Search

Forest attributes such as volume or basal area are concentrated at tree locations and are absent elsewhere. It is, therefore, more meaningful to consider the amount of forest attributes at a prefixed spatial grain, within regular plots of prefixed size centered at the points of the study area. In this way, the diversity of attributes within plots also can be considered and quantified by suitable indexes, giving rise to a diversity surface defined on the continuum of points constituting the area. We analyze the estimation of diversity surfaces when a sample of plots is selected by a probabilistic sampling scheme and diversity within nonsampled plots is estimated using an inverse distance weighting interpolator. We discuss the design-based asymptotic properties of the resulting maps when the survey area remains fixed and the number of sampled points increases. Because diversity surfaces share suitable mathematical properties, if the schemes adopted to select sample points ensure an even coverage of the study areas avoiding large portions of non-sampled zones, it can be proven that the estimated maps approach the true maps. Key words: diversity maps, inverse distance weighting interpolator, design-based consistency, spatial simulation, case study. Les attributs de la foret comme le volume ou la surface terriere sont concentres a l'emplacement des arbres et sont absents ailleurs. Par consequent, il est plus significatif de considerer la quantite des attributs de la foret a une resolution spatiale predeterminee, dans des parcelles regulieres de taille predeterminee, centrees sur les points de la zone d'etude. De cette maniere, la diversite des attributs au sein des parcelles peut egalement etre prise en compte et quantifiee a l'aide d'indices appropries, donnant lieu a une surface de diversite definie sur le continuum de points constituant la superficie a l'etude. Nous analysons l'estimation des surfaces de diversite lorsqu'un echantillon de parcelles est selectionne au moyen d'un plan d'echantillonnage probabiliste et que la diversite au sein de parcelles non echantillonnees est estimee a l'aide d'un interpolateur pondere par l'inverse de la distance. Nous discutons des proprietes asymptotiques des cartes obtenues, qui resultent du plan d'experience, lorsque la zone d'inventaire demeure fixe et que le nombre de points echantillonnes augmente. Puisque les surfaces de diversite partagent des proprietes mathematiques appropriees, si les plans d'echantillonnage adoptes pour selectionner les points d'echantillonnage assurent une couverture uniforme des zones d'etude, en evitant de laisser de larges portions non echantillonnees, on peut demontrer que les cartes estimees se rapprochent des vraies cartes. [Traduit par la Redaction] Mots-cles : cartes de diversite, interpolateur pondere par l'inverse de la distance, coherence basee sur la conception du plan d'experience, simulation spatiale, etude de cas., Introduction Spatially explicit estimates are needed in many environmental and ecological applications for making maps: the increasing practical demand is for information on the spatial distribution of environmental and ecological [...]

Non-detection of trees is an important issue when using single-scan TLS in forest inventories. A hybrid inference approach is adopted. Quoting from distance sampling, a detection function is assumed, so that the inclusion probability of each tree included within each plot can be determined. A simulation study is performed to compare the TLS-based estimators corrected and uncorrected for non-detection with the Horvitz-Thompson estimator based on conventional plot sampling, in which all the trees within plots are recorded. Results show that single-scan TLS provides more efficient estimators with respect to those provided by the conventional plot sampling in the case of low-density forests when no distance sampling correction is performed. In low-density forests, uncorrected estimators lead to a small bias (1%-6%), increasing with plot size. Therefore, care must be taken in enlarging the plot radius too much. The bias increases in forests with clustered spatial structures and in dense forests, where the bias levels (30%-50%) deteriorate the performance of uncorrected estimators. Even if the bias- corrected estimators prove to be effective in reducing the bias (below 15%), these reductions are not sufficient to outperform conventional plot sampling. Therefore, there is no convenience in using TLS-based estimation in high-density forests.Key words: plot sampling, TLS-based detection, distance sampling, hybrid inference, simulation study.La non-detection des arbres est un probleme important lors de l'utilisation d'un scanner laser terrestre (SLT) a balayage unique pour les inventaires forestiers. Une approche d'inference hybride est adoptee. En se fondant sur l'echantillonnage a distance, on peut deduire une fonction de detection permettant de determiner la probabilite d'inclusion de chaque arbre present dans chaque placette. Une etude par simulation est realisee pour comparer les estimateurs fondes sur le SLT qui ont ete corriges ou non pour la non-detection a l'aide de l'estimateur de Horvitz-Thompson, lequel est base sur un echantillonnage conventionnel de placettes dans lequel tous les arbres des placettes sont enregistres. Les resultats montrent que le SLT a balayage unique produit des estimateurs plus efficaces que ceux provenant d'un echantillonnage conventionnel de placettes dans le cas des forets a faible densite lorsqu'aucune correction d'echantillonnage a distance n'est effectuee. Dans les forets a faible densite, les estimateurs non corriges entrainent un leger biais (de 1 a 6 %), qui augmente avec la taille de la placette. Par consequent, il faut prendre garde de ne pas trop agrandir le rayon de la placette. Le biais augmente dans les forets ayant une structure spatiale regroupee et dans les forets denses, pour lesquelles la taille du biais (de 30 a 50 %) deteriore la performance des estimateurs non corriges. Meme si la correction du biais des estimateurs s'avere efficace pour reduire le biais (inferieur a 15 %), ces reductions ne sont pas suffisantes pour surpasser l'echantillonnage conventionnel de placettes. Par consequent, il n'y a pas d'avantage a utiliser une estimation fondee sur le SLT dans les forets a forte densite. [Traduit par la Redaction]Mots-cles : echantillonnage de placettes, detection par scanner laser terrestre (SLT), echantillonnage a distance, inference hybride, etude par simulation., 1. IntroductionTerrestrial laser scanning (TLS) has been demonstrated to be a promising tool for plot-level field inventories (e.g., Liang et al. 2011; Moskal and Zheng 2012; Kankare et al. 2015; [...]

