WCHR Home   WCHR 1   WCHR 2   WCHR 3   WCHR 4   WCHR 5   Search  

MODELLING GREENHOUSE CROPS: STATE OF THE ART AND PERSPECTIVES

Christian Gary
INRA, Unité de Bioclimatologie
Domaine St-Paul, Site Agroparc - 84914 Avignon CEDEX St-Paul, Site Agroparc - 84914 Avignon CEDEX 9

An overview of the short history of greenhouse crop models

Crop modelling started in the early seventies with the elaboration of an integrated and dynamic vision of the whole plant physiology (de Wit, 1970) and with the generalisation of the use of computers in all scientific fields. It developed quickly as models appeared to be powerful scientific as well as engineering tools. Crop models offer a conceptual framework for the organisation of research: for their development, various persons or groups can mobilise their different skills in a co-operative project, and different levels of organisation can be co-ordinated. Models are also justified, generally by the same research teams, by their applicability for improving management of the system they describe. They usually provide quantitative information from which decisions, such as crop timing, irrigation, fertilisation, crop protection, and climate control, can be taken at the field scale. On a regional scale, policies can be evaluated from estimations of e.g. potential yield, water needs, or fertiliser losses.

In horticulture, a recent survey of the literature (Gary et al., 1998) shows that crop modelling significantly developed in the eighties. In greenhouse cultivation systems (for vegetable and ornamental production), it has been motivated by the need for quantitaton), it has been motivated by the need for quantitative information to improve decision making with the emerging computer tools that were designed to control the shoot and root environments. In these control systems, crop models can be used at the operational level to simulate the short-term crop processes that interact with the greenhouse climate (CO₂ and water vapour exchanges) and contribute to the daily crop growth (carbon balance). At the tactical level, models are needed to relate the general policy of climate control and crop management along the crop cycle to yield formation.

The major research teams in this field belong to regions of the world where the greenhouse industry is of economic importance and/or where crop modelling had already developed on other cultivation systems: the Netherlands, England, France, Israel, the USA, Germany, Sweeden, Canada... More recently, a modelling activity is emerging in regions where greenhouse cultivation is developing, like southern Europe (Portugal, Spain, Greece) and South America (Brazil, Colombia).

Prevailing methods and tools

In greenhouse crops, priority has clearly been given to the modelling of growth, development (Marcelis et al., 1998), and transpiration (Jones and Tardieu, 1998). Few models of nutrient absorption have been published (Le Bot et al., 1998). At present, the modelling of the interactions betw. At present, the modelling of the interactions between crops and pests and diseases, and the use of genetic parameters in crop models are still in their infancy. The plant architecture is often considered as an input to models of light interception and photosynthesis (e.g. Gijzen and Goudriaan, 1989) yet the modelling of morphogenesis would be of interest for ornamental crops.

Various modelling approaches have been used. Model are function oriented when the intention is to represent the key features of the crop behaviour without getting into mechanisms. Growth models based on the radiation use efficiency approach (i.e. converting directly intercepted light into biomass; Monteith, 1977), and models of nitrogen uptake based on the decrease in nitrogen content with biomass accumulation (Le Bot et al., 1997) belong to this group. Models are process oriented when they are based on a fairly detailed description of the production and distribution of assimilates and of the plant development. A significant effort has been made to develop such mechanistic models for a few greenhouse crop species, mainly tomato (Bertin and Heuvelink, 1994; Heuvelink and Bertin, 1995) and cucumber (Marcelis, 1994). These models enable not only to simulate the formation of yield along a crop cycle, but also the short-term gas exchanges in relation with the short term changes in the crop environment.

Till now, eac in the crop environment.

Till now, each research team has used specific tools of programme development yet FORTRAN is the most common computer language in use. Several authors have pled for generic approaches that would facilitate the exchange of (sub)models and data (Acock and Reynolds, 1990; Gauthier and Zekki, 1996).

Applications

In greenhouse production, the main application of crop models is the control of the environment, at the operational and tactical levels, as mentioned earlier (Baker et al., 1995). The supply of water is based on simplified forms of transpiration models. As greenhouses are semi-closed systems, crop and climate interact. The optimisation of CO₂ concentration, temperature and humidity must be based on coupled models of mass and energy balance, and of net photosynthesis and transpiration rates. The development of the crop itself is under the grower's control. Stem density, fruit and leaf pruning, branching... are commonly used to tune the vegetative/generative balance. These techniques can be improved on the basis of appropriate models of development and assimilate production and partitioning.

