Mapping Prosopis Juliflora Aboveground Biomass Using Remote Sensing in Taveta Sub-county, Kenya

Abstract

Prosopis juliflora is a widespread invasive species listed on the Global Invasive Species Database (GISD) among the 100 most invasive species in the world. Introduced in Kenya in the 1970s–1980s, it has spread rapidly and is now found in 2% of Kenya’s landmass (KEFRI, 2020) with the ability to double its coverage every 5 years under favourable conditions (IUCN (International Union for the Conservation of Nature), 2010). Prosopis juliflora has negative impacts such as loss of biodiversity, injury to humans and livestock, and loss of livelihood for the affected communities. Its invasive characteristics make it a challenging species to control and eradicate. A Prosopis juliflora aboveground biomass map contributes to a better understanding of its spatial distribution and helps to inform control and management approaches. This study aims to evaluate the suitability of LiDAR point clouds to estimate the aboveground biomass (AGB) of Prosopis juliflora and subsequently map its AGB distribution. The AGB of Prosopis juliflora in 21 sample field plots were estimated using allometric equations. Regression models were then fitted over the field estimated AGB and LiDAR metrics to identify the significant prediction variables. These were used to model wall-to-wall AGB in the study area. To segregate Prosopis juliflora AGB from other vegetation, Sentinel 2 imagery was classified using a random forest binary classification. The results were a Prosopis juliflora AGB map and a predictive model for estimating Prosopis juliflora AGB from LiDAR data. The map could help policymakers to develop and implement effective control, management, and utilisation measures targeting areas with the highest biomass. e.g., as outlined by Adoyo et al., (2021). Uncertainties in the obtained results are due to, among others, the choice of allometric equations and classification errors attributable to the mixture of Prosopis juliflora with other vegetation species, and the spatial resolution of Sentinel 2 imagery.