EFFECTS OF SOIL AND CLIMATE IN A TABLE GRAPE VINEYARD WITH COVER CROPS. IRRIGATION MANAGEMENT USING SENSORS NETWORKS

The use of mulches in vineyards and orchards is a traditional agricultural practice used with the aim of saving moisture, reducing weed growth and improving organic matter content in the soil. In table grape vineyards trained to overhead system in Puglia region (Southeastern Italy), plastic sheets covering the canopy are often used to either advance ripening or delay harvest. In this environment, the living mulches could contribute to the modification of the microclimate around the canopy below the plastic sheets. This condition has an influence on the climatic demand and on both the vegetative and productive activities, mainly in stages with a high evapotranspiration. However, the presence of living mulches could increase the demand of available water and nutrient resources and this could cause a lower yield. The aim of this study was to acquire a suitable knowledge to manage irrigation and verify the influences of living mulches on the vine by using wireless sensor networks to measure the vapor pressure deficit, soil water potential and content


INTRODUCTION
In agriculture, weeds control is an important issue because weeds highly compete for the available nutrients and water resources, and according to some researches, they could affect yield up to the 80% (Cousens & Mortimer, 1995). Such control is traditionally undertaken by plowing (mechanical control of weeds) and/or using herbicides (chemical control). However, the consumers are asking to use less herbicides and the utilization of these products is being reduced due to environmental reasons, such as, the emergence of herbicide resistant weeds and the increase of organic food demand (Carter et al., 1991;Major, 1992).
The growth of weeds could be controlled by using organic and synthetic mulches, such as waste material of the olive oil industry (Ferrara et al., 2015(Ferrara et al., , 2012a. The utilization of specific species as living mulch, in combination with drip irrigation, provides both organic matter and nutrients (Welker and Glenn 1988). Additionally, the presence of the living mulch improves the water retention capacity of the soil, and the erosion is consequently reduced (Haynes, 1980;Merwin et al., 1995;Verdú and Mas, 2007). Hall et al. 1984 measured water runoff volume in a corn field planted on a 14% slope and grown conventionally, managed with no-till or with a crown vetch living mulch. The water runoff in the conventional corn field was reduced by 90% and 98%, in the no-tilled and mulched with crown vetch plot, respectively.
In the specific case of table grape grown in Puglia, Southeastern Italy, vines trained to overhead system are often covered with plastic sheets to either advance ripening or delay the harvest, and the presence of a living mulch could modify the microclimatic conditions of the vine with effects on vigour and, consequently, on the water needs (Hartwig and Ammon 2002).
The use of living mulch generates a specific microclimate since transpiration is higher as a consequence of the presence of a crop in the interrow. In addition, soil water conditions could also vary because of the presence of the living mulch. The parameters of such environment could be measured by using a set of specific environmental and soil sensors. These measurements can provide information about the influence of living mulch on several parameters. In particular, different sensors should be deployed in the soil with the purpose of obtaining the appropriate determination of the variables. Such sensors should be installed in different sites of the vineyards so that the whole crop is monitored as accurate as possible.
Agricultural sensors, uzed for monitoring soil, plant and climatic parameters, are usually connected to dataloggers. These devices store the acquired data so that they could be used by the operator where they are required. The main problem of these devices is that they are the key element of a wired network, where several meters of wire should be deployed between the datalogger and the corresponding sensor. It is an important issue that should be taken into account when deploying a set of distributed sensors, and it could be difficult to install such sensors at the required locations due to this limitation. Some manufactures (Spectrum Technologies Inc. Aurora, IL, USA) have developed autonomous (working with batteries) and low size dataloggers, which could acquire data of up to 4 sensors. These characteristics (in combination with a remote data accessing and the autonomous capability of managing the required energy for working), have promoted the use of both autonomous and wireless flexible communications platforms, which are named Wireless Sensors Networks (WSNs). They have been used by several authors to conduct trials in agricultural crops (Morais et al. 2008.
Wireless Sensors Network is a measurement platform used nowadays in agriculture to manage water resources , since it has great flexibility for measuring parameters in agricultural field. Soil, plant and climatic parameters can be measured using a wireless sensor network that comprises different nodes with several typical parts: a radio transceiver, an antenna, a microcontroller and a battery with an energy-harvesting system. Each node could work in different places without wires .
