Potential Evaporation Estimates Research Articles

Evaporation Estimation Using Artificial Neural Network

Precise estimation of potential evaporation, has a great significance in many water resources applications such as management of hydrologic, hydraulic and agricultural systems. Although there are empirical formulas available for Evaporation estimation, but their performances are not all satisfactory due to the complex nature of the evaporation process and the data availability. For this purpose, artificial neural networks (ANN) models was developed to estimate monthly potential evaporation in Pantagar, US Nagar (India) based on four instructive climatic factors. Observations of relative humidity, solar radiation, temperature, wind speed and evaporation for the past 19 years and 8 months (total 236 months) have been used to train and test the developed models. Results shown that the model was able to well learn the events they were trained to recognize. These encouraging results were supported by high values of coefficient of correlation and low mean square errors. The correlation coefficient was found 0.9236 and root mean square error was 0.9863 for testing data sets.

Estimating Potential Evaporation from Vegetated Surfaces for Water Management Impact Assessments Using Climate Model Output

AbstractRiver basin managers concerned with maintaining water supplies and mitigating flood risk in the face of climate change are taking outputs from climate models and using them in hydrological models for assessment purposes. While precipitation is the main output used, evaporation is attracting increasing attention because of its significance to the water balance of river basins. Climate models provide estimates of actual evaporation that are consistent with their simplified land surface schemes but do not naturally provide the estimates of potential evaporation (PE) commonly required as input to hydrological models. There are clear advantages in using PE estimates controlled by atmospheric forcings when using stand-alone hydrological models with integral soil-moisture accounting schemes. The atmosphere–land decoupling approximation that PE provides can prove to be of further benefit if it is possible to account for the effect of different, or changing, land cover on PE outside of the climate model. The methods explored here estimate Penman–Monteith PE from vegetated surfaces using outputs from climate models that have an embedded land surface scheme. The land surface scheme enables an examination of the dependence of canopy stomatal resistance on atmospheric composition, and the sensitivity of PE estimates to the choice of canopy resistance values under current and changing climates is demonstrated. The conclusions have practical value for climate change impact studies relating to flood, drought, and water management applications.

Use of soil data in a grid-based hydrological model to estimate spatial variation in changing flood risk across the UK

A grid-based flow routing and runoff-production model, configured to employ as input either observed or Regional Climate Model (RCM) estimates of precipitation and potential evaporation (PE), has previously been used to assess how climate change may impact river flows across the UK. The slope-based Grid-to-Grid (G2G) model adequately simulated observed river flows under current climate conditions for high relief catchments, but was less successful when applied to lower-relief and/or groundwater-dominated areas. The model has now been enhanced to employ a soil dataset to configure the probability-distributed store controlling soil-moisture and runoff generation within each grid-cell. A comparison is made of the ability of both models to simulate gauged river flows across a range of British catchments using observations of rainfall and PE as input. Superior performance from the enhanced G2G formulation incorporating the soil dataset is demonstrated. Following the model assessment, the observed precipitation and PE data used as input to both hydrological models were replaced by RCM estimates on a 25 km grid for a Current (1961–1990) and a Future (2071–2100) time-slice. Flood frequency curves derived from the flow simulations for the two time-slices are used to estimate, for the first time, maps of changes in flood magnitude for all river points on a 1 km grid across the UK. A high degree of spatial variability is seen in the estimated change in river flows, reflecting both projected climate change and the influences of landscape and climate variability. These maps also highlight large differences between the climate impact projections arising from the two models. The improved structure and performance of the soil-based G2G model adds confidence to its projections of flow changes being realistic consequences of the climate change scenario applied. A resampling method is used to identify regions where these projections may be considered robust. However, with the climate change scenario used representing only one plausible evolution of the future climate, no clear message can be drawn here about projected river flow changes.

