Mapping Forest Park

In September, 1999  Portland's Wildwood Trail in Forest Park was completed.  Thirty miles (30.16) of urban wilderness trail starting at Washington Park's Vietnam Veteran's Memorial.  After it  crosses Cornell Rd at Mi 4.96, it crosses a city street which has traffic only once (at Germantown Rd, Mi 24.63).  It is entirely within the city limits.  This map shows the last 2 miles.

How do we map Forest Park and Wildwood Trail?  We use magnetic compass, aneroid altimeter, and topographic map.  Here's a bit of our tracing over the USGS map of the last two miles of Wildwood trail...and the entire north to south length of Firelane 15, a bit of Firelane 12, a bit of the park boundary, a few new streets, and some streams.

The west end of Forest Park and the last two miles of the Wildwood Trail

Some of the mapping is easy.
A lot can be traced directly from the USGS map: streets, creeks, some trails, most fire lanes, plus a lot of the park boundary—we know how the boundary relates to the various section corners and other survey points shown by USGS.  (The park boundary is occasionally updated as new land is acquired.)

Some of the mapping is difficult.
The challenge is to locate those features not shown on the USGS map: mostly the trails constructed by Portland's Bureau of Parks and Recreation plus a few of the fire lanes and, of course, the numerous new streets in the housing developments that persistently press against the park.   They, too can be traced on the map because we can reliably locate certain features, and then we can trace the trails and roads according to the contours USGS has provided on their map.  When we do this, however, distances on our map and the actual distances measured on the ground do not agree.  For example:

On Firelane 15 we can easily identify (on the USGS map):  (1) the intersection of Firelane 15 with Skyline Blvd is shown on the USGS map;  (2) the ridge top near the two power transmission line towers near the Keilhorn Meadows road; (3) the ridge that FL 15 goes down on its way to WW Tr and FL 12; (4) the switchback of FL 15 directly under the west-most power line a little above the intersection with FL 12.  The point WW Tr crosses FL 15 is identifiable by the flattening of the ridge just below (and by the distance from the turn in the ridge toward the power lines).  After we measure a few altitudes, distances and bearings FL 15 is precisely placed on our map.  However...

Map distances and ground-measured distances don't agree.
Kielhorn road to WW Tr is .28 mi on the map and .32 mi on the ground.  From WW Tr to FL 12 is .32 mi on the map and .37 mi on the ground.  That much error is unacceptable.  Mapping Wildwood Trail consistently has the same problem.  The section of trail from FL 15 (mi 28.36) to Mile 28-1/2 maps to about .06 miles if we simply follow the contours.  The actual distance is, of course, 0.14 mi.  There's a deep gulch there and the USGS contours underestate its depth.

Lesson #1:
Don't oversimplify.

We must pull together different pieces of information from many sources and then balance them one against another until the map we draw makes sense.  Possible sources of errors must be examined.  Contradictions must be resolved.  Uncertainties must be evaluated—and also balanced against each other.  That makes map making a sort of  metaphor for doing science.

Map making is an exemplar for doing science and for distinguishing science from  pseudoscience.
The Singles

Some features on the USGS map can be generally considered reliable: benchmarks, stream positions, ridge and hill tops, for example.  Other feature can be considered reliable at the date the issue of the map: streets, buildings, boundaries, for example.  Other features are not so reliable.  For example, some of the power lines and fire lanes are shown in purple, indicating they were added in a later printing of the map, and their positions are only approximations.  Most significant for our purposes are the finer details of the contours.  Because of the technique used to draw the contours, rather significant features of the terrain can be missing from the map.

We must look at the evidence and study the observations that conflict and tell us something is wrong.

An easy one:  USGS shows the elevations of the bases of the towers as more than 50 feet different: the actual difference is about 15 feet.  Either the towers are shown in wrong places or the contours are incorrectly drawn here.

The persistent problem that plagues the plotting:  The plotted trails and firelanes have lengths on the map that are consistently shorter than they actually measure on the ground because many smaller gulches and ridges don't appear in the contours as drawn.  The reason is that in heavily wooded terrain such as is most of Forest Park it is the treetops that are followed when the contours are drawn and not the ground itself.  The contours are drawn from aerial photographs that are set up to present the cartographer with a stereoscopic image.

The magic of photogrammetry

NOTE: Starting in February, 2002, mapping information has been augmented with GPS data.  Most of the data have been gathered with a Garmin eTrex Vista GPS receiver.  The data are downloaded into the computer and placed on the USGS topographic 1:24000 maps (7-1/2 min series) with AllTopoMaps software.  When good satellite reception is achieved the GPS track coordinates are accurate to within about 10 feet.  However, the dense foliage and steep terrain  in most of Portland's parks often makes good GPS data collection difficult (waiting for fortuitous satellite locations can greatly improve the signals).  

We can obtain very good mapping by combining mapping techniques.  This allows us to correct some of the  errors on topographic maps due to the difficulty the photogrammetric operator has in finding the ground in the aerial  photographs--tree tops are what the operator sees.

A GPS track annotated on the topographic map.

Each yellow circle is a track point.  The software connects the points with a blue line.  The large gap at the lower end of the track (upper right) is due to lost satellite signals on the very steep slope (the satellites are mostly to the south and the signals are block by the hill).  Track points usually show a small systematic error, easily seen on the topographic map.  The entire track is then shifted to elmininate that error.  This gives a very accurate map.