Search

Budiman Minasny, Alex. B. McBratney, Brendan P. Malone, Denis A. Angers, Dominique Arrouays, Adam Chambers, Vincent Chaplot, Zueng-Sang Chen, Kun Cheng, Bhabani S. Das, Damien J. Field, Alessandro Gimona, Carolyn Hedley, Suk Young Hong, Biswapati Mandal, Ben P. Marchant, Manuel Martin, Brian G. McConkey, Vera Leatitia Mulder, Sharon O’Rourke, Anne C. Richer-de-Forges, Inakwu Odeh, José Padarian, Keith Paustian,Genxing Pan, Laura Poggio, Igor Savin, Vladimir Stolbovoy, Uta Stockmann, Yiyi Sulaeman, Chun-Chih Tsui, Tor-Gunnar Vågen, Bas van Wesemael, Leigh Winowiecki 2017. 4 per 1000 soil carbon sequestration. Proceedings of the Global Symposium on Soil Organic Carbon 2017. 21-23 March 2017. Rome, Italy. FAO.; International audience

Dominique Arrouays, Budiman Minasny, Alex. B. McBratney, Mike Grundy, Neil McKenzie, James Thompson, Alessandro Gimona, Suk Young Hong, Scott Smith, Alfred Hartemink, Songchao Chen, Manuel P. Martin, Vera Leatitia Mulder, Anne C. Richer-de-Forges, Inakwu Odeh, José Padarian, Glenn Lelyk, Laura Poggio, Igor Savin, Vladimir Stolbovoy, Yiyi Sulaeman, Dedi Nursyamsi, Gan-Lin Zhang, Mogens H. Greve, Zamir Libohova, Philippe Lagacherie, Pierre Roudier, Johan G.B. Leenaars, Gerard B.M. Heuvelink, Luca Montanarella, Panos Panagos, Jon Hempel. 2017. GlobalSoilMap for soil organic carbon mapping and as a basis for global modeling. Proceedings of the Global Symposium on Soil Organic Carbon 2017. 21-23 March 2017. Rome, Italy. FAO.; International audience

International audience; The limited availability of soil information has been recognized as a main limiting factor in digital soil mapping (DSM) studies. It is therefore important to optimize the joint use of the three sources of soil data that can be used as inputs of DSM models, namely spatial sets of measured sites, soil maps and soil sensing products.In this paper, we propose to combine these three inputs, through a cokriging with a categorical external drift (CKCED). This new interpolation technique was applied for mapping seven soil properties over a 24.6 km2 area located in the vineyard plain of Languedoc (Southern France), using an hyperspectral imagery product as example of a soil sensing data. Cross-validation results of CKCED were compared with those of five spatial and non-spatial techniques using one of these inputs or a combination of two of them.The results obtained in the La Peyne Catchment showed i) the utility of soil map and hyperspectral imagery products as auxiliary data for improving soil property predictions ii) the greater added-value of the latter against the former in most situations and iii) the feasibility and the interest of CKCED in a limited number of soil properties and data configurations. Testing CKCED in case study with soil maps of better quality and soil sensing techniques covering more area and depths should be necessary to better evaluate the benefits of this new technique.

Des zones de rendements homogènes ont été identifiées sur deux parcelles adjacentes de l’ordre de 50 ha chacune localisées sur la commune de Villamblain dans le Loiret. Le zonage a été réalisé selon deux approches à partir de mesures sur 11 années de rendements du blé et du maïs effectuées à l’aide de capteurs embarqués sur les machines de récoltes. La première approche de zonage est basée sur une analyse fréquentielle que nous avons dénommé AFQ dans ce mémoire, et la seconde approche est basée sur une analyse géostatistique mutivariable connue sous le nom d’analyse factorielle krigeante (AFK). Dans un premier temps et dans les deux cas, on parvient à obtenir des zones présentant des rendements moyens différents et de tailles suffisantes pour qu’elles soient envisagées comme des unités de gestion différenciées pour l’agriculteur. Toutefois, les résultats révèlent que le zonage par AFK paraît notamment mieux approprier dans le cas d’une culture systématiquement irriguée pour laquelle le zonage établi par AFQ est d’avantage influencé par la méthode d’irrigation. Le zonage selon l’approche AFK semble donc plus générique, il demeure néanmoins moins facile à mettre en oeuvre comparativement à l’approche par AFQ qui est plus intuitive. Dans un second temps, en utilisant une carte pédologique détaillée de la parcelle établie antérieurement, on montre que le zonage établi par les deux approches est en adéquation avec le type de sol, son épaisseur et sa réserve utile potentielle. Ce constat ouvre de nombreuses perspectives en ce qui concerne l’utilisation des cartes de zones de rendements homogènes., Homogeneous yield areas were identified on two adjacent plots of about 50 ha each, in the centre France. The zoning was done using two approaches based on measurements of 11-year yields of wheat and maize made using onboard sensors of harvest machines. The first zoning approach is based on a frequency analysis referred as FQA in this work, and the second approach is based on a multivariate geostatistical analysis known as factorial kriging analysis (FKA). Both approaches lead to areas with average yields statistically different and sufficient sizes for them to be considered as differentiated management units for farmers. However, the results show that the FKA zoning seems more appropriate in the case of a systematically irrigated crop for which the zoning established by FQA is more influenced by irrigation method. Zoning according to FKA approach seems more generic, it nevertheless remains less easy to implement compared to the FQA approach which is much more intuitive. In addition, this work revealed that the zoning established by the two approaches is consistent with soil type, its thickness and its potential available water content. This finding opens many perspectives regarding the use of maps of homogeneous yield areas.