On the use of meteorological data to assess the evaporation from a bare soil

Summary The main goal of this study was to evaluate different methods for the estimate of the evaporation from a bare soil. The starting point was the experimental data set of a field survey carried out for a period of 16 months in the area of Parco Lambro, in Milano (Italy), where meteorological and soil measurements were collected to estimate the local recharge to the phreatic aquifer. The evaporation from the nearly bare soil was assessed with two different approaches. A first, indirect approach was based on the evaluation of the actual evaporation as a fraction of the potential evaporation, which was computed with both Penman’s and Hamon’s equations. As a second approach, the Bowen ratio method, that directly estimates the actual evaporation from atmospheric data, was applied. Attention was mainly focused to the effects of the configuration of the measurement instruments (particularly the measurement heights of air temperature and humidity) on the estimate of the potential and actual evaporation and to the numerical stability of each method, in relation to the measurement uncertainty. In particular, numerical problems related to the intrinsic instability of the Bowen ratio equation were deeply investigated and an original method, that identifies the data leading to unphysical results on the basis of the error propagation, was proposed. The obtained results were discussed in order to evaluate the suitability of each approach, taking into account several aspects. Penman’s and Hamon’s methods resulted to be quite stable with respect to the measurement uncertainty and the potential evaporation estimates were not affected by the configuration of the field instruments. Conversely, the Bowen ratio estimates were affected by large uncertainties due to both the intrinsic numerical instability of the method and to the sensitivity to the heights of measurement; nevertheless, the application of a suitable criterion for the data rejection allowed for a significant reduction of uncertainties on the estimates of the actual evaporation.

Pan‐evaporation measurements and Morton‐point potential evaporation estimates in Australia: are their trends the same?

AbstractThis paper compares Australian pan‐evaporation (Epan) observations with point potential evaporation (Ep) estimates derived using the Morton method. In particular, it focuses on trends in both Ep and Epan. Ep is defined as the potential evaporation from an area which is so small that the effects of the evaporation on the overpassing air would be negligible, while Epan is evaporation measured using a standard pan. The analyses are based on monthly data from 28 sites and refer to the period 1970–2004. The results show that Ep and Epan are strongly correlated on the monthly time scale. Furthermore, the sign of the monthly trends are in agreement, on average, at 60% of the sites. Where there is agreement, positive trends outweigh negative trends. This is consistent with the fact that although trends in Epan and Ep varied from site to site and from month to month, the median trends were positive except for December. The correlation between the trends in both quantities across all sites is statistically significant (R = 0.51) which indicates that projected changes based on climate model outputs can be used to estimate changes in future potential evaporation. Copyright © 2008 Royal Meteorological Society

Calculating potential evaporation from climate model data: A source of uncertainty for hydrological climate change impacts

Evaporation is an important part of the water balance of a catchment, and estimates of potential evaporation (PE) are an important input to hydrological models. When modelling the hydrological impacts of climate change, using data from climate models, present models generally do not provide direct estimates of PE from the land surface. Estimates thus have to be calculated from other climate variables, and many existing formulae can be applied. This paper compares the well-established, more physically based but data-intensive, Penman–Monteith PE against a simple, temperature-based (T-based) PE, when calculated from readily available monthly climate model data over Britain (for five global and eight regional climate models). The performance of the two PE formulations is compared to MORECS PE (a gridded dataset derived by the UK Met Office from weather observations, using a modified Penman–Monteith formulation) for the baseline period 1961–1990, and the changes in the two PE estimates between the 1970s and the 2080s are compared. The results show that the T-based PE matches MORECS PE better than does Penman–Monteith PE, for all the climate models studied. However, the changes in the two types of PE between the 1970s and 2080s are different, for each of the climate models, and these affect the modelled hydrological impacts. This is illustrated using three example catchments spread across Britain. The uncertainty introduced by the PE formulation is less than that due to the climate model, but could still be important for some applications.

The impact of broadleaved woodland on water resources in lowland UK: II. Evaporation estimates from sensible heat flux measurements over beech woodland and grass on chalk sites in Hampshire

Abstract. The impact on recharge to the Chalk aquifer of substitution of broadleaved woodland for pasture is a matter of concern in the UK. Hence, measurements of energy balance components were made above beech woodland and above pasture, both growing on shallow soils over chalk in Hampshire. Latent heat flux (evaporation) was calculated as the residual from these measurements of energy balances in which sensible heat flux was measured with an eddy correlation instrument that determined fast response vertical wind speeds and associated temperature changes. Assessment of wind turbulence statistics confirmed that the eddy correlation device performed satisfactorily in both wet and dry conditions. There was excellent agreement between forest transpiration measurements made by eddy correlation and stand level tree transpiration measured with sap flow devices. Over the period of the measurements, from March 1999 to late summer 2000, changes in soil water content were small and grassland evaporation and transpiration estimated from energy balance-eddy flux measurements were in excellent agreement with Penman estimates of potential evaporation. Over the 18-month measurement period, the cumulative difference between broadleaved woodland and grassland was small but evaporation from the grassland was 3% higher than that from the woodland. In the springs of 1999 and 2000, evaporation from the grassland was greater than that from the woodland. However, following leaf emergence in the woodland, the difference in cumulative evaporation diminished until the following spring.

