Modelling Seasonal Dynamics from Temporal Variation in Agricultural Practices in the UK Ammonia Emission Inventory

Most ammonia (NH₃) emission inventories have been calculated on an annual basis and do not take into account the seasonal variability of emissions that occur as a consequence of climate and agricultural practices that change throughout the year. When used as input to atmospheric transport models to simulate concentration fields, these models therefore fail to capture seasonal variations in ammonia concentration and dry and wet deposition. In this study, seasonal NH₃ emissions from agriculture were modelled on a monthly basis for the year 2000, by incorporating temporal aspects of farming practice. These monthly emissions were then spatially distributed using the AENEID model (Atmospheric Emissions for National Environmental Impacts Determination). The monthly model took the temporal variation in the magnitude of the ammonia emissions, as well as the fine scale (1-km) spatial variation of those temporal changes into account to provide improved outputs at 5-km resolution. The resulting NH₃ emission maps showed a strong seasonal emission pattern, with the highest emissions during springtime (March and April) and the lowest emissions during summer (May to July). This emission pattern was mainly influenced by whether cattle were outside grazing or housed and by the application of manures and fertilizers to the land. When the modelled emissions were compared with measured NH₃ concentrations, the comparison suggested that the modelled emission trend corresponds fairly well with the seasonal trend in the measurements. The remaining discrepancies point to the need to develop functional parametrisations of the interactions with climatic seasonal variation.     

An Evaluation of Daumod Model in Estimating Urban Background Concentrations

This article presents an evaluation of the performance of the urban atmospheric dispersion model (DAUMOD) in estimating nitrogen oxides (NO_x) background concentrations in Copenhagen. Estimations of hourly average (averaged over a year), mean daily and mean monthly concentrations of NO_x are compared with observed values for two years of data. The model slightly underestimates low hourly average values and overestimates high values. The cumulative frequency distribution of mean daily concentration obtained from model estimations is in good agreement with the obtained from observed data. We performed a statistical analysis to determine the agreement between estimated and observed concentration values. The results show that 95.8% of hourly average estimations, 86.8% of mean daily and 100% of monthly average concentrations are within a factor of two of the observed values. The normalised mean square error of predictions is +0.13 for hourly average estimations, +0.22 for mean daily values and +0.02 for monthly mean concentrations. The fractional bias values are: –0.049 for hourly mean estimations, –0.047 for mean daily values and –0.053 for monthly average estimations. The values of the statistical parameters allow us to consider that though estimations are lightly larger than the observed values, the model performance is acceptable.     

Freshwater Critical Loads in Wales

The Steady State Water Chemistry (SSWC) modeland the Diatom model have been used to calculate criticalloads for acidity using annual mean chemistry for 102 acidupland streams in Wales sampled as part of the Welsh AcidWaters Survey (WAWS) in 1995. Diatom critical loads werelower than SSWC values reflecting the higher effective[ANC]_limit of the Diatom model compared to the[ANC]_limit of zero used in the SSWC model. The WAWSstream sites were all located within 41 10 × 10 km squares andeach square was assigned the lowest critical load value fromamongst the sites located within it. Comparison with valuesassigned under the UK national critical loads mappingprogramme (UKCLMAP) showed that WAWS critical loads were lowerthan UKCLMAP values in approximately 40% of squares.Differences in critical load class were variable, but exceeded2 keq ha^-1 yr^-1 in up to a maximum of seven squares.It cannot be assumed, therefore, that reducing acid depositionto the currently mapped UKCLMAP critical load will protect allstreams occurring within a given 10 × 10 km grid square inWales. The limited number of sample sites means that even inthose squares where all WAWS sites will be protected, theremay be other, more acid sensitive freshwaters with lowercritical loads. This has important implications for theinterpretation and use of critical loads data for regional andlocal environmental planning.     

