Lars Eklundh

Spring phenophases such as the beginning of leaf unfolding, measured in the Irish gardens of the International Phenological Garden (IPG) network, indicate an earlier spring occurrence hence a longer growing season. However, these measurements are limited to selected species of trees at a few point locations in the southern half of the country. The aim of this study was to develop a methodology, based on satellite remote sensing, to measure the vegetation start of season (SOS) across the whole island of Ireland on an annual basis, complementary to existing ground-based methods. The SOS metric was extracted for each year in a 7-year time series of 10-day composited, 1.2 km reduced resolution MERIS Global Vegetation Index (MGVI) data from 2003 to 2009, based on curve fitting, using the time series analysis software, TIMESAT. Spatio-temporal variability in the SOS was detected across the island on an annual basis and highlighted in a series of anomaly images showing variation from the 7-year mean SOS. The 2006 SOS was late across the island while there were strong geographical gradients to the SOS anomalies in 2009 when it occurred later in the south and earlier in the north. There was a mix of early and late anomaly values throughout the country in the other years. Qualitatively, the spatial patterns in the timing of the SOS were related to the distribution of landcover types as indicated by the CORINE Land Cover map (CLC). Three statistically separable groups of CLC classes were derived from differences in the SOS, namely agricultural and forest land cover types, peat bogs, and natural and semi-natural vegetation types. These groups demonstrated that vegetation in cultivated areas like pastures has a significantly earlier SOS than in areas of unmanaged vegetation such as peat bogs. An initial climate analysis indicated that an anomalously cold winter and spring in 2005/2006 delayed the 2006 SOS countrywide; while a cold winter followed by a mild spring in 2009 caused considerable spatial variability in the 2009 SOS across the country, ranging from later SOS in the south to early SOS in the north. This study has demonstrated the utility of 10-day MGVI composites for derivation of an SOS metric which can be used as an indicator of spatial variability in vegetation seasonality and has highlighted how SOS varies according to landcover type. The availability of longer time series in the future will allow more focused studies on the sensitivity of the SOS metric to changes in climate as well as short term weather events. (C) 2012 International Society for Photogrammetry and Remote Sensing, Inc. (ISPRS) Published by Elsevier B.V. All rights reserved.