Climate and Water - A
The objective of the Second International Conference on Climate and
Water was to review developments in the study of the impact of climate variation and
change on hydrology and water resources since the Conference on Climate and Water in
Helsinki, Finland from 11 to 15 September 1989. Research on climate change and its impacts
has intensified since 1989. The purpose of the present report is to summarise the main
conclusions and recommendations of the Second Conference as an input to the current debate
at national and international levels on the implications for science and for society of
climate variability and change and on the options for an appropriate response based on
sound hydrological principles.
For convenience, the main recommendations arising from the papers presented and the discussions held at the Conference are grouped in this executive summary into four categories appropriate to the particular interests of the reader:
The considerations underlining these recommendations are described briefly in the body of this executive summary and in more detail in the published proceedings, in which they are grouped according to the Conference topics under which they arise. It will be noted that the theme of the need for improved communication between diverse groups recurs in the above-mentioned recommendations. This is a true reflection of the emphasis on this point throughout the Conference.
The analysis of the present situation and the background to the above recommendations as presented at the Conference is summarised in the body of this report under three main headings of:
The papers presented to the Second Conference were edited by Risto Lemmel and Nea Helenius and pre-published in three volumes covering 1676 pages. They were grouped under three main topics:
These main topics were further divided into nine sub-topics. The following discussion covers all of these sub-topics but is grouped according to three general areas of responsibility:
acquisition and trend analysis
A critical situation exists in relation to data for all stakeholders concerned with water and water resources. Good sets of reliable data are essential for extending our understanding of hydrologic processes, for the detection of significant trends in the relevant variables, for the verification of a wide number of models involving water behaviour or use, for the evaluation of the impacts of both variation and change in the hydrologic variables on ecosystems and on socio-economic systems, and for the clear formulation of the alternative scenarios to be considered in key policy decisions. It is not surprising therefore that the question of data sets and their interpretation arose in a number of presentations and discussions at the Conference.
Several authors and speakers expressed their concern about the dramatic reduction in monitoring services in many countries where basic climatic and hydrological data programmes no longer receive the same financial support as previously. The comment was also made that regional and continental scale studies are often hampered by failure to process, retain and exchange basic data. Inter-governmental and specialised international agencies should do all in their power to remedy this situation. It was stressed in many of the papers and discussions at this Conference that historical records are invaluable in providing a basis for the evaluation of natural variability and the consequences of climate change. This applies to both long-term data sets and to shorter records covering anomalous periods.
It was clearly stated that the decline in ground observations cannot be compensated for by the greater availability of remote-sensing techniques. The accuracy of weather satellites is not high enough for hydrological purposes while the higher resolution of the more expensive earth resources satellites is offset by their longer return periods. Reliable conversion of radiation data at the land surface to hydrologic information is still difficult in many instances. However, satellite data have proved useful in the disaggregation of precipitation predicted by global climate models (GCMs) in order to couple the atmospheric models with high resolution hydrologic models and also in the estimation of heat fluxes at land surfaces for use in evaporation studies.
In the period 1990-1996, there were five flood disasters in different parts of the world, during which at least a thousand people lost their lives: Bangladesh, in April 1991 (140,000 killed); China, in July 1991 (3,074 killed); China, in June-August 1996 (2,700 killed); Pakistan, in October 1992 (1,500 killed); China, in May-June 1994 (1,410 killed). In the same years, there were 21 floods with materials losses exceeding one billion USD. The five biggest losses were as follows: China, in June-August 1996 (26.5 billion); USA, in June-August 1993 (16 billion); North Korea, in July-August 1995 (15 billion); Italy, in November 1994 (12.5 billion); China, in July 1991 (7.5 billion). (WMO Bulletin 1998 No 2). During the Conference week, devastating floods in China were climbing towards the top of both of these statistics (at least 3,000 victims, 20 billion USD damage). Smaller headlines were telling us that a critical water shortage prevailed in Jordan, Lebanon, Cyprus and other countries in the Eastern Mediterranean region.
Extreme hydrological events have always been with us. In the Conference, both historical events and recent extremes were analysed. The effect of the coincidence of several factors in producing extreme floods was described for a number of cases. Flooding is often the consequence of critical and unfortunate timing of several flood-generating factors such as delayed snowmelt in combination with thick snow cover and early summer rains, or heavy rainfall in combination with high antecedent moisture conditions.
