| Summary: | Quantitative monitoring of vegetation change over time is essential in understanding the environmental processes of which are important in climate change and global warming models, because vegetation change is an indicator of environmental variability. However, obtaining such information has been a challenge especially for vegetation phenology due to the lack of appropriate methods for quantitative assessment. There is therefore a need to derive methods to quantitatively characterize vegetation dynamics in order to monitor the effect of climate change on the biosphere and as inputs to global change models. The aim of this research was to test the relationships between ground-based measurement of leaf area index (LAI) and vegetation indices (VI) derived from satellite remote sensing instruments to quantitatively monitor vegetation dynamics in a broadleaf and coniferous forest in the UK. This research has four key hypotheses. First, phenological changes (which is the timing of recurring biological events in plants) in broadleaf and coniferous forest canopies may be characterized using ground-based measurement of LAI, because LAI is good proxy for vegetation phenology. Second, cloud cover frequency in the UK leads to a requirement for higher temporal resolution remote sensing data to monitor changes in vegetation phenology. Third, data from the Disaster Monitoring Constellation (DMC) satellites provides a sufficiently high temporal resolution for monitoring vegetation phenology in the UK. Fourth, vegetation indices derived from atmospherically corrected DMC data may be used to monitor vegetation phenology in the UK. Analysis of Advanced Very High Resolution Radiometer (AVHRR) and Moderate Resolution Imaging Spectroradiometer (MODIS) cloud mask showed that the average of number of cloud free days at the UK test sites in the year 2005 was five days per month with a minimum of one cloud free day per month implying that high temporal resolution satellites like the DMC will be appropriate for monitoring vegetation change. Nine DMC satellite images were acquired over 2005/2006 for the study sites plus one coincident Landsat ETM+ in 2005. Four vegetation indices (VI) were derived from the satellite data sets and were related to LAI/PAI. PAI is the plant area index defined as the total surface area of both photosynthetic and nonphotosynthetic part of plant per unit ground area. A regression model was used to predict LAI/PAI and the root mean square error (RMSE) was determined for both sites. The RMSE of the observed and predicted LAI values show that the levels of errors at Risley Moss were 0.51 for LAI, 0.52 for overstorey PAI and 0.8 for total canopy while PAI was 1.1 for Charter's Moss. Therefore, the DMC and one Landsat ETM+ data set related to LAI/PAI can adequately retrieve biophysical parameter in the deciduous woodland. However, in the coniferous canopy the numbers of observations was fewer and the measurement errors larger leading to a requirement for more data in order to establish statistically significant and ecologically useful relationships. Improvements in the accuracy of ground-based LAI/PAI measurements, radiometric and atmospheric correction of satellite data are expected to increase the accuracy of such LAI/PAI estimates in future.
|
|---|
