Abstract

Vegetation height plays a key role in many environmental applications such as landscape characterization, conservation planning and disaster management, and biodiversity assessment and monitoring. Traditionally, in situ measurements and airborne Light Detection and Ranging (LiDAR) sensors are among the commonly employed methods for vegetation height estimation. However, such methods are known for their high incurred labor, time, and infrastructure cost. The emergence of wearable technology offers a promising alternative, especially in rural environments and underdeveloped countries. A method for a locally designed data acquisition ubiquitous wearable platform has been put forward and implemented. Next, a regression model to learn vegetation height on the basis of attributes associated with a pressure sensor has been developed and tested. The proposed method has been tested in Oulu region. The results have proven particularly effective in a region where the land has a forestry structure. The linear regression model yields (r² = 0.81 and RSME = 16.73 cm), while the use of a multi-regression model yields (r² = 0.82 and RSME = 15.73 cm). The developed approach indicates a promising alternative in vegetation height estimation where in situ measurement, LiDAR data, or wireless sensor network is either not available or not affordable, thus facilitating and reducing the cost of ecological monitoring and environmental sustainability planning tasks.