Data acquired by each node are sent to a receiver node, named coordinator, which receives the information from all sensor nodes that comprise the whole WSN. Such coordinator node is in charge of both managing the information and transmitting it to the servers that store the data. WSNs have gradually evolved towards mobile networks, based on machine to machine communications (M2M), thanks to the wide increase in mobile communications, the reduction in the data rates and the growth experimented in the communication speed of such networks. This means that the data acquired by the sensor node are directly transmitted to the servers hosted in the cloud by using GSM mobile networks. In this case, the economic cost is affordable.
In this work, soil and environmental parameters have been monitored in a table grape vineyard covered with a plastic sheet, and with a living mulch in the inter-row. Specifically, wireless sensor nodes equipped with a mobile network transceiver have been used. The acquired data have been stored in remote servers using cloud computing techniques. This paper analyzes the effects of a living mulch on both soil and microclimate parameters measured by using a wireless sensors platform.

MATERIAL AND METHODS
The trial was conducted in a table grape vineyard in 2015. Italia grapevines grafted onto 1103 P of similar growth and vigour were selected for the study (application of sensors). The vineyard was located in the territory of Adelfia (Bari province), Puglia region, Southeastern Italy. Vines were spaced 2.2 × 2.8 m, trained to an ypsilon trellis system with four fruiting canes/vine (40-50 buds/vine) and were drip irrigated (3 emitters of 4L•h -1 per vine). The vineyard was covered with a plastic sheet in order to advance ripening. Specifically, the trial was conducted in two contiguous plots (see Figures 1 and 2), the first one with a living mulch (Trifolium repens) in the interrow (clover seeded on winter 2014), and the latter with a clean soil which was periodically ploughed (Figure 1 and Figure 2). The trial was conducted from budbreak up to the grape harvest.  The same number and type of sensors were installed in both plots. Such sensors and the associated measured variables are listed in Table 1.
The mentioned parameters were measured by using one GPRS communication and control sensor node in each monitoring point. Specifically, precision agriculture systems, based on wireless nodes and cloud computing approaches for both storing and processing the information in the cloud, were used (Widhoc Smart Solutions S.L., MU -Spain). The sensor nodes are autonomous from the energy point of view and maintenance is almost non-existent. They were installed above the vine canopy with the purpose of ensuring that the recharging solar system could properly work. A Pluviometer connected to the drip emitter with the purpose of measuring the volume of water supplied to the crop.
The irrigation management was carried out keeping the water potential between -20 and -30 kPa in both plots, using the measurements acquired from the matric water potential sensor (Figure 3). The purpose was to keep non-limiting conditions of the water in the soil. Using drip irrigation and considering our cultivation conditions, the highest root density is located between 5 and 35 cm deep. The tensiometer was placed at 25 cm of depth in order to reflect this root activity. Figure 3 shows the evolution of water tension and irrigation episodes. The sensor nodes were equipped with a pluviometer with the aim of knowing when the irrigation was switched on/off. On the other hand, the total amount of irrigation water was measured by using flow meters in each plot.
Data were stored in servers of Widhoc Smart Solutions S.L. (Fuente Alamo, MU -Spain) by using the M2M service provided by the Italian mobile 4G Company WIND. The data analysis was carried out by using the MatLAB software.

RESULTS AND DISCUSSION
Irrigation management was carried out by using two different premises. During the first stage , vines were irrigated with the same criteria in both plots, with the aim of studying the soil properties. From DOY 190 onward, the irrigation management was accomplished by using the information provided by the soil water potential sensors. In this case, the intended value for the soil was kept around -30 kPa, sending irrigation commands when values under such threshold were reported. Figure 3 shows the data of the irrigation management by using the matric water potential sensors and the amount of water supplied. The interface belongs to Widhoc Smart Solutions (Fuente Alamo, MU -Spain).
In order to have more information, the water volume reported by the farmer (according to local practices), was also checked. This was useful for comparing the results obtained with the use of the sensors with those provided by the farmer. In addition, midday stem water potential and fluorescence measurements (fortnightly) were also collected with the purpose of better verifying the irrigation conditions. The midday stem water potential is very representative of the level of water deficit of the vine (Choné et al. 2001).
