Vegetation height estimation using ubiquitous foot-based wearable platform
- PDF / 4,535,664 Bytes
- 19 Pages / 547.044 x 736.903 pts Page_size
- 36 Downloads / 155 Views
Vegetation height estimation using ubiquitous foot-based wearable platform Sofeem Nasim · Mourad Oussalah Kl¨ove · Ali Torabi Haghighi
· Bjorn
Received: 12 November 2019 / Accepted: 26 October 2020 / Published online: 21 November 2020 © The Author(s) 2020
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 (r 2 = 0.81 and RSME = 16.73 cm), while the use of a multi-regression model yields (r 2 = 0.82
S. Nasim · M. Oussalah () Centre of Machine Vision and Signal Processing, Faculty of Information Technology, University of Oulu, Oulu, Finland e-mail: [email protected] B. Kl¨ove · A. T. Haghighi Water, Energy and Environmental Engineering Research Unit, University of Oulu, Oulu, Finland
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. Keywords Vegetation height · Machine learning · Ubiquitous sensor platform
Introduction Vegetation height is a key indicator for many terrestrial ecosystems linked to habitats, their biodiversity, and biomass structure (Hyde et al. 2006; Dong and Wu 2008; Nilsson 1996). Indeed, vegetation height is considered one of the most important forest properties and a fundamental characteristic for several areas of ecological studies; particularly, fire modeling, biodiversity monitoring, and disaster management where it can be utilized for classification of land cover or estimating forest age and habitat quality. For instance, vegetation height is highly correlated with vegetation biomass (Hyde et al. 2006), which is the fundamental element of the carbon cycle and a substitute for fuel loading estimation (Finney 1998). Short vegetation, known as herbaceous vegetation, plays a key role in determining the confined livestock grazing and climatic variability as agents of
774
Page 2 of 19
vegetation change (Fuhlendorf et al. 2001).
Data Loading...
