Professional Documents
Culture Documents
A R T I C LE I N FO A B S T R A C T
Keywords: Irrigation water requirement is influenced by climatic fluctuations which can be simulated using agro-meteor-
Irrigation prediction ological models. This study investigates fluctuating irrigation water requirements for rice as affected by drought
Reconnaissance drought index occurrences. Irrigation requirements were simulated by the AquaCrop model in paddy fields of Guilan in
Rice ET northern Iran. Model validation was done based on field measurements during two consecutive years of 2012
and 2013 in the study area. The reconnaissance drought index (RDI), based on cumulative values of precipitation
and potential evapotranspiration, was used for drought monitoring for 1982–2014, in two time scales of 3- and
6-month. Also, irrigation water requirements (IWR) were calculated for this period. The normalized root mean
square error (NRMSE) for simulating canopy cover, biomass, rice evapotranspiration and deep percolation by
AquaCrop were obtained 7.0, 8.8, 18.45 and 26.6%, respectively. The contributions of crop evapotranspiration
and deep percolation in IWR were 70.5% and 22.9%, respectively. There are more drought occurrences after
1994. A good correlation was obtained between simulated rice IWR and RDI of July (calculated with 3-month
time scale) (R = -0.89). Maximum amounts of required irrigation water in wet, normal and dry years, were
6750, 8050 and 8760 m³ ha−1, respectively. Obtained relationship between rice IWR and RDI of July with a 3-
month time scale can be used to allocate paddies irrigation water of the studied area at transplanting time based
on seasonal forecasts of drought.
⁎
Corresponding author.
E-mail addresses: npirmoradian@guilan.ac.ir (N. Pirmoradian), n_davatgar@yahoo.com (N. Davatgar).
https://doi.org/10.1016/j.agwat.2018.10.003
Received 6 June 2018; Received in revised form 2 October 2018; Accepted 4 October 2018
0378-3774/ © 2018 Elsevier B.V. All rights reserved.
N. Pirmoradian, N. Davatgar Agricultural Water Management 213 (2019) 97–106
et al., 2014). A major advantage of water-driven models over radiation- 2. Materials and methods
driven models is the opportunity to normalize the water productivity
(WP*) parameter for climate, making these models widely applicable in 2.1. Study area location
different locations under varying climatic and spatio-temporal settings
(Steduto and Albrizio, 2005; Steduto et al., 2007). Guilan province is located at south of the Caspian Sea and bounded
Guilan province is located at the northern part of Iran (Fig. 1). Rice to Talesh Mountains westward and Alborz Mountain Range southward.
as the most important crop is produced from about 230,000 ha in The plain regions of thisprovince are located between the sea and the
Guilan (Ministry of Jahad-e-Keshvarzi, 2015). Sefidrood Dam is located mountains. Guilan’s climate is known as “moderate Caspian” which is
on Sefidrood River upstream of Guilan plain and supplies irrigation under the influence of weather from both Alborz Mountains and
water for 189,000 ha of rice cultivated area. The rest of the paddy fields Caspian Sea. The Sefidrood Irrigation and Drainage Network consist of
are irrigated by local river or groundwater resources. In recent years, three districts (East, Central and Foumanat) and 17 command areas.
various water storage and adjustments dams have been built in basins This study was conducted at Foumanat District in the eastern part of the
upstream of Sefidrood Dam, resulting in less water entering the Sefi- F1 command area (east longitude between 49˚ 38ʹ to 49˚ 47ʹ and north
drood Dam reservoir and a water supply crisis in the Guilan Province. latitude between 37˚ 16ʹ to 37˚ 26ʹ), where there were 3293 ha of paddy
Now, in the operation of Sefidrood Irrigation and Drainage Network, rice in 2012.
IWR is estimated for normal condition based on the average of a cli- The daily weather data required by AquaCrop including maximum
matic period, and does not consider the effects of wet or dry conditions and minimum temperature (Tmax, Tmin), sunshine hours (n), precipita-
on IWR, crop yield, or water productivity. It is necessary to account for tion (P), maximum and minimum relative humidity (RHmax, RHmin) and
climatic fluctuations, particular drought, in a more realistic estimation wind speed at 2 m above ground (U), were obtained from Rasht
of IWR. Agricultural Meteorological Station (Lat. 37˚12′ N; Long. 49˚39′ E;
Drought indices are important tools for monitoring and assessing Elevation of 24.6 m). Mean monthly data for the period 1982–2014 are
drought for planners, designers and water resources managers (Richard shown in Table 1.
