ul 2004 
+ TSO, 
Bartin 
-Cevre 
lantısı, 
nlarimin 
3tkileri. 
45-355. 
ng and 
IS, Ínc., 
002). A 
r Land 
XXIII 
izanimi 
Si) IL 
Mayis 
nlemler 
project 
rmany) 
project 
1guldak 
EVALUATION OF AVHRR NDVI FOR MONITORING INTRA-ANNUAL AND 
INTERANNUAL VEGETATION DYNAMICS IN A CLOUDY ENVIRONMENT 
(SCOTLAND, UK) 
S. Brand *. T.J. Malthus" 
“ Institute of Photogrammetry and Remote Sensing, University of Karlsruhe, Englerstr. 7, 76128 Karlsruhe, Germany, 
brand @ipf.uni-karlsruhe.de 
? School of GeoSciences, University of Edinburgh, Drummond St, Edinburgh EH8 9XP, UK, ijm? geo.ed.ac.uk 
KEY WORDS: Land Cover, Vegetation, Change Detection. Monitoring, Multitemporal, Interpretation 
ABSTRACT: 
Vegetation change detection has become increasingly important in understanding vegetation dynamics and its role in terrestrial and 
aunospheric systems. In the case of Scotland, there is a need to routinely monitor habitats which include native pine woodland, 
montane habitats, upland heathland and blanket bog (EC Habitats Directive). In northern temperate regions restricted and distinct 
growing seasons impart significant phenological changes in vegetation which are detectable in remotely sensed data. However, the 
high frequency of cloud cover in northern temperate climates presents a complicating factor in the use of remote sensing data for 
monitoring vegetation. High cloud probability means loss of data which makes it more difficult to compile and interpret time-series 
of data than would be the case in less cloudy areas. For this study satellite data of coarse spatial resolution and high temporal 
frequency from the NOAA/AVHRR sensor have been evaluated for their utility in monitoring interannual and intra-annual 
vegetation dynamics of semi-natural and agricultural vegetation types in Scotland. Weekly composited NDVI images for 1995 to 
1998 were used in conjunction with digital maps of vegetation distribution to extract NDVI time-series. Temporal development 
curves of the NDVI values for various vegetation cover types were visually compared to observe the phenology and to detect 
changes in NDVI response between years. The weekly time-series approach proved to be useful for understanding seasonal dynamics 
and allowed for interannual comparisons despite the high cloud-cover rate. However, it has also become evident that pre-processing 
and accurate calibration of the AVHRR imagery is extremely important. 
1. INTRODUCTION 
Since the 1970s it has been possible to receive radiometric data 
from meteorological and Earth resources satellites that allow us 
to analyse the interactions between the solar radiation spectrum 
and ocean, atmosphere and land properties. In particular. the 
high temporal resolution sensor Advanced Very High 
Resolution Radiometer (AVHRR) of the meteorological 
satellite series TIROS-NOAA (National Oceanic and 
Atmospheric Administration) has been used to gain knowledge 
of the temporal, spatial and reflective behaviour of vegetation. 
The daily coverage that this series of satellites and sensors has 
provided offers a unique and long-term dataset with which to 
investigate the magnitude of changes in vegetation at 
continental and global scales (Lambin, Ehrlich 1997: 
Malingreau, 1986). 
Vegetation change detection has become increasingly important 
because of the current interest in the establishment of human 
impacts upon the environment. Long-term monitoring 
programmes have been developed to examine and monitor these 
changes. Conservation bodies like Scottish Natural Heritage 
(SNH) are required by national and European legislation to 
routinely monitor important and protected habitats. In the case 
of Scotland, these habitats include native pine woodland, 
montane habitats, upland heathland and blanket bog, each 
recognised by the EC Habitats Directive (ECNC,. 1992. 
MLURI, 1998). Information on the phenology and dynamics of 
7 
vegetation types and the magnitude and speed of their possible 
changes needs to be provided to these conservation agencies. 
Remote sensing could play an important role in identifying 
areas of change at large scales. However, in north temperate 
climates restricted and distinct growing impart 
significant phenological change in vegetation, the magnitude 
and pattern of which needs to be understood in order that the 
more important longer-term interannual changes can be 
properly identified and understood. 
seasons 
This study is based on weekly NDVI data of the AVHRR sensor 
over Scotland from 1995 to 1998. It aims to investigate and 
assess the use of this imagery for monitoring the phenology of 
semi-natural vegetation in an environment with a high cloud 
cover rate. Most studies, especially early on, have focussed on 
monitoring and mapping of African, South-American or Asian 
vegetation communities, mostly using monthly composited 
NDVI data or single images (Malingreau, 1986; Lambin, 
Ehrlich, 1997). Several studies analysed time-series satellite 
data in order to develop intra-annual (DeFries et al., 1995) and 
inter-annual (Reed ct al. 1994) profiles of the condition of 
vegetation. Few studies have dealt with regional monitoring of 
vegetation dynamics in a northern temperate climate using 
short-term time-series of AVHRR NDVI data. The monitoring 
of vegetation with a continuous time-series of weekly NDVI in 
such a climatic environment will be made difficult by the 
frequent presence of clouds and haze. To minimise this 
problem. the composited period can be augmented. e.g. monthly