Accuracy and hydrographic realism of an under-forest DEM derived from airborne P-band radar interferometry over a wide area in the Brazilian amazon

ABSTRACT

P-band radar interferometry is a promising tool for digital terrain modelling in forested areas such as the Amazon due to the ability of P-band to penetrate densely forested areas. The quality of Digital Terrain Models (DTMs) is a primary requirement for many geoscience applications because DTM accuracy is an essential variable for characterizing landform shapes. This article presents a multicriteria evaluation of the overall quality of a digital terrain model produced using P-band SAR interferometric processing. The evaluation considers behaviours related to elevation and slope and analyses the impact of landscape properties such as slope and aspect. The multicriteria accuracy of the study showed promising results about the realism of landforms, according to applied geomorphological criteria, and also in the visual comparison between the produced DTM and the reference. The study provided an estimate of the accuracy of P-band radar DTM in terms of elevation with a standard deviation of 2.36 m and in terms of slope with a mean error of approximately −4°, a standard deviation of 4.74°, and a RMSE (Root Mean Square Error) slope error of 6.14°, using a reference DTM derived from an airborne LiDAR (Light Detection and Ranging) survey. The spatial behaviour of these errors and their sensitivity to slope and aspect were analysed. The investigation of hydrographic inconsistencies in the DTM was conducted based on various criteria, including identifying depressions along watercourses and adhering to Horton’s law. Finally, the effects of the acquisition, processing, and resampling processes are revealed through indicators such as the rose histogram. This multicriteria study demonstrates the suitability of airborne P-band radar interferometry for digital elevation modelling in tropical rainforest environments.

KEYWORDS:

• Quality
• accuracy
• multicriteria
• digital elevation model
• P-band radar
• forested areas
• landforms

Acknowledgements

The authors are grateful to SEMA (Amapá state) for the P- and X-band radar DEMs, and to CTG (French Guiana) for the LiDAR data used in this study. We also thank the researcher Mayara Cobacho Ortega Caldeira for her contributions to this work.

Disclosure statement

No potential conflict of interest was reported by the author(s).