and Heim, 2002; Hayes, 2004). Irrigation requirements vary by climatic
fluctuations that are monitored by drought indices. Therefore, drought 2.2. Field measurements
monitoring can be used to predict irrigation requirements before the
start of the growing season. One of the most widely used indices for Field measurements were conducted during two years, 2012 and
assessing the severity of the drought is Reconnaissance Drought Index
(RDI), calculated on the basis of monthly amounts of precipitation and Table 1
potential evapotranspiration. Water scarcity could be better expressed Mean monthly meteorological data in Rasht between 1982 and 2014.
using RDI with regard to evapotranspiration as compared to indices that
Month n Tmin Tmax RHmin RHmax P
are only based on precipitation. RDI can be more effectively associated
(hr/month) (˚C) (˚C) (%) (%) (mm/month)
with hydrological and agricultural drought. Also, it is an ideal index to
study the effects of climate instability (Tsakiris and Vangelis, 2005). Jan 100.6 3.0 11.2 70 97 130.6
This study was conducted to estimate IWR of paddy rice in Guilan as Feb 92.8 3.1 11.2 70 97 119.6
Mar 103.2 5.7 13.9 68 97 106.4
affected by climatic fluctuations using the FAO AquaCrop model, and
Apr 129.4 10.3 19.4 64 97 64.0
relate the simulated IWR to the corresponding RDI. The aim was to find May 180.0 15.2 24.1 63 96 44.8
a relationship between RDI and irrigation water requirements in order Jun 223.3 19.4 28.3 60 95 41.0
to be able to predict irrigation water requirements before the start of Jul 230.9 21.3 30.4 59 94 38.1
the growing season. Aug 196.0 21.2 30.4 61 95 73.3
Sep 139.7 18.5 26.9 68 97 155.5
Oct 122.7 14.0 22.4 71 98 213.6
Nov 100.4 9.0 17.2 72 98 199.7
Dec 96.4 5.0 13.2 71 97 152.3
98
N. Pirmoradian, N. Davatgar Agricultural Water Management 213 (2019) 97–106
Fig. 2. Geographical positions of the studied paddy fields and meteorological station.
2013. Equipment to measure evapotranspiration (ET) and deep perco- pan layer) was used to measure deep percolation in each field. A lid was
lation (DP) were installed in 10 paddy fields at first. The measurements placed on each cylinder to prevent evaporation from it. ETc and DP
were continued over the two years only for seven fields, and the mea- were measured at 2 or 3-day intervals. At the start, water level in the
sured data are the means for the seven fields. The geographical posi- mini-lysimeters and cylinders were adjusted to be the same as outside.
tions of the studied paddy fields and that of the meteorological station This level was marked as the reference level. In subsequent measure-
are shown in Fig. 2. ments in the absence of rain, water level in the mini-lysimeter or cy-
Before cultivation, soil samples were taken from each field at depth linder dropped due to ETc or DP. Water was then added to reach the
of 0–30 and 30–60 cm. Subsamples were randomly collected from 12 reference level and the volume added was taken as the amount of ETc or
points of each field and provided as a composite sample. Soil particles DP during that period. In the event of rain, ETc was calculated as the
size distribution was determined using the hydrometer method (Klute, sum of rain and the added water (when water was added to reach the
1986). The volumetric water content at field capacity (33 kPa matric reference level), or as the amount of rain minus the removed water
suction, FC) and permanent wilting point (1500 kPa matric suction, (when water was removed to reach the reference level). During growing
PWP) were measured using a pressure cell. Soil characteristic data are season, the reference level inside the mini-lysimeters and cylinders
shown in Table 2. were also the same as outside and water level was adjusted at 3 to 7 cm
The fields were flooded and covered by 5–10 cm water depth and in each irrigation event. In cylinders, water level after adjusting to the
puddled before transplanting. A local cultivar named Hashemi was reference level would be partially higher than the outside because of
transplanted on May 4 in both years, and harvested on July 29, 2012, ET.
and on August 4, 2013. Fertilizers were applied as 100, 50, 50, 25 kg In each field and at each sampling time, an area of 1 m2 was har-
ha−1 of N, P, K, Zn, in the form of urea, triple super phosphate, po- vested every 10 days for measuring leaf area index (LAI) and biomass.
tassium sulfate and zinc sulfate, respectively, according to re- Plants were removed with roots and transported to the laboratory with
commendations by Rice Research Institute of Iran. The irrigation was the roots immersed in water in pots in order to maintain plants fresh-
controlled by a reference mark in each field and the water level was ness. Then, the leaves, stems and panicles of plants were separated and
regulated at 3 to 7 cm in each irrigation event due to spatial variations leaf area was measured using a leaf area meter (MODEL: LI- 3100 AREA
in plot leveling. Regular field control and measurements were carried METER). LAI was calculated based on 1 m2 planted area.
out between 7 to 8:30 AM from Field 1 to Field 7 (Fig. 2), respectively. To calculate the canopy cover (CC) according to measured LAI, Eq.