About 15 papers were presented dealing with the analysis of trends in a number of hydrologic variables. The areas covered included catchments in the Balkans, the Baltic countries, Canada, Colombia, Estonia, Finland, Greece, Hungary, Italy, Mexico,
Nigeria, Poland, Russia, Scotland and Spain. The hydrologic variables included precipitation (both rain and snow), evaporation, runoff and lake levels. In different localities positive or negative trends were detected for the same or similar variables thus indicating the need for large scale data networks. Many of the papers applied classical techniques to their data sets but some developed extended techniques designed to incorporate non-stationarity into statistical hydrologic analysis.
The Conference had relatively few papers related to lakes, either their hydrological or ecological aspects. This was surprising and somewhat alarming - not only because the Conference was held in a country of 188 000 lakes, but also due to the fact that lake chemistry and ecology should be studied as an integral part of a system including the upstream watercourse and land area. Two papers explored physical and biological data from an extensive lake survey in northern Finland and neighbouring parts of Norway, 98 lakes in total. The main conclusions of these papers are that, inter alia, catchment vegetation is a very important explanatory variable for the chemical state of lake water, and that different taxa of diatoms are temperature specific to the extent that they can be used as summer temperature indicators. These organisms can also be identified in lake sediments. Another paper discussed the potential of using the levels of closed lakes as climate indicators. It also acknowledged the limits of this approach, as the lake can get dry or overflow. One paper demonstrated the potential of a simple lake model to simulate temperature variations and ice cover duration in a large and morphologically complex Finnish lake.
The following recommendation is primarily addressed to hydrologists:
Large-scale experiments have been and are being carried out across a range of scales including on a local scale to measure the energy and surface water balance and carbon balance of different land use characteristics:
Lastly, the following recommendation is
Role of modelling
In theory, the modelling of the hydrologic consequences of climate change is a simple sequence:
In reality, this procedure might not lead to useful results. The uncertainties of climate scenarios and GCM-outputs are large. The ability of GCMs models to reproduce the present situation on a regional or catchment scale is low, and for the future the local predictions differ not only in quantity but sometimes also in sign. The coarse spatial resolution of GCMs makes the outputs at the catchment scale problematic.
The uncertainties of the hydrological model predictions are also large. The main assumption in such models is that the set of hydrological model parameters is the same today and in the future under different climate scenarios, which might be far from the truth. For example, at high CO2 content of air, stomata of the plants contract to take in carbon dioxide, hence transpiration decreases, implying that water use efficiency increases. This essential factor is still often neglected in the models. Instead of the mechanical running of climate change scenarios through hydrological models, more effort should be given to determining the magnitude of the uncertainty of the hydrological response to change. These efforts may yield bigger dividends than a number of single scenario impact studies conducted throughout the world, which may in the future prove to have been ill-conceived. But without doubt, we can also benefit from the routine scenario studies. We have for example learned that a marginal shift in climate can lead to dramatic alterations in the hydrological regime. This can be the case in catchments with low precipitation and low runoff coefficient, or in catchments where significant phase shifts between rain and snow can occur.
Natural climate variability and associated hydrological response may not be well represented by current GCM-hydrological modelling efforts. Use of historical records of anomalous periods may prove valuable in understanding the potential hydrological consequences of climate change and the natural variability within the climate system.
The climate and water modelling community have been drawn together by the need to better represent the energy forcing in hydrological models and runoff in GCMs. Use of the experience of hydrological models to evaluate what the climate models are doing is proving beneficial. For example, comparisons of the outputs of a daily water balance model applied to the Baltic Sea basin were compared to ten-year climate model simulations of two European GCMs. These helped pinpoint systematic or compensating errors in the land parameterization schemes in these models, particularly the need to adequately represent the seasonal variability of precipitation.
Another fundamental barrier to progress in studies of climate-water interface is the mismatch of scales used in both areas. Spatial and temporal scales used in the atmospheric studies considerably differ from those of hydrology, where the basic unit, the catchment, itself embraces already quite a considerable range of scales. In order to match these discrepant scales it is necessary to upscale description of hydrological processes and to downscale climatic variables. Despite considerably progress achieved to date, much further effort in needed to improve further the predictions and make them credible and practically useful.
