Search

National audience

International audience

International audience; The cultivation of crops in a given area is highly dependent of climatic conditions. Assessment of how the climate is favorable is highly useful for planners, land managers, farmers and plant breeders who can propose and apply adaptation strategies to improve agricultural potentialities. The aim of this study was to develop an assessment method for crop-climate suitability that was generic enough to be applied to a wide range of issues and crops. The method proposed is based on agroclimatic indicators that are calculated over phenological periods (ecoclimatic indicators). These indicators are highly relevant since they provide accurate information about the effect of climate on particular plant processes and cultural practices that take place during specific phenological periods. Three case studies were performed in order to illustrate the potentialities of the method. They concern annual (maize and wheat) and perennial (grape) crops and focus on the study of climate suitability in terms of the following criteria: ecophysiological, days available to carry out cultural practices, and harvest quality. The analysis of the results revealed both the advantages and limitations of the method. The method is general and flexible enough to be applied to a wide range of issues even if an expert assessment is initially needed to build the analysis framework.The limited number of input data makes it possible to use it to explore future possibilities for agriculture in many areas. The access to intermediate information through elementary ecoclimatic indicators allows users to propose targeted adaptations when climate suitability is not satisfactory.

International audience; Phenology is a bio-indicator of climate evolutions. Measurements of phenological stages on perennial species provide actually significant illustrations and assessments of the impact of climate change. Phenology is also one of the main key characteristics of the capacity of adaptation of perennial species, generating questions about their consequences on plant growth and development or on fruit quality. Predicting phenology evolution and adaptative capacities of perennial species needs to override three main methodological limitations: 1) existing observations and associated databases are scattered and sometimes incomplete, rendering difficult the implementation of multi-site study of genotype-environment interaction analyses; 2) there are no common protocols to observe phenological stages; 3) access to generic phenological models platforms is still very limited. In this context, the PERPHECLIM project, which is funded by the Adapting Agriculture and Forestry to Climate Change Meta-Program (ACCAF) from INRA, has the objective to develop the necessary infrastructure at INRA level (observatories, information system, modeling tools) to enable partners to study the phenology of various perennial species (grapevine, fruit trees and forest trees). Currently the PERPHECLIM project involves 27 research units in France. The main activities currently developed are: defining protocols and observation forms to observe phenology for various species of interest for the project; organizing observation training; developing generic modeling solutions to simulate phenology; supporting the building of research projects at national and international level; developing environment/genotype observation networks for fruit trees species; developing an information system managing data and documentation concerning phenology. Finally, PERPHECLIM aims to build strong collaborations with public (Observatoire des Saisons) and private sector partners (technical institutes) in order to allow a more direct transfer of knowledge.

In the Provence Alpes Cote d'Azur region (France), overproduction of apple (Malus domestica Borkh.) can result in a significant surplus (0-230 000 Mg per year between 1986 and 1995) which has to be used or destroyed. Among the possibilities for disposing of overproduction, land application is flexible and inexpensive. A field study was undertaken to identify the environmental risks and main biogeochemical processes involved when apple waste is applied to farm land. Three experimental plots receiving 0, 500, or 1000 Mg of apple ha -1 were studied. Soil solutions were collected at three depths (0-0.3, 0.3-0.6, and 0.6-0.9 m) at different times, pH and Eh were measured, major ions, metals, and total and carbonated alkalinity were analyzed. Rapid pH and Eh decrease (from 7.0-5.5 for the pH and from +500 to -300 mV for the redox potential) were observed. The soil became reduced, thus causing the dissolution of Mn. Nitrogen immobilization and denitrification resulting from waste application explained NO - 3 depletion in the soil. The formation of organic anions complexed the metals and major cations. Finally, the oxidation of the soil during drying created conditions which polluted less, emphasizing the importance of the choice of season for land application of organic waste.