Based on recorded dates and daily temperature data, the duration of a (1) was used (Ritchie, 1972; Ritchie et al., 1985; Belmans et al., 1983):
process or the time required to reach a particular stage was computed in
CC = 1-exp (-K*LAI) (1)
growing degree day (GDD, °C day) instead of number of days (Raes
et al., 2012). Where CC is canopy cover, and K is extinction coefficient. The values of
A closed bottom cylindrical mini-lysimeter with a diameter of 56 cm extinction coefficient for rice are between 0.4–0.7 (Hay and Walker,
and depth of 60 cm was installed in each field to measure the crop 1989). Here, it was considered to be 0.55.
evapotranspiration (ETc). The mini-lysimeters set in the underground. Sampled biomass was air dried for 5 d before being oven dried at
Farming operations such as transplanting date and density (6 plants at 70 °C for 48 h.
20 × 20 cm spacing in mini-lysimeter) in mini-lysimeters were the
same as the surrounding field. Also, an open bottom cylinder with a
diameter of 20 cm and depth of 50 cm (30 cm below the soil on the hard
Table 2
Mean ( ± STDa) of some soil characteristics in the studied paddy fields.
Depth EC pH Sand Silt Clay FC PWP Ks
Cm dS m−1 % % % cm−3 cm−3 cm−3 cm−3 mm d−1
0-30 0.87 ( ± 0.22) 7.02 ( ± 0.39) 10.4 ( ± 5.8) 40.8 ( ± 6.8) 48.8 ( ± 5.7) 0.52 ( ± 0.03) 0.29 ( ± 0.02) 7.6 ( ± 0.12)
30-60 0.81 ( ± 0.27) 6.96 ( ± 0.29) 12.2 ( ± 7.0) 34.1 ( ± 5.2) 53.8 ( ± 7.7) 0.57 ( ± 0.08) 0.30 ( ± 0.04) 1.6 ( ± 0.04)
a
Standard deviation.
99
N. Pirmoradian, N. Davatgar Agricultural Water Management 213 (2019) 97–106
n
2.3. Model description ∑i = 1 (Si−Oi )2
d = 1− n
∑i = 1 (|Si−Omean| + |Oi−Omean |)2 (5)
AquaCrop simulates crop growth and yield based on atmospheric
(rainfall, temperature, reference evapotranspiration and carbon dioxide Where Si and Oi are simulated and observed values, respectively, Omean
concentration), crop (phenology, green canopy cover, root depth, nor- is the mean of observed values and n is the number of observations. The
malized water productivity (WP*), harvest index, and stress responses), values of RMSE close to zero indicate the best fit of the model. In case of
soil (saturated hydraulic conductivity, field capacity, permanent wilting NRMSE, simulation results can be considered excellent if NRMSE is
percentage), and management (plant density, irrigation, fertility) fac- smaller than 10%, good if it is between 10 and 20%, fair if it is between
tors (Raes et al., 2009a; Steduto et al., 2009). AquaCrop calculates se- 20 and 30% and poor if it is larger than 30% (Willmott, 1982; Jacovides
parately crop transpiration and soil evaporation using an approach and Kontoyiannis, 1995). EF and d, range from negative infinity to
based essentially on crop coefficients and a daily water balance. Tran- positive 1; the closer to 1, the higher the model fidelity.
spiration is related to canopy cover whereas evaporation is related to Daily reference evapotranspiration (ETo) was computed from me-
the area of soil uncovered by canopy. The crop responds to water stress teorological data (Tmax, Tmin, U, n, RHmax and RHmin) using FAO 56
through four stress coefficients (leaf expansion, stomata closure, canopy Penman–Monteith method (Allen et al., 1998) with the ETo Calculator
senescence, and change in harvest index). AquaCrop calculates the software (FAO, 2009) for the 2012 and 2013 seasons. After preparing
daily aboveground biomass from the daily transpiration and reference the other input data, simulation was run with AquaCrop version 6.0 for
evapotranspiration and WP* (Hsiao et al., 2009; Steduto et al., 2009). mentioned period.