The difficulty in direct use of GCM scenarios in hydrological studies due to scale mismatch may become alleviated by different techniques of downscaling. One of the promising options is nesting of higher-resolution regional climate models into a GCM. Sub-grid parameterization should capture intra-cell variability, with adequate representation of sensitive components such as lateral flow in addition to vertical water movement, and snow in the mountains. Temporal downscaling and disaggregation (capturing daily and diurnal variability of precipitation) are urgently needed for making inferences on extreme values. Regional climate models are quite good in simulation of some variables, yet extreme precipitation cannot be adequately simulated. Request to atmospheric scientists is that in addition to reducing uncertainty they should try to quantify uncertainties.
Scale studies are not yet capable of reconstruction of details of magnitude, location and timing of extreme events in re-analysis. Yet, they lead to useful regional results, greenhouse effect may cause increased winter precipitation (and flood danger), earlier snowmelt and, possibly, increased summer droughts. As higher annual precipitation total may fall in a smaller number of days and frequency of intensive floods may consequently grow.
Models should be sensitive to climate in a reasonable way, thus they cannot be oversimplified. Models used nowadays range from empirical formulae and regressions through conceptual lumped models and water balance descriptions to process-based approaches. Yet, temporal resolution of many models is insufficient for peak flow in large rivers. Are models appropriate? It seems that the representation of evapotranspiration is notoriously difficult to validate as is the validation of macro-scale models, based on runoff data.
There is a strong feedback between land surface processes and weather and climate; the influence of soil wetness is a key issue in GCMs. Soil moisture is the primary forcing factor over land in summer in temperate climates. Changes in land use produce changes in vegetation, which in turn alters the partition of energy and evapotranspiration through seasonal growth and carbon cycles. Consequent changes in soil moisture profiles and thus runoff need to be modelled together with the soil chemistry. While atmospheric models are based on SVAT models at the grid scale, lateral water flow must also be adequately modelled to link climate and water. The two-way coupling of land use processes and NWP (Numerical Weather Prediction) models requires the aggregation of the processes in NWP and the disaggregation of the meteorological forcing in the land surface and hydrological models. Acceptable aggregation rules exist for vegetation, but the aggregation of soil characteristics is more difficult. Aggregated soil characteristics at the grid scale can adequately model average evaporation flux, but fail in defining the average drainage below the root zone and in handling lateral flow.
The problem of parameterisation of evapotranspiration in hydrological modelling was identified as a key issue by the Conference. This includes direct effects of increased temperatures, but also indirect effects of land use changes, changing vegetation and feedback from increasing concentrations of CO2 in the atmosphere. Evapotranspiration can be modelled using geographical information system (GIS) data over complex land use and terrain, taking into account shadow effects if necessary. Evapotranspiration is most sensitive to relative humidity and is therefore highly dependent on the spatial rainfall distribution.
Hydrologists are in urgent need of climate simulations focusing on extreme precipitation and on other flood-generating factors. The lack of these simulations is a major factor preventing an efficient use of hydrological models in flood risk assessment in the changing climate. Methods to quantify uncertainties of all relevant climatological factors should also be developed. A particularly vulnerable issue related to extremes is the monsoon rains. They bring rainfall to an area which includes 60% of world's population and 80% of rural dwellers. The climate models have their weakest performance within this zone. The advance in the knowledge of El Nio is a positive sign but more work is needed.
In June 1998, there was a programme on Finnish television where people could ask questions related to El Nio. A farmer from a remote village wanted to know what role El Nio played in the occurrence of night-time frosts in Finland in late summer, particularly harmful events for farmers. The experts doubted any connection. Two months later a workshop on Lake Ice and Climate was arranged at the International Limnology Society (SIL) Conference. One of the findings, based on long ice formation and break-up series from all over the Northern Hemisphere, was that the lakes in Finland have clear El Nio signals; when this anomaly on the other side of the Earth is strong, Finnish lakes freeze later and have their break-up later than during other winters. (Proceedings of the SIL Conference, August 1998, Dublin, Ireland). The North Atlantic Oscillation Index (NAOI) was found to have some connections to Norwegian runoff, precipitation and temperature series.