Red clay soils from subtropical China (Ultisol) are not considered currently as sensitive to crusting and erosion by water. However, recent field data showed erosion can be significant. To investigate the sensitivity of red clay soil to crusting and runoff, a laboratory rainfall study was carried out. Because rainfall kinetic energy is known to have an important effect on aggregate breakdown and therefore crust formation, two treatments were applied: high kinetic energy (HKE) and low kinetic energy (LKE). Air-dried soil was packed in a soil box (50 cm × 50 cm) and was subjected to 38 mm.h-1 rain for 90 min (3 replicates per treatment). Samples were taken for preparing soil thin sections and runoff, splash, and percolation were measured during the rain. Crusts formed after 40 min and 60 min of rain for the HKE and LKE treatment, respectively. Runoff for the HKE treatment started after 25 min of rain and reached a steady state after the crust well formed (50 min). While for the LKE, runoff started after 40 min of rain and kept increasing to a final runoff rate similar to the HKE treatment. Percolation started after 57 min of rain for the LKE treatment while no percolation occurred for the HKE treatment. The total mass of splashed material was 12 times higher for the HKE treatment than for the LKE treatment and significantly more soil was eroded for the HKE treatment. Stability tests gave mean weight diameters of 0.55 mm, 1.48 mm, and 1.74 mm for the fast wetting, mechanical stirring, and slow wetting, respectively, indicating red clay soils are very sensitive to slaking. The results show that Ultisol is prone to crusting and that kinetic energy is a major driver of their crust formation. Reducing the rain kinetic energy can preserve aggregate structure and delay the development of crust, which subsequently increase infiltration and decrease runoff and erosion on red soils.

Les nouvelles possibilités d'intégration de capteurs sur des matériels de récolte ouvrent la voie de mesures de qualité intraparcellaires en temps réel. Comment valider les mesures ainsi réalisées ? Pour répondre à cette question, nous avons mis au point un dispositif de prélèvement d'échantillons géoréférencés. L'analyse en laboratoire de ces échantillons permet d'établir des cartes intraparcellaires de référence afin de contrôler les mesures en temps réel. Une première utilisation a concerné la mesure intraparcellaire de la teneur en protéines. / The new possibilities given by the sensors integration on harvest equipment open the way of real time site-specific quality measurement. How to validate the measurement done by this way? To answer this question, we have developed a georeferenced sampling system. The laboratory analyses of these samples allows to create reference site-specific maps usable to control real time measurement. A first use was about site-specific proteins level measurement.

International audience; Soil survey constitutes one of the major means of describing the spatial variations of soil cover. It is based on local observations and a landscape analysis. A first approach for spatial prediction is based upon map unit characterization and delineation. They make it possible: to choose sampling strategies (Fig. 1); to describe the main nested levels of the spatial structure (Fig. 2) ; to estima te a given property. Nevertheless, the notion of continuum in soil cover and the great spatial variability of soil properties are not taken into account. This has lead to using more and more geostatistical methods these last ten years. This second approach is based on the spatial continuity in soil mantle and privileges spatial relationships between punctual observations by using the notion of spatial structure. ft requires a good estimate of this structure (Fig. 4) and makes it possible to decrease the estimated variance (Fig. 3). Due to the complementarity of these two approaches, works of comparison or new combined approaches have recently been developed. This paper analyses how spatial prediction is solved by soil survey approaches, by geostatistical approaches and finally by new combined approaches. Three phases are considered .' sampling strategy; spatial variability analysis; statistical methods ofspatial prediction.; La carte pédologique est un des principaux moyens pour décrire l'organisation des sols dans l'espace. Elle est à la fois fondée sur des observations ponctuelles du sol et sur une analyse du paysage. Le découpage de la couverture pédologique en unités cartographiques, retenu par la carte pédologique, a classiquement conduit à donner un poids important à la définition et à la délimitation des unités cartographiques pour effectuer des estimations spatiales d'une propriété donnée. Cependant, la notion de continuum dans la couverture pédologique, associée à la grande variabilité spatiale des caractéristiques pédologiques, a imposé ces dernières années un recours important aux méthodes géostatistiques, au sens de la théorie des variables régionalisées. Ces méthodes sont fondées sur la continuité spatiale de la couverture pédologique, par la notion de structure spatiale, et privilégient ainsi les relations de voisinage entre les observations pour effectuer des estimations spatiales. Ces deux approches sont à la fois très différentes sur le plan conceptuel et complémentaires du point de vue de la prise en compte de l'organisation des sols. Or des travaux comparatifs de ces démarches et une émergence d'approches couplées apparaissent. Cet article analyse comment le problème de l'estimation spatiale est abordé dans les approches basées sur la carte pédologique, dans les approches géostatistiques, puis dans les approches couplées. Trois principales étapes de l'estimation spatiale sont abordées: la stratégie d'échantillonnage; l'analyse de la variabilité spatiale; la méthode d'estimation spatiale associée. Mots-clés : carte des sols-cartographie des sols-unité cartographique-statistique spatiale-estimation spatiale-géostatistique.