Discussion Directory UMM :Data Elmu:jurnal:A:Agricultural & Forest Meterology:Vol105.Issue4.Dec2000:

364 M.A. Sutton et al. Agricultural and Forest Meteorology 105 2000 351–369 Table 3 Comparison of mean NH 3 concentrations and fluxes from the classical aerodynamic gradient method AMANDA and the passive flux sampling approaches a Run periods compared b Measuring system χ 1 m ± S.D. mg m − 3 F ± S.D. ng m − 2 s − 1 Runs: 1, 2, 3 AMANDA 1.1 ± 0.1 20.4 ± 11.8 Runs: 1, 2, 3 Passive denuders 2.0 ± 0.8 c 25.0 ± 45.6 Run 3 AMANDA 1.2 23.3 Run 3 Passive shuttles 1.3 c 22.4 ±145 d a χ 1 m: NH 3 concentration at 1 m above the zero plane displacement height; F: net NH 3 flux; S.D.: standard deviation. b Timing of runs: Run 1: 8–1161995; Run 2: 11–1461995; Run 3: 14–1761995. Passive samplers changed at noon GMT. c Windspeed weighted concentration. d As only one run, the standard deviation refers to the scatter in the vertical concentration profile of eight samplers. one run was made, and where the scatter was much larger than for the passive denuders, for which the av- erage standard deviation of an individual profile was 35 ng m − 2 s − 1 . The NH 3 concentrations measured by the passive samplers are similar or larger than the AMANDA reference, and this may partly reflect the fact that these represent windspeed weighted con- centrations. However, if this were the case it would suggest the existence of larger NH 3 concentrations occurring during high windspeed conditions, mostly during the day. In fact there is no clear relationship be- tween χ 1 m from the AMANDA measurements and windspeed data not shown. This would suggest that the difference in χ measured by the two systems is actually not due to the difference between time aver- aged and windspeed weighted average concentration.

5. Discussion

5.1. Comparison of micrometeorological estimates and active NH 3 sampling flux estimates The comparison of the different micrometeorologi- cal estimates of u ∗ and H showed that these parameters may be measured with sufficient accuracy with both gradient and eddy covariance approaches. In princi- ple, similar uncertainties would be expected in mea- surement of λE, which in this case was limited by inadequate performance of the gradient sampling sys- tem. As has been shown before e.g. Sutton et al., 1993a, it is clear that overall uncertainties in NH 3 fluxes are dominated by estimation of the NH 3 ver- tical concentration profile. Although periods of good agreement between the two AMANDA systems were observed, there were also substantial deviations, which required a careful series of checks to explain. This illustrates again cf. Harrison and Kitto, 1990 how the practical implementation of a sampling technique is critical to its success. The comparison of the two AMANDA systems here also highlights the uncer- tainties that might be missed when measuring using a single gradient system. It is thus essential to regu- larly establish that the system is functioning reliably, e.g. by co-sampling different inlets at one height. The comparison of Fig. 4 illustrates the difficulty in giv- ing a simple statistic regarding the precision and over- all uncertainty of the flux measurements. As has been shown in other measurements applying AMANDA e.g. Wyers et al., 1993; Sutton et al., 1998, rela- tive uncertainty is dominated by the ability to measure NH 3 gradients accurately with the AMANDA system. Hence the uncertainty in the flux is a function of NH 3 air concentrations, atmospheric humidity, temperature and overall optimization of the AMANDA continu- ous flow. During sound operation of the AMANDA the flux estimates are typically precise to ±20 at 100 ng m − 2 s − 1 , although the overall uncertainty in- cluding all sources of error would be around ±30. These estimates vary and the uncertainties were larger for the filter packs and UPM AMANDA. 5.2. Diurnal behaviour of NH 3 fluxes in relation to environmental conditions Despite the uncertainties in the measured fluxes, the data reveal clear diurnal patterns in NH 3 exchange fluxes, which are broadly similar to those over other agricultural crop canopies, such as cereals and grass- land e.g. Harper et al., 1987; Sutton et al., 1993b, M.A. Sutton et al. Agricultural and Forest Meteorology 105 2000 351–369 365 1995a,b. The classic pattern of large emission in day- time and smaller deposition at night was frequently seen, although there were also clear instances of large night-time emission for the rape here. Key processes affecting the diurnal patterns of NH 3 fluxes are the variation in the turbulent diffusion coefficient K H and linked to this the atmospheric resistance R a , as well as factors affecting the affinity of the surface for NH 3 . The value of K H is typically largest in the day, which increases the magnitude of fluxes compared with night, but does not directly affect the flux direc- tion. The direction of the flux results from the balance between the air concentration, e.g. χ 1 m, and the surface concentration or compensation point of the canopy χ c , Sutton et al., 1995b. The latter is particu- larly a function of surface wetness and temperature so that deposition is generally expected in cooler, wetter nocturnal conditions. The compensation point model of Sutton et al. 1995b, which has been applied to sev- eral canopy types, deals with the competition of depo- sition to leaf cuticles and the