The ALiBi model, a Soil-Vegetation-Atmosphere Transfer model, was developed to simulate mass and energy exchanges between vegetation canopies, the soil and the atmosphere. In the present work, it was used in conjunction with remote sensing data in thermal, solar and microwave domains to assess energy balance of various vegetation canopies (two soybean crops, one wheat crop and one fallow canopy). Inversions of radiative transfer formulations were performed to retrieve model parameters: -i) canopy structure parameters (leaf area index and canopy height) from red and near-infrared reflectances, when available; -ii) surface soil moisture from microwave measurements, when available and -iii) plant water transfer parameters from thermal infrared measurements. The estimates of energy balance fluxes were in good agreement with measured values, in particular for well-developed canopies. It can be noticed that it was not necessary to know accurately surface soil moisture and leaf area index to assess energy balance fluxes. Conversely, the vegetation height must be known with a good accuracy.

This work investigates the possibility to retrieve the root zone soil moisture from the combined use of time series of surface soil moisture, that can be estimated from microwave remote sensing instruments, and Soil-Vegetation-Atmosphere-Transfer (SVAT) modeling. The analysis is based on the ISBA surface scheme and two data sets acquired over' soybean crops, in 1989 and 1990. They include very detailed measurements of soil and vegetation characteristics, mass and energy transfers in the soil-plant-atmosphere system and crop remote sensing signatures in the thermal and microwave domains during a 2- or 3-month period. The 3-month experiment in 1989, made it possible to investigate the accuracy of the soil reservoir retrievals, as a function of the time period and repetitivity of the surface measurements included in the retrieval process. This work contributes to a better definition of the requirements for the use of remotely-sensed microwave observations of surface soil moisture. This is a crucial problem, if we consider the increasing potential of microwave remote sensing technology and the fundamental needs in atmospheric and hydrological models for water transfer characterization in soil.

National audience

A non-parametric estimation method is proposed for the intensity of the Poisson process associated to a Boolean random function. The case studied is that of marks being half-spheres as primary functions, these half-spheres lying on a given horizontal plane. The proposed method is derived from the method presented in Gyorfi, Hardle, Sarda et Vieu [1989] for the case of time series. An application in soil science is presented.

International audience; The vertical distributions of heterotrophs, NO 3--reducers and denitrifiers, specific respiratory and denitrifying activities were studied in the upper vadose zone of two profiles, distinguished from each other by the risk of transitory anaerobic conditions. According to pedological observations, 3 and 4 layers were defined in the fluvic hypercalcaric cambisol and haplic calcisol profiles, respectively. Bacterial enumerations were performed using the Most Probable Number (MPN) method. Aerobic respiratory and denitrifying activities were estimated from the depletion of O 2 and the accumulation of N 2O in the presence of C 2H 2, respectively, for soil in sealed flasks incubated in optimal conditions. The denitrifier - to - heterotroph ratio indicates that the denitrifying ability favors micro-organisms in the competition for deep layer colonization in the poorly aerated site. Whereas specific aerobic respiration does not vary or increase with depth, the specific denitrifying activity decreases with depth and is mostly affected by organic-C in the poorly aerated site. The use of specific denitrifying activity should be accounted for in models of denitrification; Dénitrification dans la zone non saturée superficielle : comparaison entre un fluvic hypercalcaric cambisol et un haplic calcisol . Dans cette étude les distributions verticales des micro-organismes hétérotrophes, des réducteurs de nitrate et des dénitrifiants, ainsi que de leurs activités spécifiques respiratoires et dénitrifiantes, ont été comparées dans la zone non saturée superficielle de deux sites se distinguant par le risque de conditions anoxiques transitoires. Suite à une caractérisation pédologique, 3 et 4 horizons ont été distingués respectivement aux niveaux d'un fluvic hypercalcaric cambisol et d'un haplic calcisol. Les nombres de bactéries ont été estimés par la méthode du nombre le plus probable (MPN). Les activités respiratoire aérobie et dénitrifiante ont été caractérisées par le suivi respectif de la disparition de O 2 ou de l'accumulation de N 2O en présence de C 2H 2, pour du sol incubé en conditions optimales à l'intérieur de flacons scellés. Le rapport dénitrifiants/hétérotrophes indique que l'aptitude à dénitrifier donne un avantage dans la compétition pour coloniser les horizons profonds du site le moins bien aéré. Alors que l'activité respiratoire aérobie spécifique semble constante ou s'accroître avec la profondeur, l'activité spécifique dénitrifiante décroît. Elle est affectée par le C-organique plus faible dans le site moins aéré. L'introduction de l'activité spécifique dénitrifiante dans les modèles de dénitrification pourrait améliorer les prédictions de concentration en NO 3- arrivant à la nappe.

