A research paper submitted in partial fulfillment of the requirements for the degree of
MASTER OF SCIENCE in GEOGRAPHY Portland State University 2001



The example of snapshot data presented in this chapter evaluates the snapshot model and its ability to quantify change. Combined with background information, this study documents surface water changes for a 120-year period using 11 snapshots. The temporal information presented in this chapter is limited to the state of a particular water body based on evidence contained in the source date. This study models snapshots in a temporal order and because there is no direct evidence that allows us to say that a particular event caused a particular spatial effect. Included with these snapshots are information not directly related to the source data, but which serve the function of linking snapshots to ongoing temporal events that produce change. These events are happening outside the capacity of snapshot model to store the information.


3.1 Historic Map Sources in Portland Oregon

The fluvial system is a physical system with a history. Figure 1 illustrates the extent of surface water in Portland, Oregon in 1880, Figure 2 illustrates the extent of surface water in Portland, Oregon in 2000. Any assessment of a past condition is complicated by the fact that past conditions have been influenced in ways not detectable with map evidence alone. Maps of two-dimensional channel and other surface water forms can be used to justify changes and effectively demonstrate historic three-dimensional surface water forms (Parsons 1995). Surface water features must appear consistently across the study area. Spatial data obtained from map sources demonstrate observable and measurable change to surface water features including simple channel shifts, cutoffs, and changes in braiding (Hooke and Redmond 1989). Historic maps have a wide variety of sources, themes, and scales. The snapshot model provides the freedom to use a wide range of source material because the data are kept in separate layers. This was done with much difficulty in the days of manual map overlay, prior to GIS, but map registration, re-projection, and scaling functions are mush easier now. Surface water features digitized from historic maps are used as source data in this study. Included are historic maps originating from data collected between 1880 and 2000 with source scales between 1:20,000 and 1:62,500. The extent of the project encompasses the greater Portland metropolitan area, including the Willamette and Columbia Rivers. Map sources and map features obtained for Portland Oregon are plotted along a timeline in Figure 3 and listed in Table 4.


Map Source: Source Date: Source Scale: Features Collected
U.S. Coast and Geodetic Survey 1880 1:40,000 Shorelines, Lakes, Sloughs, and Wetlands
U.S.G.S. 1909 1:62,500 Shorelines, Lakes, Sloughs
U.S. Coast and Geodetic Survey 1922 1:40,000 Shorelines, Lakes, Sloughs
U.S. Coast and Geodetic Survey 1925 1:20,000 Shorelines, Lakes, Sloughs
U.S. Coast and Geodetic Survey 1933 1:40,000 Shorelines, Lakes, Sloughs
U.S. Coast and Geodetic Survey 1945 1:20,000 Shorelines, Lakes, Sloughs
U.S. Coast and Geodetic Survey 1951 1:40,000 Shorelines, Lakes, Sloughs
U.S. Coast and Geodetic Survey 1961 1:40,000 Shorelines, Lakes, Sloughs
U.S.G.S. 15 minute quadrangle 1971 1: 62,500 Shorelines, Lakes, Sloughs
U.S.G.S. 7.5 minute quadrangle 1985-1990 1:24,000 Shorelines, Lakes, Sloughs
Metro (RLIS LITE) 2000 1:24,000 Shorelines, Lakes, Sloughs

Table 4 Cartographic source maps used to obtain the extent of surface water in Portland, Oregon.


3.2 Converting map sources into GIS

The procedure for capturing and processing digital data involves several steps. First of all, digitizing remains the most cost effective and accurate way to capture and vectorize delineations of surface water from historic map sources.

In the digitizing process, each map, regardless of the age or scale, is prepared, registered, and essentially digitized manually. The digitizing method used for this project assures that identified features appearing on historic maps become spatial data. In the digitizing procedure, line work representing water body boundaries, shorelines, stream channels, and islands are connected to poly-line representations. Line work is edited for completeness and accuracy prior to being modeled into topological coverages. Editing decisions must be made concerning how to physically partition the surface water into discreet units for measurement purposes. The topology of surface water must be changed to accommodate a nominal classification in the database. Partition lines are added at natural intersections between features and other physiographic breaks.

For example, a line may extend from one shore to the closest point on the opposing shoreline, at the confluence of the Willamette and Columbia Rivers. As a result, the added information content goes beyond a simple return of "water". In addition to feature names, each body of water is attributed by type, separating lakes from sloughs, sloughs from rivers, and rivers from streams.


3.3 Attribute Data

The data creation process concludes when each snapshot has an attribute table populated with the names and types of each water body feature. The snapshots are stored as ESRI coverages, are attributed by date, and every feature in each snapshot layer is assumed by that date. As a result, the snapshot database does not include separately recorded events nor does it define the extent of events. For example, a short duration flood event does not get recorded even though the effects may be present on the snapshot. The flood event wouldn't be recorded as such even if it appeared on the source map because the snapshot database cannot differentiate it from other surface water.