Long Term Trends in Concentration of Major Pollutants (SO<Subscript>2</Subscript>, CO,NO, NO<Subscript>2</Subscript>, O<Subscript>3</Subscript> and PM<Subscript>10</Subscript>) in Prague – Czech Republic (Analysis of Data Between 1992 and 2005)

To assess the effect of changes in traffic density and fuels used for heating at the beginning of the 1990s, 1992–2005 monthly averages of PM₁₀, SO₂, NO₂, NO, CO and O₃ from Prague, the Czech capital, were analyzed together with long term trends in emissions of major pollutants, fuel consumption and number of vehicles registered in Prague. The data from all monitoring stations were retrieved from the database of the state automated monitoring system. Correlation coefficients between ambient monthly averaged temperature and all pollutants of concern showed distinct seasonal trends. The results showed that while SO₂ and to some extent also CO concentrations dropped namely in the first half of the analyzed period (1992–1997) as a result decreased fossil fuel consumption for local heating, the behaviour of other pollutant concentrations followed a different pattern. PM₁₀ concentrations decreased during the beginning of the 1990s but showed a sign of increase after 2000. Concentrations of ozone and NO₂ did not reveal any significant change throughout the whole studied period. It can be concluded that during the studied period traditional urban sources of pollution, such as coal and oil combustion, lost their importance but were simultaneously substituted by pollutants from automotive transport (namely PM and NO₂) making the problem of air quality even worse.     

New Numerical Model Study on a Tidal Flat System – Seasonal,Daily and Tidal Variations

We developed a new numerical model to investigate the dynamics of tidal flat ecosystems and their role in water quality in terms of the carbon cycle. This model was applied to Isshiki, a natural tidal flat area, which is the largest in Mikawa Bay, Japan.This model dealt with variations of biochemical or physical interaction among dissolved oxygen and C–N–P species (comprised of carbon, nitrogen, and phosphorus elements) both on a short-time scale (<24 h) as well as over a long-time scale (seasonal variation). The model indicated that the time dependence on phytoplankton, NH₄–N, NO_x–N, and PO₄–P are more sensitive to daily environmental variation than to seasonal environmental variation. This means that the rotation speed of these materials in the tidal flat area is fast. Here, we defined the rotation speed as the ratio of total fluxes of substance to the mass of the substance. Phytoplankton with a high rotation speed in the tidal flat area means that the tidal flat has the potential to recover from rapidly increasing phytoplankton: red tide. The model also indicated that the peculiar feature of the tidal flat is the mineralization of organic material. The effect on a long term base, is that it prevents the accumulation of sediment, which results in controlling the increase of oxygen consumption in benthic system, which is the cause of oxygen depleted water.     

Sensitivity of Water Quality Model Results for the Dutch Coastal Zone to the Distribution Coefficient of Cadmium

Results of a water quality model of the Dutch coastal zone appeared to be highly sensitive for the distribution coefficient particulate/dissolved (K_d) of cadmium.Field data of the Dutch coastal zone were used to calculate the annual and seasonal trend in the distribution coefficient of cadmium over the years 1983–88. A strong seasonal and spatial gradient in the distribution coefficient was found with relatively high values in summer and lower values in winter (K_d=3.0–7.0 log l/kg). Near the coast (2 km) the K_d was lower than more offshore (70 km from the coast). In addition, values for the distribution coefficient for cadmium were extracted from the literature (K_d=2.9–4.7 log l/kg).The range of K_d values obtained from the field data was used to perform model simulations for cadmium, in order to determine the sensitivity of the model to the distribution coefficient. The modelled yearly averaged concentrations of dissolved cadmium at one location 10 km from the coast, ranged from 0.005–0.035 μg/l, depending on the magnitude of the K_d used in the simulation. These concentrations are low compared to measured values (0.053 μg/l) due to an underestimation of the cadmium input to the North Sea, or possibly the occurrence of bottom-water exchange processes which the model does not include.     

Uncertainties in Predictions of Surface Water Acidity using the MAGIC Model

In this study, we have used the MAGIC model together with data from the Birkenes catchment in Norway, at which 27 years of data (1974–2000) are available. We calibrated the MAGIC model to the five year observed average chemistry around 1990, and then used the data from the five year period around 1980 to refine the calibration. From 1990, forecasts were run for the different sets of inputs and parameters, and the sets of inputs and parameters were further refined using observations for the period 1996–2000. Through an automatic calibration routine, the model was calibrated a large number of times with different sets of input data to account for the uncertainties in the observed data using a Monte Carlo set-up. The results show that the uncertainty in the model predictions decreases as more observed data from different points in time are used in the model calibration. The results also show that when usingthe time series data in calibration, the distribution of the forecastchanged. The distribution of the predicted Acid Neutralisation Capacity (ANC) in the future is lower for the more refined model calibration. The 10 and 90 percentiles of predicted ANC in 2010 are –3 to 21 μeq L^-1 when only a five-year average is used for calibration, but are –7 to 9 μeq L^-1 when data from the three different time periods are used.     

