AZOFERT : A NEW DECISION SUPPORT TOOL FOR FERTILISER N ADVICE BASED ON A DYNAMIC VERSION OF THE PREDICTIVE BALANCE SHEET METHOD

Transcription

1 AZOFERT : A NEW DECISION SUPPORT TOOL FOR FERTILISER N ADVICE BASED ON A DYNAMIC VERSION OF THE PREDICTIVE BALANCE SHEET METHOD J. M. Machet 1, P Dubrulle 1, N Damay 2, R Duval 3, S. Recous 1, B. Mary 1 1 INRA, Unité d Agronomie LaonReimsMons, Rue Fernand Christ, 27 Laon Cedex, France, unite.lrm@laon.inra.fr, 33.() , 33.() Laboratoire Départemental d Analyse et de Recherche, Rue Fernand Christ, 27 Laon Cedex, France, ldar@cg2.fr, 33.() , 33.() Institut Technique de la Betterave, 45 Rue de Naples, 758 Paris, France, itb@itbfr.org, 33.() , 33.() Abstract AzoFert is a new software for N recommendations of main field crops, integrating the results of research done over the past fifteen years on the dynamics of organic matter in soils and on the fate of fertiliser N. AzoFert is based on a complete inorganic N balance sheet. At the opening of the balance sheet (end of winter for winter crops, at sowing for spring crops), the soil inorganic N pool is measured at the rooting depth. In order to take into account the various contributions of crop residues, catch crops and organic products previously applied to the residual mineral N (varying with climate and characteristics of added organic matter), the decomposition of the different organic sources are simulated (using observed climatic data) from harvest of the previous crop, until the opening of the balance sheet. Decomposition is expressed over time using a normalised time, based on temperature and soil moisture functions. From the opening of the balance sheet to the harvest of the crop, the subsequent net contribution of the crop residues or organic products and the net mineralisation of the humified organic matter are simulated using normalised days calculated from the past years mean climatic data of the area. The model takes into account nitrate leaching and includes new functions as volatilisation of ammonia and microbial immobilisation at the expense of fertiliser N. Data processing of the software allows users to easily modify the parameters in order to adapt them to a large range of pedoclimates and cultural situations. The present version gives N advice for 4 annual crops (cereals, industrial and vegetable crops). A validation work, carried out by the French Technical Institute for Sugar Beet, presents N recommendations accuracy. Introduction The increasing demand for high quality crops (protein content of cereals, technological quality of sugar beet, nitrate rate of vegetables) and protection of the environment (minimising nitrate leaching and gaseous losses) on the one hand, the evolution of agricultural practises with increasing and diversification of organic applications on the other hand, require an adaptation of reasoning and a rigorous management of the N fertilisation, as well as an evaluation of environmental impacts. Until now, a static predictive balance sheet method for calculating fertilisern rates applied to annual crops (Meynard et al., 1997) was the basis of the Azobil model (Machet et al., 19), the software most commonly used in France by soil laboratories and advisors. The results of research done over the past fifteen years on the dynamics of organic matter in soils and on the fate of fertiliser N led us to develop a dynamic approach of the balance sheet method. AzoFert is a new software for N recommendations of main field crops.

