skip to primary navigation skip to content
 

SPRI Review 2001: Remote Sensing Group

Remote Sensing Group

Dr W.G. Rees

Professor P.J. Williams, Dr O.V. Tutubalina, J.S. Ash, M. Balshi, A.J. Fox

Spatial variation of the rate of change of vegetation vigour

Changes in vegetation cover are among the more obvious of the alterations that mankind has wrought on the global environment. Historically these changes have been local or regional in effect, although widespread. For example, it has been estimated that 15-20% of the world's forested area has been lost through human action since 1700. More recently, concern has been expressed that human activity is altering the global climate system, with potential consequences for the global distribution of vegetation. In fact, the vegetated part of the biosphere contributes to several important feedback processes in the global climate sys-tem, for example through the carbon and water cycles and the effect on albedo. Thus vegetation is both driven by and drives the climate.

For many reasons, the detection and monitoring of the dynamics of vegetation is highly desirable. This is especially true of the tundra biome, between the polar treeline and the cryosphere. Here the environment is harsh, biodiversity is low, and recovery times after disturbance are unusually long. The tundra biome is mostly vegetated but treeless, consisting of grasses and sedges, mosses, lichens, and dwarf shrubs. In many places the tundra is boggy or swampy, a consequence of poor drainage due to the underlying permafrost.

As well as the potential effects of global climate change, the Arctic tundra is also in places subjected to pollution and other local and regional-scale disturbances. The most important types of pollution disturbance arise from contamination by heavy metals and acidification, both principally a consequence of the extraction and processing of non-ferrous metals, and by oil leaks at production fields and from pipelines. These phenomena are particularly concentrated in the Russian Arctic. Another, perhaps rather unexpected, source of disturbance is grazing by reindeer. A large area of the Eurasian tundra, perhaps as much as two million square kilometres, as well as a significant area of the adjacent boreal forest, is used for reindeer herding or is grazed by wild or feral deer. This raises the possibility of overgrazing and of long-term changes in the vegetation cover. Preliminary work suggests that the tundra lichens are undergoing a sustained and very widespread decline, probably at least in part attributable to grazing.

It is becoming increasingly clear that any attempt to interpret changing vegetation patterns in the Arctic in terms of global climate change must take into account these various more localised influences. This is especially so within northern Russia, although the phenomena are probably circumarctic in nature. The Remote Sensing in NDVI with regional anomalies, particularly in the Nenets Autonomous Okrug and north central Siberia. The causes of these variations are not known, but are under investigation. It has been suggested that the spatial anomalies may be due in part to changes in the duration of the snow-free period, and this possibility is being investigated as part of the Remote Sensing Group's study of snow, not discussed here. However, it is probably significant that the spatially largest anomaly includes the area around the city of Noril'sk. This, the largest settlement in the Arctic, is a major source of atmospheric emissions of sulphur dioxide and of heavy metals through the smelting of nickel and other non-ferrous metals. These heavy emissions (annual emissions of SO2 can exceed two million tonnes) have had a major impact on the vegetation around the city, particularly on the boreal forest to the south. ASTER satellite image of Noril'sk and its surroundings, covering an area roughly 100 x 100 kmThe figure to the right shows part of an ASTER satellite image of Noril'sk and its surroundings, covering an area roughly 100 x 100 km. It is a grey-scale representation of a false-colour infrared image, in which vigorous green-leaved vegetation appear as a mid-grey. Part of Lake Pyasino can be seen at top centre, visualised by plumes of sediment, and the Karaelach Mountains are visible at top right and the Lontokoisky Group is engaged in a long-term programme of developing and applying techniques for monitoring vegetation dynamics in response to specific disturbances, using primarily data from satellite imaging systems. This work is mainly concentrated in Russia, taking advantage of a long-standing collaboration with the Geography Faculty of Moscow State University.

The figure at the top of the page was derived by Michael Balshi as part of his MPhil research entitled 'Satellite monitoring of the response of circumarctic vegetation to environmental and anthropogenic forcing.' It was calculated from a 20-year series of satellite data and shows the spatial variation of the rate of change of vegetation vigour, derived using the Normalised Difference Vegetation Index (NDVI), during this period. The lightest grey indicates a mean annual increase of the NDVI of greater than 0.008, with progressively darker shades indicating rates of 0-0.008, -0.008-0, and less than -0.008, respectively. The figure shows a general latitude-dependent increase Kamen' ridge at bottom right. The city of Noril'sk is clearly visible at the northern end of the ridge, as is a smoke plume from one of the smelters. At the right centre of the image, one can identify an area of very poor vegetation, similar in appearance to the surrounding mountains. In fact this is the northern extremity of a vast area of anthropogenic disturbance, where pollutants have been carried southwards by prevailing winds. Quantitative measurements based on the detailed analysis of satellite images of this area dating back to 1972 reveal that an area of about 2000 square kilometres, and perhaps more, now consists of dead or severely damaged vegetation.