Estimation of Potential Evaporation: Artificial Neural Networks Versus Conventional Methods

Accurate estimation of potential evaporation, especially in arid regions such as Saudi Arabia, has been of a great concern to many researchers. Its importance is obvious in many water resources applications such as management of hydrologic, hydraulic and agricultural systems. For this purpose, four three-layer backpropagation neural networks were developed to forecast monthly potential evaporation in Riyadh, Saudi Arabia, based on four explanatory climatic factors. Observations of relative humidity, solar radiation, temperature, wind speed and evaporation for the past 22 years have been used to train and test the developed networks. Results revealed that the networks were able to well learn the events they were trained to recognize. Moreover, they were capable of effectively generalizing their training by predicting evaporation for sets of unseen cases. These encouraging results were supported by high values of coefficient of correlation and low mean square errors reaching 0.98 and 0.00015 respectively. The study has also evolved a comparison with traditional methods and has proven that the developed neural networks were superior.

Real-Time Variational Assimilation of Hydrologic and Hydrometeorological Data into Operational Hydrologic Forecasting

Variational assimilation (VAR) of hydrologic and hydrometeorological data into operational hydrologic forecasting is explored. The data assimilated are the hourly real-time observations of streamflow and precipitation, and climatological estimates of potential evaporation (PE). The hydrologic system considered is a single headwater basin for which soil moisture accounting and routing are carried out in a lumped fashion via the Sacramento model (SAC) and the unit hydrograph (UH), respectively. The control variables in the VAR formulation are the fast-varying SAC soil moisture states at the beginning of the assimilation window and the multiplicative adjustment factors to the estimates of mean areal precipitation (MAP) and mean areal potential evaporation (MAPE) for each hour in the assimilation window. In a separate application of VAR as a parameter estimation tool, the estimation of empirical UH is also explored by treating its ordinates as the control variables. To evaluate the assimilation procedure thus developed, streamflow was forecast with and without the aid of VAR for three basins in the southern plains under the assumption of perfectly forecast future mean areal precipitation (FMAP). The streamflow forecasts were then compared with each other and with those based on persistence and the state space-based state-updating procedure, the state-space Sacramento model (SS-SAC). The results indicate that the VAR procedure significantly improves the accuracy of the basic forecast at short lead times and compares favorably with SS-SAC.

Evaporation characteristics of wetlands: experience from a wetgrassland and a reedbed using eddy correlation measurements

Abstract. Measurements of evaporation were made from July to November 1999 using the eddy correlation method on two wetland types – wet grassland and reedbeds – in south west England. The evaporative water use of a reed bed exceeded that of the grassland wetland by 15% (or 50 mm over the 5 months). The evaporation rates at both sites exceed of the Penman Potential Evaporation estimates calculated for this area. The difference between sites results from the higher roughness length of the reed bed and the lower effective surface resistance of the reed/open water assemblage. At the grassland site, a significant relationship between the surface resistance and water table level has been demonstrated. The water table at this site is managed to maintain the plant diversity and allow some agricultural access. This regime specifies a water table below the surface during the summer period, which results in higher surface resistances and lower evaporation. The results have important implications for local water resources management, especially where wetlands are maintained by pumping from rivers or groundwater. Keywords: wetlands, evaporation, eddy correlation, wet grassland, reedbed, vapour pressure, water table level

A general model for estimating actual evaporation from non-saturated surfaces

Based on energy balance equation and mass transfer equation, a general model to estimate actual evaporation from non-saturated surfaces was derived. Making use of two concepts, “relative evaporation” and “relative drying power”, a relationship was established to account for the departure from saturated conditions. Using this model, the actual evaporation (evapotranspiration) can be calculated without the need of potential evaporation estimation. Furthermore, the model requires only a few meteorological parameters that are readily and routinely obtainable at standard weather stations. Based on nearly 30 years data of 432 meteorological stations and 512 hydrological stations in China, in combined with GIS, nine typical river basins were selected. Using the data of the selected river basins, the model was tested. The results show that the actual evaporation rate can be estimated with an error of less than 10% in most areas of China, except few years in the Yellow River Basin.