Named water bodies give the snapshot database additional information content that wouldn't otherwise exist in the snapshot database. These snapshot data contain attributes including unique identification, date, and size. The attribute table (Table 5) contains attributes for surface features such as: dry land, island, lake, river, and sloughs, and for place names of features such as: Sauvie Island, Vancouver Lake, and Willamette River.

Field Description
Acres Total area of individual features in acres.

Total area of individual features in square feet.

Class Designated type of feature: Dry land, Island, Lake, River, Slough
Name Feature's common map name. Name that appears on the source map.
Date Date of map version

Table 5 Attribute descriptions for snapshot data layers


Attribute queries on the snapshot database provide useful information about object states for comparison purposes. Information about the size of any land or surface water feature is obtained by an attribute query. For example, the size of Ramsey Lake in1880 is obtained by querying the 1880 snapshot database. An "identify query", like the kind commonly used in spatial data viewers like Arc-View, return fields including [Name] = "Ramsey Lake" and [Acres] = 724. Another query on [Name] = "Ramsey Lake" contained in the 1961 snapshot database returns [Acres] = 268, a different value than the one obtained from the 1880 snapshot database for the size of Ramsey Lake. Comparing the information obtained from the two queries, we can see that the surface area of Ramsey Lake has reduced in size by 456 acres during a 81 year time span.

Spatial queries with the snapshot database provide information about the change in shape of features from one snapshot to the next. For example, to estimate the change in shape of Ramsey Lake between 1880 and 1961 requires that two snapshots are spatially joined. The commonly shared areas of overlap similar to those depicted in Figure 4 become spatially joined. The spatial join indicates that in addition to a change of 456 acres in size, Ramsey Lake has changed shape. A spatial query on the spatially joined database identifies 1 polygon, 456 acres in size as the extent affected by Ramsey Lake changing shape during a 81 year time span. Still, temporal information pertaining to the exact date of the incremental changes to Ramsey Lake does not exist in the snapshot database.


3.4 Temporal Joining

The discrete arrangement of spatial features is represented as a map state for each time slice. No evidence to explain what events begin and end these states is present in the database. Sometimes we do have information on events for which no spatial representation exists, but which are time-stamped. This ancillary information can be temporally joined with the snapshot database, based on time. Much like a spatial join links attributes of two tables without a common key attribute by assigning location in common, so too, the temporal states of a map state attributed by date can be linked to a representation of that feature's temporal state. Only in this way may we get an idea of how duration in the snapshot model might be bounded by temporal events.

Temporally joining information about historic events that have an impact on surface water gives relevance to mapped features that have changed location, shape, or identity. The majority of surface water changes contained in the snapshot data are directly related to human induced events. To better understand these events in terms of duration, purpose, and magnitude, the following information accompanies this surface water data.

An example of temporal joining involves the explanatory text describing continuous events that have occurred. Over the last 120 years the U.S. Army Corps of Engineers removed 635 acres of river and 965 acres of bottomland have been filled adjacent to the Willamette River (Figure 5). Since 1853 the surface area of the Willamette River has been reduced by 14% between the Selwood Bridge and the mouth of the Willamette (WMP 1973). When depths were last deepened from 35 feet to 40 feet in 1970, over 400 million cubic yards of dredge materials was generated. Of this total, 11 million cubic yards was generated from the Portland Harbor alone (WMP 1973).

Comparing snapshots with temporal joining adds value to the information content of the present snapshot database. The extent of the progressive elimination of natural surface water due to dredging and urbanization are apparent in both description of the events that have taken place and the changes to objects contained in the database. This method shows us that the overall changes to the size and shape of surface water can be linked to events that are continuously modifying the natural landscape to an urban landscape.


3.5 Evaluating the data model

Snapshot data increase our knowledge of the change to surface water over time by indicating the past location and extent of surface water compared to the present or other periods in the past. These snapshot data also allow queries to determine a time interval that defines a cartographic state. Limitations to the snapshot model are a result of the way time is stored in the database. The temporal content of the snapshot database is limited to the date the source map evidence was collected.

One limitation that affects the temporal information content of snapshot data involves the use of valid time. In the snapshot database the exact date of the final disappearance of Ramsey Lake does not appear because it was not recorded within the source data. Figure 4 contains 3 snapshots overlain to demonstrate the reduction in surface area of Ramsey Lake between 1880 and 1961. The lake appears to get smaller between 1880 and 1961 until it does not appear at all on the 1977 snapshot. The snapshot's on Figure 4 explain the behavior of the surface water as changes in size and shape until it finally disappears after 1961 and before 1977. The final disappearance date of Ramsey Lake according to these snapshot data is between 1961 and 1977. There are two sources of temporal information to model temporal relationships between snapshots: the independent time reference that links each snapshot with a source date and the temporal link to an event database. Snapshots are an effective way of representing "time slice" data for themes such as surface water in Portland, Oregon because temporal joining provides additional information that bounds each snapshot. The temporally joined information gives additional clues about events that did not appear in the source data. Combining source date information from mapping with temporally joined information aids in the interpretation of historic events that are responsible for change.

<-- Chapter 2
Return to Menu
Chapter 4 -->