THE DEVELOPMENT OF A "CROP COMPONENT" BY THE SEMoLa SIMULATION ENVIRONMENT F.Danuso, M.Zuliani, S.Barbieri DISAA – University of Udine, Via delle Scienze 208, 33100 Udine, Italy
[email protected] Traditional agronomic experimental techniques and empirical knowledge on agroecosystems are not sufficient to fill out the complex matrix of indicators that decision makers require to manage the whole modern agricultural system, both at the administrative and farm level (Jones et al., 2003). To fill this gap many software tools have been developed starting from early eighties. These applications, mainly based on approaches driven by specific problems, encounter some difficulties both when models had to be improved and used within larger simulation frames. The SIPEAA project (Donatelli et al., 2002) aims to build software components to simulate the different farming aspects (biological, physical and economical); these component are thought to be integrated in a software tool for planning and scenario analysis. This should have a multilevel structure where every single component is an independent, updateable, specific and totally replaceable problem solver (Donatelli et al., 2004). In this context the simulation of the cropping system plays a central role and has been the target of many efforts in the last two decades (Williams et al., 1989; Boote et al., 1998; Jones et al., 2001; Brisson et al., 2003; Stockle et al., 2003). The advancing in models capability to describe in a better details cropping, or even more complex, systems seems to be slacken, highlighting the need of more powerful concepts, tools and simulation environments. The introduction of the OOP paradigm in agroecological modelling (f.i, ModCom) and the many efforts in creating simulation environments (e.g., ACSL, MatLab/Simulink, ModelMaker, SEMoLa , Simile, Stella, VenSim, VisSim, etc.) demonstrate the importance of this topic, seen as a real bottleneck for the improvement of agroecological models. An important issue is also the need to create and to maintain an updated model documentation. In this paper we describe the use of the SEMoLa simulation environment (Danuso, 2003) to generate, from the same declarative source code in the SEMoLa meta- language, both the crop module of CSS (Cropping System Simulator; Danuso et al., 1999) and the CSS_Crop component (a generic crop simulator in the DLL format) for the SIPEAA project. Methods CSS_Crop is a module that simulates crop growth dynamics and phenology to be used in a decision support system for farming activities. The general structure of the crop subcomponent CSS_Crop (in figure, a UML diagram that illustrates elements and relationships) contains two modularity levels. CSS_Crop, in fact, is a module for a system requiring a crop simulation process and it is composed, on its turn, by two groups of replaceable elements with specific tasks. CSS_Crop calculates the modifications rates of: i) yield, aerial and root system biomass growth, ii) leaf area and iii) crop phenology. Current amounts of biomass, LAI and thermal accumulation (driving the phenological development) are calculated by an external integrator sited in the SIPEAA System and using rates exposed by CSS_Crop. Input requirements for CSS_Crop are specifiers of management events and state and rate variables from the system (e.g. soil water and nitrogen content, mean air temperature, reference evapotranspiration, solar radiation). Since no integration takes place in the component, CSS_Crop is a single time step model, which requires updated state values as input. CSS_Crop contains a static element (DB_crop) where parameters specific for the modelling approach are stored. Parameters are editable and modifiable using specific methods.
Documentation about ranges, type, units and definitions of variables used in CSS_Crop are retrievable from the dCrop component and are directly and automatically generated from the SEMoLa code, according to the meta-data structure defined in the project. The Crop_phenology routine simulates the crop development based on thermal accumulation. Photoperiod and vernalization are also accounted for in the model as modifiers of the thermal time accumulated. At the end of calculation step, crop stage and daily increment of thermal accumulation are exposed. Crop_biomass routine calculates rate variations of biomass accumulation, in the different crop structures (leaves, stem, storage organs and roots). Three methods are implemented to estimate canopy growth: the minimum between transpiration and radiation dependent growth (Stockle et al., ibidem), the radiation-dependent growth approach of SUCROS (Supit et al., 1994), and the simpler relative growth rate approach. User can set the modelling method at simulation start. Temperature, water and nitrogen stresses reducing potentia l crop growth are also included. Crop_biomass calculates and exposes leaf area index (LAI) as well; LAI is calculated using current leaves biomass amount. Sub-components are built as DLL using the SEMoLa meta- language designed to integrate knowledge (simulation models, fuzzy logic expert systems, neural networks, regression models) into deliverable applications. In this environment, mainly by means of a nonprocedural language, implementation and system analysis concepts overlap. Hence SEMoLa could be an easy to use modelling tool that do not requires advanced computer programming skills. Moreover, this platform can be seen as a tool for software modules development and maintenance. Such modules can be combined into more complex software systems. Conclusions SIPEAA Project adopts a new planning pattern in the traditional working behavior of modellers. A right balance between previous skills and new paradigms became a must, trying to avoid legacy or dogmatic approaches. In this context the usual modelling practice of our Department, based on SEMoLa environment (dimensional check, sensitivity analysis, calibration, automatic documentation, etc.) was exploited as a tool for an easy packaging of knowledge also into SIPEAA project. At present, improvements to the CSS_Crop module are made in the SEMoLa code and then directly transferred to the CSS_Crop DLL. References Brisson N. et al, 2003, Eur. Jou. Agr, 18:309-332. Boote K.J, et al.,1998.. In: Peart R.M., Curry R.B. (Eds.), Agric. Syst. Modeling and Simul. Dekker, New York, pp. 651-692. Danuso F. et al., 1999. Proc. Int. Symp."Modelling cropping systems", ESA, Lleida, 21-23 June, 1999, Catalonia, Spain, 287-288 (http://www.dpvta.uniud.it/~Danuso/docs/CSS/CSS_Home.html) Danuso F., 2003. Proc. XXXV SIA Congress, Napoli, Italy, 16-19 September 2003, 283-284 (http://www.dpvta.uniud.it/~Danuso/docs/Semola/homep.htm). Donatelli M. et al. 2002 Proc. of the VII ESA Congress, Cordoba, Spain, 271-272 Donatelli M et al. 2004 These Proceedings Jones J.W. et al., 2001. Agric. For. Meteorol. 70:421–443. Jones J.W. et al., 2003,. Eur. Jou. Agr., 18 (3-4): 235-265 Stöckle, C. O. et al, 2003, Eur. Jou. Agr. , 18: 289-307. Supit I., Hooijer A.A., Van Diepen C.A., 1994. JRC European Commission, Brussels. Williams, J. R. et al. 1989. The EPIC crop growth model. Trans. of the ASAE, 32(2): 497-511. Acknowledgements Research under the auspices of MiPAF, paper SIPEAA no. 20