Evaporative demand and water requirements of the principal crops of the Guadalentín valley ( SE Spain ) in drought periods

The drought periods that affect the province of Murcia, especially the Guadalentín Valley, are aggravated by an increase in evaporative demand. The aim of the present study was to characterize the increased water demand of woody and herbaceous crops during drought periods in the Guadalentín Valley, an agricultural zone with an excellent climate for specialty crops, which is of great economic importance for Murcia. After defining the drought periods of the last three decades in time and space by means of the standard index of rainfall drought (IESP), several methods were used to determine the reference evapotranspiration (ETo): the Penman-Monteith model (ASCE and FAO models for grass), the Hargreaves method (ETo-ASCE for alfalfa), and ETo using the FAO Radiation method. Finally, the crop water requirements for each to crop type and area of cultivation were estimated using monthly crop coefficients (Kc) and the mean monthly evaporative demand values were obtained by the best fitting method. The increase in the evaporative demand reflected the increased water deficits that occur in the drought years, both in summer and winter (1.23 hm yr). Drought periods are also responsible for reducing the areas dedicated to horticultural crops, because of their high water demands and the additional costs involved, resulting an aggravated socioeconomic position and increased unemployment. Additional key words: crop coefficients (Kc); lysimetric measurements; standardized index of rainfall drought; woody and herbaceous crops.


Introduction
Due to the great importance of the agricultural sector in the province of Murcia (SE Spain), especially in the Guadalentín Valley, where the intensive cultivation of horticultural crops has replaced traditional dry land agriculture, there is great concern about the lack of water and the constant need for the same (Gil-Olcina, 2004).
The principal use of water in the area for woody crops is to irrigate citrus trees (17%), almond trees (13%) and table grapes (5.2%), stone fruit trees and grapevines for wine hardly present in this district.Among herbaceous plants, alfalfa is the main receives of water (14%), while lettuce (7%) and tomato (5%) are also important in this respect (Portero-Faus, 2000).
The yields of these crops are continuously affected by a pronounced annual water deficit (about 600 mm) as a result of the scant rainfall the area receives (annual mean of 350 mm) and the high rate of potential evapotranspiration (annual ETP > 900 mm) (Sánchez-Toribio et al., 1996).To these scant resources we can add other factors, both physical (geological, hydromorphological and edaphological) and socioeconomic, which render the area particularly sensitive to the twin processes of degradation and desertification (López-Bermúdez et al., 1996).
The sequences of drought occurring between 1980 and 2009 were defined by reference to the Standardized Index of Rainfall Drought (IESP in Spanish) (Pita-López, 2001).The overall and per sector evaporative demands were calculated for this whole period, but especially during the severe bouts of drought experienced, The climatological data used were those of the Murcia/Alcantarilla, Librilla, Alhama «Huerta Espuña», Totana, Lorca, Puentes and Puerto Lumbreras meteorological stations belonging to the Spanish Meteorological Agency (AEMET) and the Murcia Institute of Investigation and Agrarian Development (IMIDA, Comunidad Autónoma de la Región de Murcia) (Fig. 1).To calculate evapotranspiration the Penman-Monteith-FAO method (Monteith, 1965) was used after checking the goodness of its fit for the ET of grass in this region by lysimeter (Sánchez-Toribio et al., 1990).In other geographical locations, especially the USA, it is more common to apply the Penman Monteith ASCE equation using alfalfa as crop reference (ASCE-EWRI, 2005).Although the use of gramineae or alfalfa as reference crop has been widely discussed (ASCE 28), the advantage of the former is that it is easy to maintain a green surface throughout the year, with a LAI (leaf area index) of about 3 (De Santa Olalla, 2005), which is particularly useful when using lysimeters.Among the most used species are ryegrass (Lolium perenne L.) or festuca (Festuca arundinacea L.), and among the least suitable those which exercise strong biological control over transpiration, such as Bermuda (Cynodon dactylon L.).
In normal stress conditions, the value of the plant cover is low, so the crop coefficient (K c ) will be less affected by the conditions of the soil in which these crops and natural cover grow.Equations that express the fit as a function of the plant cover fraction and LAI are described in Allen et al. (1998).
The values of K c are equally necessary to establish, in conjunction with the ETo, the water requirements of the different crops and covers.The usefulness of the «K c • ETo» method has already been demonstrated at a regional scale (ASCE28) (Allen et al., 1998).In this case, the applications correspond basically to the field of hydrology and the management of water resources, and are usually directed at ascertaining the water balance at aquifer or hydrographical basin scale.This type of analysis is strongly conditioned by the level of detail of the knowledge we have of the area and soils at the scale used in each study.In the case of the Guadalentin Valley, the K c values comes from the Irrigation Assessment Programme of IMIDA, the results of which form part of a specific data base which can be managed by a Geographic Information System.
The objective of the present study was to determine, as a function of the cultivated area, the evaporative demand and the different crop coeff icients, the in-