2 AzoFert description : the new agronomical aspects AzoFert is based on a complete inorganic N balance sheet. The following equation is used to predict fertilisern rates, expressing that the variation of soil inorganic N between opening and close of balance sheet equals the difference between N inputs and outputs : Rf Ri = (Mn + X + Ap + Fns + Fs + Ir) (Pf Pi + Ix + Gx + Lx + Gs + Ls) With Mn = Mh + Mr + Ma + Mci + Mp Rf : soil inorganic N at close of balance sheet (at harvest), Ri : soil inorganic N at opening of balance sheet (end of winter for winter crops, sowing for spring crops), Mn : net mineralisation from humus (Mh), crop residues (Mr), organic products (Ma), catch crops (Mci) and meadow (Mp) residues, X : amount of fertiliser N, Ap : N wet deposition, Fns : non symbiotic fixation, Fs : symbiotic fixation, Ir : N irrigation contribution, Pf : total N uptake by crop at close of balance sheet, Pi : N uptake by crop at opening of balance sheet, Ix : fertiliser N immobilised, Gx : fertiliser N lost as gas, Lx : fertiliser N lost by leaching, Gs : soil inorganic N lost as gas, Ls : soil inorganic N lost by leaching between opening and close of balance sheet It is assumed that Fns compensate Gs and that Lx is close from zero. AzoFert integrates a dynamic simulation of soil N supplies. At the opening of the balance sheet (end of winter for winter crops, at sowing for spring crops), the soil inorganic N pool is measured at the rooting depth. In order to take into account the various contributions of crop residues, catch crops and organic products previously applied to the residual mineral N (varying with climate and characteristics of added organic matter), the decomposition of the different organic sources are simulated (using observed climatic data) from harvest of the previous crop, until the opening of the balance sheet. Decomposition is expressed over time using a normalised time, based on temperature (T) and soil moisture (W) functions : Normalised time (day i, day j) = Σ ij f(t) * g(w) Normalised time takes into account climatic variations and determines a potential rate of decomposition. From the opening of the balance sheet to the harvest of the crop, the subsequent net contribution of the organic residues or wastes and the net mineralisation of the humified organic matter are simulated using normalised days calculated from the past years mean climatic data of the area (Figure 1). Soil inorganic N measurement Sowing Crop residues management, covercrops, organic manure Previous crop harvest Harvesting FertiliserN applications Figure 1. Different dates and cultural techniques are taken into account for the dynamic approach applied in Azofert AzoFert is composed of a new module for evaluating N mineralisation from soil and organic matters. Net N mineralisation in the soil is the sum of humified organic matter

3 mineralisation and contribution of different organic sources. The humified organic matter is mineralised in the upper layers. The mineralisation rate of these layers is function of a potential rate depending on the humified organic nitrogen pool and soil texture (clay and limestone contents) and of temperature and moisture soil conditions (Mary et al., 1999). The mineralisation rate also integrates the effects of cultural techniques (applications and types of organic products, frequency and species of catch crops, no tillage). The decay of crop residues and organic amendments in the soil results in net mineralisation or net immobilisation of soil nitrogen. Each crop residue and organic product is characterised by a specific kinetic curve of decomposition according to N and C. The decay rate of these products depends on the nature of organic residues (chemical characteristics and C:N ratio) and temperature and moisture soil conditions (Nicolardot et al., 21). AzoFert takes into account processes affecting the availability of fertilisern. The model includes new functions as volatilisation of ammonia and microbial immobilisation at the expense of fertiliser N. Tracers from fertiliser experiments using 15 N for different crops in the temperate climate of Northwest Europe showed that immobilisation of N by the soil heterotrophic microflora and gaseous losses compete with plant uptake for fertilisern (Recous et al., 1997). Volatilisation of ammonia, estimated by a simple model expressing the effects of soil factors (ph, cationic exchange capacity), fertiliser N form (physical and chemical) and application method (at soil surface or into the soil), and crop status at date of fertiliser N application. The amount of fertiliser N immobilised is calculated for the upper cm layer from the availability of C (calculated by the model at any time as C coming from rhizodeposition, crop residues and organic amendments), and an immobilisation ratio (immobilised N:decomposed C) that varies with N availability (Mary et al., 1996). The present version gives N advice for 4 annual crops : cereals (winter and spring cereals, maize), rape, industrial crops (sugar beet, flax) and field vegetable crops (potatoes, carrots, onions, beans, chicory, spinach, ), for which the N requirements and the cycle of vegetation are known. For certain species as potatoes, new functions integrating the length of the crop cycle and production objective led to evaluate N requirements differently from constant value (Chambenoit et al., 22). AzoFert description : computer characteristics and functioning AzoFert software developed in Visual Basic operates under Windows 9x and XP environments. Users can easily modify the parameters in order to adapt them to a large range of situations and pedoclimates with a Graphical User Interface managing different catalogues (soils, crops, organic amendments, crop residues, etc) and grids. Therefore, users become responsible for the N fertiliser advice given to the farmers. The software is designed to be easily integrated into the data management system of a laboratory (a LIMS for example) by using input/output files. The user s system constitutes the input file for Azofert and reads the output file from Azofert in order to publish a report of results, including interpretation and fertiliser N recommendations. For that reason, the laboratory has to create an interface between the Laboratory Information Management System (LIMS) and Azofert software. So the Laboratoire Départemental d'analyses et de Recherche (LDAR) made Azolims, a system of management for the user. It is interfaced with the Software package Solution Laboratory and the Module interpretation of Azofert. Azolims has different functionalities : Agronomical data capture thanks to a dataprocessing form, with several data entry screens,