Yield is obtained by multiplying biomass by harvest index. The ad-
justment of HI in relation to the available water depends on the timing, 2.5. Model estimation of IWR
severity and duration of water stress (Hsiao et al., 2009; Raes et al.,
2009a; Steduto et al., 2009). The default reference HI in AquaCrop for The validated model was used to estimate IWR for a previous
rice is 43 percent, but can be cultivar specific (Hsiao et al., 2009). period, from 1982 to 2014. Reference ET for those 33 years were cal-
culated the same way as already described. AquaCrop features a way to
calculate IWR, but not for flooded cultivation. It was necessary to
2.4. Model validation generate an irrigation schedule to approximate intermittent flooding
using the generation of irrigation schedule feature of the model. First,
Model validation was conducted by using the default conservative the start of simulation was set at 1 day before transplanting, with initial
parameters (Hsiao et al., 2009; Raes et al., 2009b) calibrated by FAO for condition of soil water set at 50% total available water (TAW), ac-
rice, along with local management dependent parameters and phono- cording to regional soil water condition in May, to account for water
logical stages for the local cultivar, as listed in Table 3. The local needed for pre-transplanting flooding. The irrigation schedule in
parameters were gained from the measured data during growing sea- AquaCrop can be generated based on time and depth of irrigation.
sons of 2012 and 2013. Reference HI, though nominally considered to Under “Time Criterion” in the model, the allowable depletion of readily
be conservative, is taken to be 40 percent, slightly less than the 43 available water (RAW) was set to zero. Under “Depth Criterion”, the
percent default value. amount of water applied was set at “back to FC + 50 mm” to keep the
Validation of the model was done based on comparing simulated soil saturated and flooded, from transplanting to 10 days before har-
and measured crop evapotranspiration (ETc), deep percolation (DP), vesting. At 10 days before harvesting, “Time Criterion” was set to de-
canopy cover (CC), biomass (BM), and grain yield in 2012 and 2013 plete RAW 100% and “Depth Criterion” was set as “back to FC -100
growing seasons. The comparison criteria were the root mean square mm”, to simulate drying of paddy fields for harvesting. Bund height in
error (RMSE), normalized root mean square error (NRMSE), Nash- field management menu was set at 0.25 m and water depth between soil
Sutcliffe model efficiency coefficient (EF) and Willmotts index of bunds in initial condition menu set as zero. The saturated hydraulic
agreement (d), as follows: conductivity (Ksat) was taken to be the measured values (Table 2).
n
1
RMSE =
n
∑ (Si−Oi )2 2.6. Drought assessment
i=1 (2)
RDI as a drought characterization index is proposed by Tsakiris and
NRMSE =
( 1
n
n
)
∑i = 1 (Si−Oi )2 100 Vangelis (2005). RDI is based on the ratio of cumulative precipitation to
cumulative potential evapotranspiration (ak), over the same time in-
Omean (3) terval, and is denoted by ak with k indicating a certain month during a
year. The Standardized RDI (RDIst) is computed by Eq. (6).
n
∑i = 1 (Si−Oi )2 yk −y¯k
EF = 1− n RDIst (k ) =
∑i = 1 (Si−Omean )2 (4) σˆk (6)
Where yk is the ln ak of a given year, ȳk and σ̂k are the arithmetic mean
Table 3 and the standard deviation of ln ak for all the years of the climate record
Value of some measured local parameters used in AquaCrop to simulate rice data. It is advisable to use periods of 3, 6, 9 and 12 months in calcu-
growth and yield. lating RDI as a general indicator of meteorological drought (Tsakiris
Description Measured value and Vangelis, 2005). The mentioned time scales mean that ak is for a
period ending at the end of month k plus including the time of 2, 5, 8, or
Initial canopy cover, % 1.25
11 months preceding the month k, with total coverage of 3, 6, 9 or
Plant density, plants ha−1 250000
Maximum canopy cover (CCx) in fraction soil cover 0.9 12 months.