A simple two-term model of the radar backscattering coefficient of crops, designed for the retrieval of the amount of water in the canopy, is described and analyzed. The principle of the method is to calibrate the simple model from the simulations of a discrete first-order radiative transfer model during crop development. The canopy structure is taken into account in the discrete model to compute the relationships between a) the vegetation direct contribution to backscattering σ ° v and the optical depth τ and b) the optical depth τ and the amount of water in the canopy. The two-term model is tested against C-band radar data acquired over a soybean crop during the whole vegetation cycle. The simulations correlate well with the measurements and the retrieval of the amount of water in the canopy Wc (kg/m 2 ) can be carried out. Accurate temporal information on the crop growth could be derived from the radar data. Ancillary information about soil moisture are required, but it is found that rough estimates on a 4–5 day basis are sufficient.

We studied the possibility of estimating root zone soil moisture through the combined use of a time series of observed surface soil moisture data and soil-vegetation-atmosphere transfer modeling. The analysis was based on the interactions between soil- biosphere-atmosphere surface scheme and two data sets obtained from soybean crops in 1989 and 1990. These data sets included detailed measurements of soil and vegetation characteristics and mass and energy transfer in the soil-plant-atmosphere system. The data measured during the 3-month experiment in 1989 are used to investigate the accuracy of soil reservoir retrievals, as a function of the time period and frequency of measurements of surface soil moisture involved in the retrieval process. This study contributes to better defining the requirements for the use of remotely sensed microwave measurements of surface soil moisture.

Soil Vegetation Atmosphere Transfer (SVAT) models have been implemented to simulate energy and mass fluxes between soil, vegetation and atmosphere of various ecosystems. Usually, these models are simple, but they use realistic descriptions of radiative, turbulent and water transfers. These include description of stomatal control of transpiration and CO/sub 2/ fluxes. They can be used for assimilating remote sensing data and derive vegetation canopy evapotranspiration or photosynthesis. Various remote sensing data may provide useful information to drive SVAT models. Surface temperature may be used through inversion procedures to retrieve parameters related to stomatal conductance or root zone soil moisture. Parameters related to vegetation structure (LAI, vegetation height) may be retrieved from reflectance measurements in the solar domain, either through direct relationships with some vegetation index or by inverting radiative transfer formulation against spectral reflectance measurements. The microwave data contribution has not been studied very often in the case of vegetation canopies, but they were proposed for estimating surface soil moisture. In this paper, inversions of the ALiBi model were performed to retrieve canopy evapotranspiration from thermal infrared, spectral reflectances and microwave data on two water stressed soybean crops. In a previous study, thermal infrared data alone were used to invert the model on plant water status parameters, while other parameters, related to canopy structure and soil surface water status, were prescribed from in situ measurements. In the present study, spectral reflectance and radar measurements were used to retrieve canopy structure parameters (LAI and vegetation height) and surface soil moisture by inverting radiative transfer models.

