THE IMPACT OF CLIMATE VARIABILITY ON WATER FOOTPRINT COMPONENTS OF RAINFED WHEAT AND BARLEY IN PROVINCE OF

RASTA NAZARI1, HADI RAMEZANI ETEDALI1, BIJAN NAZARI1 AND BRIAN COLLINS2

1Department of Water Sciences and Engineering, Imam Khomeini International University, Qazvin, Iran 2The Centre for Crop Science, The University of Queensland, Toowoomba, Australia

ABSTRACT

Due to the shortage of precipitation, rainfed farming is facing numerous challenges in Iran. Understanding the impact of climate variables on the production of rainfed crops in each region is of utmost importance for dry farming. Based on the 11-year (2004-2015) data from synoptic stations in of Iran, a comprehensive simulation analysis was conducted with AquaCrop-GIS to study the yield and the green and gray water footprint (WF) of the main rainfed crops (wheat and barley) along with the correlation of the target variables (yield and WF components) with the selected climate variables. Based on the estimated values of green and gray WFs, planting of rainfed wheat and barley in Qazvin and Moalem Kalayeh stations with lower total WF will be more beneficial than in other stations. Regression analysis showed that in most stations, reference evapotranspiration had a direct effect on wheat total WF (TWF) while precipitation has a positive effect on barley TWF in Qazvin and Moalem Kalayeh. The regression equation of barley green WF in the Qazvin station showed the highest correlation with climate variables (R2 = 0.98). TWF of wheat and barley in had the highest correlation with climate variables (R2 = 0.73 and 0.85, respectively). Finally, it was concluded that in arid regions, the variability in TWF of rainfed products was heavily influenced by spatiotemporal variations of climate variables.

† L’impact de la variabilité climatique sur les composantes de l’empreinte hydrique du blé et de l’orge pluviaux dans la province de Qazvin en Iran This is the author manuscript accepted for publication and has undergone full peer review but  Dr. Hadi Ramezani Etedali. Imam Khomeini International University, Department of Water Sciences has not been through the copyediting, typesetting, pagination and proofreading process, which and Engineering, Qazvin 34149-16818, Islamic Republic of Iran, T: +982818371279. E-mail: may lead to differences between this version and the Version of Record. Please cite this article [email protected]. as doi: 10.1002/ird.2487

1 This article is protected by copyright. All rights reserved. KEY WORDS: AquaCrop-GIS, climate variability, crop modelling, water footprint, regression analysis.

RĖSUMĖ

En raison du manque de précipitations, l’agriculture pluviale est confrontée à de nombreux défis en Iran. Comprendre l’impact des variables climatiques sur la production de cultures pluviales dans chaque région est de la plus haute importance pour l’agriculture sèche. Sur la base des données sur 11 ans (2004-2015) des stations synoptiques de la province de Qazvin en Iran, une analyse de simulation complète a été réalisée avec AquaCrop-GIS pour étudier le rendement et l’empreinte en eau verte et grise (WF) des principales cultures pluviales. (blé et orge) ainsi que la corrélation des variables cibles (composantes rendement et WF) avec les variables climatiques sélectionnées. Sur la base des valeurs estimées des WF vertes et grises, la plantation de blé et d’orge pluviaux dans les stations Qazvin et Moalem Kalayeh avec une WF totale plus faible sera plus bénéfique que dans d’autres stations. L’analyse de la régression a montré que dans la plupart des stations, l’évapotranspiration de référence avait un effet direct sur la WF total du blé (TWF) tandis que les précipitations avaient un effet positif sur le TWF de l’orge à Qazvin et Moalem Kalayeh. L’équation de régression de la WF verte de l’orge dans la station Qazvin a montré la corrélation la plus élevée avec les variables climatiques (R² = 0,98). La TWF de blé et d’orge de Buin Zahra présentait la corrélation la plus élevée avec les variables climatiques (R² = 0,73 et 0,85, respectivement). Enfin, il a été conclu que dans les régions arides, la variabilité de la TWF des produits pluviaux était fortement influencée par les variations spatio-temporelles des variables climatiques.

MOTS CLÉS: AquaCrop-GIS; variabilité climatique; modélisation des cultures; empreinte hydrique; analyse de régression.

INTRODUCTION

Wheat is the most important cultivated crop in Iran. Wheat, barley, lentil and chickpea account for about 99% of land area and 92% of the rainfed agricultural production in Qazvin province in . Iran is in the list of countries with water scarcity. The country’s renewable water resources are expected to drop to less than 1,500 m3 per capita by 2030 (Yang et al., 2006).

