No abstract

. 2006. Towards optimum sampling for regional-scale N 2 O emission monitoring in Canada. Can. J. Soil Sci. 86: 441-450. There is an increasing need for field monitoring studies of N 2 O emissions to assess the reliability of process models. Our goal is to review the issues surrounding the design of monitoring and regional upscaling of fieldmeasured N 2 O emissions for Canadian conditions. Management history creates a range of controlling conditions and emission responses for each land use present in the study region and multiple fields should be sampled within each land use class. The requirement for multiple sample fields necessitates chamber-based sampling designs (ideally in conjunction with site-specific micrometeorological measurements). Uniformly minimum variance unbiased (UMVU) estimators have been recommended for estimation of the mean and variance of the sample distributions, but use of these estimators is limited where zero or negative values occur. Three basic annual emission patterns are observed across a range of N 2 O studies: background, seasonal, and event + seasonal. The event + seasonal pattern (where short duration, high emission events are superimposed on an underlying seasonal pattern) is common in Canadian agroecosystems. Background and seasonal annual patterns can be effectively captured with biweekly or even triweekly samplings. For the event + seasonal patterns more frequent samplings at periods when emission are believed to be higher (e.g., snow melt or post-fertilization events) are required. Linear interpolation of emissions between the sporadic measurements is the simplest and most reproducible method of temporal interpolation. Spatial extrapolation of the measured emissions is typically done by a measure and multiply approach, where the measured emissions for a given class are multiplied by the area of the class. The methods used to define the soil wetness, land use, and management classes should be clearly defined and consistently applied. The lack of information on the duration and magnitude of event-based emissions and the diurnal pattern of emissions is a major limitation to temporal interpolation. Les auteurs examinent les difficultés associées à la conception de modèles permettant de suivre régionalement les émissions de N 2 O mesurées sur le terrain dans les conditions typiques au Canada. La succession des pratiques culturales engendre diverses conditions régulatrices et réactions au niveau des émissions pour chaque type de terre présent dans la région à l'étude et l'on devrait échantillonner de nombreux terrains pour chaque classe d'utilisation des terres. L'échantillonnage de nombreux terrains nécessite le prélèvement d'échantillons en cellule (de préférence avec la prise de relevés micrométéorologiques sur le site même). On recommande l'utilisation d'estimateurs uniformes non biaisés à variance minimum pour estimer la moyenne et la variance de la distribution des échantillons, mais l'utilité de tels estimateurs est limitée avec les valeurs nulles ou négatives. On remarque ...