Our new website. TerrainWorks is the next step in our business evolution. We’ve worked since 1997 as a 501(c)3 nonprofit research organization (Earth Systems Institute), and a large part of our efforts over the last 7 years has been development of software tools for spatial data analysis from a landscape perspective via a platform called NetMap. We’ve been successful in funding development of NetMap, but not so much in funding implementation, maintenance, updates, and user support. This is where TerrainWorks comes in. TerrainWorks is a for-profit company whose role is to implement, maintain, and update the software and data sets, and provide user support, funded via data sales, software subscriptions, and implementation services.

A key component of NetMap is a software platform we refer to as a “Digital Landscape”. The digital landscape provides a computer implementation of our conceptual models of how landscape processes interact over space and time. It provides a convenient way to look at linkages and interactions between disparate data and disparate processes.

Integral to the digital landscape is a synthetic channel network. This is a network derived from flow directions inferred from digital elevation data. Within the digital landscape, every point is explicitly linked to the channel network via surface flow paths. In the digital landscape, the synthetic channel network provides the spatial framework for linking different processes.

A quantitative model allows us to make predictions, to test and improve  conceptual models.

The design, in terms of data structures, computer languages, software, and user interfaces must anticipate how a digital landscape will and could be used.

The digital landscape consists of interlinked data and software. A digital channel network is an integral component of a digital landscape.

We’ll go through all these steps in subsequent slides.

Mosaic prior to any coordinate-system projections. Be sure to use bilinear interpolation when projecting. Warping and subsampling done with Fortran code.

Warping over 3000-m transition zone. Warping matches mean elevations over a specified radius (150 m) from each DEM. High-resolution detail maintained throughout overlap, except for a 100-m transition zone. Subsampling of warped DEMs using bilinear interpolation.

As we progress with this project, we are finding a lot of spatial variability in channel extent. We are expanding the initiation criteria to accommodate greater spatial variability based on landform mapping, which primarily delineates areas with differing soil and subsurface characteristics.

Red and blue zones show areas meeting different area-slope thresholds.

We’re finding that we cannot resolve all channel initiation sites, even with 1-m LiDAR. For example, channels that appear within wetlands don’t have much topographic signature. So we added the ability to specify specific channel initiation points using a point shapefile.

We seek the highest channel density resolved by the DEM. Then we find other data sources to classify channels. For example, if there are topographic thresholds that distinguish ephemeral from seasonal from perennial channels, the network can be divided into these separate classes and further analyses focused on the channel types of interest. It is always possible to remove channels from the synthetic network; not so easy to add them (because that can require re-directing flow directions).

We will take advantage of whatever data sources we can.

For the MatSu, we will probably use option 2 above, digitized short line segments.

Someday we’ll figure out how to get the computer to map road prisms and identify road crossings automatically. Not quite there yet.

The “bomb-crater” strategy allows us to use a point file of culverts, for example, that has some spatial error. The culvert points don’t have to be perfectly located over the road prism.

Note that the direction of flow in the right panels is actually downward, not upward as the arrow indicates.

A variety of options exist for summarizing other raster and vector data, e.g., field surveys, land cover, fish presence.