4 Tests of coherence of the data thanks to an expert system, and storing these data in the specific nitrogen data base, Importation of data of the LIMS towards the specific nitrogen data base, like the administrative data concerning the customers or the results of analyses of the laboratory, Constitution of the input file necessary so that the Azofert software can run, Importation of the output file of Azofert in the specific nitrogen data base Impression of the report of results and interpretation is managed by Azolims, Advanced management by Azolims (choice of the model of report for example) Input data The input data allowing the constitution of the input file come either from laboratory results or from an information sheet filled by the farmer. For the soil, the content of true clay, the content of sands, the content of total limestone, organic carbon, total nitrogen, the ph of the surface horizon, the percentage of stones, the depth of ploughing, the depth of rooting of the culture are given. The farmer is asked if it is accustomed to plowing, how he manages crop residues, which are his practices about organic manure and the establishment of catch crops and finally if there was or not a meadow before. About the previous crop, the nature of the culture, its yield, the quantity of Nfertiliser applied, the date of harvest, the managing of crop residues (incorporated, exported or left on the ground) and the date of incorporation of these residues are indicated If there is an organic fertilisation, the nature of organic material, the quantity brought and date of spreading are required. If there is an analysis of the product, the percentages of carbon, total nitrogen and mineral nitrogen are also indicated. For a catch crop, the nature of the intermediate culture, the date of establishment and the date of destruction are indicated. About the crop to be fertilised, the nature of the culture, the date of establishment, the date of harvest, the type of manure brought and the mode of spreading, information on the irrigation, the yields aimed for the cultures for which the needs are proportional to the yield, the stage of development for winter cereals and the quantity of nitrogen already absorbed by the rape are required. Output data The reports consist of 4 pages of A4 format. On the first page are indicated administrative information concerning the farmer, the results of analyses of the laboratory, in the form of table and of graph, as well as the level of nitrogen amount to be brought on the culture. On the second one, the informations provided by the farmer and also comments about these. The third presents the recommendations about nitrogen fertiliser rate for different yield objectives, timing to put the fertiliser in the field and comments about the interpretation made by Azofert are published. On the last page is printed the plan of manure estimated, which takes the various parts of the nitrogen balance equation used for the calculation of nitrogen fertiliser amount. There are also comments on the environmental impact of the fertilisation. Validation of AzoFert for sugar beet crop A validation work was carried out by the French Technical Institute for Sugar Beet (ITB) in order to test N recommendations. In France, more than 6 % of sugar beet fields are sampled at the end of winter for measurement of soil mineral nitrogen before calculation of a N dose by the predictive balance sheet method. Moreover, situations in which sugar beet is cultivated can be very different (soil type, various organic products, catch crop). The database used for the study has been built with 9 years of N trials driven by ITB from 19 to 1998, in the different french beet areas. Only trials for which the information was complete to use AzoFert software were taken into account. In all 18 field trials have been maintained. These were set in farms fields in order to get a good representation of various