The water consumption of grapevines is one of the main factors influencing wine quality. The general characteristics of the vines (deep and sparse rooting system, heterogeneity of canopy size) and of vineyards (sloping ground, stony soils, small fields) make it difficult to estimate the actual evapotranspiration of the vineyard using the soil water balance or energy balance methods. In this paper these two methods are compared to a method relying on separate measurements of soil evaporation and plant transpiration. An experiment was conducted for two months in 1994 in a one-hectare vineyard of Southern France in which five monitoring plots had been installed. In each plot, soil moisture was monitored down to 260 cm under the row and between the rows to take into account the heterogeneity of the rooting system. Water flux at 260 cm depth was computed from measurements of soil water potential and hydraulic properties. Evapotranspiration was then obtained from the soil water balance. Transpiration was monitored with two sap flow gauges per plot and evaporation with three mini-lysimeters installed at different distances from the row. The energy balance was determined with micrometeorological measurements taken in the field. The reliability of the measurements of evaporation from the soil and of plant transpiration was satisfactory. First, soil evaporation computed from mini-lysimeter measurements compared well with that calculated from the water balance of the soil surface. Second, the sap flow followed the diurnal pattern of net radiation at a time scale of less than one hour. The three methods for estimating actual evapotranspiration gave comparable mean field values and showed a similar evolution through the summer. Statistical tests showed that the observed differences between the estimated values were not significant, whatever the length of the period considered (3–31 days) and the rate of evapotranspiration (1.8–3.5 mm d −1 ). The experiment shows the inherent limits and advantages of the three methods. The soil water balance method must be used over periods of longer than one week to provide acceptable precision in estimating field evapotranspiration. The energy balance method gives small time scale information but requires constant technical follow-up. Measuring evaporation and transpiration separately to estimate vineyard evapotranspiration is easily adaptable to a large range of vine-growing conditions. Moreover, it proved to be more precise than the other two methods.

Ary Bruand : Present Address : Institut des Sciences de la Terre d'Orléans; International audience; This study discusses the use of a deterministic model, at a regional scale, when the soils have small heterogeneity. Long-term soil drainage was simulated on 3500 ha in Beauce (France), taking into account variation in soil thickness, agricultural practices, and climate characteristics for a 32-yr period. The model was based on a numerical solution of Richards' equation, using a finite element method. Hydraulic properties were determined at four sites that are 4 km apart. The results show a small variability of hydraulic parameters, with coefficient of variation (CV) ranges from 1 to 16% for the α parameter, the empirical constant n, the saturated water content θs, and the residual water content θr, and from 60 to 75% for the saturated hydraulic conductivity Ks. Hence, the spatial and temporal variability of drainage are mainly related to soil thickness, agricultural practices, and climate characteristics. Assigning small spatial variability to hydraulic properties, we analyzed the role of soil thickness and initial water content (IWC), induced by agricultural practices and climate characteristics, on drainage prediction. Five IWC were defined: 100, 90, 80, 70, and 55% of field capacity (FC). Four soil thicknesses were studied: 30, 60, 100, and 150 cm. The IWC appeared as a significant factor in drainage prediction for 60% of the 32 yr, regardless of soil thickness. Soil thickness became the most significant factor (20% of the 32 yr) for low IWC. The model was spatially extrapolated by combining the calculated drainage (which varied with soil thickness and IWC) and the spatial distribution of crops and soil thickness classes. Simulated drainage was closely related (r = 0.90) to groundwater recharge during the 32 yr, indicating that the model could be applied to other areas with appropriate parameters.

Solving soil unsaturated flow problems requires knowledge of the water retention θ(h) and unsaturated hydraulic conductivity K(θ) relationships. The purpose of this study was to investigate more thoroughly the properties and accuracy of an evaporation method described by Wind (1969) for determining θ(h) and K(θ) from laboratory cores. Evaporation from a vertical column of soil was first simulated using numerical solution of Richards equation for a given set of soil hydraulic properties. The simulated data were then used to evaluate the ability of Wind's method to provide estimations of the retention and unsaturated conductivity curves when measurement errors were taken into account. The main sources of error were (i) errors due to the position of the tensiometers in the sample, (ii) errors due to the calibration of the transducers used for the pressure head measurements, and (iii) errors due to layering in the soil column. The estimated water retention curves were sensitive only to soil layering. On the other hand, small uncertainties in the position of the tensiometers (1 or 2 mm) and in the calibration curve of the transducers for the pressure head measurements (1-5%) had a great influence on the estimated hydraulic conductivity curves. A correction procedure was proposed and was satisfactory when errors of position in the tensiometers were taken into account. Results also showed that temperature corrections related to viscosity of liquid water were large. Finally, this method gives poor estimates of hydraulic conductivities when raw tensiometric data are corrupted with small errors in their position or calibration.