bi-directional exchange through stomata with a compensation point χ s of leaf tissues. The resistance for cuticular deposition is described as a function of humidity, while the stomatal flux depends on temperature affecting the magnitude of χ s and stomatal resistance, as well as the calcu- lated magnitude of χ c . Since nocturnal stomatal emis- sions are unlikely given stomatal closure and a lack of diurnal variability in tissue ammonium concentrations Husted et al., 2000, the net emissions observed at night here must result from another source. As further analysis indicates Nemitz et al., 2000a,b, the extra source for the rape canopy is decomposing litter on the soil surface, which can largely explain the night-time emissions of NH 3 from the oilseed rape. Attempts to model the net fluxes, including component emis- sions from foliage, siliques seed cases and fallen leaf litter, are described by Nemitz et al. 2000b. 5.3. Comparison of NH 3 fluxes before and after cutting There is an abundance of literature indicating larger NH 3 emissions from senescing than from green leaves e.g. Farquhar et al., 1979; Parton et al., 1988; Schjo- erring et al., 1998, and this includes measurements of increased compensation points for attached senescing leaves of oilseed rape Husted and Schjoerring, 1996, as well as literature on NH 3 volatilization from crop plant residues Mannheim et al., 1997. There is, how- ever, little information on emission of NH 3 following cutting of crops including oilseed rape or from decom- posing leaf litter. The measurements here clearly show an increase in NH 3 emissions following cutting. From Table 2, it is seen that the net emission flux increased by more than a factor of 3 from 16 to 57 ng m − 2 s − 1 from the pre-cutting to the post-cutting measurement periods. This may result from the increased aerody- namic access of the fallen leaf litter to the atmosphere, as well as from increased turbulence from the aerody- namically rougher cut field. The first effect is expected to be most important, particularly as this would add to cutting induced senescence emissions from the cut crop itself. The effect of increased turbulence is prob- ably a secondary effect and would, e.g. not explain the increased surface NH 3 concentrations during the post-cutting period Section 5.5. Average values of χ c were 1.3 and 6.8 mg m − 3 for the first and second measurement period, respectively, and the larger value of χ c for the second period supports this conclusion. 5.4. Overall ammonia emissions from the canopy The fluxes reported here for oilseed rape are larger than has typically been reported for other crop canopies in northern Europe. Typical midday emis- sion fluxes here of 50 ng m − 2 s − 1 for the ripening canopy and 100–200 ng m − 2 s − 1 after cutting, com- pare with daytime emissions from wheat of typically 10–30 ng m − 2 s − 1 Sutton et al., 1995a. Although larger spring emissions from wheat were measured by Sutton et al. 1995a, these emissions were related to residues of urea, which are known to cause large NH 3 losses. Fig. 8 indicates the average daytime and night-time NH 3 fluxes recorded throughout the measurement period. Integration of these values for the campaign as a whole provided a net emission of 0.22 and 0.44 kg N ha − 1 for the pre- and post-cutting measure- ment periods, respectively, representing 20 and 11 days. If these figures were extrapolated to the sum- mer months of May–August, this would represent net emissions of 0.95 and 1.6 kg N ha − 1 , or a total of approximately 2.5 kg N ha − 1 . Although extrapola- tion beyond these periods would be very speculative, 366 M.A. Sutton et al. Agricultural and Forest Meteorology 105 2000 351–369 it is likely that net emission also occurs for spring. The largest uncertainties apply for autumn and win- ter months, especially in relation to ploughing and variability in timing of fertilization. Despite the un- certainties, the measurements here have implications for regional scale emission inventories, which have often subsumed crop NH 3 emissions into estimates of losses from fertilizer application e.g. Sutton, 1996. The measurements suggest that for oilseed rape, this may be an underestimate. Expressed as a percentage of the N applied to the crop this may account for an additional 1–2 emission as ammonia, compared with the existing CORINAIREMEP total loss esti- mate of 2 applied N where fertilizer is applied as ammonium nitrate Sutton, 1996. 5.5. Interactions between bi-directional fluxes and NH 3 air concentrations A particularly interesting feature of the data here, of relevance to both modelling exchange and atmospheric budgets, is the link between fluxes and NH 3 concen- trations. In modelling dry deposition, it is common to infer fluxes from a resistance model and air concen- tration estimates ‘inferential approach’, e.g. Fowler and Unsworth, 1979; Sutton et al., 1993a; Erisman et al., 1994. While, for periods of deposition, larger deposition fluxes are linked to larger air concentra- tions, the opposite holds for periods of NH 3 emission from the canopy. Here larger air concentrations of- ten correspond to larger NH 3 emissions cf. day and night fluxes; pre- and post-cutting. If NH 3 fluxes were viewed as linking statically to a constant canopy com- pensation point χ c , then larger emissions would be expected during periods of small NH 3 air concentra- tions. The