It may seem surprising to the layman that the academic disciplines of hydrology and meteorology are not inextricably intermingled, but a lesson from a recent joint RMS and BHS discussion meeting is that they are still disconcertingly far apart. On 17^th March, a joint RMS and BHS discussion meeting entitled: Water-Land-Atmosphere interactions was held at the Zoological Society of London Meeting Rooms. The first talk, by Dr. Nick Chappell, served as an introduction to how new research in meteorology can help further the research in hydrology and vice versa, setting the scene for the more specialist talks that followed, which were given by: Dr. David Grimes, Prof. Nigel Arnell, Dr. Peter Cox, Dr. Eleanor Blyth and Dr. Mike Bonell. What follows is a summary of each of these talks. The actual presentations can be accessed in full from the following web site: http://www.es.lancs.ac.uk/BHS_RMetS/.

    The meeting was opened and chaired by Chris Collier, Vice-President of the Royal Meteorological Society (University of Salford) and then introduced by Nick Chappell (University of Lancaster) with a talk entitled :-

An introduction to coupling meteorology and hydrology. Nick identified four areas where new meteorological research is assisting hydrology. First, there is value in improving understanding of meteorological processes, notably rainfall. Clearly, the spatial and temporal distributions of rainfall are the key control for river discharges, so that flood prediction will be improved by more accurate estimates of rainfall. Secondly, it is important to have the latest understanding of climate dynamics (e.g., ENSO), as these changes can mask or magnify the effects of land-use changes on the hydrological system. Thirdly, there has been a vast improvement in the availability of meteorological data for the hydrologist to utilise. The quality of both satellite and radar data have improved, whilst using merged precipitation data sets, e.g., the Global Precipitation Climatology Project (GPCP) data have improved the description of the global precipitation field. This new data has provided a useful tool for the evaluation of GCMs and for continental scale hydrological analysis (e.g., the Global Energy and Water Cycle Experiment - GEWEX). Finally, the frequent use of GCMs is forcing hydrologists to think about the scaling issues involved with describing processes in a lumped manner, at the scale of 1o or more, rather than at the scale of an experimental catchment.

    The hydrological research community is beginning to supply to meteorologists:

(i) a new understanding of large-scale hydrological phenomena and
(ii) new large-scale data, notably global riverflow, though issues remain with regard to large-scale evaporation and subsurface water data.

i) the quality of the GCM derived water-fluxes within the tropics is far from ideal, at the regional scale,
ii) there is a lack of agreement between rainfall databases,
iii) Land Surface Schemes are over-simplified, and consequently, incorrectly generate large quantities of overland flow,
iv)there are currently no large-scale evaporation data sets, and
v) scaling issues need further research

    David Grimes (University of Reading) gave the first specialist talk entitled Application of satellite-based rainfall estimates to river flow forecasting in Africa. This presentation discussed the use of satellite rainfall data to drive a river-flow forecasting model for the Bakoye catchment (Mali), where the rainfall is seasonally variable and dominated by the ITCZ, giving intense and often-localised convective rainfall. There were only ever 10 raingauges available within the 100, 000 km2 Bakoye catchment, therefore, localised events were frequently missed by the gauge network.
    The alternative to raingauges is satellite derived rainfall data to drive the hydrological model. This paper investigated driving the model with: (i) Infrared (IR) satellite data, (ii) IR satellite data and some extra meteorological information supplied by the ECMWF reanalysis 'data set'. Their method uses a typical IR algorithm, which relates cloud-top temperature to rainfall. This introduces errors, but they are smaller for convective than stratiform rainfall. The second method adds extra information on the storm type and the phase of the African Easterly Wave to the algorithm (though factors such as vertical wind speed, near-surface relative humidity were found to have little relationship with rainfall). A phase number was assigned to each position on the Easterly Wave (e.g., 1 for ahead of the ridge), each of which has a unique rainfall index, with higher values used to indicates higher rainfall totals. Similarly, a storm index was added for different storm types (e.g., an isolated storm or a mature squall line). When the satellite data was supplemented by the ECMWF information, the derived rainfall data then showed a significant improvement over the standard satellite based rainfall.
    The rainfall derived from the two satellite methods and from the raingauge network was used to drive the 'Pitman hydrological model'. When the model was driven by the enhanced satellite data the fit to the observed discharges was better than when the model was forced with either the raingauge data or the satellite data alone.