5 crop conditions in all regions. Most of the trial sites were maintained several years (in different fields of the same farm). A synthetic description of the different situations is given in table 1. Table 1: Main characteristics of the trials in the ITB data base Région Trials number Soil type Irrigation Intercrop management Champagne 4 Chalky soil : 37 All trials without Bare soil : 21 Loamy soil : 3 irrigation Catch crop : 18 Picardie Normandie Beauce Burgundy 83 Loam: 37 Silt clay : 21 Sandy loam : 17 Sandy : 4 Calcareous loam: 4 51 Clay : 8 Silt clay : 33 Loam: 7 Sandy loam : 1 Calcareous clay : 2 All trials without irrigation irrigated : 34 climatic : 17 Bare soil : 59 Catch crop : 24 Bare soil : 48 Catch crop : 3 Previous crop Cereals : 37 Others : 3 Cereals : 74 Peas : 4 Flax : 4 Others : 1 Cereals : 48 Others : 3 Organic product Vinasses : 14 Manure : 3 Pig slurry : 2 None : 21 Manure: 1 Pig slurry : 5 Sewage sludge : 4 None : 8 manure: 8 vinasses : 5 Poultry manure : 2 None : 35 On each trial a response curve is carried out with 4 to 6 N doses (always including a control treatment kg N ha 1 ). The calculated dose (DC) (Azobil calculation at the time) is framed with 2 doses (DC+4 kg N ha 1 and DC4 kg N ha 1 ) and the device is, for a part of trials, completed by 2 other complementary doses. These were randomised microplots trials, with 4 replicates. The optimum dose has been established with the determination key used by ITB for N trials. This one corresponds to the lowest dose which reaches the maximal yield of sugar. Some specific rules, established from a mean response curve, were applied for cases in which 2 successive doses lead to little yield differences, or when the range of doses tested in the trial didn t correspond to the optimal zone. The N dose calculated by AzoFert is situated on the N response curve and compared to the optimum. The results have been sorted out according to the region where the trial was set. Three main regions have been considered, which contrast on climate and soil types : PicardieNormandie in the West and North of France (deep loamy soils, relatively wet oceanic climate, frequent applications of organic products), Champagne in the East (white chalky soils, semicontinental climate), BeauceBurgundy (clay and siltyclay soils, low rooting depth, need of irrigation). Since the trials had been set up for testing the previous software Azobil, it gives the opportunity to compare the 2 softwares : AzoFert and Azobil. In order to give a synthetic view of the results, the difference between the calculated dose and the optimal dose is evaluated for each trial. The differences were distributed into classes with a step of kg N ha 1. Figure 2 presents the distribution of trials in each class (percent) for the Beauce Burgundy region. The use of AzoFert leads to an improvement of the advice accuracy : more trials are in the target zone (target zone corresponds to the best situation, with a difference between calculated and optimum doses less than kg N ha 1 ), 57 % of trials are close to the optimum for AzoFert against 37 % for Azobil. The number of overfertilisation cases significantly decreases. A possible explanation for BeauceBurgundy is the taking into account of irrigation in soil N supplies in the new software. Figure 2 : Trials repartition for AzoFert and Azobil in BeauceBurgundy region (51 trials)

6 % of trials Différence ( Azofert dose optimal dose) (kg/ha) Différence ( Azobil dose optimal dose) (kg/ha) AzoFert constitutes a new decision support tool for fertiliser N advice based on a dynamic version of the predictive balance sheet method. The introduction of new terms, of a dynamic simulation of soil N supplies allow its application to a larger range of cultural situations and pedoclimates. The integration of real data characterising the climate, soil type and cultivation practices leads to a significant improvement of N recommendations accuracy at field scale. References Chambenoit, C., Laurent, F., Machet, J.M., & Scheurer, O. 22. Fertilisation azotée de la pomme de terre. Guide pratique (Eds AgroTransfert, Inra, Itcf/Itpt), 128 p. Dubrulle, P., Machet, J.M., N. & Damay, N. 23. Azofert : a new decision support tool for fertiliser N recommendations. Abstracts for the 12 th Nitrogen Workshop, 21 st 24 th September, Exeter, Devon, UK. Machet, J.M., Dubrulle, P., & Louis, P. 19. Azobil : a computer program for fertiliser N recommendations based on a predictive balance sheet method. Proc. of 1 st Congress of the European Society of Agronomy, S2 P21. Mary, B., Recous, S., Darwis, D., & Robin, D Interactions between decomposition of plant residues and nitrogen cycling in soil. Plant and Soil, 181, Mary, B., Beaudoin, N., Justes, E., & Machet, J.M Calculation of nitrogen mineralisation and leaching in fallow soil using a simple dynamic model. European Journal of Soil Science, 5, Meynard. J.M., Justes, E., Machet, J.M., & Recous, S Fertilisation azotée des cultures annuelles de plein champ. In : Maîtrise de l azote dans les agrosystèmes (eds. G. Lemaire and B. Nicolardot). Les colloques de l INRA, 83, Nicolardot, B., Recous, S., & Mary, B. 21. Simulation of C and N mineralisation during crop residue decomposition : A simple dynamic model based on the C:N ratio of the residues. Plant and Soil, 228, Recous, S., Loiseau, P., Machet, J.M., & Mary, B Transformations et devenir de l azote de l engrais sous cultures annuelles et sous prairies. In : Maîtrise de l azote dans les agrosystèmes (eds. G. Lemaire and B. Nicolardot). Les colloques de l INRA, 83, 15.