Reference Harvest Index (HIo) (%) 40 In this study, potential evapotranspiration was taken to be the re-
From transplanting to recovered transplant, GDD 50 ference ET calculated, based on FAO Penman–Monteith method (Allen
From transplanting to maximum canopy cover, GDD 525 et al., 1998). The RDI was calculated for the period 1982–2014 in two
From transplanting to start senescence, GDD 1090
time scales of 3- and 6-month regarding the period of agricultural ac-
From transplanting to maturity, GDD 1410
Maximum effective rooting depth, m 0.30 tivities on the paddy fields. Then, the best month and time scale for RDI
to estimate IWR is obtained based on the correlation between RDI and
100
N. Pirmoradian, N. Davatgar Agricultural Water Management 213 (2019) 97–106
Fig. 3. Rainfall and simulated and observed ETc and deep percolation (DP) in paddy fields during growing season for 2012 and 2013. Error bars indicate the standard
deviation across measurements on 7 fields.
estimated IWR from AquaCrop. Also, the RDI and corresponding IWR According to Fukuda and Tsutsui (1979), DP in paddy fields in Japan
were classified for drought severity, with RDI ≥ 0.5 being considered as was 1 to 3 mm d−1 for clay loam. Pirmoradian et al. (2004) obtained
wet, −0.5 < RDI < 0.5 as normal, and RDI ≤ -0.5 as dry. The stan- 3.5 mm d−1 of DP in continuous flooding irrigation of a paddy field
dardized RDI (RDIst) behaves in a generally similar way to the Stan- with a silty clay soil in southern Iran. In Guilan province, the amount of
dardized Precipitation Index (SPI) (McKee et al., 1993) and therefor, DP reported by Herve (1996) was 1.9 to 4.2 mm d-1 in Foumanat district
the interpretation of the results is similar since the same threshold as for continuous flooding irrigation. In this study, soil textures of paddy
SPI can be used (Tsakiris and Vangelis, 2005). fields were clay (Table 2) and water depth in irrigation events of the
fields regulated between 3 and 7 cm. Therefore, the measured DP were
lower than those obtained.
3. Results and discussion
Maximum measured LAI during growing season were 4.29 and 4.06
in 2012 and 2013, respectively. The observed yield and BM at harvest
3.1. Model validation
were obtained 4.52 ( ± 0.3) and 10.86 ( ± 0.49) t ha−1 in 2012 and
4.76 ( ± 0.31) and 11.56 ( ± 0.33) t ha−1 in 2013, respectively. Grain
The occurred rainfall and simulated and observed ETc as well as DP
yield was estimated with an absolute error of 6.0%.
in the paddy fields, during growing season 2012 and 2013 are shown in
The statistical indications of model performance in simulating ETc,
Fig. 3. As expected, there was an increasing of DP and a decreasing of
DP, CC, and BM are shown in Table 4. Based on the average of NRMSE,
ETc with occurrence of rainfall. The simulated and observed CC and BM
there was an excellent performance (NRMSE < 10%) in simulating CC,
are shown in Figs. 4 and 5, respectively. For each parameter, the
BM, and yield. Also, simulation results of ETc and DP by AquaCrop were
standard deviation across measurements on 7 fields per day were cal-
good (10% < NRMSE < 20%) and fair (20% < NRMSE < 30%), re-
culated in EXCEL software and are shown as the error bars in the Fig-
spectively. There was an overestimate of DP according to a mean of
ures. There is quite a good fit between observed and simulated data.
relative error as -0.165. Water level in the cylinder during days after
Measured ETc varied between 2.4 to 6.9 mm d−1 for both years. Mea-
that regulation would be partially higher than the outside because of
sured DP ranged between 0.7–2.3 mm d−1 in the paddy fields.
101
N. Pirmoradian, N. Davatgar Agricultural Water Management 213 (2019) 97–106
Fig. 4. Comparison of simulated and observed canopy cover (%) during growing season in studied paddy fields for 2012 and 2013. Error bars indicate the standard
deviation across measurements on 7 fields.
Fig. 5. Comparison of simulated and observed biomass during growing season in studied paddy fields for 2012 and 2013. Error bars indicate the standard deviation
across measurements on 7 fields.
Fig. 6. Simulated irrigation water requirement and total occurred rainfall during rice growing season for the studied years.
102
N. Pirmoradian, N. Davatgar Agricultural Water Management 213 (2019) 97–106
Fig. 7. Simulated transpiration, evaporation and deep percolation during rice growing season for the studied years.
Fig. 8. Variations of the calculated RDI during 1982–2014 in 3- and 6-Month time scales.
103
N. Pirmoradian, N. Davatgar Agricultural Water Management 213 (2019) 97–106
Table 6
Correlation coefficients between IWR of rice and RDI for 3- and 6-month time scales.