2 This article is protected by copyright. All rights reserved. Therefore, management of agricultural water consumption, the largest consumer of water in the country, is of utmost importance. The concept of ‘water footprint’ (WF) was introduced by Hoekstra (Hoekstra, 2003) and is an efficient tool for managing water resources in water scarce areas. This term is an indicator of the allocation of freshwater resources to various sections of the production process (Ababaei and Ramezani Etedali, 2014, 2017). This concept has been adopted in numerous studies on WF (e.g. Hoekstra and Hung, 2002; Hoekstra and Chapagain, 2007, 2008; Liu et al., 2007; Hoekstra and Mekonnen, 2012; Chenoweth et al., 2013; Hoekstra, 2017; Ababaei and Ramezani Etedali, 2014, 2017). Crop models are one useful tool to predict crop growth and development under various management and climate scenarios and to understand the way water is uptaken and appropriated during the production process. AquaCrop, introduced by Food and Agriculture Organization of the United Nations (FAO), is widely used as it requires less input data compared with most other well-known crop models (Raes et al., 2009). The benefits of adopting AquaCop include the flexibility of the model in implementing various management solutions, irrigation method, and the ability to simulate the effects of environmental stresses such as water stress, logging, salinity, fertility and heat. AquaCop has been used in different regions and for various purposes and crops (Farahani et al., 2009; Garcia-Vila et al., 2009; Geerts et al., 2009; Heng et al., 2009; Hsiao et al., 2009; Tavakoli et al., 2010; Andarzian et al., 2011; Salemi et al., 2011; Alizadeh et al., 2010; Babazade and Saraee Tabrizi, 2012; Ramezani Etedali et al., 2016). Knowledge of the causes of variability in WF is important to obtain information on water requirements corresponding to crop production and helps understand the impact of climate variables on WF components, which is especially important in rainfed cropping. Precipitation and temperature, as the most important climate variables, have significant effects on agricultural production. Precipitation plays an important role in the cultivation of rainfed products and is especially important to determine the appropriate time and place for cultivation of a specific crop in order to receive maximum precipitation and obtain highest yields. Therefore, the objectives of this study were: i) to quantify the variability of the yield of rainfed wheat and barley in the Qazvin province of Iran; ii) to estimate the variability in the WF of the studied crops; iii) to analyse the correlation between yield and climate variables.

MATERIALS AND METHODS

Study area

3 This article is protected by copyright. All rights reserved. The Qazvin province is located in the northwest of Iran. With an area of ~15,820 km2, Qazvin is located between N 36° 15΄and E 50° 00΄. The counties of the Qazvin province are Qazvin, , Buin Zahra, , and County (Figure 1). The province consists of two main basins of Shur-Rud and Sefid Rud. The Shur-Rud basin is the largest water catchment area in the province and includes 72.4% of the area of the Qazvin plain. Given its history, agricultural products and animal species and with an area of ~65,000 ha, the Qazvin plain is of an economic and historical significance in central Iran. The plain has an advanced irrigation network that, with an age of over 35 years, includes 1,122 ka of concrete canal. The irrigation and drainage network of Qazvin plain covers an area of 17,000 km2. Based on De Martonne aridity index, this region’s climate is semi-arid. The overall pattern of cultivation in the study area is 50% autumn crops, which are mostly wheat and barley. In addition, fruit gardens, forage and corn maize, forage crops and oilseeds are among the dominant cultures of the study area.

[Figure 1 is here]

According to the De Martonne climagram, the semi-arid cold climate is the largest climatic zone in Qazvin, Abyek and Takestan. In the highlands and in the north parts of these counties, with a decrease in average temperature, the semi-arid ultra-cold climate is observed. The driest region in the province is Buin Zahra and its surrounding areas in the east and south have an arid cold climate. Meanwhile, in the highlands of Avaj, ultra-cold wet and ultra-cold semi-humid climates are dominant. The annual precipitation of the province varies between 210 mm in the east and more than 550 mm on northeast mountains. The meteorological stations are presented in Table I.

[Table I is here]

AquaCrop-GIS To investigate the spatial variations of the target variables, AquaCrop-GIS (Lorite et al., 2013) was adopted. AquaCrop-GIS and AquaCrop-Data connect the AquaCrop model with ArcGIS enabling the simulation of numerous spatial units under various scenarios. A study was conducted to simulate the impacts of climate change on wheat yield in southern Spain by AquaCrop-GIS and AquaCrop-Data (Lorite et al., 2013). The results showed that AquaCrop- GIS is a powerful tool at the district, basin, and regional scales. In another study, the impacts of soil fertility and climate change in southern Alberta, Canada on the yield of various crops were

4 This article is protected by copyright. All rights reserved. simulated with AquaCrop-GIS (Langhorn, 2015). The model was able to adequately capture the impact of climate change on crop yield. Jiang et al. (2015) simulated the impact of spatial variability of irrigation water on the yield of various crops in China using SWAP-EPIC and ArcGIS. Raes et al. (2013) concluded that the prediction of crop yield with AquaCrop considering spatial variations in the quantity and quality of irrigation water, soil fertility, soil physical and chemical characteristics, groundwater level and quality and climate required about 1,000 hours of work, while AquaCrop-GIS and AquaCrop-Data helped save 99% of this time. AquaCrop-GIS (v2.1), with the possibility of executing high-level simulations, benefits from all the features and facilities of the AquaCrop crop model (Lorite et al., 2015). It requires weather data, including reference evapotranspiration, minimum and maximum temperature, precipitation values, and annual average atmospheric CO2 concentration. The weather data were obtained for the period of January 2004 to December 2015. Soil input parameters such as texture, thickness, field capacity (FC) and permanent wilting point (PWP) were considered as loam, 0.20 m, 32.2 and 16.1%, respectively (Ramezani Etedali et al., 2016). Calibrated crop parameters used in the model are presented in Table II (Ramezani Etedali et al., 2016).

