The Performance of the Man-Kendall Test in the Analysis of Coastal Changes along Cliff Sections on the Baltic Sea

Airborne Light Detecting and Ranging (LiDAR) surveys became essential in tracking the evolving coastal landscapes of Mecklenburg-Vorpommern on the Baltic Sea for more than one decade, producing a data series of Digital Terrain Models (DTMs) crucial for estimating coastal erosion along the exposed cliffs. Although change detection based on differences between these DTMs is supposed to represent erosion and deposition accurately, a detailed analysis indicates that the initial and final DTMs in the data series sometimes fail to capture the full extent of changes due to various factors. So, natural phenomena, such as the movement of cliff materials (rolling, sliding, creeping), human activities aimed at coastal protection, and errors in DTM processing may disturb clear trends, introducing uncertainties and, in particular, making the data series appear alternating.

To address these issues, we proposed to apply the robust Mann-Kendall test, a non-parametric statistical method used to identify trends in a data series without assuming any particular data distribution. It focuses on determining the direction and consistency of trends (ascending or descending), rather than the change’s magnitude. By implementing this approach, we can pinpoint areas that exhibit clear trends, thereby significantly improving the accuracy of coastal retreat estimations. In regions where trends are not readily apparent, it becomes crucial to investigate potential contributing factors thoroughly by exploring natural environmental dynamics, assessing the impact of human activities, and scrutinizing any errors in data processing. Such a comprehensive analysis ensures a more holistic understanding of the factors influencing these zones.

We employed the proposed approach across four distinguished shore areas characterized by the distinct geological composition of the cliffs, delving into the trends of coastal retreat over the past ten years. As expected for areas with clear trends, the estimation of the dimensions of the recent coastal retreat was in good agreement with historically recorded data. Additionally, in areas exhibiting no discernible trends, we were able to identify the underlying reasons, shedding light on the intricacies of coastal dynamics.