Ary Bruand : Present Address : Institut des Sciences de la Terre d'Orléans; International audience; The volumetric water content of a silty clay loam soil was measured using the capacitance probe method (CPM) for three successive winter-spring periods. The results were compared with those obtained using the neutron scattering method (NSM). The CPM was calibrated using gravimetric water content measurements, whereas the NSM was calibrated using a theoretical calibration which requires chemical analysis of soil samples. The calibration curves of CPM were linear for volumetric water content ranging from 0.25 to 0.40 m3 m−3. The correlation coefficients (r) ranged from 0.87 to 0.98. The advantages of the CPM were discussed, and it was demonstrated that, for low differentiated soils, this method can be a better alternative to the NSM, which is commonly used for the determination of soil water content. The CPM and NSM measurements were compared and root mean squared errors between these two methods ranged from 0.0098 to 0.0159 m3 m−3. The precision of NSM and CPM in a field calibration exercise was calculated. For CPM, this study discusses unbiased and biased statistical treatments to establish the calibration curves. The unbiased treatment greatly decreased the total variance in volumetric water content. The standard deviation of volumetric water content ranged from 0.0194 to 0.0248 m3 m−3 and from 0.0108 to 0.0136 m3 m−3 for biased and unbiased statistical method, respectively. Finally, the soil water storage and associated errors were calculated using CPM and NSM.

This paper deals with the prediction of the soil water retention h(S) and the soil unsaturated hydraulic conductivity K(S) functions of a clay-loam soil at a field scale (1 ha) where the variable S represents water saturation. The Van Genuchten model and the corresponding Mualem-Van Genuchten model were used to predict h(S) and K(S) functions respectively. The field data (tensiometric and neutron probe measurements) used in this study were provided by the soil water balance (four neutron sites, 0.35 to 1.55 m soil layer) of a soybean crop over a 78 days growing season. The advantages of the scaling approach for describing the field variability of the h(S) function were confirmed. The scaling approach accounted for 73% of the field variability of the soil matrix potential. A simple procedure was proposed in order to predict the K(S) function using scaling theory. This was done by simultaneously applying a ``zero flux method'' and ``deep flux method'' to compute the soil water balance and fit the saturated hydraulic conductivitiy (Ksat), the only unknown parameter in K(S).

International audience

Supplement 2; International audience

When the soil water balance method is applied at a field scale, estimation of the spatial variability and confidence interval of actual evapotranspiration is rare, although this method is sensitive to the spatial variability of the soil, and thus to the sampling strategy. This work evaluated the effect of soil sampling strategies for soil water content and water flux at the bottom of the soil profile on the estimation of the daily and cumulative evapotranspirations. To do that, according to the statistical properties of daily measurements in a field experiment with a soybean crop, the water content and flux through the base to the soil profile in space (field scale) and time (daily scale) were simulated. Four different sampling strategies were then compared, and their effects on daily and seasonal cumulative evapotranspirations quantified. Strategy 1 used ten theoretical sites randomly located in the field. The daily water content estimates were assumed to be available each day from these same ten locations, which were located from 0.15 m to 1.55 m in depth, with space steps of 0.10 m. Strategy 2 assumed that daily water content estimates combined two sources: in the 0.00–0.20 m soil layer, ten theoretical sites were selected but changed every day, with thin soil layers for soil moisture sampling, from 1 to 5 cm in thickness. In the 0.20–1.60 m soil layer, the daily water content estimates were assumed to come from the same ten locations (the first soil moisture estimate was located at 0.25 m, and the others were located every 0.10 m until 1.55 m). Strategy 3 used ten theoretical sites located in the field, as in strategy 1, however the water content estimates in the 0.00–0.20-m soil layer were assumed to come from accurate water content measurements (soil layers from 1 to 5 cm in thickness), while for the 0.20–1.60 m soil layer, the strategy was similar to strategies 1 and 2. Strategy 4 used 10 new theoretical locations of measurement every day. Precise water content estimates for thin layers were assumed to be available in the 0.00–0.20 m soil layer as in strategy 2. The layers for water content estimates in the 0.20–1.60 m were similar to those of strategies 1, 2, and 3. Results showed that the spatial variability of the daily actual evapotranspiration may not be negligible, and differences from approximately ±1.0 mm d −1 to ±3.0 mm d −1 were calculated between the four sampling strategies. Strategy 1 gave the worst results, because variations in the water content of the top soil layers were neglected, and thus the daily evapotranspiration was underestimated. Strategy 2 led to a considerable variability for estimating daily evapotranspiration which was explained by the effect of the spatial variability due to the daily site sampling for the top soil layers (0 to 0.2 m). Strategy 3 appeared to be the best practical compromise between practical field considerations and the necessity to obtain accurate evapotranspiration measurements. The accuracy of daily evapotranspiration could reach ± 0.5 mm d−1, and could be further improved by increasing the number of measurement sites. The best results were obtained with strategy 4, although such a destructive and time-consuming strategy is not likely to be practical.

