Two years of collaboration

May 2009, Wallingford - This interdependent project brings together different expert groups (hydrologists, climatologists, water use experts etc.) divided into different WorkBlocks. Overall progressed in the second year of the project has been excellent. The development of global data sets, both for driving models and their verification has been good. In particular the intercomparison project is generating considerable interest both within and outside WATCH and will provide a platform for the integrated modelling system and uncertainty analyses. Outputs have been substantial, as evidenced by the 18 WATCH technical reports and over 30 presentations of WATCH results, spanning both scientific and stakeholder forums.

The 20th Century
Work has continued on generating new data products in three main areas: Satellite-derived products; consolidated European Streamflow data; and a new half-degree forcing data set.

An innovative method was developed to estimate the global monthly distribution and area of land-surface open water from multi-satellite observations. This method used a group of complementary satellite observations. The use of satellite data with different sensitivities to the various parameters (e.g. vegetation, roughness, water presence) makes it possible to isolate the contribution of the surface-water in the signal.

Global satellite-derived mean fractional inundation at annual maximum in % of the equal area 773 km2  pixels (0.25°x0.25° at the equator) for the 1993-2000 period

Global satellite-derived mean fractional inundation at annual maximum in % of the equal area 773 km2  pixels (0.25°x0.25° at the equator) for the 1993-2000 period.

The “WATCH forcing data” for the twentieth century is being generated in two stages: a) 1958-2001 based on the three-hourly ECMWF ERA40 reanalysis data and b) 1901-1957 by downscaling CRU (Climate Research Unit) monthly gridded observations via a “weather generator” using the means and variability (i.e. spectral-) characteristics of the data obtained in the first stage.

The new data are based on one-degree ERA40 reanalysis data down-scaled to half-degree spatial resolution (i.e. 0.5o x 0.5o). All the data are provided in NetCDF format using the CRU half-degree land-sea mask, only at land points and exclude Antarctica. The initial forcing data has allowed the first land surface model and hydrological model intercomparison runs.

The UK Met Office land surface scheme (JULES) was extended to include a simple representation of groundwater, to incorporate this into the JULES land surface model, a TOPMODEL-type approach was applied. For instance, changing the topography has a large impact on both deep and surface runoff. As the mean water table rises, the region over which the water table reaches (or exceeds) the surface also increases, resulting in more surface runoff. Changes in deep runoff drive changes in surface saturation which, in turn, are responsible for smaller changes in surface runoff. Where by an increase in deep runoff will lead to an increase in total runoff (and vice-versa). Incorporating a groundwater model gives significantly different results in surface and sub-surface runoff components. The groundwater model is sensitive to landscape and the reduction of porosity with depth. Including the groundwater model produces long term mean runoff results that are closer to what is observed over many regions.

Known withdrawals and consumptive use of groundwater (GW) are important to realistically determine GW availability at a particular point in time, as well as the longer-term water balance, and the sustainability of the GW resource. A preliminary
modelling scheme uses GW withdrawal statistics plus indicator driven GW use as drivers to allocate withdrawals. These values per grid square drive the withdrawal components of the model.

Groundwater withdrawal likelihood modelling for the irrigation water use sector for 3000+ US counties
Groundwater withdrawal likelihood modelling for the irrigation water use sector for 3000+ US counties.

Partners in WATCH have also developed a new representation of crops and irrigation. The impact of irrigation on crop biomass has significant spatial variation. Irrigation causes an increase in net carbon uptake only in areas with high crop densities. On a global scale the interaction of an increased carbon sink with climate is small and depends on the fate of the crop. Irrigation is costly in terms of water consumption, but can be highly efficient (see figure below). Areas which have the potential to gain most from irrigation in terms of agricultural output require a significant input of water (>1800 kg m-2 yr-1). However, in these areas nearly 100% of water added through irrigation is used by crops. Areas with low  irrigation efficiency, have other factors that limit growth thus water added during irrigation is lost via runoff.