[Table II is here]

Calculation procedure Our objective was to analyse the spatial distribution of the green and gray water footprints (WF) of barley and wheat in the case study area. Various WF components were estimated over the period from 2004 to 2015 using the framework proposed by Hoekstra et al. (2011) and modified by Ababaei and Ramezani Etedali (2014, 2017). The total green and blue crop water use (CWU) were expressed in mm year-1 or m3 ha-1 year-1 (1 mm = 10 m3 ha-1). The

3 -1 green CWU (CWUGreen, m ha ) was calculated as the minimum of ETc and Peff. ETc is the actual crop evapotranspiration (ETc, mm) and Peff is the effective precipitation (Peff, mm). The calculation of green CWU under rainfed conditions (CWUGreen, RF) was done using the following equation:

CWUGreen, RF min ET c, P eff  10  P eff (1) where RF denotes rainfed production systems. The factor 10 converts water depths from

3 -1 3 -1 millimetres into water volumes per land surface (m ha ). The blue CWU (CWUBlue, m ha ) was determined as the minimum of ETc and the effective irrigation supply (a portion of the irrigation water supply which is stored in the rooting profile). No blue water was considered

5 This article is protected by copyright. All rights reserved. under rainfed conditions. The blue WF was calculated using Eq. 2 under rainfed condition:

CWU Blue, RF  0 (2)

3 -1 Next, the green WF (WFGreen, m ton ) was calculated as the ratio green CWU to the actual crop yield (ton ha-1) under rainfed condition, following Ababaei and Ramezani Etedali (2014):

CWUGreen, RF WFGreen, RF  (3) YRF

-1 3 -1 where YRF (ton ha ) is crop yield under rainfed conditions. The gray CWU (CWUGray, m ha ) is the amount of freshwater required to dilute leached loads of pollution to a suitable level. In this research, for comparison pollution source in agricultural area, nitrogen and phosphorus application are studied.

3 -1 The gray WF (WFGray, m ton ) under rainfed condition was calculated using Eq. 4:

 AR RFRF WUGray, RF CCMax Nat WFGray, RF  (4) YYRF RF where AR (Kg ha-1) is the ratio of the application rate of chemicals per hectare (nitrogen or phosphorus and so on) which was obtained from the Ministry of Agriculture Jihad, α is the

-1 leaching-run-off fraction under rainfed conditions, CMax (mg l ) is the maximum allowable

-1 concentration, and CNat (mg l ) is natural concentration for the pollutant considered. The nitrate-

-1 N was considered as the characteristic pollutant and CMax was set at 10 mg l , according to the

-1 USEPA standard. CNat was conservatively set at zero mg l while α was set as 5 percent under

-1 rainfed conditions. For phosphorus (P), CMax and CNat were assumed to be 2 and 0 mg l respectively, according to the FAO (1994) water quality guidelines for agriculture, while α was set at 0.03 as suggested by the Gray Water Footprint Accounting: Tier 1 Supporting Guidelines

(Franke et al., 2013). The gray WF was calculated for nitrogen (WFGray (N)) and phosphorus

(WFGray (P)) fertilizers separately. To calculate the total WF (TWFRF), maximum value of gray WF for two types of fertilizers was considered:

WFGray = MAX (WFGray (N), WFGray (P)) (5)

6 This article is protected by copyright. All rights reserved.

Finally, TWFRF is the sum of the green and gray WF components:

nn (6) TWFWFWFRF Green,, RF Gray RF ii11

3 -1 where TWFRF (m ton ) denotes the total WF in rainfed lands, and n is the number of barley and wheat producing cities (5 cities) in Qazvin province.

Statistical analysis Statistical analyses were performed with SPSS (v16.0) and Minitab (v17.0). Pearson, Spearman (Pearson’s r on ranks) and Kendall’s Tau correlation coefficients were adopted to investigate the correlations between TWF and climate variables. The Pearson correlation is the most frequently used coefficient for normally distributed data (Chok, 2010). On the other hand, nonparametric measures such as Spearman’s rank-order and Kendall’s tau correlation coefficients are usually recommended for non-normal data with both correlation coefficients being based on ranks and suitable for data with skewness. Therefore, they are resistant to effects of outliers (Helsel and Hirsch, 2002). Tau (τ) measures all monotonic correlations (linear and nonlinear; Helsel and Hirsch, 2002). The regression equations between climate variables and yield and WF components along with the associated coefficient of determination (R2) are presented.

RESULTS AND DISCUSSION

Precipitation and reference evapotranspiration

Figure 2 shows precipitation (PRE in mm) and reference evapotranspiration (ETo in mm) during the growing season of autumn-sown wheat and barley across five weather stations over

2004–2015. In all stations, the average annual ETo was higher than precipitation for rainfed wheat and barley and this gap was wider for wheat than barley and in Buin Zahra than in the other stations for its drier climate.

[Figure 2 is here]

Crop yield

7 This article is protected by copyright. All rights reserved. The grain yields of the rainfed wheat and barley simulated with AquaCrop-GIS are present in Figure 3. On average, barley yields were lower than wheat in all five stations. For both crops, the yield was lowest in Buin Zahra and highest in Moalem Kalayeh.

[Figure 3 is here]

Zoning average annual yield and harvest index by AquaCrop-GIS The average annual yield and harvest index were zoned with AquaCrop-GIS across the five counties (Figure 4). The average simulated yield of barley ranged from 0.7 to 1.0 ton ha-1, and harvest index from 38.9 to 53.2% (Figure 4a, b). The average yield of wheat ranged between 1.1 and 1.7 ton ha-1 and harvest index varied between 37.4 and 54.1% (Figure 4c, d). For rainfed barley, the average yield and harvest index at Moalem Kalayeh, Qazvin and Avaj were higher than in other two stations while Moalem Kalayeh and Qazvin had the highest simulated yield and harvest index among the five stations.