Measured and estimated evaporation and soil moisture deficit for growers and the water industry

AbstractMeasurements and estimates of evaporation have been routinely carried out for over a century. In the United Kingdom the most well‐known method of estimating evaporation and its derivatives such as soil moisture deficit (SMD) is the Meteorological Office Rainfall and Evaporation Calculation System (MORECS) which is based on the Penman‐Monteith equation. Direct measurements of evaporation have become less common in recent years. A comparison of measurements and estimates of evaporation and SMD at a single site and over two catchment areas is presented in this paper. Measurements of actual evaporation from a grassed lysimeter have been achieved for the period 1995–98 and they agree well with catchment losses over the same time period in two natural catchments, the Upper Brue and Upper Cary. In contrast to this, MORECS estimates of actual evaporation are different by an average of +20% for the Upper Brue catchment and +27% for the Upper Cary catchment. The largest difference is nearly +40% for the Upper Cary catchment. Similarly, there are substantial differences in the estimates of SMD, often up to 50 mm. When local rainfall is used together with MORECS estimates of potential evaporation the discrepancy is reduced but significant differences remain during much of the growing season, although improvements do occur. The results from this paper will be of great importance to growers, water resource planners, engineers who give flood warnings, and to river flow modellers. Copyright © 2002 Royal Meteorological Society.

Parameterising the impact of shelter on crop microclimates and evaporation fluxes

This paper presents the modelling framework and parameterisations adopted to enable the effects of wind shelter on crop microclimates and growth to be simulated. Two modelling components were required: the first, SCAM (Soil Canopy Atmosphere Model), is a biophysical land surface model used to simulate the effects of shelter on the plant canopy and soil energy and water exchanges. The second is the Agricultural Production Systems Simulator (APSIM) configured for crop modelling (Carberry et al. 2002). This paper describes the link between the two and the parameterisations developed to simulate the effects of shelter and provide potential evaporation estimates for input to APSIM. A series of sensitivity analyses using SCAM demonstrate, as expected, that over the course of the growing season and during meteorological conditions typical of agricultural regions in south-eastern Australia, plant evaporation is relatively insensitive to wind speed. This is consistent with theory and illustrates that reducing wind speeds will only reduce evaporation fluxes when the (unsheltered) canopy evaporation rate exceeds the equilibrium evaporation rate. In other words, reductions in water loss as a result of shelter will arise from one or a combination of the following: (i) reduced soil evaporation following rainfall; (ii) reduced evaporation of rainfall intercepted by the plant canopy; and (iii) reduced plant transpiration during wet periods and/or periods of very high demand when transpiration exceeds the equilibrium evaporation rate. Increased storage of soil moisture as a result of shelter early in the season may sustain crop transpiration later in the growing season. While this may not modify seasonal water use, increased soil water availability for the growing crop may confer an advantage in terms of final yield. The paper then describes the development and validation of a bulk scalar transfer coefficient, or resistance, based on wind tunnel experiments investigating scalar (heat and water vapour) transport in windbreak flows. This coefficient is used in SCAM to parameterise the effect of shelter on the exchanges of heat and water between the atmosphere and the soil and/or plant canopy. A simple parameterisation for this coefficient is developed and implemented. The potential effects of shelter on evaporative demand and hence plant growth are then quantified by using the predictions of plant potential evaporation, calculated from a modified version of SCAM as input to APSIM. The simulations show that events where reduced evaporative demand benefits plant growth can occur intermittently in the latter part of the growing season (i.e. post-anthesis for cereals) for some climatic regimes, as illustrated by the simulations for the Roseworthy, Esperance and Atherton sites used in the National Windbreaks Program.

Rainfall interception in laurel forest in the Canary Islands

Throughfall, stemflow, incident rainfall and meteorological variables necessary for estimation of potential evaporation were monitored over the period December 1994 to December 1995 in a laurel forest plot on Tenerife Canary Islands). Observed interception losses were compared with those predicted using the Rutter model and the Gash model. All approaches predicted interception losses reasonably, the best prediction was obtained with the Gash model (only 2% over-estimation of the measured total for the study period), with the method of estimation of r a having little effect. Application of the Gash model with 30 years’ rainfall data (1967–1997) from the study area indicated a mean interception loss of 30% of rainfall, with a maximum of 41% in 1995. The marked differences between our results and those of previous studies are attributable to both canopy characteristics (notably the high canopy storage capacity, 2.45 mm) and climatic characteristics (notably the low mean evaporation rate during post-saturation rainfall, 0.028 mm h −1, due to the presence of orographic cloud during rainfall).