Irrigation efficiency (%): water lost via transpiration as a % of water added through irrigation

Irrigation efficiency (%): water lost via transpiration as a % of water added through irrigation

The results highlight the important links between food, water and climate change and shows how models can be used in the assessment of climate adaptation strategies. Climate interaction is potentially significant in areas with high irrigated crop coverage.

Capturing human impacts
This section aims to deliver datasets on population; land cover and use; and sectoral water demands. Data on spatially explicit estimates of present and past domestic water use has been compiled to validate the domestic water use model. For instance the basic approach of the domestic water use model is to first compute the domestic water intensity (m3/cap-year) and then to multiply this by the population of water users. Changes in water use intensity can be expressed by structural changes and structural change is based on the observation that, as average income increases, water consumers tend at first towards a more water-intensive lifestyle. Finally a maximum level is reached after which per capita water use is either stable or declines.

Crop water deficits have been calculated globally by estimating the water a crop uses that is in excess of what is supplied by rainfall. In addition, the potential production increase for an area which could be fully irrigated has been calculated. The next step is to compare the water deficit with water availability.

Mean annual water deficit 1901-1995 in areas with cultivation

Mean annual water deficit 1901-1995 in areas with cultivation

In order to produce a spatially explicit global database of water use in industry and energy production data was compiled and calculated on water use in the manufacturing sector. Here the manufacturing water use refers to the annual amount of water withdrawn (and consumed) in production processes. In addition water use for producing electricity for each year assesses the amount of water withdrawn (and consumed) for cooling purposes of thermal power plants in the electricity sector. Finally a variety of existing maps, data and information, notably the Global Lakes and Wetlands Database (GLWD) has been chosen to represent the best available information on global dams. This information on water use reconstructions and scenarios will be incorporated into the projects data framework.

The 21st Century
Future climate model scenarios depend crucially on their adequate representation of the hydrological cycle. Within WATCH special care is taken to couple state-of-the-art climate model output to a suite of hydrological models, this is expected to lead to a better assessment of changes in the water cycle. Methodologies were developed to adequately handle biases in climate model output and to quantify the resulting uncertainties in estimating future global water cycle components. Moreover comprehensive analyses of projected changes in the hydrological cycle over Europe have been conducted, because they are represented in existing global and regional climate model simulations. Further work includes discussions on defining scenarios for the 21st century so that regional and global changes in socio-economic conditions are reflected. This shall be used to prescribe temporal, and spatially distributed, boundary conditions for global and regional hydrological sensitivity simulations.

Analysis of 20th Century Floods and Droughts
Work on developing a river basin datasets, includes a comprehensive metadata catalogue, for the WATCH river basins (see below). These datasets are used to study processes generating drought, including the propagation from meteorological drought into hydrological drought as well as spatial and temporal patterns of drought at the river basin scale.

Location of the selected river basins for hydrological extremes study

Location of the selected river basins for hydrological extremes study

The time series were analysed to investigate droughts in precipitation, recharge and groundwater discharge. The study confirmed that climate (including snow accumulation and melt), soils and the responsiveness of the catchment (e.g. aquifer characteristics) have a major influence on drought generation, and show that natural variability in the climate system causes droughts to occur irregularly in time and space at the regional scale.

In a special study on the Pang catchment (Thames, UK) emphasis is on how a meteorological drought (deficit in rainfall) is propagated in the hydrological cycle to appear as a drought in groundwater recharge, hydraulic head and groundwater discharge. Rather large differences in the spatial and temporal characteristics of drought for the different variables were revealed. Meteorological droughts frequently cover the whole catchment and last for a short time, whereas droughts in recharge and hydraulic gradient typically cover a smaller area and last longer.