[Figure 4 is here]

Relationships between yield and climate variables Regression analysis was performed in order to investigate the impact of spatiotemporal variability of climate variables on yield of wheat and barley (Tables III and IV). First, the correlation between yield and each of the climate variables was examined (Table III). The relationship between yield and precipitation in all stations was positive for both crops while the relationship between yield and evapotranspiration was negative in all stations except for Buin Zahra. Precipitation for barley in Buin Zahra and evapotranspiration for wheat in Takestan had the highest correlation coefficients (R2 = 0.78) at a 1% significant level. Minimum and maximum temperatures had the highest correlation coefficient (R2 = 0.67 and 0.76, respectively) at a significant level of 1% for wheat in Buin Zahra. In Qazvin, Avaj and Moalem Kalayeh, the correlations between the target climate variables and barley yield was higher than the correlations with wheat yield. However, the correlations between climate variables and wheat yield were stronger in Takestan and Buin Zahra than with barley. The yields of wheat and barley in these two counties were lowest due to their warm and dry climate (Figure 4). For wheat, regression equation in Buin Zahra, with positive effect of precipitation (PRE), evapotranspiration (ETo) and minimum temperature (Tmin) and negative effect of maximum

2 temperature (Tmax), had the highest coefficient of determination (R = 0.88) than the other equations.

8 This article is protected by copyright. All rights reserved.

[Tables III and IV are here]

Total water footprint of rainfed wheat and barley The average yield and fertilizer consumption (nitrogen and phosphorus) for each crop are presented in Table V. The gray WF(P) is higher than that the gray WF(N) in all stations. Thus, TWF was derived from the sum of green WF and the gray WF for phosphorus fertilizers.

nn (7) TWFWFWFRF Green, RF Gray ( P ), RF ii11

The green and gray WFs for wheat and barley are shown in Figures 5 and 6, respectively. Values related to barley were estimated to be generally higher than those for wheat in all counties. The green and gray WFs for both crops in Qazvin were lower than in the other counties and Buin Zahra showed the highest values among the studied stations. Distribution of green WF of both crops in Buin Zahra and for barley in Takestan was skewed due to the existence of outliers while gray WF of barley was not normally distributed in most stations.

[Table V is here] [Figure 5 is here] [Figure 6 is here]

The average green and gray WFs and TWF in Buin Zahra were highest while Qazvin and Moalem Kalayeh had the lowest values (Figure 7). Therefore, planting of rainfed wheat and barley in Qazvin and Moalem Kalayeh with lower TWF (Figure 8) would lead to higher water productivity than in other counties of the Qazvin province. Except for outliers caused by low yields in Buin Zahra in 2005-2006 for wheat and barley and in Takestan for barley in 2008- 2009, the time series of both crops generally had a clear and regular trend. In all stations, the highest value of TWF for rainfed wheat was estimated in the 2014-2015 season, which can be attributed to a low yield in this year. That was also the case for wheat and barley in Buin Zahra in 2005-2006 and for barley in Takestan in 2008-2009.

[Figure 7 is here] [Figure 8 is here]

9 This article is protected by copyright. All rights reserved. Relationships between total WF and climate variables In order to study the relationships between TWF and climate variables, the distribution of TWF was first examined at a 5% significant level (Figure 9). The probability distribution in Qazvin, Avaj and Moalem Kalayeh were normal for both crops, while it was normal for rainfed wheat in Takestan at a 1% significant level and was not normal in Buin Zahra. The distribution of TWF for rainfed barley in Takestan and Buin Zahra were not normal as well.

[Figure 9 is here]

The correlations between TWF and climate variables are presented in Table V. In stations with normal distributions (Figure 9), the Pearson’s correlation coefficient was significant at a

5% significance level in some cases (PRE for barley at Qazvin and Moalem Kalayeh and ETo for wheat at Takestan). In other cases (ETo for wheat in Avaj and Moalem Kalayeh and PRE for barley in Moalem Kalayeh), Spearman’s correlation falls in between the Pearson’s and Kendall’s coefficients at a 5% significance level. For stations with non-normal distributions (Figure 9), Spearman and Kendall tau were less sensitive to outliers while Pearson correlation coefficients were not reliable for both crops in Buin Zahra. Only in one case (PRE for barley in Moalem Kalayeh) the results of all three coefficients agreed and were significant at P = 0.05.

In most stations, ETo had a direct effect on TWF of wheat and precipitation had a positive effect on TWF of barley in Qazvin and Moalem Kalayeh. Thus, unlike the relationship between yield and evapotranspiration (Table III), TWF of wheat increased with increasing evapotranspiration. In Buin Zahra and due to low yields (Figure 4), TWF increased with decreasing evapotranspiration. In addition, TWF of barley had a positive correlation with evapotranspiration, except in cases where yield was too low. In Qazvin, The green WF regression equations showed higher correlations with climate variables for both crops (Table VI). The R2 of the equation was estimated to be 0.98 for barley. In Moalem Kelayeh, the R2 of the green WF equation was estimated to be 0.88. Generally, The correlation between climate variables and the gray WF of wheat was less than those of the green WF, except in Takestan. The TWF regression equation in Buin Zahra has the highest correlations with climate variables for wheat and barley with the R2 estimated to be 0.73 and 0.85, respectively. Buin Zahra had lower yield than other stations, TWF increased with decreasing yield under the influence of climate variables (Table VI). On average over a period of 11 years, the green WF correlation of barley yield to climate variables was stronger than that of wheat yield.

