Classification of multi-temporal spectral indices for crop type mapping: A case study in Coalville, UK

Palchowdhuri, Y.; Valcarce-Diñeiro, R.; King, P.; Sanabria-Soto, M.

Abstract

Remote sensing (RS) offers an efficient and reliable means to map features on Earth. Crop type mapping using RS at various temporal and spatial resolutions plays an important role spanning from environmental to economical. The main objective of the current study was to evaluate the significance of optical data in a multi-temporal crop type classification-based on very high spatial resolution and high spatial resolution imagery. With this aim, three images from WorldView-3 and Sentinel-2 were acquired over Coalville (UK) between April and July 2016. Three vegetation indices (VIs); the normalized difference vegetation index, the green normalized difference vegetation index and soil adjusted vegetation index were generated using red, green and near-infrared spectral bands; then a supervised classification was performed using ground reference data collected from field surveys, Random forest (RF) and decision tree (DT) classification algorithms. Accuracy assessment was undertaken by comparing the classified output with the reference data. An overall accuracy of 91% and κ coefficient of 0·90 were estimated using the combination of RF and DT classification algorithms. Therefore, it can be concluded that integrating very high- and high-resolution imagery with different VIs can be implemented effectively to produce large-scale crop maps even with a limited temporal-dataset.

Citation

Palchowdhuri, Y., Valcarce-Diñeiro, R., King, P., & Sanabria-Soto, M. (2018). Classification of multi-temporal spectral indices for crop type mapping: A case study in Coalville, UK. Journal of Agricultural Science, 156(1), 24-36. https://doi.org/10.1017/s0021859617000879

• Journal Article Type: Article
• Online Publication Date: Jan 18, 2018
• Publication Date: Jan 1, 2018
• Deposit Date: Nov 27, 2024
• Journal: Journal of Agricultural Science
• Print ISSN: 1916-9752
• Publisher: Canadian Center of Science and Education (CCSE)
• Peer Reviewed: Peer Reviewed
• Volume: 156
• Issue: 1
• Pages: 24-36
• DOI: https://doi.org/10.1017/s0021859617000879
• Keywords: Crop phenology; Decision tree classification; Machine learning; Random forest classifier; Vegetation indices
• Public URL: https://hull-repository.worktribe.com/output/4716187