Time Scale (Month) Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
a
Correlation is significant at the 0.05 level (2-tailed).
b
Correlation is significant at the 0.01 level (2-tailed).
Fig. 9. Variation of RDI for July with 3-month time scale during 1982–2014.
time scale the drought duration and severity as RDI values tend to in-
crease. The periods of droughts in Fig. 8 were often interrupted by
normal rainfall which slightly relieved the drought. As shown, there are
more drought occurrences after 1994. The percentages of wet, normal
and dry conditions during 1982–2014 and separately for 1982–1994
and 1995–2014 periods are presented in Table 5. The frequency of
drought had more than doubled during 1995–2014 compared to
1982–1994. This is the consequence of 3-month decreased precipitation
and increased ETo for 1982–1994 (P = 363 mm, ETo = 208 mm)
compared to 1995–2014 (P = 317 mm, ETo = 230 mm).
104
N. Pirmoradian, N. Davatgar Agricultural Water Management 213 (2019) 97–106
the paddies irrigation water of the studied area based on drought pre- Fukuda, H., Tsutsui, H., 1979. Rice irrigation in Japan. JAJA 155.
diction studies. Hay, R.K.M., Walker, A.J., 1989. An Introduction to the Physiology of Crop Yield.
Longman Scientific and Technical, New York, pp. 292.
RDI is calculated from measured weather data in monthly time Hayes, M., 2004. Drought Indices. National Drought Mitigation Centre. http://www.
scale. On the other hand, water allocation is usually made ahead of drought.unl.edu/whatis/indices.htm.
planting time for the crop, so RDI must be predicted for the future. Herve, P., 1996. Guilan, a successful irrigation project in Iran. Irrig. Drain. Sys. 10,
95–107.
Then, IWR can be estimated based on RDI with 3-month time scale and Hsiao, T.C., Heng, L.K., Steduto, P., Rojas-Lara, B., Raes, D., Fereres, E., 2009. AquaCrop-
used by water resources planners. They can decide to apply water al- the FAO crop model to simulate yield response to water: III. Parameterization and
location policies in drought conditions, such as water supply turn- testing for maize. Agron. J. 101, 448–459.
Jacovides, C.P., Kontoyiannis, H., 1995. Statistical procedures for the evaluation of
around or the management of transplanting date in the study area. evapotranspiration computing models. Agric. Water. Manage. 27, 365–371.
Drought forecasting is generally based on drought indicators com- Klute, A., 1986. Methods of Soil Analysis, Part 1, Physical and Mineralogical Methods.
puted using dynamic or statistical model simulations of drought-related Madison, Wisconsin, USA.
Maniruzzamana, M., Talukderb, M.S.U., Khanc, M.H., Biswasd, J.C., Nemese, A., 2015.
variables (AghaKouchak, 2014). Chhinh and Millington (2015) focused
Validation of the AquaCrop model for irrigated rice production under varied water
on relationships between drought occurrences in Cambodia and (i) regimes in Bangladesh. Agric. Water. Manage. 159, 331–340.
damage to the annual rice harvest, and (ii) the Niño 3.4 index. There Matthews, R., Wassmann, R., 2003. Modelling the impacts of climate change and methane
was achieved a statistically significant correlation (r = −0.455) be- emission reductions on rice production: a review. Eur. J. Agron. 19, 573–598.
McKee, T.B., Doesken, N.J., Kleist, J., 1993. The relationship of drought frequency and
tween Niño 3.4 and 12-month SPI (Standard Precipitation Index) values duration to time scales. Proceeding of the 8th Conference on Applied Climatology
lagged by three months. That indicated the importance of ENSO lin- 179–184.
kages in explaining drought in that region. Mei, Y.F., Cheng, Q.P., Hua, Z.J., Da, L.E., Vijendra, B., 2011. Uncertainties in assessing
the effect of climate change on agriculture using model simulation and uncertainty
Prediction of RDI can be made based on the probability of RDI va- processing methods. Atmos. Sci. 56 (8), 729–737.
lues derived from past weather records, or on weather forecasted for the Ministry of Jahad-e-Keshvarzi, 2015. Statistical Book of Agriculture. Ministry of Jahad-e-
full growing season. The validity of the seasonal forecasts depends on Keshvarzi Publication, Tehran, Iran (In Persian).
Nash, J.E., Sutcliffe, J.V., 1970. River flow forecasting through conceptual models part I-
the ability to model processes on the scale of months. For example, a discussion of principles. J. Hydrol. 10, 282–290.