Guadalentín River Basin
crease in water requirements of the main woody and herbaceous crops of the Guadalentin Valley, an agricultural zone with excellent temperatures for high yield crops and of great economic importance for the province, even during droughts.The frequent occurrence of such events, with the severe water stress that they involve, makes these drought periods ideal states for applying and comparing methods such as those proposed.

Study area
The Guadalentin Valley was chosen as study area because of the constant water stress it suffers, which is aggravated by frequent drought periods.The valley begins near the middle of the Guadalentín river/wadi and lies within the province of Murcia.The upper reaches (205 km 2 ), drained by the Viznaga wadi, include a small ENE-WSW corridor of tectonic origin between Puerto Lumbreras and Lorca.It is separated from the coast by the Carrasquilla and Almenara Sierras Downstream of Totana, the valley follows the tectonic control of the Bajo Guadalentín, bounded to the north by the mountain ranges of Espuña, El Cura y La Muela, and to the south by the Sierra de Carrascoy.Although the great geographical diversity of the Guadalentín basin implies climatic diversity (Navarro-Hervás, 1991;Martín-Vide and Olcina-Cantos, 2001), its situation in the south east of the Iberian Peninsula and the orographic shelter conferred by the Bettics give the whole valley a semiarid Mediterranean character.

Delimitation of the drought sequences
The dry sequences, present in all climatic dominions, particularly affect semiarid regions marked by a strong inter-annual variation in rainfall patterns.Special attention to dry period lasting more than one year has been taken, since they produce the most important damage in zones like the Iberian Peninsula, familiar with this type of event and adapted to shorter periods with almost total impunity.
Sequences of drought need to be translated into a numerical format if their characteristics are to be determined.From a social-economic point of view, which is what determines whether an extreme climatic event (drought) is counted a natural risk, several parameters should be mentioned: magnitude, duration, frequency, time taken to impose itself, extension and spatial dispersion.To quantify such parameters mathematical calculations are necessary to establish in detail to what extent there has been drought and what the temporal limits of the drought episode are (Marcos, 2001).
Several indices exist (percentage of mean rainfall, quantiles, Palmer PDSI etc.).In this work the standardised index of rainfall drought derived from rainfall but applicable to identifying drought sequences has been applied, since it precisely reflects the beginning and end of each sequence, its duration and intensity at any given moment.The index is the result of standardising accumulated monthly rainfall anomalies, interrupting the accumulation when a negative rainfall anomaly appears, that is, the beginning of each drought sequence.In this way, the effect of minimizing droughts as a consequence of the accumulation of preceding excess (positive) rainfall is avoided.The IESP has been shown to be effective for analyzing the long drought sequences that characterize Mediterranean dominions (Pita-López, 2000, 2001).The index is based on the calculation of accumulated monthly rainfall anomalies following these steps: i) Calculate the rainfall anomaly of each month of the series (AP i ) using the expression: where AP i = monthly rainfall anomaly; P i = monthly rainfall and P MED = mean of the corresponding month.
ii) Obtain accumulated rainfall anomalies from the f irst month when there is a negative anomaly (AP i < 0) to -as a result of the accumulations-a positive accumulated rainfall anomaly occurs (AP i > 0).This is when the drought ends (APA > 0 marks the end of the rainfall deficit).The following drought sequence begins when a month showing a negative rainfall anomaly occurs: where APA i = accumulated rainfall anomaly of the month.
iii) Standardize the accumulated anomalies by converting into Z scores.This reveals the intensity of the drought and will be used to identify the drought sequences: where ZAPA i = standardised rainfall anomaly of the month; APA --= mean of the accumulated rainfall anomalies for all the months of the series and σ APA = typical deviation of the accumulated rainfall anomalies for all the months of the series.
The method can be applied to long time series and, simultaneously, to different observatories, so that the risk of drought can be analysed for different sectors of the Valley.