The aim of this paper is to give an overview of some recent proposals to support national forest inventories. The reviewed literature is strictly of design- based nature, i.e., uncertainty only stems from the sampling scheme actually adopted in the survey, rather than being assumed or modeled as in model- based approaches. National forest inventories are viewed as two-phase sample surveys to estimate at the same occasion the extent of the continuous population of points constituting the forest cover and the total of a forest attribute (e.g., volume or biomass) in the discrete population of trees for several forest types and/or administrative districts. The first phase is performed from remote sensing imagery while the second phase is performed on the field, possibly adopting the information acquired in the first phase as auxiliary information. A novel methodology is adopted based on Monte Carlo integration methods, which leads to a very general estimation strategy. Some recent proposals are considered in which remote sensing information acquired in the first phase is used to assess some physical characteristics of non-forest resources, such as woodlots, tree-rows and isolated trees outside the forest without additional field work. Finally, a new proposal is discussed in which canopy height from laser scanning is adopted as auxiliary information to recover missing data occurring when some sampled points cannot be reached because of hazardous terrain.

The present study investigates some sampling strategies for the estimation of abundance and canopy cover of trees outside forest (TOF) over large areas. A collection of about 53 000 TOF units in Central Italy was acquired by visual, on-screen interpretation of aerial orthophotos and was taken as the reference population with the purpose of investigating: (i) one-phase inventories with sample points located by means of the tessellation stratified sampling (TSS), which involves covering the study region by a grid of regular polygons of equal sizes and randomly and independently selecting a point in each of them; (ii) two-phase inventories with the one-per-stratum stratified sampling adopted in the second phase to select a sample of polygons from the grid and then visit only the points contained in those polygons. Uniform random sampling is also considered in the first phase as a benchmark for tessellation stratified sampling. The sampling schemes adopted to select TOF units at the sample points are as follows: (i) point sampling, (ii) centroid-based plot sampling with plot radius of 50 m (CPLS50) or 100 m, and (iii) plot intersect sampling with plot radius of 50 or 100 m. CPLS50 under single-phase TSS proves to be a promising strategy to large-scale TOF inventories. Key words: design-based estimation, tessellation stratified sampling, two-phase sampling, one-per-stratum stratified sampling. Cette etude porte sur quelques strategies d'echantillonnage pour estimer l'abondance et le couvert de la cime des arbres situes hors des forets (AHF) sur de grandes superficies. Un ensemble d'environ 53 000 unites d'AHF situe dans le centre de l'Italie a ete acquis a l'aide d'une interpretation visuelle d'orthophotos aeriennes digitalisees pour former une population de reference dans le but d'etudier (i) des inventaires a une phase avec des points d'echantillonnage localises au moyen de l'echantillonnage stratifie par tessellation (EST), qui consiste a couvrir la region d'etude avec une grille de polygones reguliers de taille egale, et de selectionner aleatoirement et independamment un point dans chacun d'eux; et (ii) des inventaires a deux phases a l'aide d'un echantillonnage stratifie avec un echantillon par strate utilise dans la deuxieme phase pour selectionner un echantillon de polygones a partir de la grille et visiter seulement les points contenus dans ces polygones. Un echantillonnage aleatoire uniforme est aussi considere dans la premiere phase comme reference pour l'EST. Les schemas d'echantillonnage adoptes pour selectionner les unites d'AHF aux points d'echantillonnage sont (i) l'echantillonnage par point, (ii) l'echantillonnage fonde sur les centroides avec des placettes de 50 ou 100 m de rayon, (iii) l'echantillonnage aux points d'intersection avec des placettes de 50 ou 100 m de rayon. Cinq tailles d'echantillon sont evaluees pour la deuxieme phase et correspondent a environ 50 %, 25 %, 10 %, 5 % et 1 % des polygones selectionnes a partir de la grille. Les strategies testees sont evaluees en termes d'effort et d'erreur d'echantillonnage. [Traduit par la Redaction] Mots-cles : estimations fondees sur le plan de sondage, echantillonnage stratifie par tessellation, echantillonnage a deux phases, echantillonnage stratifie avec un echantillon par strate., 1. Introduction Trees outside forest (TOF) are prominent features in many landscapes worldwide, including natural, cultural, and recently modified landscapes (de Foresta et al. 2013). However, characteristics and dynamics of [...]