10 This article is protected by copyright. All rights reserved. [Tables VI and VII are here]

CONCLUSION

In this study, an analysis was conducted with the aim of estimating grain yield along with green and gray WF of main rainfed wheat and barley and their spatiotemporal correlations with climate variables. The results showed that the average yield of barley was lower than wheat in all stations over the period of 2004-2015. For both wheat and barley, yield was lowest in Buin Zahra (driest region in the province) and highest in Moalem Kalayeh. The relationship between yield and precipitation in all stations was positive for both crops, while the relationship between yield and evapotranspiration was negative in all stations except for Buin Zahra. Generally, we found that any increase in evapotranspiration would lead to an increase in the total WF unless grain yield was too low. This result confirms that yield variability plays a significant role in the variability of water footprint of field crops. It was determined that in arid regions the total WF undergoes significant changes due to the significant impact of climate variables on yield. Thus, planting rainfed wheat and barley in Qazvin and Moalem Kalayeh with lower WF is recommended due to higher water productivity than in other regions in the province. We conclude that with proper management in the agricultural sector, it is possible to reduce the additional pressure on limited water resources in the Qazvin plain.

REFERENCES

Ababaei B, Ramezani Etedali HR. 2014. Estimation of water footprint components of Iran’s wheat production: comparison of global and national scale estimates. Environmental processes, 1(3), 193-205. Ababaei B, Ramezani Etedali HR. 2017. WF assessment of main cereals in Iran. Agricultural Water Management, 179, 401-411. Alizadeh H, Nazari B, ParsiNejhad M, Ramezani Etedali H. 2010. Evaluation AquaCrop model in deficit irrigation management wheat and barley in , Irrigation and Drainage, 4: 283-273 (in Persian). Andarzian B, Bannayan M, Steduto P, Mazraeh H, Barati ME, Barati MA, Rahnama A. 2011. Validation and testing of the AquaCrop model under full and deficit irrigated wheat production in Iran. Agric. Water Manag. 100 (1): 1-8.

11 This article is protected by copyright. All rights reserved. Babazade H, Saraee Tabrizi M. 2012. AquaCrop model evaluation under deficit irrigation management Soybean.Water and Soil Agricultural Sciences and Technology, 2:329-339 (in Persian). Chenoweth J, Hadjikakou M, Zoumides C. 2013. Review article: Quantifying the human impact on water resources: A critical review of the WF concept. Hydrol. Earth Syst. Sci. Discuss, 10, 9389–9433. Chok NSh. 2010. Pearson’s Versus Spearman’s and Kendall’s Correlation Coefficients for Continuous Data. Master’s Thesis, University of Pittsburgh. Pennsylvania, USA. Food and Agriculture Organization of the United Nations (FAO). 1994. Water quality for agriculture. R.S. Ayers and D.W. westcot. Irrigation and Drainage Paper 29 Rev.l. FAO, Rome, Italy. Farahani HJ, Izzi G, Oweis TY. 2009. Parameterization and evaluation of the AquaCrop model for full and deficit irrigated . Agron. Agronomy journal, 101(3), 469-476. Franke NA, Boyacioglu H, Hoekstra AY. 2013. Gray water footprint accounting: Tier 1 supporting guidelines (Rep. 65). UNESCO-IHE. Delft, the Netherlands. García-Vilaa M, Fereresa E. 2012. Combining the simulation crop model AquaCrop with an economic model for the optimization of irrigation management at farm level. Europ. J. Agronomy 36, 21– 31. Geerts S, Raes D, Garcia M, Miranda R, Cusicanqui JA, Taboada C, Mendoza J, Huanca R, Mamani A, Condori O, Mamani J, Morales B, Osco V, Steduto P. 2009. Simulating yield response to water of Quinoa (Chenopodium quinoa Willd.) with FAO-AquaCrop. Agron. J. 101, 499-508. Helsel DR, Hirsch RM. 2002. Statistical methods in water resources: U.S. Geological Survey Techniques of Water-Resources Investigations, book 4, chap. A3, 524 p. Available online at http://water.usgs.gov/pubs/twri/twri4a3/ Heng LK, Hsiao TC, Evett S, Howell T, Steduto P. 2009. Validating the FAO AquaCrop model for irrigated and water deficient field maize. American Society of Agronomy, 101, 488- 498. 30. Hoekstra AY, Chapagain AK. 2007. WFs of nations: water use by people as a function of their consumption pattern. Water Resour Manag, 21(1): 35-48. Hoekstra AY, Chapagain AK. 2008. Globalization of water: Sharing the planet’s freshwater resources. Blackwell Publishing, Oxford, United Kingdom. Hoekstra AY, Hung PQ. 2002. Virtual water trade: A quantification of virtual water flows between nations in relation to international crop trade. Value of Water Research, Report Series No 11, UNESCO-IHE. Delft, the Netherlands.