A methodology for understanding and assessing some of the main processes for generating major floods has been investigated, focusing on two main aspects: (i) the severity of rainfall preceding or during a major flood event; and (ii) the main atmospheric conditions prevailing before and during a major flood event. This is made up of three main steps: the first one aims to identify the flood events and to build a flood series. The second step establishes the preceding conditions of the flood events, either in terms of rainfall or in terms of atmospheric circulation type. The final stage consists of analysing the two series (flood and preceding conditions) in order to identify the main processes that could be the cause of large floods.

The methodologies that quantitatively describe the space-time development of drought and large-scale floods are applied at different scales, and focused on Europe. Information on the space and time scales as well as severity of major historical drought and floods also contributes to the drought and flood catalogues. Preliminary result shows some atmospheric conditions, as described by circulation and weather types, are more often associated to drought and flood events than others.

Comparative research on a regional scale is being undertaken to assess hydrological change in small basins at the sub-grid scale of climate models. The newly assembled and updated stream-flow data set for small basins across Europe now iincludes recent severe drought periods and covers the WATCH focus region "Europe".

Map of the river basins considered for the flood study

Map of the river basins considered for the flood study

Methodologies to correct biases in climate model output, and to quantify and handle resulting uncertainties in the water cycle, have been described. A methodology was developed to predict the spread in hydrological model output resulting from the use of forcing data from an ensemble of climate models. These approaches address the propagation of uncertainties in the model chain: climate-hydrology-detection of extremes.

Quantifying Feedbacks in the system
Work has progressed on methods to assess feedbacks in the water cycle and its impact on global water resources. Feedbacks associated with changes in land use have expanded in scope to include global landcover-change feedbacks and the impact of wetlands on regional rainfall generation in West Africa. Additionally climate change is likely to impact on agricultural outputs and consequently food supplies. Agricultural practices such as irrigation, may help adapting to change in areas with a lot and unreliable rainfall. However, irrigation has implications on water resources and feedbacks to climate.

In terms of water consumption irrigation is costly, but can be highly efficient. In areas with low irrigation efficiency other factors limiting growth, and the water added during irrigation is lost via runoff. Adding moisture to the soil cools the land surface, decreasing the heat flux to the atmosphere and increasing evaporation. In North West India where the coverage of irrigated crops is greater than 75%, the surface cooling effect is considerable (4°C- see figure below). Such modifications may have important interactions with the climate by affecting theformation of cloud.

             Effect of irrigation on surface temperature (°C)                    

Effect of irrigation on evapotranspiration (kg H20 m-2 yr-1)                

Effect of irrigation on surface temperature (°C) (left) and evapotranspiration (kg H20 m-2 yr-1) (right)

Irrigation is an important tool in improving crop yield, but increasing agricultural output comes at the price of increased water consumption, and potential climate interactions. The results highlight the important links between food, water and climate change and shows how models can be used in the assessment of climate adaptation strategies. Climate interaction is potentially significant in areas with high irrigated crop coverage.

Water resources vulnerability
Assessing vulnerability of water resources in the future century is undertaken as models are improved for the climate change impact and vulnerability analyses. The model intercomparison exercise will guarantee that results from the different models within the modelling framework will be comparable, providing a better indication of the impact and uncertainties of future climate change on water resources. A key issue is to develop the first global model including surface water quality indicators. In the past year continued development of the WaterGAP water quality model and the further compilation of the data for running and testing the model took place.

A number of test basins across Europe are used to translate water resources applications from the global water cycle system to basins. With in the last year, at the test basins, model and tool development occurred to study the impact and uncertainty of future climate change on water resources. All test basins have collected local data and have, until now, focused mainly on using “local” climate change scenarios. A more overarching protocol is being developed on how to use these datasets within WATCH to see how the global and regional analyses of water can be translated to local basins. For instance the first areal pictures on precipitation, evapotranspiration, soil moisture and runoff distribution were obtained for the upper Nitra basin:

              Spatial distribution of precipitation                        

  Spatial distribution of runoff                

Spatial distribution of precipitation (left) and runoff (right)