As for El Nio/Southern Oscillation (or ENSO), the focus was in the Southern Hemisphere, although teleconnections were also discussed e.g. between the ecosystem of Lake Baikal and the Southern Oscillation. The utilisation of ENSO in the predictionforof precipitation was analysed for Argentina, Colombia, Ethiopia and India. Despite its accepted importance, the ENSO-related variables indicated only modest ability for seasonally precipitation forecasts in these countries. In semi-arid region in Argentina, ENSO produced identifiable signals only in four months of the year. As to India, the findings indicated that the effect of El Nio is more severe during periods when monsoon rainfall is below normal. An above-normal period seems to have started around 1990; thus the authors speculate that India may be more vulnerable to floods than droughts during the next decade or two. The river flows in Australia and in Southern Africa were rather clearly linked to ENSO cycles. In Australia, the serial correlation of runoff and ENSO can be used to forecast runoff several months ahead. Researchers and water authorities are beginning to assess the benefits and risks of using these forecasts.
All the components of cryosphere - glaciers, sea ice, lake ice, permafrost, snow cover - were dealt with at the Conference, although only in a handful of papers, with the exception that snow played a considerable role in many flood-related presentations. Among the extreme views put forward was the prediction that a 'catastrophic’ warming of 5 degrees centigrade would lead to the disappearance of the Novaya Zemlya ice sheet (7,300 km3) within 130 years, with a consequential sea level rise
of 2 cm. With predicted warming in northern Europe, the Baltic ice season was estimated to shorten by one month by the year 2050. The amount of snowfall and magnitude of oceanic heat flux played a major role in the ice conditions on the Okhotsk Sea. On account of higher air temperatures, the snow cover over large areas over the northern Eurasia disappears nowadays earlier than a few decades ago, in spite of increased winter precipitation. In the Swiss Alps, a considerable reduction in snow water equivalents and consequently in the profitability of skiing resorts will be expected. As to lake ice impurities, it was indicated an ecological risk in spring with a very low pH and enrichment of pollutants in the surface water layer.
The Conference did not have a special focus on groundwater issues, but they were dealt with in some papers. In Estonia, a considerable increase of groundwater formation is expected as a consequence of climate change. This would be very beneficial, because the productivity and reliability of shallow wells could be significantly improved. In the Netherlands, increasing groundwater table together with expected land subsidence would have serious consequences.
What is really happening to water during its underground paths? True physical mechanisms of runoff generation are today much better understood than two decades ago, but more experimental investigations are still needed. Besides, the increased knowledge of runoff formation is not necessarily reflected in hydrological modelling. This is a particular problem, when the results should be valid also in chemical and biological applications, but it can also lead to misinterpretations in relation to GCMs.
In the session on paleoclimatology, two papers discussed paleogroundwater. The EU-supported project PALAEAUX has focused on the origin of palaeowaters in Europe, their present distribution at a continental scale, their importance both as archives of former climatic and environmental conditions, as well as their potential in certain areas as valuable sources of good quality drinking water. This project is a good example of efficient integration of different study methods including isotope measurements, geochemical analyses, borehole geophysical logging as well as geochemical and hydraulic modelling. Preliminary results were presented of a case study within the PALAEAUX project. The groundwater in the Estonian Cambrian-Vendian aquifer contains a unique isotope record indicating its glacial origin. The whole complex of isotope and geochemical methods, including noble gas analysis has been used to interpret the formation mechanisms of these waters. Subglacial drainage of meltwater from the Scandinavian Ice Sheet to the aquifer seems here to be the most realistic explanation of paleogroundwater formation. Local human impact on the quality of water in this aquifer through intense exploitation was demonstrated. Recommendations for sustainable use of paleogroundwater as a high quality non-renewable resource were made.
The Conference noted the various findings described above and recommended to hydrologists and to research managers as follows:
Variety of impacts
The papers and posters on the evaluation of impacts covered a large band of different climates from semiarid regions in Africa up to wetland regions in northern climatic zones. Research focused mainly on the influence of climate change on water resource management and on agriculture. Some authors used combined hydrological and socio-economical and other authors ecological and agricultural model approaches. A number of them discussed their hydrological findings in view of some potential impacts in their respective countries.