The Link Between the Exceedance of Acidity Critical Loads for Freshwaters,Current Chemical Status and Biological Damage: A Re-Interpretation

Critical loads have for several years been employed bypolicymakers to aid in the development of strategies for aciddeposition abatement. They provide an effects-based approachwhereby an acid deposition flux greater than the critical load(known as critical load exceedance) implies that long-termharmful effects on a selected target organism will occur.Implicit in this approach are two assumptions: first, theexceedance of a critical load will harm the target organism,and second, the severity of biological impact is related to themagnitude of exceedance. However, static models give noindication of when the predicted damage might occur. One suchmodel, the Steady-State Water Chemistry (SSWC) model, employs aseries of empirical relationships to derive the pre-industrial,baseline leaching rate of base cations from measured waterchemistry using the so-called `F-factor'. The SSWC model setsthe critical load relative to pre-industrial base cationleaching (a permanent buffer of acid deposition) and a selectedacid neutralizing capacity (ANC) value which corresponds with aknown likelihood of damage to a biological target organism.Here we interpret the meaning of critical load exceedance as aprediction of steady-state ANC, and explore the relationshipbetween exceedance of the critical load and current chemistry. We demonstrate that a critical loadexceedance with the SSWC model does not necessarily indicatethat the critical chemical threshold (zero ANC) has alreadybeen crossed, and there may be no correlation betweenexceedance and biological status. A reformulation of the SSWCmodel is proposed which provides a direct link between currentdeposition and current chemical conditions, and is thereforemore likely to indicate current biological damage. Thereformulation illustrates the discrepancy between currentchemical status and that predicted by the SSWC model atsteady-state, which is a function of the `F-factor'.     

A Comparison of Loch Chemistry from 1955 and 1999 in the Cairngorms,N.E. Scotland

The Cairngorms in north-east Scotland is remote from pollutant sources although it currently receives ca. 10 kg ha¹ yr¹ S and ca. 11 kg ha¹ yr¹ N deposition from the atmosphere.In 1955, 15 lochs (lakes) at a range of altitudes were sampled and analysed for major ion concentrations. A new survey of these and an additional 23 lochs and their catchment soils was conducted in 1999 to determine the impact of acid deposition, and the changes in loch chemistry since the 1955 survey. The bedrock geology of this region has a strong influence on the loch chemistry. Surface waters were generally more acidic in high altitude areas due to predominantly poorly buffered, thin alpine soils developed on granitic parent material (mean acid neutralising capacity (ANC) for 23 lochs = 30 eq L¹). At lower altitudes where the geology is dominated by Dalradian metamorphic rocks surface waters are comparatively base rich and have higher ANC (mean ANC for 15 lochs = 157 eq L¹). Surface water nitrate concentrations show a negative relationship with soil C:N status, in that higher nitrate only occurs at low soil C:N ratios. A comparison of data for 1955 and 1999 shows that sulphate concentrations are significantly lower (67.8 and 47.5 eq L¹, respectively), and pH has improved (pH 5.6 and 5.9) in response to decreased S deposition since the mid 1970s. However, mean nitrate concentrations were found to increase from 2.48 >eq L¹ in 1955 to 5.65 eq L¹ in 1999. Differences in the sampling and laboratory methods from 1955 and 1999 are acknowledged in the interpretation of data.     

Integrated Nitrogen and Flow Modelling (INCA) in a Boreal River Basin Dominated by Forestry: Scenarios of Environmental Change

A new version (v1.7) of the Integrated Nitrogen in CAtchments model(INCA) was applied to the northern boreal Simojoki river basin (3160 km²) in Finland. The INCA model is a semi-distributed, dynamic nitrogen (N) process model which simulates N transport and processes in catchments. The INCA model was applied to model flow and seasonal inorganic N dynamics of the river Simojoki basin over the period 1994–1996, and validated for two more years. Both calibration and validation of the model were successful. The model was able to simulate annual dynamics of inorganic N concentrations in the river. The effects of forest management and atmospheric deposition on inorganic N fluxes to the sea in 2010 were studied. Three scenarios were applied for forestry practices and two for deposition. The effects of forest cutting scenarios and atmospheric deposition scenarios on inorganic N flux to the sea were small. The combination of the maximum technically possible reduction of N deposition and a decrease of 100% in forest cutting and peat mining areas decreased NO₃ ^--N flux by 6.0% and NH₄ ⁺-N flux by 3.1%.     