Episodic sulphate export from wetlands in acidified headwater catchments: Prediction at the landscape scale

Sulphate (SO4-2) concentrations in 34 intensively measured Canadian Shield streams near the Dorset Research Centre, central Ontario, were used to test a hydrogeologic model that uses simple measures of wetland area and till depth to identify catchments that produce SO4-2 pulses. Mean annual measured maximum SO4-2 concentrations were significantly greater in shallow till ( 1 m depth) containing wetlands or catchments with no wetlands. Average maximum SO4-2 concentrations in wetland catchments during years with dry summers were >20 mg/L in 19 of 20 catchments with average till depths of 1 m. Peaks in mean annual maximum SO4-2 concentrations from wetland catchments with shallow till occurred during summers with rain fall 150–200 mm less than potential evaporation estimates. There were no significant differences in mean average annual SO4-2 concentration among the different catchments during wet summers, with SO4-2 concentrations ranging from 6 to 13 mg/L. These observations suggest that a large portion of the temporal and spatial variation in SO4-2 chemistry and export can be predicted in headwater catchments of the Canadian Shield and perhaps in other landscapes where till depth influences upland-wetland hydrologic connections.

Potential evaporation functions compared on US watersheds: Possible implications for global-scale water balance and terrestrial ecosystem modeling

Estimates of potential evaporation E p are commonly employed in terrestrial water balance and net primary productivity models. This study compared a set of 11 E p methods in a global-scale water balance model (WBM) applied to 3265 0.5° (lat. × long.) grid cells representing the conterminous US. The E p methods ranged from simple temperature-driven equations to physically-based combination approaches and include reference surface E pr and surface cover-dependent E ps algorithms. Cover-dependent parameters were assigned a priori based on grid cell vegetation. The WBM applies mean monthly climatic drivers and other biophysical inputs to compute water budgets on individual grid cells using a quasi-daily time step. For each E p method water budgets were computed and compared against mean monthly and annual streamflow from 679 gauged watersheds, assumed to be representative of the grid cells in which they reside. Procedures were developed for excluding watersheds for which this assumption was questionable, and 330 of the original 1009 watersheds were removed from further analysis. Among E pr methods, the range of mean bias relative to observed runoff, and thus simulated actual evapotranspiration E s, varied from approximately −100 to +100 mm yr −1; E ps methods had a substantially smaller range, from about −50 to +50 mm yr −1. These results agree well with previous E p intercomparison studies at the point scale. Some individual methods from both the E pr and E ps groups yielded relatively small overall bias when compared with observed discharge data, suggesting the utility of simple as well as physically-based evaporation functions in continental- and global-scale applications. For any individual method, the spatial distribution of E s across the US was significantly altered relative to that of E p due to moisture-induced limits on soil drying. These limitations were most pronounced in hot, dry areas, where differences among E p methods in excess of 700 mm yr −1 were reduced to differences of less than 200 mm yr −1 in E s and runoff. There was a correspondingly higher sensitivity of E s to the choice of E p in more humid regions. These findings suggest that predictions made by macro-scale hydrology models like the WBM can be sensitive to the specific E p method applied and that this sensitivity results in bias relative to measured components of the terrestrial water cycle. The adoption of particular E p functions within such models should be conditioned upon the comparison of water budget calculations to suitable records of observed discharge.

Sources of solar radiation estimates and their effect on daily potential evaporation for use in streamflow modeling

The accurate estimation of potential evaporation (PE), to in turn calculate evapotranspiration, is an important step in many hydrologic models. The National Weather Service (NWS) has used PE to obtain daily estimates of mean evapotranspiration in continuous rainfall-runoff models for river forecasting. The daily PE estimates are derived mainly from meteorological data gathered on a regular basis throughout the country. Solar radiation is one of the required input variables. Because of its widespread availability, sky cover is now used almost exclusively by NWS to estimate solar radiation. Over a period of time, a bias has developed between the long-term mean PE (computed using a combination of historical observed pan evaporation data and meteorological data) and PE estimated operationally using real-time meteorological data. This difference is a result of the use of sky cover based solar radiation estimates. These biased solar radiation estimates translate into long-term means of PE which are significantly lower than values using corresponding direct measurements of solar radiation or estimates of solar radiation using percent sunshine. A standard for PE has been established and verified to which long-term means can be compared. PE estimates derived from sky cover can be corrected to the standard using a ratio of long-term means. Many meteorological variables which have been measured or observed manually in the past are being converted to automatic observations. With the advent of automated sensors, which do not duplicate the manual sky cover observations, another source of solar radiation is necessary to model PE for use in river forecasting. Satellite estimates of solar radiation are compared with other means of measuring and estimating solar radiation. Available on a nationwide basis, satellite estimates produce values of solar radiation comparable to those obtained by direct measurement. Based on availability and accuracy, satellite estimates of solar radiation are recommended for use in the estimation of PE instead of estimates derived from sky cover or hours of sunshine (percent sunshine). This assumes that no direct observations of solar radiation are available.