It is expected that climate change will effect aquatic ecosystems. Especially in higher latitudes, changes in regional climate patterns are likely to have significant negative effects on aquatic flora and fauna. These areas have large wetlands which are very susceptible to climate fluctuations and which are habitat for a large number of birds and other species. Northward migration of permafrost would dramatically alter regional hydrology and cause massive terrain slumping. Possible water level changes would affect inland navigation, threaten fisheries and impact the ecological stability of the lakes. Improved regional climate models with sufficient spatial resolution are required that can be coupled with hydrologic models for better quantitative prediction of ecosystem effects. The rate and degree of change are unprecedented and prediction on how ecosystems will respond cannot easily be extrapolated from present understanding. Only with improved understanding can the questions be addressed whether needed or desired favourable aspects will persist without intervention and how dynamically changing aquatic ecosystems should be managed and what actions and financing are required. Another problem, which has not been taken enough into consideration, is the influence of possible water level changes on the ecosystem in the areas with shallow water table. To understand possible impacts of climate and environmental change, more quantitative data are needed.
From a few studies on the influence of possible impacts of climate changes on the aquatic system of rivers it can be seen that the water quality and the biodiversity may be affected. Studies in some areas show on one side an increase of water temperature together with increased levels of ammonium- and ammonia-nitrogen, on the other side a decrease of dissolved oxygen especially in dry and warm seasons. This will increase water pollution in rivers. In those countries in which much has been done to reduce water pollution, additional measures will have to be undertaken to keep the water quality at its present level. There are still high deficits in river water quality modelling. Not all chemical and biological processes in the aquatic zone can be modelled and it is at present not easy to couple river water quality models with terrestrial ecosystem models. Especially the modelling of the interference between the aquatic and the terrestrial phase is essentially an unsolved problem. Many efforts are necessary in this regard. To solve this problem an interdisciplinary cooperation between hydrologists, water chemists, biologists and ecologists is necessary. Interdisciplinary international programs including special workshops are badly needed.
At present there is also an obvious imbalance between different climatic zones as to the number and quality of investigations, which have been undertaken to study the impacts of climate change on the ecosystems. Relevant studies in all geographical regions are necessary. The common finding at the Conference was that in almost all investigated regions the already existing or expected problems will increase. Expected higher temperatures and less precipitation will lead to increased water demand for domestic water supply and irrigation water in regions, which are already suffering from droughts. Some authors state that the effect of climate change is an additional increase of water demand beyond the increase which is expected due to population growth and economic development. In a number of other countries, the latter two factors are considered far more important than the consequences of climate change. In some regions, runoff increases due to snowmelt and higher temperatures combined with increased precipitation and changes in snowmelt may lead to severe flooding. The same mechanism can enhance groundwater recharge, which may raise the groundwater table and shrink the aerated zone. This affects agriculture and may even lead to extension of wetland areas. In coastal zones the expected sea level rise may be a reason for an increased salt concentration in the groundwater. This affects drinking water quality and may damage agricultural production. There are also regions where no major effects on hydrology and hence on water resources are expected - compared with the natural variations of the past decades. This is especially the case in regions with present high availability of water.
All authors on these topics concluded that measures should be taken in strengthening water resources management, in improving water legislation, in enforcing economic use of water and generally increasing the sustainability of water management. Many also raised the point of uncertainty at the regional scale. Some papers described how different GCMs led to contradictory results. A further critical element is the unknown amount of water, which is 'unavailable' due to pollution, and the uncertainty in the projections of socio-economic development.
In the northern latitudes and mountainous catchments, changes in the magnitude and timing of spring runoff can lead to a need to alter operating rules and target storage levels. There could also be requirement to alter the reservoir operating guidelines to reflect a shift in the balance between flood storage and water conservation, particularly during the spring runoff period. Reductions in the low flow regime could have a negative impact on the reliability of a reservoir system for providing water supply. In the future, the planning of reservoir systems must consider that the hydrological regime may not be constant throughout the design life of a reservoir. This implies fundamental changes in the way that reservoir inflows and demands are modelled in the process of determining an appropriate reservoir capacity and also in the identification of reservoir operation policies. It is clear that there is a need for a paradigm shift in both reservoir operation and planning. The Conference also pointed out the diversity of large drainage basin characteristics and hence potential response to climate change and variability. However, it is readily apparent that there exists a general lack of scientific harmonisation across this body of work as a whole, making an objective assessment of their collective value to the policy-making community difficult at best.