Simulation of Nitrogen Dynamics and Fluxes in Contrasting Catchments in Norway by Applying the Integrated Nitrogen Model for Catchments (INCA)

The process-based INCA model was applied to Dalelva Brook (3.2 km²) and the Bjerkreim River (685 km²) including several subcatchments, in order to test the model's ability to simulate streamwater nitrate (NO₃ ^-) dynamics and output fluxes under highly contrasting climatic conditions and nitrogen (N) loading. The simulated runoff volumes and mean NO₃ ^- concentrations at Dalelva and Bjerkreimwere within +2 to +10% of the measured average during 1993–1995 (–19 to +31% within individual years). INCA to a great extent also reproduced the observed streamwater flow dynamics at both study sites (coefficient of determination, r ² > 0.70). Temporal variation of streamwater NO₃ ^- during 1993–1995 was captured quite well by the model, especially at small catchments with a distinct seasonal NO₃ ^- pattern (r ² = 0.46–0.68). At the Bjerkreim River outlet, the relationship were somewhat weaker (r ² = 0.26, p < 0.01). Despite a few situations where the model failed to capturethe streamwater NO₃ ^- dynamics, INCA proved to be a quite robust tool for simulating NO₃ ^- dynamics and output fluxes in the two study catchments.     

SEASONAL AND DIURNAL METHANE EMISSIONS FROM A LANDFILL AND THEIR REGULATION BY METHANE OXIDATION

Rates of methane emission from a Swedish landfill, measured by chamber technique and permanent frames, ranged between 0.034 and 20 mmol CH₄m⁻². h⁻¹on average. The emissions followed a seasonal pattern, with the highest fluxes occurring between September and May. Methane concentrations in soil also followed a seasonal pattern, with a marked decrease during summers. Using the means of methane emission rates from frost-free periods, a stepwise regression model was made, that could explain 95% of the variation. Soil temperature turned out to be the dominating factor, explaining 85% when transformed to a second-degree function. Methane emissions were negatively correlated with soil temperature, which strongly suggests that biological methane oxidation is an important regulating factor. The activity of methane-oxidizing microorganisms was greatest around 0.5–0.6 m depth in the soil profile, and moisture at this level enhanced emissions. The tendency for methane emissions to be higher at night was probably due to the inhibitory influence of low soil temperatures on methane-oxidizing microorganisms.     

Modelling Acidification and Recovery of Swedish Lakes

The MAGIC model was calibrated to 143 lakes in Sweden, all of which are monitored in Swedish national monitoring programmes conducted by the University of Agricultural Sciences (SLU). Soil characteristics of the lake catchments were obtained from the National Survey of Forest Soils and Vegetation also carried out by SLU. Deposition data were provided by the Swedish Meteorological and Hydrological Institute (SMHI). The model successfully simulated the observed lake and soil chemistry at 133 lakes and their catchments. The fact that 85% of the lakes calibrated successfully without being treated in an individual way suggests that data gathered by the national monitoring programmes are suitable for modelling of soil and surface water recovery from acidification. The lake and soil chemistry data were then projected into the future under the deposition scenario based on emission reductions agreed in the Gothenburg protocol. Deposition of sulphur (sea salt corrected) was estimated to decrease from 1990 to 2010 by 65–73%; deposition of nitrogen was estimated to decrease by 53%. The model simulated relatively rapid improvements in lake water chemistry in response to the decline in deposition from 1990 to 2010, but the improvements levelled off once deposition stabilised at the lower value. There was a major improvement of simulated lake water charge balance acid neutralising capacity (ANC) from 1990 to 2010 in all lakes. The modelled lakes were divided into acidification sensitive and non-sensitive. The modelled sensitive lakes are representative of 20% of the most sensitive lakes in Sweden. By 2010, the ANC in the sensitive lakes was 10 to 50 μeq L^-1 below estimated pre-industrial levels and did not increase much further from 2010 to 2040. Soils at the majority of the modelled catchments continued to lose base cations even after the simulated decline in acid deposition was complete, i.e. after the year 2010. Based on this model prediction, the acidification of the Swedish soils will in general not be reversed by the deposition reduction experienced over the last 10 years and expected to occur by the year 2010.     