International audience

On teste en laboratoire et in situ les performances d’un capteur capacitif destiné à la mesure de l’humidité du sol. En laboratoire, on montre que les dérives thermiques propres à l’appareillage sont négligeables, que la mesure de permittivité diélectrique relative est dépendante de la qualité du contact sol-électrodes, et que le volume de mesure a des dimensions de l’ordre de quelques cm3. Les relations linéaires permittivité relative-teneur en eau volumique sont dépendantes de la texture, de la structure (agrégats de 2-3 mm et 4 mm disposés autour des électrodes ou structure continue), de la température (la variation de la permittivité relative en fonction de la température peut atteindre 0,20 par °C dans la gamme de 0 à 45 °C), de la salinité (la mesure de permittivité diélectrique relative est peu sensible à la concentration saline jusqu’à des valeurs de conductance électrique de l’ordre de 2 à 3 mS). In situ, deux expérimentations sont réalisées, l’une sur un sol nu présentant un état structural homogène, l’autre sur un sol couvert (soja) présentant un état structural plus hétérogène. Pour chaque expérimentation, on réalise pendant plusieurs jours (21 et 41 j respectivement) des mesures simultanées de permittivité diélectrique, de température, de conductance électrique, et des contrôles de teneur en eau dans les couches superficielles du sol (0 à 20 cm). Les relations d’étalonnage permittivité-humidité sont généralement linéaires et variables d’un capteur à l’autre. Cette variabilité est probablement due à la variabilité spatiale de la teneur en eau du sol, de la structure à proximité immédiate du capteur, et/ou du contact sol-électrodes. Par ailleurs, les effets thermiques sur la permittivité mesurée sont comparables à ceux obtenus en laboratoire, et la gamme de conductances électriques rencontrées in situ semble compatible avec celle ne présentant pas d’influence notable sur la mesure de permittivité et estimée en laboratoire., The performance of a capacitive probe for soil moisture measurement was studied under laboratory and field conditions. The probe (21-mm ED cylinder) was composed of 2 electrodes connected to an electronic oscillator (38 MHz) located in the probe. Four soils were used to analyse the ’dielectric permittivity-volumetric water content’ relationships under laboratory or field conditions. The sensitivity of the probe including its electronic components in relation to the variations of the temperature (0-45 °C) was very low (< 0.5 permittivity unit). Permittivity measurements were dependent on the quality of the soil-electrodes contact, and on the soil structure near the electrodes. The order of magnitude of the volume of influence was a few cubic centimeters, and the area of the soil most affected by the permissibility measurements was located in the vicinity of the electrodes. The ’dielectric permittivity-volumetric water content’ relationships were linear (r > 0.99), and dependent on soil texture and structure. Moreover, soil temperature influenced the measurements since increases of the order of magnitude of 0.20 permittivity unit per °C could be observed. Permittivity measurements were not sensitive to the concentration of the solution when the electric conductance was < 2-3 mS. Two field experiments were performed using 2 different soil textures. The first soil (C = 10.5%, L = 50.6%, S = 38.8%) was a bare soil, and showed a quite homogeneous soil structure, the second (C = 27.2%, L = 61.7%, S = 11.1%) was cultivated with soybean, and showed a more heterogeneous soil structure. For each experiment field, simultaneous measurements (1 or 2 times per d) of relative dielectric permittivity, soil temperature and gravimetric soil water content were made in the vicinity of the probes (over a 21 and 41 d period, for the 1st and 2nd experiment, respectively). The variation of the relative dielectric permittivity versus time was very similar to the variation of the measurement soil water content. The calibration relationships between the volumetric water content and the relative dielectric permittivity were linear for each capacitive probe. The regression lines for the different capacitive probes exhibited variations between probes, probably due to the spatial variability of the water content, soil structure and/or soil-electrode contact for each sensor. The effects of the soil temperature on the relative dielectric permittivity measurements were similar to the previous results obtained under laboratory conditions, and effects of electrical conductance appeared negligible under our conditions.