Another application of interest is yield prediction. The time-course of production has to fit with the market requirements, in terms of quantity and quality (fruit size, maturity, shelf life...). Models of prediction of yield anty, shelf life...). Models of prediction of yield and quality become of strategic importance in the relations between producers and buyers. They can also be a tool for the estimation of potential production and for the identification of the main limiting factors in an area of production (Challa and Bakker, 1998).

Models can also be used for the education of students and workers. The greenhouse production systems have become very complex, and many decisions have to be taken daily. Training can be faster by using simulators that enable the users to compare an unlimited number of policies of climate or plant control. The success of the first examples of such simulators proves that education is a promising application of crop models (Gary et al., 1998).

Perspectives: research, methods, cooperation

After about two decades of activity, greenhouse crop modellers have to face several challenges. The first approaches have been developed on a limited number of species and functions of the crop physiology, and with some simplifications. Yet the number of species cultivated under greenhouse is large, particularly among ornamentals. Models should become generic enough to enable the quick modelling of new species. The definition of a limited number of groups of species presenting a similar morphogenesis would help. Within a species, the number of cultivars may be large and their ecies, the number of cultivars may be large and their turnover in commercial greenhouses may be high. Genetic parameters should be identified clearly, and estimated easily and early in the development of new cultivars. In general, crop models simulate an average plant whereas greenhouse crops are often heterogeneous (the environment itself is more heterogeneous than in the open field). The variability among plants may be of economic importance, in pot plants for example. Some features of the crop development are still poorly addressed in most crop models, such as nutrient uptake, morphogenesis, and quality formation. The increasing concern about the problems of pollution and food (or product) quality justify a significant effort in these fields. At last, there is still a gap between the development of models and decision making. Models can help optimising the greenhouse and crop management, provided they are part of a decision support system. This may generate constraints on the building of the models.

It is clear that progress is possible only in the framework of a multi-disciplinary approach. Crop physiologists, plant breeders, pathologists and zoologists, agricultural engineers, scientists involved in the study of farming systems should co-operate to widen the scope of greenhouse crop models. Interestingly, such networks develop, at least partly, in different countries (see for example Bakker et al countries (see for example Bakker et al., 1995 in the Netherlands and Baille, 1997 in France). These efforts should be supported as they will generate new approaches of crop modelling that will fit with the new challenges of greenhouse cultivation. The greenhouse system is complex and its study requires several scientific skills whereas the number of scientists involved is limited in comparison to other cultivation systems. The ISHS working group "Modelling plant growth, environmental control and greenhouse environment" organises workshops and symposia that strengthen links at international level. International co-operation is often based on bilateral programmes of scientific exchanges. It is desirable that national and international institutions understand the interest of, and support these exchanges.

Table A. Main research teams involved in the modelling of greenhouse crops (teams participating to the ISHS meetings on greenhouse crop modelling).

Teams or groups of teams	Cooperation (bilateral prog)	Crops	Modelling approach	Plant functions	applications
Canada: Université Laval123>Canada: Université Laval, Québec	1	Greenhouse (vegetable) crops	Mechanistic, functional	Growth, development	Climate control, crop management, yield prediction
France: INRA, Dpt of Agronomy and Environment, Avignon, Antibes INRA, Dpt of Farming Systems, Alenya INRA, Dpt of Biometrics and Artificial Intelligence, Toulouse	1,2,3	Greenhouse (vegetable + ornamental) crops	Mechanistic, functional	Growth, development, mineral uptake	Climate control, crop management, yield prediction
Germany: University of Hannover		Greenhouse (vegetable) crops	statistical	Growth, development	Climate control, crop management, yield prediction
Israel:Technion, Haifa , Dpt of Agr Eng ARO, Volcani Centre, Bsor Exp Station	2,4,5	Greenhouse (vegetable) crops	Mechanistic, functional	Growth, development, Mineral uptake	Climate control, crop management
The Netherlands: The Netherlands: WAU, Dpt Horticulture +WAU, Dpt Agric Eng +AB-DLO Wageningen + PBG Naaldwijk + Aalsmer	3,4	Greenhouse (vegetable + ornamental) crops	Mechanistic, functional	Growth, development, Mineral uptake	Climate control, crop management, yield prediction
Sweeden: University of Alnarp		Greenhouse (ornamental) crops	statistical	Growth, development	Climate control, crop management, yield prediction
USA: University of California, Davis		Greenhouse (ornamental) crops		Growth, development	Climate control, crop management, yield prediction
University of Florida, Gainesville	5	Greenhouse (vegetable) crops	Mechanistic, functional	Growth, development	Climate control, crop management