Recovery of Acidified Streams in Forests Treated by Total Catchment Liming

Reduced emissions of acidifying pollutants have changed the acidification process, and as a result, forest soils and surface waters are slowly recovering in Sweden. However, model calculations show that some areas may never recover completely unless further measures, such as liming, are undertaken. Liming of surface waters (lakes, rivers and wetlands) has been successfully practised in Sweden since the 1970s, but repeated treatments are necessary. A full recovery of acidified lakes and streams without frequent liming is however not possible until soil acidification is reversed in the most strongly affected areas. In this study, the recovery of acidified streams was examined using ‘the total catchment approach’ i.e. treatment of both recharge and discharge areas. The aim was to compare the quantitative effect of different treatments on run off chemistry and the recovery of brown trout. Catchments in southwest Sweden were treated with a combination of 2 tons of wood ash and 4, 6 or 12 tons of crushed limestone per hectare in 1998/1999. Treatment of both recharge and discharge areas resulted in fast and significant changes in stream water quality, e.g. increased concentrations of calcium, higher pH and ANC and a decreased concentration of inorganic aluminium. The initial changes were dependent on the distribution of the applied lime between discharge and recharge areas rather than the average dose on the total catchment. Treatment of recharge areas only, resulted in smaller but still significant effects on calcium, pH and ANC in stream water. Furthermore, there was an initial leaching of nitrate but it was only minor compared with the elevated leaching that occurs after a clear-cut. As a result of the treatments, brown trout is now successfully reproducing. Olle Westling (deceased).     

Use of the Dynamic Model `MAGIC' to Predict Recovery Following Implementation of the Gothenburg Protocol

Reduction of sulphur deposition causesrecovery of acidified surface waters. Processes in thecatchment delay recovery. The acidification model MAGICwas applied to the Vikedal and Tovdal rivers in southernNorway. Response in water chemsitry is delayed by 10–20 yr. The delay is due to release of old sulphate atVikedal and cation exchange at Tovdal. Assuming that theGothenburg protocol is fully implemented by the year2010, much of the predicted increase of ANC will occur inthe next 10 yr with a levelling off by about 2040. Ifnitrogen leaching increases in the future, however,recovery of ANC will not be as rapid, nor as complete.Critical load for acidity calculated by steady-statemodels is confirmed by the MAGIC predictions. Futurerequirement for mitigation measures such as liming willdecrease in the future as acid deposition decreases. Bythe year 2046 the liming requirement will be reduced byabout 45% at Vikedal and 65% at Tovdal. One of the mainpurposes of the Norwegian national monitoring programmeis to provide documentation of changes in environmentalquality due to long-range transported air pollutants.Modelling applications such as this clearly show that thedata fill this purpose.     

ROADWAY-2: A Model for Pollutant Dispersion near Highways

The formulations and evaluation of ROADWAY-2, a near-highway pollutant dispersion model, are described. This model incorporates vehicle wake parameterizations derived from canopy flow theory and wind tunnel measurements. The atmospheric velocity and turbulence fields are adjusted to account for velocity-deficit and turbulence production in vehicle wakes. A turbulent kinetic energy closure model of the atmospheric boundary layer is used to derive the mean velocity, temperature, and turbulence profiles from input meteorological data. ROADWAY-2 has been evaluated using SF₆ tracer data from General Motors Sulfate Dispersion Experiment. The model evaluationresults are presented and discussed.     

