Catchment properties in the Kruger National Park

derived from the TanDEM-X Intermediate Digital Elevation Model (IDEM)

1 Introduction and Aims

In 2014, the first Intermediate DEM (IDEM) was released for scientific users providing

a subset of the new global Digital Surface Model (DSM) with up to 12 m geometric

resolution and 2 m relative height accuracy. This dataset is based on TanDEM-X

acquisitions during the first year only (Moreira et al. 2004).

Here we report on the application of this fascinating new data in the framework of

hydrological and geomorphological investigations in Kruger National Park (KNP).

References

Baade, J. & C. Schmullius (2014): Uncertainties of a TanDEM-X derived Digital Surface Model. A Case Study from the RodaCatchment, Germany. - Proceedings IGARSS 2014: 4327-4330..

Moreira, A., G. Krieger, I. Hajnsek, D. Hounam, M. Werner, S. Riegger, and E. Settelmeyer (2004): TanDEM-X: A TerraSAR-X Add-On Satellite for Single-Pass SAR Interferometry. IGARSS 2004 Proceedings, Vol. 2: 1000 – 1003.

Acknowledgement

IDEM data was provided by the Deutsches Zentrum für Luft- und Raumfahrt (DLR) under the grant IDEM_Other0118 ‚Kruger

National Park Erosion Research DEM (KNPErosIDEM)‘. First ground truething results were obtained in the framework of the

KNP Erosion Research Project funded by the German Research Foundation (Deutsche Forschungsgemeinschaft, DFG, grant:

BA1377/12-1) and conducted in close cooperation with SANParks Scientific Services in Skukuza.

Jussi Baade ,1

1

Christiane Schmullius & Izak Smit1 2

Department of Geography, Friedrich-Schiller-University, D-07743 Jena, Germany, Scientific Services, South African National Parks, Skukuza 1350, South Africa

Contact: Jussi.Baade@uni-jena.de

2

13th Annual Savanna Science Network Meeting, March, 8 - 12, 2015, Skukuza

4 Conclusions

The application of the new satellite-based global digital elevation model (IDEM)for hydrological and geomorphological studies provides evidence for the greatpotential of the data set. However, there are as well some limitations: the highresolution requires adequat computing resources, and the fact that the datarepresents the canopy might not the benefical to all applications. Further work isneeded to disentangle canopy and terrain heights in the framework ofcatchment property characterization.

Fig. 1: IDEM04-based Hillshade representation of the Silolweni catchment. Catchment boundaries werederived from the IDEM10 (magenta) and the KNP 20-m-DEM (yellow) using ArcHydro tools

(data sources: , KNP 2008).©DLR 2014

3 Results

3.1 IDEM10-based catchment delineation

3.2 IDEM04 and terrain point cloud comparison

Figure 1 provides evidence for the unprecedented geometric resolution of thesatellite-based IDEM04 digital elevation data set. The undulating character of theterrain is well visible. However, vegetation structure is represented as well in thisDSM. For the delineation of channels and catchments this is a disadvantage,especially when channels are lined by riparian forests. Work to separate the canopysurface from the terrain in the high-resolution IDEM04 is under way.

Drainage network and catchment delineation for the southern part of KNP (Fig. 2)

was conducted based on the lower resolution IDEM10 after attempts to use the high

resolution IDEM04 in one go for the whole area failed due to computer resources

limitations. Compared to the previously used 20 m DTM, the IDEM10 clearly pics up

more details of the undulating landscape (max. relief ~ 100m) in Silolweni catchment

(Fig. 1). However, the difference in catchment size between the two approaches is

less than 1 % for Silolweni and less than 5 % in any other study site.

Figure 3 shows the comparison of IDEM04 raster height values (m HAE) with the

RTK-GNSS derived terrain point cloud (N = 1087, mapped in 2014) for the vicinity of

Silolweni reservoir. Basically, the comparison provides the expected results. Outside

of the meanwhile dried out reservoir the differences are predominantly positive, due

to the IDEM data representing the bush and tree canopy (Baade & Schmullius 2014).

Roads and the grassland surface of the former reservoir basin is often characterized

by negative values (differences up to 2,5 m) or by differences smaller than the IDEM

inherent height error of ~ 0.5 m.

2 Material and Methods

Due to time and computer resources restrictions, catchment delineation was based

on the 1 arc sec (~ 30 m) IDEM10 and compared to results derived from the existing

NGI 20 m Digital Terrain Model (DTM). Detailled analysis is based on the 0.4 arc sec

(~ 12 m) IDEM04 and RTK-GNSS-based terrain point clouds. The Silolweni

catchment located in the Tshokwane section of KNP is used as a case study to

highlight detailed results (Fig. 1).

Fig. 2: Drainage network in the southern part of KNP and reservoir siltation study sites derived from theIDEM10 using ArcHydro tools (data source: DLR 2014).©

Fig. 3: Height difference (m HAE) from IDEM04 raster values and RTK-GNSS terrain point cloudmeasurements (N = 1087) in the vicinity of Silolweni reservoir. The mean height error of IDEM04

is 0.40±0.08 m and 0.04(data source: DLR 2014, NGI 2009).

andpoint cloud data ±0.01 m, respectively. Please note: Silolweni reservoir was

breached in 2010 and dry at the time of TanDEM-X acquisition ©