The implications of climate change for water resources management were discussed in the Conference, but only few examples could be given where the possibility of future climate change impacts had already influenced management praxis. The interest is strong, but there is still hesitation about how to quantify climate change impacts and on how to assimilate the increasing uncertainty that has emanated. It was consequently concluded that it is necessary to find ways to account for global warming and its effects in water resources management praxis. Several papers explicitly considered the issues of water engineering (e.g. river diversion, irrigation, impoundment and hydropower) either as a factor influencing the water balance of large basins or from the viewpoint of vulnerability to climate change. The increasing recognition of the role of water engineering works in defining the hydrology of drainage basins will require a concerted effort to link the engineering and hydrological communities. An economic dimension to this question is also apparent and policy-makers are urged to foster the appropriate multi-disciplinary understanding of the issues.
There are many competing demands on high-level policy-makers, especially politicians with short-term agendas. Their attention is likely to be focused on public welfare, economic stability and foreign policy. They are certainly not interested in the hydrological cycle. For them, water is a resource or a source of potential disasters, as in floods and droughts. There are also likely to be barriers to the implementation of policies considered desirable by the scientific community. These may result from established institutional structures, systems of taxation and land tenure, for example. While it is reasonable to ask decision-makers to explain their policies and their basis for developing new policies, it is not reasonable to expect them to come all the way to meet the scientists in all topics. It is necessary to offer good arguments for the importance of climate change and its impact on water resources. In this, it is necessary to work with and through relevant ministries and other institutions that advise the policy-makers. There will still remain the problem of enforcing any policy that is announced. Studies on the current state of the environment can illustrate why we face a problem and the nature of that problem. Analyses based on our knowledge of natural sciences and risk assessment can lead to conclusions as to whether the problem is serious and the social and economic disciplines can offer options for ameliorating the problem. Unfortunately, the natural sciences currently offer advice and information on mean values, when the decision-maker would be more concerned with extremes; or on the more likely state, when the concern is for certain thresholds; or on an equilibrium future state, when the rate of changes is the controlling factor; or on the magnitude of a change, when the need is to handle a risk. One policy option in the face of the lack of data on uncertainty in prediction is to take no action. But this may lead to the need for major investment at a later date. It can be argued that at least some investment should be made now, such as data collection and research, so as to reduce the risk of major costs later. The problem is - do we have enough information as yet to be sure that these initial investments will be effective?
One major problem, and a long-standing one, is needed to explain uncertainty to the public and policymakers. This includes uncertainty in further climate and the inadequacy of our knowledge on that future. This leads to an appeal for government policy to support continued research, data collection and access to existing data. Often the greatest concern is to match long-term climate variability with expected population and economic growth, climate change being of much less concern. Once again this raises the need for better data and a greater understanding of climatic and hydrological extremes. The changes of these extremes will be the most critical feature of any change in climate.
The scientific community must also put its case to the local community, who are not familiar with the contents of international professional journals. This is the public affected by climate impacts and the voting public who can influence politicians and their senior advisers. One problem here is that science is not so well respected now as it was and it will be necessary to present clear explanations and convincing arguments. The best science and wisest policies are of no use if the public oppose or ignore them. It would be wise to work at the local community level to explain the current scientific understanding and seek local advice as to what studies might be undertaken to answer local concerns at local and regional level, what are most critical specific rates of change or critical levels to which the community is vulnerable. The risk of surpassing these site-specific levels is what concerns the community, not the results of global scenario studies.
The round table on Education and Training dealt with a wide range of topics. As in other sessions at the Conference, there was emphasis on the need to take account of the social sciences as well as the natural sciences (both physical and biological) and to impart knowledge based on both theory and practice. It was stressed that a key problem was to find the best system which would combine modern learning techniques with local circumstances and traditional values. The point was made that the concept of technology transfer had been damaged in the past in developing countries due to the interference of political and commercial interests. The discussion covered both general sessions in schools to promote awareness at an early stage and more specialised instruction in higher education based on best modern practices. It was suggested that UNESCO and WMO could play a role in the preparation of pilot courses for the Internet. It was emphasised that all projects in water education required: (a) careful planning based on a realistic evaluation of costs and potential benefits; (b) implementation based on local circumstances and available resources; and (c) regular updating based on feedback and rigorous assessment.