the scale of the original PLS survey, and the obscuring effects of subsequent history. However, throughout the study area, the PLS and Coast Survey documents would have underestimated beaver ponds because the beaver trade had depleted the region of beavers. The fur-trade in the first decades of the nineteenth century was almost entirely prior to the PLS and Coast Survey mapping; by 1840, the beaver trade was in rapid decline, in part because beaver had been trapped out (Mackie 1997). Beaver dams commonly fail soon after abandonment, so that the landscape described by the early surveyors would have had fewer beaver dams and ponds, and would possibly have shown the effects of meadow and forest succession following beaver dam abandonment.

The intensity of 20th century land use can locally affect the refinement of mapping in the geodatabase. For example, extensive land grading associated with urban development compromises the utility of 20th century sources (e.g., lidar, 1930s-era aerial photographs, soils mapping) for supplementing, cross-referencing, and most commonly, refining by 19th century source maps. For example, as a result of extensive development in the Green River valley south of Seattle, the boundaries of wetlands in our mapping crudely drawn, with poorly-defined boundaries, as shown in the original PLS survey.

Mapping scale

As discussed previously, where the PLS is the main information source (i.e., upland of the area encompassed by Coast Survey mapping), small features such as ponds and wetlands could “fall between the cracks” –in this case, between the survey lines. While supplementary use of 20th and 21st century sources as supplements—1930s aerial photographs, lidar, soils and wetland mapping—identify some of these features, some were certainly missed.

Inconsistency in surveys

PLS surveyors were charged with record inundation depths and frequency as part of their duties spelled out by the Swamp Lands Act of 1850, which was extended to Oregon Territory in 1860 (White 1983). We were not able to determine the seasonality and depth of wetland inundation in some parts of the study area. (Data for the fields “SUMMER_INUND” and “WINTER_INUND” are not completed and have not been entered into the geodatabase at this time; discussion of these fields has not been included in the current draft of this document.) This could be because some surveys were carried out prior to or in the first years after the Swamp Lands Act was extended to Oregon Territory; in general, field notes are more complete in this regard in the northeast study area (Snohomish, Stillaguamish, Skagit, and Nooksack river drainages), which were completed later than the rest of the area. It also might be that some surveyors, for whatever reason, did not pay as much attention to inundation characteristics as other surveyors.

Ambiguous symbols on T-sheets for smaller nearshore features

Many nearshore features were mapped from the T-sheets without the benefit of corroborating sources. For these features mapping relies heavily on the interpretation of T-sheet symbols. Many mapping units on the T-sheets are very small, and symbols are subtle and difficult to interpret, creating inherent uncertainty that is generally greater for the smallest features.

References

Anderson, J. R., E. E. Hardy, J. T. Roach, and R. E. Witmer, 1976. A Land Use and Land Cover Classification System for Use with Remote Sensor Data. US Geological Survey Professional Paper 964, 28 p.

Cazier, L. 1976. Surveys and surveyors of the public domain, 1785-1975. US Govt. Printing Office.

Collins, B. D., D. R. Montgomery, and A. D. Haas. 2002. Historical changes in the distribution and functions of large wood in Puget Lowland rivers. Canadian Journal of Fisheries & Aquatic Sciences 59: 66-76.

Cowardin, L. M., V. Carter, F. C. Golet, E. T. LaRoe. 1979. Classification ofwetlands and deepwater habitats of the United States. U. S. Department of the Interior, Fish and Wildlife Service, Washington, D.C.

Grossinger, R. 1995. Historical Evidence of Freshwater Effects on the Plan Form of Tidal Marshlands in the Golden Gate Estuary. Masters Thesis, University of California, Santa Cruz. 130 pp.

Morse, E. 1885. Tide lands of Washington Territory, chapter 7 in Nesbit, D. M. 1885. Tide marshes of the United States. U. S. Department of Agriculture Miscellaneous Special Report No. 7.

Shalowitz, A.L. 1964, Shore and Sea Boundaries, Volume 2, U.S. Department of Commerce, U.S. Coast and Geodetic Survey, Washington, DC. http://chartmaker.ncd.noaa.gov/ocs/text/shallow.htm.

White, C. A., 1983. A history of the rectangular survey system. U. S. GPO, Washington, D. C.

Questions or comments

Contact: riverhistory@ess.washington.edu

Brian Collins, September 5, 2008