Digital Surface Models (DSMs) are commonly employed to investigate topographical characteristics and processes; however, the presence of canopy and infrastructure in urban and forested areas can lead to height biases and inaccuracies. In this study, I aim to correct such biases by applying a deep learning approach known as Residual U-Net to remove the selected pixels and generate Digital Terrain Models (DTMs) that accurately represent the Earth's surface without canopy and infrastructure influence.

The Residual U-Net model was trained and tested on a dataset of DSM and DTM pairs, which were acquired from resampled AHN4. The model was evaluated on its ability to predict DTMs from DSMs, and its performance was compared with other existing methods. Additionally, the model was tested on different resolutions and the Copernicus DEM to assess its adaptability and generalization capabilities.

The results indicate that the Residual U-Net model outperforms conventional techniques, effectively reducing the influence of canopy and infrastructure, and resulting in DTMs with enhanced precision. The study also explores the errors in detail and identifies the model's error causes, highlighting its limitations and areas for potential improvement.

This study concludes by demonstrating the efficacy of applying deep learning techniques, such as Residual U-Net, to correct global elevation models for canopy and infrastructure. The results indicate that the model is a promising tool for topographical investigation in both urban and woodland situations, offering a versatile solution for generating accurate DTMs from DSMs.

With the increasing pressure on global forests due to deforestation and habitat loss, it is more important than ever to understand our forest ecosystems on a larger scale. Canopy gap fraction is an indicator used for estimating forest biomass and better understanding of ecosystem functioning. For decades there have been research on computing canopy gap fraction using ground measurements as well as Airborne Laser Scanning (ALS). These approaches, however, are limited by the data available. A global coverage of data on canopy gap fraction could be made available by space-based laser altimetry mission Ice, Cloud and land Elevation Satellite (ICESat-2). Although its main scientific objectives are focused on polar areas, it has already proven to facilitate broader scientific disciplines. The ATL08 data product is focused on land-vegetation and already provides global data on canopy heights. However, there is ongoing research for using ATL08 data for canopy gap fraction estimation. Although some approaches have been suggested in the literature, a tested workflow to achieve this goal has not been published.

This thesis tests two methods for estimating canopy gap fraction from ICESat-2 ATL08 data and evaluates the results against openly available ALS data. First, a simple method of using canopy to total photon ratio is used. Then, an alternative method that aims to correct for the surface reflectivity is tested. The results from both methods are similar, therefore the computationally less expensive method is recommended. Although this thesis does not achieve to present a sufficiently accurate approach for using ICESat-2 ATL08 data for canopy structure estimation, it is shown that further research is needed and the results are promising. Furthermore, it is demonstrated that annual trends in canopy gap fraction can be seen in ATL08 data. Considering the global coverage of ICESat-2 data, it is concluded that despite the accuracy not meeting the expectations, using ATL08 for studying canopy gap fraction on a global scale and through time has high value and great potential for environmental research.