12 This article is protected by copyright. All rights reserved. Hoekstra AY. 2003. Virtual water trade. In: Proceedings of the International Expert Meeting on Virtual Water Trade, Delft, the Netherlands, 12–13 December 2002; Value of Water Research Report Series No. 12; UNESCO-IHE. Delft, the Netherlands. Hoekstra AY. 2017. WF assessment: Evolvement of a new research field. Water Resour. Manag. 31, 3061–3081. Hoekstra AY, Chapagain AK. 2007. WF of nations: Water use by people as a function of their consumption pattern. Water Resour. Manag, 21, 35–38. Hoekstra AY, Chapagain AK, Aldaya MM, Mekonnen MM. 2011. The WF assessment manual: setting the global standard. Earthscan. London, United Kingdom. Hoekstra AY, Mekonnen MM. 2012. The WF of humanity. Proc. Natl. Acad. Sci. USA, 109, 3233–3237. Hsiao TC, Heng LK, Steduto P, Rojas-Lara B, Raes D, Fereres E. 2009. AquaCrop-the FAO crop model to simulate yield response to water, III: Parameterization and testing for maize. Agronomy Journal, 101, 448-459. Jiang Y, Xu X, Huang QZ, Huo ZL, Huang GH. 2015. Assessment of irrigation performance and water productivity in irrigated areas of the middle Heihe River Basin using a distributed agro-hydrological model. Agr. Water Manage, 147, 67-81. Liu J, Williams JR, Zehnder AJB, Yang H. 2007. GEPIC – modeling wheat yield and crop water productivity with high resolution on a global scale. AgrSyst, 94:478-493. Lorite IJ, Garcia-Vila M, Fereres E. 2015. AquaCrop-GIS. Version 2.1. Reference manual. Rome, Italy. Lorite IJ, García-Vila M, Santos C, Ruiz-Ramos M and Fereres E. 2013. AquaData and AquaCrop-GIS: Two computer utilities for temporal and spatial simulations of water - limited yield with AquaCrop. Computers and Electronics in Agriculture 96: 227-237. Ramezani Etedali H, Liaqat A, Parsinejhad M, Tavakoli A. 2016. AquaCrop in irrigation management model calibration and verification ofimportant cereals, Journal of Irrigation and Drainage, 3:389-397 (in Persian). Salemi HR, Soom MAM, Lee TS, Mousavi SF, Ganji A, Yusoff MK. 2011. Application of AquaCrop model in deficit irrigation management of winter wheat in arid region. African J. Agric. Res. 610, 2204-2215. Tavakoli AR, Oweis T, Ashrafi Sh, Asadi H, Siadat H, Liaghat A. 2010. Improving rainwater productivity with supplemental irrigation in upper Karkheh river basin of Iran. International Centre for Agricultural Research in the Dry Areas (ICARDA). Aleppo. Syria. Yang H, Wang L, Abbaspour KC, Zehnder AJB. 2006. Virtual water trade: an assessment of

13 This article is protected by copyright. All rights reserved. water use efficiency in the international food trade. Hydrology and Earth System Sciences, 10: 443–454, DOI: 10.5194/hess-10-443-2006.

14 This article is protected by copyright. All rights reserved. Table I. Meteorological stations across the Qazvin province of Iran Code Weather station parameters Station (Figure 4) Longitude (N) Latitude (E) Altitude (m) 1 Qazvin Airport 50° 03΄ 36° 15΄ 1279.2 2 Moalem Kelayeh 50° 29΄ 36° 37΄ 1629.2 3 Takestan 49° 42΄ 36° 03΄ 1283.4 4 Buin Zahra 50° 04΄ 35° 46΄ 1225 5 Avaj 49° 13΄ 35° 34΄ 2034.9

Table II. AquaCrop parameters used for simulations Crop characteristics Rainfed wheat Rainfed barley Planting method Direct sowing Planting date 6-Nov 1-Nov Base temperature (°C) 0 0 Upper temperature (°C) 26 26 Crop coefficient 1.10 1.10 Water productivity (g m-2) 15 15 Maximum effective rooting depth (m) 1.20 1.00

Reference Harvest Index (HIo) (%) 50 49

15 This article is protected by copyright. All rights reserved. Table III. Regression coefficients between climate variables and yield of wheat and barley (2004-2015)

Station Crop Variable Coefficient R2 PRE 0.00157* 0.53

ETo -0.00139* 0.38 Wheat Tmin 0.120 0.13 Tmax 0.0596 0.07 Qazvin PRE 0.00210 0.24

ETo -0.00425* 0.54 Barley Tmin 0.220 0.22 Tmax 0.083 0.07 PRE 0.00109* 0.40

ETo -0.00229** 0.55 Wheat Tmin 0.112 0.19 Tmax 0.0292 0.02 Avaj PRE 0.00582* 0.54

ETo -0.00845* 0.44 Barley Tmin 0.536 0.22 Tmax 0.267 0.08 PRE 0.00121 0.33

ETo -0.00162* 0.50 Wheat Tmin 0.0898 0.10 Tmax 0.0243 0.01 Moalem Kelayeh PRE 0.00178* 0.37

ETo -0.00251 0.24 Barley Tmin 0.095 0.07 Tmax 0.012 0.00 PRE 0.00272** 0.59

ETo -0.00234** 0.78 Wheat Tmin 0.080 0.03 Tmax 0.0395 0.02 Takestan PRE 0.00791 0.17

ETo -0.0112** 0.62 Barley Tmin 0.402 0.08 Tmax 0.204 0.06 Buin Zahra Wheat PRE 0.00183 0.30

16 This article is protected by copyright. All rights reserved. ETo 0.00261 0.32 Tmin -0.154** 0.67 Tmax -0.11** 0.76 PRE 0.00639** 0.78

ETo 0.00162 0.03 Barley Tmin -0.179 0.17 Tmax -0.139 0.23 PRE 0.00808* 0.46

ETo -0.00712 0.28 Wheat Tmin 0.167 0.01 Qazvin Province Tmax -0.094 0.01 (Average of 11 years) PRE 0.000116 0.01

ETo 0.00023 0.02 Barley Tmin -0.0241 0.04 Tmax -0.0322 0.15 PRE: precipitation (mm), ETo: reference evapotranspiration (mm), Tmin and Tmax: minimum and maximum temperature (˚C), ‘**’ and ‘*’ show the significance level of 99% and 95% respectively.

