Estimation of Leaf Area Index using geospatial methods-A review

Abstract

Leaf Area Index (LAI) is an essential vegetation leaf structure parameter for forest and agricultural ecosystems which can be estimated using different means. Remote sensing (RS) methods provide cost-effective alternatives to conventional field-based methods for LAI estimation. This paper presents the concept of LAI and methods for its estimation, with specific emphasis on geospatial methods. It also reviewed the concept of image classification in LAI estimation and the most commonly used image classification algorithms, which include random forest, supervised vector machine, artificial neural network, Bayesian, CARS-SPA, boosting, genetic, lookup tables, K-nearest neighbour, and modified Knearest neighbour, while also highlighting some current research issues associated with LAI estimation from remotely sensed data. The findings of the study suggested that LAI can be accurately estimated from UAV imagery and that random forest, modified KNN, and conventional KNN algorithms are the most suitable classification algorithms for accurate LAI estimation using a remote sensing approach, especially when UAV images are used. The need for considering data science techniques for validation of the image classification approach for LAI estimation was recommended