DEM resolution, topographic analysis, and erosion modeling

DRAFT

HELENA MITASOVA, LUBOS MITAS, WILLIAM M. BROWN, DOUGLAS M. JOHNSTON
University of Illinois at Urbana-Champaign

1. Test DEM

We will illustrate the issues of resolution, noise and systematic errors in DEMs and topographic analysis using the 30m DEM for a test area:

fly-by over a DEM

The entire area covers more than 3000 sq. km and the 30m DEM is 1800*2300 grid cells. The analyses and simulations were performed at selected subareas A,B,C,D,L:

2. Topographic analysis

GRASS GIS allows us to compute several important topographic parameters describing various geometrical properties of terrain, as illustrated for an area C for the original 30m DEM:

slope aspect

profile tangential curvatures

The curvatures reveal significant noise in the original DEM.

3. Smoothing and resampling

We will illustrate some issues of resolution, noise and systematic errors in DEMs in more detail using terrain with draped tangential curvature. Tangential curvature for 30m DEM shows acceptable structure in mountainous area while significant noise and systematic errors (stripes) are present in lowland.

original Kt

After smoothing and resampling to 10m resolution using the RST interpolation method (regularized spline with tension) the noise is reduced and both the major topographic features and systematic errors become more visible:

a) DEM and K_t with smoothing 0.1, ten=80,
b) DEM and K_t with smoothing 1.0, ten=60
c) DEM and K_t with smoothing 10.0, ten=50

These images clearly demonstrate that the need for precision and accuracy is spatially variable, with flatter areas much more sensitive than mountains. Note how the artificial structure continuously transforms into the real terrain structure, this is especially "dangerous" phenomenon if DEM is going to be used for simulations, as the artificial structure can be mistaken for the real topographic feature. Comparison with the higher resolution DEM from a different source can reveal more details about the difference between the DEM artifacts and true topographic features.

To highlight the main topographic features, the curvature from smoothed DEM (c) was draped over the DEM with the smallest smoothing (a).

smoothed Kt

4. Flow pattern analysis

Flow pattern analysis is used to identify potential streams. The results are significantly influenced by flow tracing algorithm and resolution as well as quality of the DEM as illustrated by the following example:

Upslope area from the original 30m DEM,
Upslope area from a 10m reinterpolated and smoothed DEM with smoothing 0.1, ten=80,
Upslope area from a 10m reinterpolated and smoothed DEM with smoothing 1., ten=60,

Visualizing the exagerated water surface over terrain shows nicely the relationship between the terrain and water flow accumulation and also shows that smoothing with the RST reduced the pits in the original DEM and allowed water flow to create continuous streams:

Model of steady state water flow

5. Topographic potential for erosion and deposition estimated by the USPED model

To illustrate the impact of resolution on modeling landscape processes we have computed and visualized transport capacity of water flow (function of slope and uplsope area, also used as modified LS factor for erosion index based on USLE/RUSLE) first for the smoothed DEM in area C using the water flow presented above:

and then also for areas A, B, C, D using DEMs at 90,30 and 10m resolutions:

Computation of net erosion/deposition as a change in transport capacity is very sensitive to artifacts in DEM as illustrated by the following examples (overview image). However, smoothing and reinterpolation to 10m resolution with RST allow us to obtain realistic pattern of erosion/deposition:

6. Water and sediment flow simulation by SIMWE

Results for process based water flow and erosion simulation for uniform rainfall, soil and cover conditions, by the SIMWE model for area D:

steady state water flow ,

It is also possible to simulate the dynamics of water flow during a storm - this animation shows the development of water depth during and after an uniform stationary storm which lasted approx. 5hrs (add hydrographs):

movie