Spatially Disaggregated Inventories of Soil NO and N2O Emissions for Great Britain

Estimates of soil N₂O and NOemissions at regional and country scales arehighly uncertain, because the most widely usedmethodologies are based on few data, they do notinclude all sources and do not account forspatial and seasonal variability. To improveunderstanding of the spatial distribution of soilNO and N₂O emissions we have developedsimple multi-linear regression models based onpublished field studies from temperate climates.The models were applied to create spatialinventories at the 5 km² scale of soil NOand N₂O emissions for Great Britain. The N₂O regression model described soilN₂O emissions as a function of soil N input,soil water content, soil temperature and land useand provided an annual N₂O emission of 128 kt N₂O-N yr^-1. Emission rates largerthan 12 kg N₂O-N ha^-1 yr^-1 werecalculated for the high rainfall grassland areasin the west of Great Britain.Soil NO emissions were calculated using tworegression models, which described NO emissionsas a function of soil N input with and without afunction for the water filled pore space. Thetotal annual emissions from both methods, 66 and7 kt NO-N yr^-1, respectively, span the rangeof previous estimates for Great Britain.     

How are Results from Critical Load Calculations Reflected in Lake Water Chemistry?

The Steady State Water Chemistry Model (SSWCM) is a common method for determinations of critical loadof acid and subsequently of critical loadexceedance for lakes. One way to verify the modeloutput, is to compare with chemical indicatorssuch as pH-value, alkalinity or ANC. When themedian chemical status (as ANC) is used 65% ofthe lakes responded according to the exceedancevalue. For these the calculated exceedanceresulted in violation of the critical chemicalvalue while non-exceedance gave no violation.Since biota react on extreme conditions a morecorrect evaluation should be based on minimumvalues for the chemical indicator. This raises thefraction of lakes responding to 78%. Non-exceedance is seldom found inlakes with acid conditions. The evaluationindicates that the calculation of critical load ofacidity by means of SSWCM is very reliable.     

Analysis of the St. Petersburg Traffic Data using the OSPM Model

Since October 1998 two DOAS instruments were installed at the level of the first floor and at the top of a building located in St. Petersburg at Pestelya Street. The collected datacovers the time period of December 1998–March 2001, and include concentrations of benzene, toluene, NO and NO₂, ozone and SO₂. There is also an additional information about the traffic intensity and meteorological conditions. The results of the analysis of this data set, using the OSPM model, are presented here with the goal to understand the features of the air pollution dispersion in this street canyon and to analyse the information about the emission factors of the vehicles. In particular, the model results are used for the solution of the inverse problem of reconstructing the emission factors from measured concentrations. The results obtained indicate that most of the concentrations are well inside the Russian standards with the only exception of NO₂ (mean and 98-th percentile are equal to 57.8 and 119.2 g m^-3 for the street level). The same values for benzene are 18.5 and 62.6, respectively. Emission estimates show that there is a possibility that the NO_x and benzene basic emission factors recommended by the Russian national guidelines could result in overestimating the traffic emissions. These considerations are supplemented with the model sensitivity tests carried out in connection with the problem of predictability of NO₂ concentrations in the street canyon. Tests indicate that NO₂ concentrations are not very sensitive to NO_x emissions because of the usually low urban background ozone levels.     

Correction Factors for Covariance between Concentration and Deposition Velocity on CASTNes HNO3 Deposition Estimates

The covariance between hourly concentration (C) and depositionvelocity (V) for various atmospheric species may act to bias the dry deposition (D) computed from the product of the weeklyaverage C and V. This is a potential problem for the CASTNet filter pack (FP) species, nitric acid (HNO₃). Using ozone (O₃) behavior as a surrogate for HNO₃, correctionfactors (CF) are developed to estimate this bias. Weekly CF for O₃ depend on both site and season, and seasonal average weekly CF for O₃ at a given site may be as high as 1.25.The seasonal magnitude of this CF is generally largest in summerand is ordered: summer fall spring > winter. The CF is drivento a large extent by the diurnal correlation between C andV (i.e., both are generally higher during the day and lower at night). However, since the diurnal C profile at elevated sites is relatively constant, the resulting correlation between C and V is small, and the CF at montane sites is generally negligiblysmall. The sampling protocol using daytime integrated sampling for a week and nighttime integrated sampling for a week capturesthe diurnal correlation between C and V adequately and may be used to aggregate relatively unbiased weekly D estimates. Day-night CF for O₃ are close to unity, and limited results suggest similar behavior for HNO₃. Using these limited FP results, the site- and seasonally-specific weekly CF forO₃are refined to estimate the corresponding CF for HNO₃. Worst-case adjustments for HNO₃ as high as 30% are indicated for summer periods at a given site.