Estimation of evapotranspiration
The concept of evapotranspiration has always been considered slightly ambiguous due to the type of vegetal cover being considered.In the 1970s, the concept of «reference evapotranspiration» (ETo) was developed and the term «potential evapotranspiration» (ETP) was coined, since when many calculation systems have been developed.
To fit the estimations of FAO methods (Doorenbos an Pruitt, 1977), there are numerous procedures for calculating ETo with parameters adapted or modified in accordance with local conditions.To confirm the validity of these models in the USA, the American Society of Civil Engineers (ASCE) compared the results of 20 widely used methods using alfalfa as reference crop (Jensen et al., 1990).A similar study for the European Commission evaluated different methods of calculating ET 0 using lysimetric measurements.Both studies confirmed the excellent behavior of the model adapted by Penman-Monteith for the FAO, both in humid and arid climates (Allen et al., 1998).Such were the conclusions of the FAO in a report published in 1990 in collaboration with the International Commission on Irrigation and Drainage (ICID) and the World Meteorological Organization (WMO).The report recommended adoption of the Penman-Monteith (ETo-PM, FAO) method for reference ETo and included calculation procedures revised and normalized in Smith et al. (1996).
In our case, the reference cover is a hypothetical cover of grass with a height of 0.12 m, a constant value of the canopy resistance of 70 s m -1 and an albedo of 0.23.This fixed resistance of 70 s m -1 implies a moderately dry soil with a weekly irrigation frequency (Allen et al., 1998): where ETo = ET of the reference crop; R n = net radiation calculated (MJ m -2 d -1 ); G = heat flux in the soil (MJ m -2 d -1 ); T = mean temperature for the period at a height of 1.5 -2.5 m (°C); U 2 = mean wind speed (m s -1 ); e s = vapor pressure at saturation (kPa), calculated at height of 1.5 to 25 m; e a = present vapour pressure at 1.5-2.5 m (kPa); ∆ = slope of the vapour pressure at saturation-temperature (kPa °C-1 ); γ = psicrometric constant (k Pa °C-1 ).
The method described above enables us to predict the effects of climate on the evapotranspiration of the reference crop (ET).However, to relate ETo with crop evapotranspiration, ET (crop), it is possible to use coefficients of culture (K c ), which represent the evapotranspiration of a crop in optimal conditions.From the product ETo • K c the ET representing the water needs can be obtained for different crops of the area.
Obtaining suitable values of K c is a complex task since many factors intervene: crop characteristics, planting or sowing date, crop growth rhythms and length of vegetative period, soil and climatic conditions, especially during growth, rainfall frequency or supply of irrigation water.The values used are those of the Irrigation Assessment Program from IMIDA, which were obtained indirectly in field experiments and from the bibliography referring to area similar to those of the Valley Guadalentín (Jiménez-Montesinos and De-Luna, 1984;Fereres, 1987;Faci and Martínez-Cobb, 1999).

