Terrestrial Laser Scanning (TLS) has great potential in creating high resolution digital elevation models (DEMs). However, little is known about the properties of TLS derived DEMs covering several hectares in heterogeneous environments compared to conventional airborne laser scanning (ALS) based models and their influence on derived products. We investigated the accuracy of DEMs with different resolutions derived from TLS and high quality ALS on a study site with complex micro-topography covered by dense forest and ground vegetation. We further examined the effect of these DEMs on predicted topsoil pH using linear regression models built on terrain attributes. We show that at high resolutions (∼1 m), TLS based DEMs performed better than ALS DEMs, which directly translated into significantly better pH models, the best of which showing an R2 of 0.62. The use of TLS therefore improves the quality of terrain attributes, which are the foundation for many ecological and hydrological applications.