opposite relation observed here indicates that χ c is more variable than χ 1 m; through the ef- fect on the emission flux, χ c is affecting the mag- nitude of χ 1 m. Fig. 9 demonstrates the positive correlation between F χ and χ 1 m for both measure- ment periods over the oilseed rape at North Berwick. In this analysis the average flux for different χ 1 m classes is plotted, and this shows how the net NH 3 flux for all NH 3 air concentration classes was emissioned. Following the above argument, larger NH 3 concen- trations correspond to increased emission fluxes i.e. χ 1 m is the dependent variable rather than F χ . In- Fig. 9. NH 3 flux as a function of NH 3 air concentration during the two measurement periods, contrasted with the behaviour for a semi-natural heathland at Elspeet, Netherlands from Nemitz, 1998. Each dot represents a block average of 100 observations for different NH 3 concentration classes 50 for North Berwick, post-cutting. terestingly, despite the larger emissions after the cut, the slope of the relationship remained more or less the same, which may reflect the broadly similar u ∗ values during the two parts of the experiment. In Fig. 9 the findings over the oilseed rape at North Berwick are contrasted with measurements of NH 3 deposition to a semi-natural heathland at Elspeet, NL Nemitz, 1998. In this case the deposition flux increased with air con- centration, showing the dependence of F χ on χ 1 m as is normally recognized in the inferential approach. An approximate indication of the daytime air con- centration enhancement due to the crop emissions may be given by estimating the air concentration at a height little affected by the surface emissions. Using a simple resistance analogy it can be shown that χ 10 m = χ 1 m − F NH3 R a 1 m, 10 m 8 During daytime the turbulent atmospheric resistance R a z 1 , z 2 , for the layer between 1 and 10 m above d of the rape canopy, was typically 25 s m − 1 . Taking M.A. Sutton et al. Agricultural and Forest Meteorology 105 2000 351–369 367 daytime values of χ 1 m and F from Table 2, pro- vides an average χ 10 m during the first period of 0.54 mg m − 3 and during the second of 0.52 mg m − 3 . Although this analysis is very approximate, it broadly supports an unaffected background air concentra- tion of ≈0.5 mg m − 3 and enhancements 1 m above the canopy of 0.6 and 2.1 mg m − 3 for the pre- and post-cutting periods, respectively. There were, nev- ertheless, periods during which NH 3 advection from local point sources temporarily increased the air con- centration at North Berwick. While the effect of an emission source on air con- centrations is well known for classical NH 3 sources such as animal manures, it is a new step to make this link for the bi-directional exchange over croplands. Since NH 3 concentrations in arable areas are partly a response to the crop emissions, it becomes difficult to use inferential models to infer net bi-directional ex- change from monitoring data. The way forward must be to incorporate the bi-directional NH 3 resistance models developed from flux measurements directly into atmospheric transport models, and these must give particular attention to predict ground level NH 3 con- centrations. In this way, the interactions between NH 3 concentrations and fluxes become integrated for both large livestock sources and bi-directional exchange with vegetation. 5.6. Assessment of the passive flux sampling approach In addition to provide a detailed picture of the NH 3 exchange fluxes, the availability of the continuous AMANDA measurements provided the facility to test the passive flux sampling method for determining ver- tical NH 3 fluxes. Although intensive measurements were made with the passive denuders 16 samplers per profile with duplicate profiles, short sampling periods led to substantial scatter in the results, and this was even more extreme for the shuttles. This is because a shorter sampling time allows for a less capture of NH 3 in the samplers thereby increasing uncertainties. As has been shown by Schjoerring 1995, for a sampling period of 7 days, a mean flux of 15 ng m − 2 s − 1 would be required to maintain a coefficient of variation in the passive denuder flux estimates at 20. On average, the fluxes measured here agreed closely within 25, but given the degree of scatter this must be considered as fortuitous. The clear message is that much longer duration inter-comparisons of several months are re- quired to assess the passive approach e.g. Hansen et al., 1999. A much longer inter-comparison period would have the advantage that each run could be of greater duration improving precision of gradient determination, and also provide a larger population of sampling runs for comparison with the reference. Recognizing the costs of providing continuous flux estimates for reference, this necessarily becomes a major task. The present results also highlight the scale limitation of the passive approach. For typical air concentrations and fluxes with the same degree of scatter, the passive method cannot be expected to provide precise flux estimates with less than 10–20 days resolution. These provide important messages to other researchers wishing to evaluate further the pas- sive flux sampler approach. Such studies must move beyond a campaign approach and should ideally fo- cus on performance between different seasons and years.

6. Conclusions