Eugénio Araújo Faria

Neural network models in greenhouse air temperature prediction

Abstract The adequacy of radial basis function neural networks to model the inside air temperature of a hydroponic greenhouse as a function of the outside air temperature and solar radiation, and the inside relative humidity, is addressed. As the model is intended to be incorporated in an environmental control strategy both off-line and on-line methods could be of use to accomplish this task. In this paper known hybrid off-line training methods and on-line learning algorithms are analyzed. An off-line method and its application to on-line learning is proposed. It exploits the linear–non-linear structure found in radial basis function neural networks.

Effectiveness of different foliar iron applications to control iron chlorosis in orange trees grown on a calcareous soil

The effectiveness on controlling Fe chlorosis in orange trees grown on calcareous soils was tested. The treatments were Fe(II) sulfate (500 mg Fe L−1), sulfuric acid (0.5 mM H2SO4), Fe(III)-chelate (Hampiron 654 GS, 120 mg Fe L−1) and distilled water as a control. A non-ionic wetting agent was used in all treatments. The use of frequent foliar sprays alleviated Fe chlorosis in orange trees. Sprays of Fe(II) sulfate increased the concentrations of chlorophyll, Fe and zinc in leaves and improved fruit size and quality compared to fruits of control trees. Sprays of Fe(III)-chelate also increased leaf chlorophyll and Fe concentrations and improved fruit quality, but did not increase fruit size. Sprays of sulfuric acid alone slightly increased leaf chlorophyll and Fe concentrations, without improving fruit size and quality. These results suggest that foliar sprays with Fe could help to avoid yield and quality losses caused by Fe chlorosis in citrus orchards. Furthermore, these treatments could be done with relatively cheap materials such as solutions containing Fe(II) sulfate.

RESPONSES OF “NEWHALL” ORANGE TREES TO IRON DEFICIENCY IN HYDROPONICS: EFFECTS ON LEAF CHLOROPHYLL, PHOTOSYNTHETIC EFFICIENCY, AND ROOT FERRIC CHELATE REDUCTASE ACTIVITY

Orange (Citrus sinensis L. Osb. cv. ‘Newhall’) plants grafted on Citrange troyer rootstock were grown in nutrient solution with 0, 5, 10, or 20 μM iron (Fe), with and without calcium carbonate. Calcium carbonate was added in order to mimic the natural conditions in calcareous soils. Leaf chlorophyll concentration was estimated every 3–4 days using the portable instrument SPAD-502 meter. Chlorophyll fluorescence parameters, photosynthetic capacity estimated from oxygen evolution, leaf Fe concentrations, and root tip ferric chelate reductase activity were measured at the end of the experiment. Plants from the 0 and 5 μM Fe treatments showed leaf chlorosis and had decreased leaf chlorophyll concentrations. Leaves of plants grown in the absence of Fe in the solution had smaller rates of oxygen evolution both in the presence and absence of calcium carbonate, compared with plants grown in the presence of 10 μM Fe. In the absence of calcium carbonate the photosystem II efficiency, estimated from fluorescence parameters, was similar in all treatments. A slight decrease in photosystem II efficiency was observed in plants grown without Fe and in the presence of calcium carbonate. A 2.5-fold increase in root tip ferric chelate reductase activity over the control values was found only when plants were grown with low levels of Fe and in the presence of calcium carbonate.

The use of floral analysis to diagnose the nutritional status of orange trees

A field experiment was conducted in a commercial orange orchard (Citrus sinensis L. cv. ‘Valencia late’), established on a calcareous soil in the south of Portugal, to investigate if flower analysis could be used to diagnose the nutritional status of the trees and predict fruit quality. In April 1996, during full bloom, flowers and leaves were collected from 20 trees. Leaf samples were again collected in June and August from the same trees. In February 1997 the fruits were harvested and their quality evaluated. The concentrations of phosporous (P) and magnesium (Mg) in flowers were correlated with those in leaves, and were also predictive of the chlorophyll content of leaves 60 days later. However, by 120 days after full bloom these effects were masked, possibly by a fertilizer application carried out by the farmer. The iron (Fe) concentration in flowers was correlated with chlorophyll measured either 60 or 120 days after full bloom. These parameters can therefore be used to predict the appearance of iron chlorosis. Fresh weight and diameter of fruit were related to flower P, Mg, and manganese (Mn) concentrations, while the percentage of citric acid in the fruit juice was inversely correlated with Fe flower concentrations.

The recovery of citrus from iron chlorosis using different foliar applications. Effects on fruit quality

The response of Encore trees grown on a calcareous soil to different foliar applications to offset iron deficiency. Four treatments were tested: distilled water (control); iron (II) sulphate (500 mg Fe.L−1); sulphuric acid (0.5 mM H2SO4) and Fe-EDDHA (120 mg Fe.L−1). The recovery from iron chlorosis was evaluated with the SPAD apparatus and the values converted to total chlorophyll density. We also evaluated effects of the treatments on some physical and chemical characteristics of the fruit. Chlorophyll density in the leaves, and the total sugar content of the fruits, were greater in all experimental treatments, compared with control, but with no significant differences between treatments. The concentration of citric acid decreased in the treated plants. The greatest diameter and fresh weight of fruits were obtained in the treatment with iron chelate. Foliar applications of iron sulphate or sulphuric acid led to values of these parameters that were intermediate between those of the control and the iron chelate treatments.

Application of radial basis function neural networks to a greenhouse inside air temperature model

Abstract The problem of the adequacy of radial basis function neural networks to model the inside air temperature as a-function of the outside air temperature and solar radiation, and the inside relative humidity in an hydroponic greenhouse is addressed. This type of network is structurally simple and suitable to be integrated in real-time greenhouse environmental control systems. Due to the time variability of the process, training methods with on-line adaptation capabilities are needed Three of such methods are analysed in terms of fitness and network size. The model-predictive outputs obtained showed very close fittings to the measured values.