References

Acock, B. and Reynolds, J.F., 1990. Model structure and data base development. In: R.K. Dixon, R.S. Meldahl, G.A. Ruark and data base development. In: R.K. Dixon, R.S. Meldahl, G.A. Ruark and W.G. Warren (Editors), Process modeling of forest growth responses to environmental stress, Timber Press, Portland (Oregon), pp. 169-179.

Bakker, J.C. , Bot, G.P.A., Challa, H.  and van de Braak, N.J., 1995. Greenhouse climate control - an integrated approach. Wageningen Pers, Wageningen.

Bertin N., Heuvelink E., 1993. Dry matter production in a tomato crop: comparison of two simulation models. J. Hortic. Sci., 68, 995-1011.

Challa, H. and Bakker, J., 1998. Potential production within the greenhouse environment. In: Z. Enoch and G. Stanhill (Editors), Ecosystems of the World. The Greenhouse Ecosystem. Elsevier, Amsterdam, in press.

Baille, A., 1997. Actes du Séminaire de l'AIP intersectorielle "Serres", INRA, Avignon.

De Wit, C.T., 1970. Dynamic concepts in biology. In: I. Setlik (Editor), Prediction and measurement of photosynthetic activity. Pudoc, Wageningen, pp. 17-23.

Gary, C., Charasse, L., Tchamitchian, M., Bertin, N., Rebillard, A., Boulard, T., Cardi, J.P. and Baille, A., 1998. SIMULSERRE: an educational software simulating the greenhouse-crop system. Acta Hortic., in press.

Gary, C., Jones, J.W. and Tchamitchian, M., 1998. Crop modelling in horticulture: state of the art. Sci. Hortic., in press.

Gauthier, L. and Zekki, H., 1996. An object-oriented framework for crop growth a Zekki, H., 1996. An object-oriented framework for crop growth and development simulation models. In: F.S. Zazueta (Editor), Proceedings of the Sixth International Conference on Computers in Agriculture, ASAE, Chicago, pp. 1022-1037.

Gijzen, H. and Goudriaan, J., 1989. A flexible and explanatory model of light distribution and photosynthesis in row crops. Agric. For. Meteorol., 48: 1-20.

Heuvelink E., Bertin N., 1994. Dry matter partitioning in a tomato crop: comparison of two simulation models. J. Hortic. Sci., 69, 885-903.

Jones, H.G. and Tardieu, F., 1998. Modelling water relations of horticultural crops. Sci. Hortic., in press.

Le Bot, J., Adamowicz, S. and Robin, P., 1998. Modelling of plant nutrition in horticultural crops. Sci. Hortic., in press.

Le Bot, J., Andriolo, J., Gary, C., Adamowicz, S. and Robin, P., 1997. Dynamics of N accumulation and growth of tomato plants in hydroponics: an analysis of vegetative and fruit compartments. In: G. Lemaire and I.G. Burns (Editors), Diagnostic procedures for crop N management, INRA Editions, Paris, 37-51.

Marcelis, L.F.M., 1994. A simulation model for dry matter partitioning in cucumber. Ann. Bot. 74: 43-52.

Marcelis, L.F.M., Heuvelink, E. and Goudriaan, J., 1998. Modelling of growth and yield in horticultural crops: a review. Sci. Hortic., in press.

Monteith, 1977. Climate and the efficiency of crop production in Britain. Phi Climate and the efficiency of crop production in Britain. Philos. Trans. R. Soc. London B, 281: 277-294.