Palaeohydrology and future climate change

Abstract

Since the mid-1980s there has been growing concern that increasing concentrations of so-called greenhouse gases will lead to global warming, changes to regional climates, and hence impacts on the environment, society and economy. The United Nations Environment Program and the World Meteorological Organization in 1987 set up the Intergovernmental Panel on Climate Change (IPCC), which has reported on climate change predictions (IPCC 1990a, 1992a), the possible impacts of climate change (IPCC 1990b, 1992b) and strategies for mitigating the effects of climate change (IPCC 1991). The IPCC Second Assessment is scheduled for publication in 1996. Climate change science is essentially predictive: it is trying to predict conditions during the twenty-first century. By far the most common approach is based around the use of global climate models (GCMs). These numerical simulation models are used to predict the global and regional climatic effects of changing greenhouse gas concentrations, and many climate change impact studies use scenarios based in some way on GCM simulations. These scenarios are then used to perturb current climatic time series, and fed through a catchment hydrological model to simulate river flows and other hydrological properties (e.g. Bultot et al, 1988; Lettenmaier & Gan 1990; Arnell & Reynard 1993, and many others). The two major problems with this approach lie in the definition of credible catchment-scale scenarios from GCM simulations, and the development of realistic hydrological simulation models. The latter problem is fundamental to hydrological simulation, whilst the former is peculiar to climate change impact assessments and arises for two reasons. First, GCMs do not at present represent all the climatic processes in a realistic manner, particularly those relating to the development of clouds and the interactions between the atmosphere and the land surface, and second GCMs operate at a very coarse spatial resolution. Some important atmospheric processes, such as the development of meso-scale circulation patterns and convective storms, are therefore not simulated particularly well, and whilst GCMs can simulate large-scale atmospheric features well, regional and local climates are often not well reproduced. The coarse spatial resolution also means that GCM output has to be interpolated down to the catchment scale. A variety of techniques of varying degrees of sophistication have been used or proposed, ranging from simple statistical interpolation through empirical relationships between large-scale and local climate to the use of nested regional climate simulation models, but all rely ultimately on the reliability of the GCM simulations of large-scale climatic features (Arnell 1995a). The other popular approach to climate impact assessment uses the past as an analogue for the future. One variant uses the instrumental period, in practice the last century, another uses historical data, and a third uses palaeoclimatic reconstructions.