Brian OConnor Presentation
On a methodology to extract measures of seasonality from MERIS reduced resolution data to characterise seasonal trends in Irish vegetation.
Brian O’ Connor1, Ned Dwyer1, and Fiona Cawkwell2
1. Coastal & Marine Resources Centre, Environmental Research Institute, University College Cork; email@example.com 2. Department of Geography, University College Cork
Information extracted from the reflectance characteristics of optical satellite data may offer a complementary approach to ground-based methods of monitoring vegetation seasonality in Ireland. For this study, ten-day composites from 2003 to 2008 of the Fraction of Absorbed Photosynthetically Active Radiation (FAPAR), derived from the European Space Agency’s ENVISAT-MERIS Global Vegetation Index data are used to study spatio-temporal seasonality trends across the island.
Initial work has focused on deriving the start of growing season (SOS) for ten vegetation cover types defined by the CORINE 2000 landcover dataset. The time-series, curve-fitting software, TIMESAT, was used to extract automatically SOS dates for each pixel in each growing season according to cover type. This was done by non-linear least-squares fitting of the time series data to either asymmetric Gaussian or double logistic smoothing functions.
For pasture, the SOS had occurred in 88% of the pixels by 1st-10th April 2007, similar to 82% for the same period in 2003, compared to 67% in 2006 and 61% in 2004. The SOS also varies according to landcover. On average over the 6 year period, SOS occurred in 76% of the pasture pixels by 1st-10th April, compared to just 42% of the coniferous forest pixels. Confidence in these results stems from the fact that the image derived SOS for the Valentia phenological garden site occurred between the first and last observed leaf unfolding dates for all six growing seasons. A seasonality-climate correlation study using variables such as air temperature will be now undertaken to investigate phenology-climate relationships.
Access presentation here (2 meg)
Return to Presentation Programme Page