Wetterhall et al. (2015) used a probabilistic seasonal forecasting system Pirmoradian, N., 2016. Modeling of Water Delivery Scheduling in Sefidrood Irrigation
and investigated how well seasonal forecasts predict dry spells over a Network Paddy Fields Using Geographic Information System (GIS) to Perform of
Intermittent Irrigation. Research Final Report. Regional Water Company of Guilan,
basin during the rainy season with lead times from 0 to 4 months. The Rasht, Iran, pp. 173 (In Persian).
forecasts showed skill in predicting dry spells in comparison with a Pirmoradian, N., Sepaskhah, A.R., 2007. Rice optimal water use in different air tem-
climatological ensemble based on previous years. Also, AghaKouchak peratures at flowering, nitrogen rates and plant populations. Pak. J. Biolog. Sci. 10
(23), 4197–4203.
(2014) used a probabilistic framework for drought forecasting based on
Pirmoradian, N., Kamgar-Haghighi, A.A., Sepaskhah, A.R., 2004. Lateral seepage, deep
the persistence property of accumulated soil moisture and showed that percolation, runoff, and the efficiencies of water use and application in irrigation rice
the US drought of summer 2012 was predictable several months in in Kooshkak region in Fars province, I.R. of Iran. Iran Agric. Res. 23 (2), 1–8.
advance. Therefore, RDI can be predicted based on seasonal forecasts Popova, Z., Pereira, L.S., 2008. Irrigation scheduling for furrow-irrigated maize under
climate uncertainties in the Thrace plain, Bulgaria. Biosyst. Eng. 99, 587–597.
that provide information on the development of the climate up to 3 Raes, D., Steduto, P., Hsiao, T.C., Fereres, E., 2009a. AquaCrop-the FAO crop model to
months. Then, water allocation can be planned at transplanting time for simulate yield response to water: II. Main algorithms and software description.
paddy fields. Agron. J. 101, 438–447.
Raes, D., Steduto, P., Hsiao, T.C., Fereres, E., 2009b. Crop Water Productivity. Calculation
Procedures and Calibration Guidance. AquaCrop Version 3.0. FAO, Land and Water
Acknowledgment Development Division, Rome.
Raes, D., Steduto, P., Hsiao, T.C., Fereres, E., 2012. Reference manual of AquaCrop model.
Chapter 2. Users Guide. FAO Land and Water Division, Rome Italy, pp. 164.
This work was financially supported by the Research Committee of Richard, R., Heim, J., 2002. A review of twentieth-century drought indices used in the
the Regional Water Company of Guilan, Iran (No. GII89034). United States. Bull. Am. Meteo. Soc. 83, 1149–1165.
Ritchie, J.T., 1972. Model for predicting evaporation from a row crop with incomplete
cover. Water Resour. Res. 8, 1204–1213.
References
Ritchie, J.T., Godwin, D.C., Otter-Nacke, S., 1985. CERES-Wheat: A Simulation Model of
Wheat Growth and Development. Texas A. & M. Univ. Press, College Station.
Aggarwal, P.K., Mall, R.K., 2002. Climate change and rice yields in diverse agro-en- Saadati, Z., Pirmoradian, N., Rezaei, M., 2011. Calibration and Validation of AquaCrop
vironments of India. II. Effect of uncertainties in scenarios and crop models on impact Model in Rice Growth Simulation under Different Irrigation Managements. ICID 21st
assessment. Clim. Change 52, 331–343. International Congress on Irrigation and Drainage. Publ. no R.56.5/Poster/14,
Aghakouchak, A., 2014. A baseline probabilistic drought forecasting framework using Tehran, Iran, pp. 15–23 October, 2011.
standardized soil moisture index: application to the 2012 United States drought. Shrestha, N., Raes, D., Vanuytrecht, E., Sah, S.K., 2013. Cereal yield stabilization in Terai
Hydrol. Earth Syst. Sci. 18, 2485–2492. (Nepal) by water and soil fertility management modeling. Agric. Water Manage. 122,
Allen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop evapotranspiration - guidelines 53–62.
for computing crop water requirements. FAO Irrigation and Drainage Paper 56. Food Smith, P.C., Calanca, P., Fuhrer, J., 2012. A simple scheme for modeling irrigation water
and Agriculture Organization of the United Nations, Rome, Italy, pp. 293. requirements at the regional scale applied to an Alpine River Catchment. Water 4,
Azam, A., Crout, S.N., Bradley, R.G., 1994. Perspectives of modelling resource capture by 869–886.
crops. In: Monteith, J.L., Unsworth, M.H., Scott, R.K. (Eds.), Resource Capture by Steduto, P., 2003. Biomass water-productivity: comparing the growth-engines of crop
Crops. Proceedings of the 52nd Univ. of Nottingham Eastern School. Nottingham models. FAO Expert Meeting on Crop Water Productivity under Deficient Water
University Press, pp. 125–140. Supply. pp. 26–28 Rome, February.