Delimitation of drought sequences by IESP
Although drought is a widespread phenomenon, inter-territorial comparisons may serve to better manage the consequences of the same, as long as compensatory mechanisms exist between sectors.However, a small spatial co-variation in the case of drought does not guarantee in itself that amendment mechanisms can be established.For this to be possible, the areas which supply water must always be in a situation of excess so that they can temporarily export water.
Application of the IESP to the weather stations of the Guadalentín Valley (Fig. 2) identif ied various drought sequences (negative Z values) that approximately coincide (García-Marín and Conesa-García, 2006).As regards their most significant characteristics and, especially, their spatial extension, the drought periods of the 1980s and 1990s are of note, followed by the drought phase at the turn of the millennium (1999-   (2004)(2005)(2006). of the 1980s (1981-1985) According to both integral and regional studies, the 1980s began with one of the longest drought periods of the XX century throughout mainland Spain.In the Guadalentín Valley, rainfall deficits exceeding four years were recorded (e.g. in Lorca from September 1980 to October 1985), and Z values below -2 and even below -3 were recorded due to the accumulation of an extraordinary number of consecutive months with a rainfall deficit.The severity of this prolonged drought was evident in the strong impact observed on the water ecosystems of the river Segura and the socioeconomic repercussions generated, especially as regards agricultural production (Zapata-Nicolás et al., 1990).(1993)(1994)(1995)(1996)(1997) The first part of the last decade of the XX Century was marked by an intense period of drought through most of the Iberian Peninsula, although it reached the southeast slightly later, but lasted longer.Certain stations (e.g.Lorca) registered more than 50 months of drought, while others registered between 30 and 50.The intensity was quite pronounced in most of the basin even though it lasted less than the previous drought.Suffice to give one example, that of Murcia/Alcantarilla, which registered 41 consecutive months (June 1993-October 1996) of strong intensity (-2.04).

Droughts of the 1990s
Despite lasting less time, this drought was especially harsh in several areas.After the summer of 1905, one of the driest of the XX century, the Spanish government put into effect the so-called «Plan Metasequía», which included emergency measures to guarantee the immediate and future water supply to several Spanish cities.These included the drilling of wells, the use of depurated water and the construction of desalination plants ( Morales-Gil and Olcina-Cantos, 1999).
Both droughts effects (April 1999-October 2001and August 2004-March 2007) were felt along the Guadalentín Valley although with differing degrees of intensity, the most severely affected being Alhama followed by Alcantarilla.

Evaluation of evaporative demand
The rainfall drought and, as a consequence, water shortage was aggravated by increased evaporative demand (Fig. 3).An increase in evaporation over the extensive surface area of exposed water in the region (regulation reservoirs, and the water stored behind small dams and retention dykes for storing occasional water), together with irrigation reservoirs, led to the loss of a considerable amount of water, seriously damaging the potential yields of the abundant forced crops fed by drip irrigation.

Behaviour of the different methods for measuring evapotranspiration
The ETo, ETo Penman-Monteith models (ASCE and FAO models for grass) and Hargreaves Samani's ETh model show very similar daily mean evapotranspiration values, while the results of the ETr model (ASCE for  alfalfa) are closer to obtained with ETr (Radiation FAO) (Table 1).
The Blaney-Criddle-FAO, Radiation-FAO and ETr (ASCE alfalfa) equations show higher values of ET, especially during summer months.Sánchez-Toribio (1990) demonstrated the overestimations that result from using the Radiation and Blaney-Criddle-FAO models compared with lysimetric measurements made in the lower stretches of the Guadalentín Valley.A compari-son of the theoretical values of the models used confirms the good fit of ETh, based only on temperature, with the physical ETo method of Penman-Monteith (Fig. 3).
Comparing the evapotranspiration between a rainy period (1989-93) and a dry period  in the Alcantarilla station (Lower Valle of Guadalentín) (Fig. 4), the annual evapotranspiration rate gradually increased in inverse relation with rainfall, although not at the same rhythm.J a n -8 9 J u ly -8 9 J a n -9 0 J u ly -9 0 J a n -9 1 J u ly -9 1 J a n -9 2 J u ly -9 2 J a n -9 3 J u ly -9 3 J a n -9 4 J u ly -9 4 J a n -9 5 J u ly -9 5 J a n -9 6 J u ly -9 6 ETh Rainfall ETh tendency Rainfall tendency The values of ETo Penman-Monteith during the dry periods exceeded those of wet or normal periods (Fig. 5).The differences were especially evident after the spring months, and reach a maximum in July.From September to February the values of both the dry and normal months evened out.

