Incorporating plant phenological trajectory in exotic saltcedar detection with monthly time series of Landsat imagery

Chunyuan Diao, Le Wang

Research output: Contribution to journalArticle

Abstract

Over the past century, non-native saltcedar (Tamarix spp.) has expanded into most of riparian zones of the southwestern United States and posed significant threats to the native biotic communities. Repeated monitoring of saltcedar distribution over region-wide geographic areas is urgently essential for conservation agencies to develop cost-effective control strategies. Current studies have mostly concentrated on the mapping of saltcedar distribution with a single remote sensing image acquired during its leaf senescence. Given the phenological variation within saltcedar and the spectral confusion between saltcedar and natives, our ability to detect saltcedar with a single-date image is still limited. The objective of this study was to develop new intra-annual phenology-based strategies to detect exotic saltcedar with monthly time series of Landsat imagery. Several temporal phenology-based detection strategies (i.e., phenological bands, phenological NDVI, and phenological metrics) that could track the intra-annual phenological trajectory of plants were devised. With the proposed detection strategies, crucial months and phenological stages in saltcedar detection were investigated. Results indicated that the proposed strategy of phenological bands could accommodate the phenological variation within saltcedar and improve the classification accuracy significantly. Several phenological stages (e.g., flowering and leaf senescence) were deemed as important in discriminating saltcedar from other riparian plants at the Landsat scale. The proposed strategy was found to be relatively robust to the lack of a single Landsat image. It is concluded that monthly time series of Landsat imagery are promising in facilitating the long-term mapping of saltcedar distribution over extended areas.

Original languageEnglish (US)
Pages (from-to)60-71
Number of pages12
JournalRemote Sensing of Environment
Volume182
DOIs
StatePublished - Sep 1 2016
Externally publishedYes

Keywords

  • Feature selection
  • Image classification
  • Intra-annual phenology
  • Landsat
  • Random Forest
  • Saltcedar

ASJC Scopus subject areas

  • Soil Science
  • Geology
  • Computers in Earth Sciences

Fingerprint Dive into the research topics of 'Incorporating plant phenological trajectory in exotic saltcedar detection with monthly time series of Landsat imagery'. Together they form a unique fingerprint.

  • Cite this