Water resources in the desertification-threatened Messara Valley of Crete: estimation of potential lake evaporation

Solar and terrestrial radiation flux models are used with long-term monthly-averaged climatological data to estimate the seasonal variation in downward surface radiation fluxes in central Crete. Assuming a water surface, Fresnel reflection is used to compute the albedo of the surface in order to estimate the net heating and hence potential lake evaporation. The long-term monthly variation in the surface radiation budget is then compared with measurements by recently installed automatic weather stations, one on the north coast and another in the Messara Valley in the south. As expected, over monthly timescales the surface radiation budget does not differ appreciably between the two locations and good agreement is found between the predictions of the models and the measured fluxes. An estimate of the monthly variation of potential evaporation is obtained and it is shown that a water surface in the Valley undergoes relatively large variations in potential evaporation, from 2 mm per day during its typically Mediterranean wet winter to 7 mm per day during its dry summer. A simple first estimate of the Valley's actual evapotranspiration gives about 2 mm per day average during the wet winter months and about 2–3 mm per day during the dry summer months for drip-irrigated olive groves, where water supply is limited by the pumping of the available groundwater.

Calibration of a modified SFB model for twenty-five Australian catchments using simulated annealing

Simulated annealing is used to calibrate a conceptual rainfall-runoff model (a modification of Boughton's SFB model) by fitting modelled monthly runoff to historical data. The problem to be solved is that of finding the global minimum of a multivariate function that has many extraneous local minima, a situation in which conventional optimisation methods are ineffective. The model estimates runoff from daily rainfall and daily potential evaporation estimates for a catchment. The 25 catchments used for the study cover a wide range of climatic types, including humid tropical, humid subtropical, arid, semiarid, Mediterranean and marine west-coast. The catchments differ in the number and type of model processes activated, which is indicated by the parameters fitted. A parsimonious approach was adopted where only the sensitive model parameters for a catchment were fitted. The objective function which quantifies discrepancies between the computed and observed streamflows was carefully selected to satisfy the least squares assumptions. Finally, a case study is presented to illustrate the fitting process and highlight some of the difficulties that may be encountered. It is shown that simulated annealing is able to fit six or seven sensitive parameters of the modified SFB model by locating the global minimum of the objective function.

The Plynlimon water balance 1969-1995: the impact of forest and moorland vegetation on evaporation and streamflow in upland catchments

Abstract. The Plynlimon experiment in mid-Wales, designed to determine the extent to which coniferous plantation increases evaporation losses and reduces streamflow relative to upland grassland, has now been yielding data since 1969 from the grassland Wye and the 67% forested Severn catchments. Water balance analyses of the early data indicated significantly higher evaporation rates from the forested catchment and studies of the hydrological processes involved attributed this to the high loss rates of precipitation intercepted by the forest canopy. Models based on these process studies predicted losses from the forested catchment that were similar but marginally higher than those determined by the catchment water balance. As the data sets from the catchments increased in length and a detailed reassessment of the ratings of the streamflow gauging structures was completed the updated water balances continued to show a significantly greater evaporation loss from the forested catchment, but the gap between the forest water balance and the model predictions widened. Furthermore Hudson and Gilman (1993), using the best data sets then available, identified downward trends in the evaporation from both catchments which the models did not reproduce and for which no obvious physical or physiological explanation was forthcoming. This dictated a major reassessment of the longer data sets, using the more powerful data processing techniques now available, to identify and eliminate any errors and inconsistencies. This paper describes the reassessment of the precipitation data and the estimates of potential evaporation and presents the water balance results emerging from the revised data sets. The revised results indicate that the evaporation losses from the grassland Wye catchment remained broadly similar to the potential evaporation estimates throughout the 1969-1995 period. The losses from the forested area of the Severn catchment declined from a level some 61% above that of the grassland in 1972 to a level only 18% higher before the start of felling in 1985. This downward trend continued as the felling and re-planting progressed. Over the period since 1990 the forest catchment losses appear to have stabilised at some 5-10% below those of the grassland catchment. Using the revised precipitation and potential evaporation data, process based models over-predict the forest catchment evaporation throughout the period and do not mirror the re-felling decline. Possible reasons for this apparent decline in evaporation rates are discussed.