Land use inventories are sound measures to provide information on the area occupied by different land use or land cover types and their changes, although less widespread than traditional mapping; as such, they are distinctively well-established tools for generating statistics on the state and the dynamics of land use in the European Union. Italy has recently set up a land use inventory system (IUTI) as a key instrument for accounting removals and emissions of greenhouse gases (GHG) associated to land use, land use change and forestry (LULUCF) activities elected by Italy under the Kyoto Protocol. IUTI adopts a sta­tistical sampling procedure to estimate the area covered by LULUCF land use categories in Italy, and associated uncertainty estimates. Estimates of land use have been so far processed for the period 1990-2008 and highlight three interlinked land use change patterns in Italy: (i) increase in forest land for a total uptake of 1.7% of the Italian territory; forest cover estimates, with a standard error of 0.1%, indicate an annual increase of forestland higher over the period 1990-2000 (32 901 ha year-1) than in 2000-2008 (22 857 ha year-1); surpri­singly, also a significant deforestation rate is observed (-7000 ha year-1), due to forest land conversion mainly into artificial areas; (ii) consumption of arable land (-4.2% of the Italian territory) primarily due to land uptake by urban areas and to conversions to permanent crops (mainly orchards and vineyards); (iii) urban sprawl uptakes 1.6% of the Italian territory in this period, with a total coverage of settlements reaching 7.1% of total land surface in Italy in 2008. Overall, land use dynamic results in land uptake by forest land is of the same magnitude of land uptake by urban areas, but the effects of these processes on GHG removals (by forest sinks) and emissions (by urban areas) is expected to be significantly different. In a broader perspective, IUTI methodology, by providing reliable estimates and well-defined levels of statistical uncertainty for assessing stocks and flows of land use at national level, can be further implemented to frame other key questions for sustainable development policies, like the set up of environmental-economic accounting systems.

The k-nearest neighbours (k-NN) method constitutes a possible approach to improve the precision of the Horvitz- Thompson estimator of a single interest variable using auxiliary information at the estimation stage. Improvements are likely to occur when the neighbouring structure in the space of auxiliary variables is similar to the neighbouring structure in the space of the survey variables. Populations suitable for k-NN can be identified via the scores of the first principal component computed on the variance-covariance matrix of auxiliary variables. If the first principal component explains a large portion of the whole variability, distances among scores provide good approximations of distances in the space of auxiliary variables in such a way that the effectiveness of k-NN can be assessed by plotting the first principal component scores versus the sampled values of each of the interest variables. Monotone relationships with high values of Spearman's correlation coefficients should denote effectiveness. Otherwise, when the first principal component explains small fractions of the total variation, an index that directly quantifies the similarity between the neighbouring structure in the space of interest and auxiliary variables is proposed. The validity of the proposed diagnostics is theoretically argued and empirically proven by a simulation study performed on a wide range of artificial and real populations. Resume: La methode des k plus proches voisins (k-NN) pourrait ameliorer la precision de l'estimateur de Horvitz-Thompson d'une variable d'interet unique en utilisant une information auxiliaire au stade de l'estimation. Les ameliorations devraient survenir quand la structure des voisins dans l'espace des variables auxiliaires est semblable a la structure des voisins dans l'espace des variables inventoriees. Les populations qui se pretent a la methode des k-NN peuvent etre identifie es par les scores de la premiere composante principale calcules a partir de la matrice des variances et covariances des variables auxiliaires. Si la premiere composante principale explique une grande proportion de la variabilite totale, la distance entre les scores fournit une bonne approximation de la distance dans l'espace des variables auxiliaires, si bien que l'efficacite de la methode des k-NN peut etre evaluee en mettant en relation les scores de la premiere composante principale et les valeurs echantillonnees de chaque variable d'interet. Des relations monotones associees a de fortes valeurs du coefficient de correlation de Spearman devraient etre une marque d'efficacite. Par ailleurs, lorsque la premiere composante principale explique une faible proportion de la variation totale, nous proposons un indice qui quantifie directement la similarite entre la structure des voisins dans l'espace d'interet et les variables auxiliaires. La validite des diagnostics proposes est justifiee theoriquement et prouvee empiriquement par une etude de simulation realisee sur une large gamme de populations artificielles et reelles. [Traduit par la Redaction], Introduction Several techniques for assessing natural resources couple information achieved from remotely sensed imagery with information collected from ground surveys. Of these methodologies, the k-nearest neighbours (k-NN) technique is becoming [...]

The numerous sigma ([sigma]) factors present in Mycobacterium tuberculosis are indicative of the adaptability of this pathogen to different environmental conditions. In this report, we describe the M. tuberculosis [[sigma].sup.B] regulon and the phenotypes of an M. tuberculosis sigB mutant strain exposed to cell envelope stress, oxidative stress, and hypoxia. The sigB mutant was especially defective in survival under hypoxic conditions in vitro, but it was not attenuated for growth in THP-1 cells or during mouse and guinea pig infection.