Kol'skiy Poluostrov

This type of satellite-based analysis of vegetation change as a result of atmospheric pollution has been carried out both for the Noril'sk region and on the Kol'skiy Poluostrov, the other main source of SO2 and heavy metal pollution (again from the smelting of non-ferrous metals) in the Arctic. The next step is to develop a quantitative understanding of the relationship between the atmospheric emissions and the nature and extent of the damage. This is complicated by the topography of the area, which modifies the manner in which movement of the air distributes the pollutants. The figure above, a photograph taken during a field trip by Dr Gareth Rees and Dr Olga Tutubalina to the Kol'skiy Poluostrov in summer 2001, illustrates this problem.

The source of pollution, the smelters at Monchegorsk, lies a few kilometres to the north of the point from which the photograph was taken, behind the mountain at the right. All vegetation on this mountain has been utterly destroyed indeed, even the soil has been destroyed but the mountain at left of the photograph provides effective shelter for some regions to the west of the smelter. The area in the immediate foreground is extensively damaged. In general, we observe that the extent of the damage is governed by distance from the smelter, elevation, and aspect. We have begun to develop a model that combines these factors.

The second main strand of the Remote Sensing Group's current research on high-latitude vegetation dynamics is to investigate the nature of the changes induced in tundra vegetation as a result of grazing by reindeer. The figure on the following page shows one main study area, in the Nenets Autonomous Okrug, Russia (a new study was initiated in 2001 in the Komi Republic). This map shows an area 72 x 40 km, on which has been indicated the four main pasture regions used by a single brigade of the Indigsky Sovkhoz (state farm). The herd managed by this brigade consists of about 1000 animals. The regions are labelled according to the approximate months of the year during which they are used for pasture.

Analysis of field data and satellite imagery from this area, spanning the period 1988-2000, shows that in the winter, spring, and summer pastures a more or less consistent change in vegetation has occurred. There has been a marked decline in the extent of the highest-biomass tundra vegetation (healthy dwarf shrub), in favour of (a) a lower-biomass version of what is essentially the same vegetation type, and (b), perhaps somewhat paradoxically, a vegetation type (grass and birch/willow scrub) with a higher biomass. Both of these phenomena can be explained in terms of heavy grazing and trampling by reindeer: the first is obvious, and the second represents a succession vegetation following heavy disturbance and manuring. Interestingly, the autumn (August-October) pasture shows a significantly different pattern of change during the same period, and although we are not yet convinced of the interpretation to be put on these results they seem to suggest that the vegetation state here is improving, or at least declining substantially less rapidly than in the other pasture areas. Although these results require corroboration, they are thought likely to be representative of a large area of the Eurasian tundra zone.

Mapping vegetation from satellite imagery can be performed using in situ observations of the distribution of vegetation types within selected 'training areas' and extrapolating from these using the statistical properties of the image. In essence, the in situ information is used to identify the spectral signature (variation of reflectance with wavelength) of a given vegetation type as it appears in the image, and a computational algorithm is then used to identify other regions of the image that have similar signatures. However, substantially greater understanding of, and accuracy in, this classification process can be obtained if the spectral signatures can be related directly to the physical (including physiological) properties of the vegetation or other target material. Little information is currently available concerning the spectral reflectance properties of Arctic vegetation species, especially lichens, through the optical and near-infrared regions of the electromagnetic spectrum. In the summer of 2001 Dr Rees and Dr Tutubalina attempted a start on this problem by arranging to borrow a GER3700 spectroradiometer from NERC and deploying it in the field on the Kol'skiy Poluostrov. This work would have been undertaken in collaboration with botanists from Moscow State University, allowing a comparison of reflectance properties with phytomass and plant health. Unfortunately the Russian customs authorities refused to allow the GER3700 in the country. We now plan to carry out a variant of this work in Sweden next year, while we attempt to solve the increasingly difficult problem of importing scientific equipment into Russia.

Map image

This brief discussion of research into Arctic vegetation dynamics began with the circumarctic view and has been brought down to the scale of individual plants. It is useful to consider how the research can be placed in a wider context. Firstly and most obviously, the development of increasingly robust methods for identifying localised 'hotspots' of vegetation change around what are essentially point sources of pollution introduces the possibility that these methods can be applied systematically throughout the Arctic. This would allow a more accurate and wider ranging assessment of pollution impact to be made, and could also potentially inform decisions about the allocation of resources towards remediation. The ability to recognise and model locally generated disturbances to vegetation will also be important in the assessment of possible alterations brought about by global climate change. It is thus desirable to integrate the kind of research that has been outlined here into suitable large international multidisciplinary programmes. One example of this type of programme in which we have been involved is the recently completed BASIS (Barents Area Impact Study). This was a regional study of the potential impacts of global climate change on the physical, biological, and socioeconomic characteristics of the area, and of the linkages between them. A successor to this project, BALANCE, is currently under consideration by the EC. SPRI has also been heavily involved in setting up a new international collaboration called the Tundra-Taiga Initiative (informally, the 'Arctic Treeline project'). All of these collaborative projects, and others, are informed by the need to understand local processes before one can properly interpret global ones.