University of Twente Student Theses


Estimation of leaves nutrient content in seagrass using spectral data: The case of Halodule uninervis

Serusi, Simona (2010) Estimation of leaves nutrient content in seagrass using spectral data: The case of Halodule uninervis.

[img] PDF
Abstract:Seagrasses are plants living under water in coastal areas. They are threatened by many factors due to human activities, being eutrophication one of the main causes of their disappearance. This work focuses on a small species located in tropical areas, Halodule uninervis. Leaf material was collected for chemical and spectral analysis in the shallow water surrounding Derawan Island, East Kalimantan, Indonesia. Chemical analysis gave the total carbon, nitrogen and phosphorus. Descriptive statistics was carried out to explore the data. Spectral data were obtained in laboratory with an ASD spectrometer. Finally, cross validated stepwise multiple linear regression was applied on reflectance, first derivative and continuum removal in order to estimate nutrient content on entire and ground leaves. Carbon, nitrogen and phosphorus content resulted to be lower than those reported in other areas; in particular P might indicate some limitation in the environment. The species was not limited by N. Results revealed that difference in spatial distribution of the seagrass parameters' are not significant however all of them shown higher average in the area further from the human settlement. Best model for nitrogen prediction was found applying first derivative transformation on mill dry leaves (r2 = 63%, RMSE = 0.098); best model for phosphorus was also obtained with first derivative but on entire semi dry leaves (r2 = 38%, RMSE = 0.03), both of them using wide range of the spectrum. Low r2 have been obtained for the semi fresh leaves and on the visible part of the spectrum. Selected bands were often in agreement with absorption features due to the chemical of interest. These results seem encouraging also because predictors were often bands related to absorption features; however the application to remote sensing data requires approaches that predict better, and also correction for the presence of the atmosphere and the water column. Keywords: seagrass, Indonesia, spectroscopy, nutrients, stepwise regression
Item Type:Essay (Master)
Faculty:ITC: Faculty of Geo-information Science and Earth Observation
Programme:Geoinformation Science and Earth Observation MSc (75014)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page