17 This article is protected by copyright. All rights reserved. Table IV. Regression equations between climate variables and yield of wheat and barley (2004- 2015) P- Station Crop Regression equation R2 value

Yield = -0.72 + 0.00133 PRE - 0.000641 ETo - 0.148 Tmin Wheat 0.64 0.135 + 0.158 Tmax Qazvin Yield = 3.41 + 0.00092 PRE - 0.00305 ETo + 0.317 Tmin - Barley 0.73 0.065 0.105 Tmax

Yield = 2.38 + 0.000610 PRE - 0.00081 ETo + 0.150 Tmin Wheat 0.63 0.145 - 0.071 Tmax Avaj Yield = -2.63 + 0.00478 PRE - 0.00307 ETo + 0.174 Tmin Barley 0.74 0.057 + 0.236 Tmax

Yield = 1.99 + 0.000517 PRE - 0.00132 ETo - 0.009 Tmin Wheat 0.54 0.249 Moalem + 0.023 Tmax

Kalayeh Yield = 3.26 + 0.00169 PRE - 0.0023 ETo + 0.032 Tmin - Barley 0.59 0.191 0.025 Tmax

Yield = 2.14 + 0.00108 PRE - 0.00183* ETo + 0.026 Tmin Wheat 0.86 0.009 + 0.0158 Tmax Takestan Yield = 3.15 + 0.00566 PRE - 0.011** ETo + 0.417 Tmin - Barley 0.81 0.022 0.011 Tmax

Yield = 2.79 + 0.00106 PRE + 0.00129 ETo + 0.303 Tmin Wheat 0.88 0.007 - 0.268 Tmax Buin Zahra Yield = 1.37 + 0.00592* PRE - 0.00045 ETo + 0.093 Tmin Barley 0.79 0.032 - 0.097 Tmax

Qazvin Yield = 12.12 + 0.00759 PRE + 0.00162 ETo + 1.20 Tmin Wheat 0.55 0.238 Province - 0.870 Tmax

(Average of Yield = 1.989 + 0.000067 PRE + 0.000667 ETo Barley 0.40 0.469 11 years) + 0.188 Tmin - 0.1406 Tmax Yield (Kg ha-1), PRE: precipitation (mm), ETo: reference evapotranspiration (mm), Tmin and Tmax: minimum and maximum temperature (˚C), ‘**’ and ‘*’ show the significance level of 99% and 95% respectively.

18 This article is protected by copyright. All rights reserved. Table V. Water footprint (WF) components of rainfed wheat and barley (2004-2015) AR (kg ha-1) WF (m3 ton-1) Crops Station Yield (kg ha-1) Gray N P Green Total N P Qazvin 1530 937 167 394 1330 Avaj 1330 1210 194 450 1660 Moalem Kelayeh 1550 50 39 1010 166 388 1400 Takestan 1310 946 197 467 1410 Wheat Buin Zahra 1190 1400 230 531 1930 Average 1380 1100 191 446 1550 CV (%) 11 18 14 13 16

Min 1190 937 166 388 1330 Max 1550 1400 230 531 1930 Qazvin 949 1460 206 583 2040 Avaj 934 1710 210 595 2300 Moalem Kelayeh 950 39 37 1600 206 581 2190 Takestan 855 1440 240 714 2150 Barley Buin Zahra 807 2310 305 774 3090 Average 899 1700 233 649 2350 CV (%) 7 21 18 14 18 Min 807 1440 206 581 2040 Max 950 2310 305 774 3090 AR: the chemical application rate to the field per hectare (nitrogen or phosphorus) was obtained from the Ministry of Agriculture Jihad.

19 This article is protected by copyright. All rights reserved. Table VI. Relationships between total water footprint (TWF) and climate variables (2004-2015) Station Crop Variable Pearson’s r Spearman’s Rho Kendall’s Tau PRE -0.236 -0.318 -0.236

ETo 0.266 0.336 0.273 Wheat Tmin -0.283 -0.228 -0.212 Tmax -0.214 -0.261 -0.262 Qazvin PRE 0.647* 0.482 0.345

ETo -0.117 0.045 0.018 Barley Tmin -0.335 -0.326 -0.250 Tmax -0.331 -0.247 -0.224 PRE -0.289 -0.273 -0.127

ETo 0.530 0.682* 0.455 Wheat Tmin -0.385 -0.477 -0.374 Tmax -0.131 -0.046 -0.073 Avaj PRE 0.100 0.227 0.164

ETo 0.111 0.191 0.127 Barley Tmin -0.259 0.112 0.098 Tmax -0.430 -0.415 -0.257 PRE -0.195 -0.336 -0.200

ETo 0.579 0.664* 0.455 Wheat Tmin -0.321 -0.336 -0.267 Tmax -0.189 -0.352 -0.241 Moalem Kelayeh PRE 0.691* 0.736* 0.564*

ETo 0.143 0.191 0.164 Barley Tmin -0.262 -0.313 -0.267 Tmax -0.357 -0.338 -0.204 PRE -0.093 0.236 0.127

ETo 0.654* 0.445 0.309 Wheat Tmin 0.050 0.032 0.019 Tmax -0.097 0.014 0.019 Takestan PRE 0.023 0.318 0.236

ETo 0.185 0.318 0.200 Barley Tmin -0.188 -0.060 -0.057 Tmax -0.121 -0.092 -0.057 PRE -0.161 -0.327 -0.200

Buin Zahra Wheat ETo -0.654* -0.082 -0.055 Tmin 0.789** 0.270 0.173

20 This article is protected by copyright. All rights reserved. Tmax 0.824** 0.464 0.345 PRE -0.281 -0.064 -0.018

ETo -0.744** -0.382 -0.273 Barley Tmin 0.894** 0.540 0.443 Tmax 0.904** 0.173 0.091 PRE -0.182 -0.082 -0.073

ETo 0.208 0.182 0.091 Wheat Tmin -0.047 0.041 0.094 Qazvin Province Tmax 0.141 0.179 0.132 (Average of 11 PRE 0.345 0.382 0.236 years) ETo -0.401 -0.318 -0.236 Barley Tmin 0.128 0.014 0.019 Tmax 0.244 0.101 0.057 PRE: precipitation (mm), ETo: reference evapotranspiration (mm), Tmin and Tmax: minimum and maximum temperature (˚C), ‘**’ and ‘*’ show the significance level of 99% and 95% respectively.

