An accurate retrieval of leaf water content from mid to thermal infrared spectra using continuous wavelet analysis

S. Ullah, A.K. Skidmore, M. Naeem, M. Schlerf

Research output: Contribution to journalArticleAcademicpeer-review

36 Citations (Scopus)

Abstract

Leaf water content determines plant health, vitality, photosynthetic efficiency and is an important indicator of drought assessment. The retrieval of leaf water content from the visible to shortwave infrared spectra is well known. Here for the first time, we estimated leaf water content from the mid to thermal infrared (2.5-14.0µm) spectra, based on continuous wavelet analysis. The dataset comprised 394 spectra from nine plant species, with different water contents achieved through progressive drying. To identify the spectral feature most sensitive to the variations in leaf water content, first the Directional Hemispherical Reflectance (DHR) spectra were transformed into a wavelet power scalogram, and then linear relations were established between the wavelet power scalogram and leaf water content. The six individual wavelet features identified in the mid infrared yielded high correlations with leaf water content (R 2=0.86 maximum, 0.83 minimum), as well as low RMSE (minimum 8.56%, maximum 9.27%). The combination of four wavelet features produced the most accurate model (R 2=0.88, RMSE=8.00%). The models were consistent in terms of accuracy estimation for both calibration and validation datasets, indicating that leaf water content can be accurately retrieved from the mid to thermal infrared domain of the electromagnetic radiation.
Original languageEnglish
Pages (from-to)145-152
JournalScience of the Total Environment
Volume437
DOIs
Publication statusPublished - 2012

Keywords

  • remote-sensing imagery
  • hyperspectral data
  • mu-m
  • spatial heterogeneity
  • reflectance data
  • transform
  • vegetation
  • leaves
  • classification
  • compression

Fingerprint Dive into the research topics of 'An accurate retrieval of leaf water content from mid to thermal infrared spectra using continuous wavelet analysis'. Together they form a unique fingerprint.

  • Cite this