Bachelet, D., Gay, C.A., 1993. The impacts of climate change on rice yield comparison of 4 Steduto, P., Albrizio, R., 2005. Resource-use efficiency of field grown sunflower sorghum,
model performances. Ecol. Model. 65, 71–93. wheat and chickpea. II. Water use efficiency and comparison with radiation use
Belmans, C., Wesseling, J.G., Feddes, R.A., 1983. Simulation model of the water balance efficiency. Agric. Meteorol. 130, 269–281.
of cropped soil: SWATRE. J. Hydrol. 63, 71–286. Steduto, P., Hsiao, T.C., Fereres, E., 2007. On the conservative behaviour of biomass
Bos, M.G., Kselik, R.A.L., Allen, R.G., Molden, D.J., 2009. Water Requirements for water productivity. Irrig.Sci. 25, 189–207.
Irrigation and the Environment. Springer, Dordrecht, pp. 174. Steduto, P., Hsiao, T.C., Raes, D., Fereres, E., 2009. AquaCrop-the FAO crop model to
Bray, E.A., 1997. Plant responses to water deficit. Trends Plant Sci. 2, 48–54. simulate yield response to water. I. Concepts and underlying principles. Agron. J.
Challinor, A.J., Wheeler, T.R., 2008. Crop yield reduction in the tropics under climate 101, 426–437.
change: processes and uncertainties. Agric. Forest Meteorol. 148, 343–356. Tanner, C.B., Sinclair, T.R., 1983. Efficient water use in crop production: research or re-
Chhinh, N., Millington, A., 2015. Drought monitoring for rice production in Cambodia. search? In: Taylor, H.M., Jordan, W.A., Sinclair, T.R. (Eds.), Limitations to Efficient
Climate 3, 792–811. Water Use in Crop Production. A.S.A., Madison, Wisconsin, USA, pp. 1–27.
De Datta, S.K., 1981. Principles and Practices of Rice Production. IRRI, Los Baños, Todorovic, M., Albrizio, R., Zivotic, L., Saab, M.T.A., Stöckle, C., Steduto, Pasquale, 2009.
Philippines, pp. 618. Assessment of AquaCrop, CropSyst, and WOFOST models in the simulation of sun-
Ewert, F., Rodriguez, D., Jamieson, P., et al., 2002. Effects of elevated CO2 and drought flower growth under different water regimes. Agron. J. 101 (509), 521.
on wheat: testing crop simulation models for different experimental and climatic Tsakiris, G., Vangelis, H., 2005. Establishing a drought index incorporating evapo-
conditions. Agric. Ecosyst. Environ. 93, 249–266. transpiration. Eur. Water 9 (10), 3–11.
FAO, 2009. ETo calculator version 3.1. Evapotranspiration from Reference Surface, FAO, Tuong, T.P., 1999. Productive water use in rice production: opportunities and limitations.
Land and Water Division. Rome, Italy. J. Crop Prod. 2 (2), 241–264.
105
N. Pirmoradian, N. Davatgar Agricultural Water Management 213 (2019) 97–106
Vanuytrecht, E., Raes, D., Steduto, P., Hsiao, T.C., Fereres, E., Heng, L.K., Vila, M.G., Earth. Syst. Sci. 19, 2577–2586.
Moreno, P.M., 2014. AquaCrop: FAO’s crop water productivity and yield response Willmott, C.J., 1982. Some comments on the evaluation of model performance. Bull. Am.
model. Environ. Modell. Softw. 62, 351–360. Meteo. Soc. 63, 1309–1313.
Wetterhall, F., Winsemius, H.C., Dutra, E., Werner, M., Pappenberger, E., 2015. Seasonal Wriedt, G., van der Velde, M., Aloe, A., Bouraoui, F., 2009. Estimating irrigation water
predictions of agro-meteorological drought indicators for the Limpopo basin. Hydrol. requirements in Europe. J. Hydrol. 373, 527–544.
106