21 This article is protected by copyright. All rights reserved. Table VII. Regression equations between climate variables and water footprint (WF) components (2004-2015) Station Crop Regression equation R2

WFGreen = 1560 + 0.929 PRE + 0.875 ETo + 76 Tmin - 92 0.44 Tmax 0.22 Wheat WFGray = 143 - 0.94 PRE - 0.12 ETo - 207 Tmin + 96 Tmax 0.14 TWF = 1700 - 0.01 PRE + 0.75 ETo - 130 Tmin + 5 Tmax

Qazvin WFGreen = 385 + 3.92 PRE + 1.08 ETo + 51.1 Tmin - 27.0 0.98 Tmax 0.29

Barley WFGray = 894 - 0.89 PRE - 0.71 ETo - 348 Tmin + 127 Tmax

TWF = 1280 + 3.03 PRE + 0.37 ETo - 297 Tmin 0.58 + 100 Tmax

WFGreen = -12 + 0.286 PRE + 0.69 ETo - 133 Tmin 0.41 + 64 Tmax 0.22

Wheat WFGray = -1720 + 0.17 PRE + 2.06 ETo - 51 Tmin + 22 Tmax 0.35

TWF = -1730 + 0.46 PRE + 2.75 ETo - 184 Tmin + 86 Tmax

Avaj WFGreen = 778 + 1.35 PRE + 1.211 ETo + 2.4 Tmin - 0.76 18.2 Tmax 0.30

WFGray = 4020 - 0.88 PRE + 0.48 ETo + 221 Tmin - Barley 265 Tmax 0.30 TWF = 4800 + 0.47 PRE + 1.69 ETo + 224 Tmin - 283 Tmax

WFGreen = -19 + 0.753 PRE + 0.954 ETo - 26.0 Tmin 0.52 + 16.0 Tmax 0.32

Wheat WFGray = 797 - 0.112 PRE + 1.39 ETo + 63 Tmin - 102 Tmax 0.40

TWF = 778 + 0.64 PRE + 2.34 ETo + 37 Tmin - 86 Tmax Moalem Kelayeh WFGreen = 482 + 1.890PRE + 0.138 ETo - 41.6 Tmin 0.88 + 45.0 Tmax 0.24

Barley WFGray = 2530 - 0.090 PRE + 0.29 ETo - 18 Tmin - 101 Tmax 0.66

TWF = 3010 + 1.8 PRE + 0.43 ETo - 59 Tmin - 56 Tmax

WFGreen = -589 + 1.99 PRE + 1.400 ETo - 17 Tmin + 5 Tmax 0.52

WFGray = 1410 + 1.46 PRE + 3.086 ETo + 447 Tmin - 0.72 Wheat Takestan 345 Tmax 0.76 TWF = 817 + 3.44 PRE + 4.49 ETo + 431 Tmin - 340 Tmax

Barley WFGreen = -1410 + 4.00 PRE + 2.02 ETo - 735 Tmin 0.15

22 This article is protected by copyright. All rights reserved. + 327 Tmax 0.14

WFGray = 2840 - 2.07 PRE + 2.20 ETo + 41 Tmin - 151 Tmax 0.08

TWF = 1420 + 1.9 PRE + 4.22 ETo - 695 Tmin + 176 Tmax

WFGreen = -4090 + 1.48 PRE - 0.03 ETo - 251 Tmin 0.77 + 357 Tmax 0.39

WFGray = -1490 - 0.22 PRE - 0.59 ETo - 331 Tmin Wheat + 248 Tmax 0.73 TWF = -5580 + 1.26 PRE - 0.61 ETo - 582 Tmin Buin Zahra + 606 Tmax

WFGreen = -11000 + 3.24 PRE - 0.03 ETo - 311 Tmin 0.89 + 780 Tmax 0.58 WF = -2120 + 0.72 PRE - 0.13 ET - 138 Tmin Barley Gray o + 195 Tmax 0.85 TWF = -13100 + 3.97 PRE - 0.15 ETo - 449 Tmin + 975 Tmax WFGreen = -511 - 0.35 PRE - 1.63 ETo - 377 Tmin 0.49 + 290 Tmax 0.20 Wheat WF = -543 + 0.01 PRE + 1.34 ET - 42 Tmin + 8 Tmax Gray o 0.14 TWF = -1050 - 0.34 PRE - 0.29 ET - 418 Tmin + 298 Tmax Qazvin Province o WF = -4070 + 1.86 PRE - 4.85 ET - 892 Tmin 0.71 (Average of 11 Green o + 712 Tmax 0.38 years) WF = 1800 + 0.26 PRE - 0.39 ET - 242 Tmin Barley Gray o + 32 Tmax 0.54 TWF = -2270 + 2.12 PRE - 5.24 ETo - 1134 Tmin + 744 Tmax PRE: precipitation (mm), ETo: reference evapotranspiration (mm), Tmin and Tmax: minimum and maximum temperature (˚C), WFGreen: green WF (m3 ton-1), WFGray: gray WF (m3 ton-1), TWF: total WF (m3 ton-1).

23 This article is protected by copyright. All rights reserved.