Estimation of overall water demand
The traditional methods for calculating the water requirements of a plant are based on ETo values.All microclimatic methods have been developed for total cover crops and their use in fruit and citrus trees presents problems of accuracy: representativity, leaf area gradients, crown density, etc.Such estimates of the water requirements are also limited because they are not directly related with the water functioning of the plants, especially in def icit irrigation situations.Consideration of the monthly crop coefficient (K c ) for different plant cover types obtained from IMIDA and the bibliography, together with the mean monthly evaporative demand values, permitted us to determine the water requirements of the irrigated areas of the zone.Given the importance of agriculture in the same, this will be of undoubted interest for the regional government.The formula used (monthly ETo • monthlyK c ), taking the K c as suggested by the FAO (2006) (0.75 during rainy months and 0.8 during dry months) provides good estimates both in normal periods and droughts.Among the crops of the area that show greatest vulnerability to water stress are citrus trees, which are particularly sensitive during flowering.Given that the K c of this crop with most of the trees in full cover is around 0.8-0.9, the irrigation in this semiarid area needs to count on a daily application of 5-10 mm.
The water requirements of each crop, multiplied by the respective areas they covered, will provide the overall water demands for the zone.The final result for the water requirements is expressed in hm 3 in order to simplify and improve the interpretation of possible comparisons.From the values obtained, it is inferred that the overall water demand for the principal crops of the Guadalentín Valley should be increased by 8.35 hm 3 during drought periods, tomato, pepper, artichoke, cauliflower and broccoli being the crops that need the greatest amount (Table 2).The same data point to quite different increased water requirements.In peppers grown to produce paprika, for example, an increase of 290 m 3 ha -1 year -1 is assumed, while the figure for lettuce and cauliflower does not exceed 12 m 3 ha -1 year -1 .
Woody crops increase their water demands during drought to a lesser extent (6.15 hm 3 ) (Table 2).Citrus (lemon and orange) and table grapes suffer most during low rainfall periods, especially given the higher evaporative demand as a result of increased insolation.
In total, the increased annual demand estimated for the main crops of the Guadalentín Valley during drought periods is some 14.5 hm 3 , a volume of water that will be difficult to provide, given the surface water deficit and already overexploited aquifers.This explains the decrease in the areas devoted to certain crops, especially herbaceous crops, that have been recorded when the persistence of rainfall drought gives way to a situation of hydrological drought (Fig. 6).
The area devoted to agriculture during the different stages of drought closely reflects the insufficient capacity of the Guadalentín Valley to survive increased annual water deficits.This is particularly clear in the reduction in the number of hectares dedicated to irrigated herbaceous crops.

Conclusions
As computed for other zones of the province of Murcia (Sánchez-Toribio and López-Bermúdez, 1996;Sánchez-Toribio and Romero-Díaz, 2006), the reference alfalfa and grass evaporation models, especially the exclusively temperature-based model of Hargreaves-Samani (1995), showed a good fit with the physical based models (ETr, ETo-Penman-Monteith).The increase in the evaporative demand reflected the increased water deficits that occur in the drought years, both in summer and winter.An increase in evaporation over the extensive surface area of exposed water in the studied zone (regulation reservoirs, irrigation dams and retention dykes) led to the loss of a considerable amount of water, which lay an even greater threat to the irriga-tion of forced crops, economically important for the region (León et al., 1986).Drought periods are also responsible for reducing the areas dedicated to horticultural crops, because of their high water demands and the additional costs involved, resulting an aggravated socioeconomic position and increased unemployment.Figure 6.Evolution of total cultivated area dedicated to herbaceous and woody crops in Guadalentín Basin.Note the sharp decline during the drought period .Data coming from CARM (1996).

Figure 4 .
Figure 4. Comparison of ETh and rainfall considering a drought period following a drought period.Alcantarilla, Lower Guadalentín Valley.

Figure 5 .
Figure 5. Monthly ETo-PM values in periods of normal rainfall and droughts.Guadalentín Valley.

Table 2 .
Main horticultural crops and irrigated woody crops of Guadalentín Valley and their annual water requirements during normal and drought periods (hm 3 ).Data were calculated following the model of ETo-Penman Monteith