    Nigel Arnell (University of Southampton) presentation addressed Hydrological change from climate model simulations. Nigel stated that the predicted changes in the climate are highly uncertain, hence, their impacts on the hydrological system are even more uncertain. To assess how the hydrological cycle will respond to a changed climate regime, 'climate change scenarios' are generated in three main ways and then form the inputs to a hydrological model by:

i) Taking the simulated run-off from a climate model.
ii) Using the climatic information generated by the climate model
iii) Perturbing a measured rainfall time-series.

    Peter Cox (Hadley Centre) began with a presentation on GCM land-surface schemes: Limitations and developments. This presentation was used to outline the state of the Hadley Centre Land Surface Scheme (LSS) as it was in 1997, how it has been improved and future developments. In GCMs, a LSS is used to generate the lower boundary conditions for the atmosphere, with the principal tasks being:

i)To partition both the incoming rainfall from the atmospheric model into an evaporation and a runoff component and the outgoing energy flux into a sensible and a latent heat flux, and
ii)Update the state variables that influence the energy and water fluxes, e.g. soil temperature, soil moisture content and canopy water content.

    Eleanor Blyth (CEH Wallingford) presented a talk entitled Land-surface evaporation: How observations inform the models. Eleanor demonstrated that there has been a large increase in the sophistication of evaporation data collection and the modelling of the evaporative process. Both fields are moving at great pace but there is a need for greater collaboration between workers within the two fields. Currently, there is no global evaporation database, mainly due to the difficulty of obtaining accurate estimates from remote locations, but the aim is to produce one at the 1 km scale by generating a typical parameter set for each broad land-cover type.
    Evaporation is often modelled using a SVAT model. When the SVAT models are run off-line, they are much more sensitive to parameter changes than when they are fully integrated within a climate model because the boundary layer feedback mechanism damps the response.
    Eleanor stated that evaporation is highly heterogeneous even at the smallest of scale, so that the community needs to decide whether a large-scale estimate of evaporation is:

i) A lumped measurement taken above the vegetation canopy / land-surface or
ii)The spatial heterogeneity is accounted for by generating an evaporation estimate for each land cover type and then summing the totals in proportion to the area occupied by that land cover.

    Mike Bonell (UNESCO) presentation on Rainfall-runoff processes in the tropics: latest findings and links with synoptic climatology completed the collection of presentations. His presentation focused on a much smaller scale than the preceding presentations - the hillslope, the scale at which the hydrological mechanisms within GCM Land Surface Schemes were originally identified. It also focused on the tropics where climate changes have significant impacts on the global climate. Tropical rainfall-runoff processes were discussed with reference to a series of example studies. In the tropics, Mike stated that there are many more storm runoff pathways than in temperate latitudes because tropical soils are often deeper than temperate soils. The governing parameter controlling these pathways is the saturated hydraulic conductivity (K_sat), which is measured at the lab scale, however, it is highly heterogeneous, particularly in the vertical direction. At present, the measurements of Ksat provided by field hydrologists are at an inappropriate scale for hydrological modellers, let alone climate modellers.
    The maximum intensity of a tropical storm is typically an order of magnitude higher than one in the mid-latitudes. During intensive storms, particularly in cyclonic rainfall regions, surface runoff can occur, however, there still exists a large water flux into the deep subsurface, something that is not present in many models.
    To conclude Mike stated that very local variations in rainfall pattern and intensity are critical for catchment modelling, and hydrologists would greatly benefit from meteorological information on this topic.

    To summarise the meeting, the presentations have shown that the increased collaboration between hydrologists and meteorologists has advanced research considerably, however there are still many issues to be resolved. Meetings such as this are vital in promoting further collaboration and demonstrating to the two communities the ways in which they can assist each other to pursue their research within a larger integrated land-atmospheric interactions community.

    Martin Fowell, University of Lancaster