Universal kriging with a topographic variable used as external drift was compared with ordinary kriging, ordinary kriging with external drift, and universal kriging for predicting the thickness of a superficial deposit on part of the Beauce of central France. Thickness was recorded at 62 sites and at another 25 sites to obtain a sample for validation. A Digital Elevation Model (DEM) was computed from 484 sites where relative elevation was measured. The slope gradient which is significantly correlated to thickness was used as an external drift variable for kriging thickness. The sample for validation was used for testing the four prediction methods. In testing each of the methods, two indices were computed from the validation sample (observed data) and predicted values. These two indices are the mean error ( M E ) and the root mean square error (RMSE). The results show that universal kriging with the slope gradient used as external drift was more precise than the other kriging methods. Further analysis based on the mean ranks of performance confirmed these results.
Amelioration du Krigeage d'une variable du sol par l'utilisation de I'intensitC de la pente comme dCrive externe
RCsumCUne estimation par krigeage universel avec un attribut topographique utilisC comme dCrive externe est comparde aux estimations rCalisCes par krigeage ordinaire, krigeage ordinaire avec dCrive externe et krigeage universel. L'exemple porte sur l'estimation de 1'Cpaisseur du limon recouvrant partiellement la rCgion de Beauce au centre de la France. L'Cpaisseur est mesurCe en 62 sites auxquels sont ajoutCs 25 sites pour la validation. Un Modkle NumCrique de Terrain (MNT) est Ctabli k partir du relev6 altimCtrique de 484 sites. L'inclinaison de la pente est significativement corrClCe h 1'Cpaisseur du limon. Elle est utilisCe comme une dCrive externe dans I'opCration de krigeage. L'Cchantillon de validation est utilisC pour comparer les quatre mCthodes. Pour chacune d'entre elles, deux indices sont calculCs pour comparer les valeurs mesurCes aux valeurs predites. Ces indices sont I'erreur moyenne et la racine carrCe de I'erreur quadratique moyenne. Les rCsultats montrent que le krigeage universel avec I'inclinaison de la pente utilisCe comme dtrive externe est plus prCcise que les autres mCthodes de krigeage. Une analyse complCmentaire de la performance des mCthodes k I'aide du calcul du rang moyen confirme les rksultats.
Dune pattern, grain-size gradients and geochemistry were used to investigate the sources and dynamics of aeolian deposition during the last glacial in southwest France. The coversands form widespread fields of low-amplitude ridges (zibars), whereas Younger Dryas parabolic dunes mainly concentrate in corridors and along rivers. Spatial modelling of grain-size gradients combined with geochemical analysis points to a genetic relationship between coversands and loess, the latter resulting primarily from dust produced by aeolian abrasion of the coversands. The alluvium of the Garonne river provided also significant amounts of dust at a more local scale. The geochemical composition of loess shows much lower scattering than that of coversands, due to stronger homogenisation during transport in the atmosphere. Overall, sandy loess and loess deposits decrease in thickness away from the coversands. Dune orientation and grain-size gradients suggest that the efficient winds blew respectively from the W to the NW during the glacial, and the W-SW during the Younger Dryas. A comparison between the wind directions derived from the proxy data and those provided by palaeoclimatic simulations suggests a change of the main transport season. Ground surface conditions and their evolution throughout the year, i.e. the length of the season with snow and frozen or moist topsoil, and the seasonal distribution of wind speeds able to cause deflation are thought to have been the main factors that controlled the transport season in the study area.
<p style="text-align: justify;"><strong>Aims</strong>: This paper presents a study of spatial and temporal variations in solar radiation for the Bordeaux winegrowing region, over a 20 year period (1986-2005).</p><p style="text-align: justify;"><strong>Methods and results</strong>: Solar radiation data was retrieved from the HelioClim-1 database, elaborated from Meteosat satellite images, using the Heliosat-2 algorithm. Daily data was interpolated using ordinary kriging to produce horizontal solar radiation maps at a 500 m resolution. Then using a digital elevation model, 50 m resolution daily solar radiation maps with terrain integration were produced for the period 2001-2005. The long term (20 year) analysis of solar radiation at low spatial resolution (500 m) showed a west to east decreasing gradient within the Bordeaux winegrowing region. Mean August-to-September daily irradiation values, on horizontal surface, were used to classify the Bordeaux winegrowing region into three zones: low, medium, and high solar radiation areas. This initial zoning was downscaled to 50 m resolution, applying a local correction ratio, based on 2001-2005 solar radiation from the inclined surface analysis. Grapevine development and maturation potential of the different zones of appellation of origin of Bordeaux winegrowing regions are discussed in relation with this zoning.</p><p style="text-align: justify;"><strong>Conclusion</strong>: Solar radiation variability within the Bordeaux winegrowing region is mainly governed by terrain slopes and orientations, which induce considerable variations within the eastern part of Bordeaux vineyards. Significance and impact of study: Solar radiation has a major impact on vineyard water balance, grapevine development and berry ripening. However, irradiation data is seldom available in weather stations records. This paper highlights the need for high resolution mapping of solar radiation that uses remote sensing and terrain effect integration for agroclimatic studies in viticulture.</p>
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.