The use of crop simulation models on a large scale for agrometeorological purposes is often limited by their inputs being non-routinely collected data, especially with regard to their soil water balance compartment. The objective of this study is to develop a water balance sub-model which can be run with readily available inputs. The model predicts water use, soil evaporation and crop transpiration throughout the growing season. Physiological reduction factors, as influenced by water stress, are derived from the soil water availability. This is achieved by using empirical relationships such as the reservoir analogy to assess water availability in relation to root development. The framework of applicable conditions is assessed by sensitivity analyses performed on inputs: the model can be run with a time step of ten days and using soil information given by soil maps, i.e. soil texture and depth, which fit agrometeorological purposes. Moreover, the model is shown to describe realistically soil water depletion, crop evapotranspiration and rooting depth. However, wetting processes are not correctly simulated, especially when large amounts of water are supplied. This limitation is emphasized for ten day time steps. Therefore incorporating effective rainfall simulation, i.e. both runoff and rainfall interception by the canopy, would improve the model.

International audience

On se propose de calculer des estimateurs non biaisés de la régression linéaire dans le cas où les variables (x) et (y) de la régression, sont entachées d’erreurs. Dans une première partie, on présente les bases théoriques de la réduction du biais des estimateurs. Ces bases permettent d’aboutir à la construction d’estimateurs sans biais à l’ordre 1. De plus, on décrit la méthodologie permettant de quantifier rigoureusement l’ensemble des sources d’erreur à prendre en compte successivement pour les variables (x) et (y). Dans une seconde partie, ces résultats théoriques sont appliqués à deux instruments de mesure nécessitant un étalonnage préliminaire, un gamma densimètre à transmission (étalonné en laboratoire) et une sonde à neutrons (étalonnée in situ). Les résultats numériques montrent que les erreurs de mesure sont négligeables pour les variables (x) (taux de comptage) et (y) (masse volumique sèche) dans le cas du gamma densimètre à transmission (≤ 0,4 p. 100). Dans le cas de la sonde à neutrons, il en est de même pour la variable (x) (taux de comptage), mais les erreurs deviennent non négligeables (9 à 16 p. 100) pour la variable (y) (teneur en eau volumique). Lorsque les erreurs de mesure restent faibles (cas du gamma densimètre à transmission), il y a peu de différences entre les estimateurs de la droite d’étalonnage avec ou sans prise en compte des erreurs de mesure. Par contre, les différences deviennent plus sensibles lorsque les erreurs deviennent plus importantes (cas de l’humidimètre à neutrons), notamment en ce qui concerne la précision de l’estimation de valeurs prédites à partir de l’étalonnage., This paper proposes unbiased estimators of linear regression when the variables (x, y) present measurement errors. In the first part, theoretical calculations are described to obtain first-order unbiased estimators, and an error analysis is developed to take into account all the error components involved in each variable (x or y). In the second part, these theoretical results were applied to the calibration of a gamma-ray attenuation probe (laboratory calibration) and a neutron probe (field calibration). Numerical results showed that the measurement errors were negligible in the case of the gamma-ray attenuation equipement (≤ 0.4 %), but were greater (9 to 16 %) for the neutron probe, particularly for variable (y). When the measurement errors are small (gamma-ray attenuation probe), there is no difference between the parameters of the calibration estimated according to the classical approach or the proposed model. But when the measurement errors are greater (neutron probe), the differences between these two calibrations become higher especially for the precision of new predicted values.

International audience

National audience