A threshold value of -0,6 MPa was defined in order to ensure the appropriate water potential for the vine, since values 0.5-0.8 MPa can be considered indicative of no or very little stress (Tramontini et al., 2013). Figure 4 shows the evolution of the air vapour pressure in both treatments (tilled and mulched) during a period of time with high evaporative demand (DOY 209-221). Several differences were observed, mainly during the hours of highest evapotranspiration. In particular, the vapour pressure was higher in the living mulch plot, with values that exceeded 3 kPa, whereas the plot without mulching hardly reached such value. This fact did not involve a higher water consumption. Specifically, a value of 1296.52 m 3 ha -1 was the irrigation volume in the plot with mulching during a period of 60 days, since the irrigation management procedure started (coincident with the maximum transpiration demand, DOY 205-265). On the other hand, during the same period, the volume irrigation in the ploughed soil reached 1448.66 m 3 ha -1 . Although the difference between the two measurements was not significant (barely a 10%), the expected results would be a higher volume of water in the plot with Trifolium repens. The data of the soil water content profiles were necessary to perform a temporal study of the soil water retention capacity in each soil. For this purpose, several humidity sensors located at different depths were used.
The variables were analysed using the same irrigation cycles, that is, in the same days and with the same duration . The measurements were taken between repetitions of each treatment. The analysis was carried out by using sensors at 25 cm and 50 cm for each repetition. During such period, 12 irrigation cycles were analysed. Figure 5 shows the evolution of the most representative volumetric soil water content sensor at 25 cm and 50 cm in both plots, after an irrigation episode in one of the two repetitions measured. Evolução do teor volumétrico de água do solo em ambas as parcelas durante dois episódios de rega mais representativos a 25 cm de profundidade (esquerda) e 50 cm de profundidade (direita). Azul para a parcela com Mulch e vermelho para a parcela Mobilizada.
Results clearly showed the beginning of irrigation episodes in both plots.
Furthermore, after analysing these data, results also showed a very marked gradient in the tilled plot (redcoloured), immediately after irrigation. This effect can be observed at 25 cm clearly. In fact, a rapid decrease of the water content was observed, probably due to the limited soil retention capacity. However, it was slightly different at 50 cm depth, where this gradient after irrigation becomes similar. The time that the variable θv takes to reach the same value that has been measured before the irrigation episode is also shown in Figure 5 for two most representative irrigation episodes.
The problem of these representations is the offset of the lines (θv at the Y axis). Really, in order to perform the temporal study of the water in the soil, it is not necessary to know the absolute value of the θv, but the gradient or evolution of the water content during the time. This gradient is the indicator of how much time the water is kept in the soil. At both soil depths, 25 and 50 cm, the reduction of water content is clear and the faster reduction at 50 cm can be explained by the soil characteristics in the vineyard. In particular, this soil has been subjected to rock fragmentation and grinding, as usual procedure in table grape vineyards in Puglia (Ferrara et al., 2012b). At 50 cm there are more skeletal fractions of medium size, which cause a faster drainage of the water with respect of the 25 cm profile (Ferrara et al., 2012b).
In order to show this feature, two most representative graphical representations of the 12 irrigation episodes are presented in Figure 6 (a) and (b) (where each one covers an irrigation cycle) at 25 and 50 cm respectively. The minimum moisture value of each curve is taken, and subtracted from each curve (remove offsets). A cubic interpolation is performed to obtain a higher density of work points (especially interesting in the area of the rise). The factor parameter indicates the number of points to have for each of the original points.
The enclosed area is calculated under each of the curves (with the offsets removed). Both curves are normalized by dividing each of their values by the area.
These curves are already normalized and can be compared directly. Two types of parameters are taken to characterize the terrain.
The process to analyze and draw the curves of a more appropriate mode is as follows.
At the first, the offset values of the curves are removed: offset=min ( Time parameters: Time in hours that the water volumetric content is present in a given proportion and at the concrete depth (25 and 50 cm) with respect to the volume of water supplied in the irrigation episode and measured as the area of the normalized curve (25%, 50% y 75%). (TABLE II)