Drought and water availability analysis for irrigation and household water needs in the Krueng Jrue sub-watershed

Times and site

The research was conducted in 23,218.06 ha of the Krueng Jrue sub-watershed located at 5°12′–5°28′N and 95°20′–95°32′E, which is part of the Krueng Aceh watershed, Aceh Province, Sumatra, Indonesia (Fig. 1). It was conducted from January to December 2019.

Data collection

Materials needed included maps (administrative, topography, soil type, and land use), and monthly rainfall, evapotranspiration, irrigation area, and population data for 2008–2017. A land map was obtained from the Krueng Aceh Watershed Management Board and climatology data were from Blang Bintang Meteorological, Climatological, and Geophysical Agency. While this study used data from the Indrapuri rainfall station, as shown in Fig. 1, monthly river discharge data were collected from the Center of River Basin Sumatera-I. There are three climatological stations in the Krueng Aceh watershed, but only one is located in the Krueng Jrue sub-watershed. Furthermore, the discharge data were collected from Kr. Meulesong hydrometry station using an automatic water level recorder (AWLR) that was set up in the field.

Drought indices

The Z-score for precipitation (ZSP) was used in this study to evaluate the meteorological drought, as shown in Eq. (1). (1)${\displaystyle ZSP=\frac{P_i-P_{avg}}{S},}$ where, P_i = precipitation (mm); P_avg = average of precipitation; S = standard deviation of precipitation.

Using the same concept, the Z-score for discharges (ZSD) was calculated to evaluate hydrological drought using the formula shown in Eq. (2). (2)${\displaystyle ZSD=\frac{D_i-D_{avg}}{Sd},}$ where, D_i = discharge (m³s⁻¹); D_avg = average of discharge; S_d = standard deviation of discharge. Drought criteria to justify the drought class for ZSP and ZSD are shown in Table 1.

The Mock model

The basic approach of the Mock model is to consider factors such as rainfall, evapotranspiration, and water balance at the soil surface, as well as groundwater content. Monthly rainfall data are needed to analyze the river’s water availability and are the primary input of the Mock method. The longer the recording period, the better the results will be. Many researchers (Setyawan, Lee & Prawitasari, 2016; Putro, 2016; Sebayang & Trianing, 2018; Chandrasasi, Limantara & Juni, 2020; Maulana, Suhartanto & Harisuseno, 2019; Krisnayanti et al., 2019; Dinar, Agus & Sarino, 2020) have used the Mock model to assess water discharges or availability of watersheds. Generally, these studies found the Mock model suitable for evaluating the water balance in specific watersheds. The equations used to calculate water balance parameters by the Mock model (Mock, 1973; Umum, 1986) are as follows:

Evapotranspiration: (3)${\displaystyle E{=ET}_0-\Delta E}$ (4)${\displaystyle \Delta E{=ET}_0\left(m1/20\right)\left({18-n}_1\right),}$ where, ΔE = the difference between potential and actual evapotranspiration (mm month⁻¹); ET₀ = potential evapotranspiration (mm month⁻¹); m₁ = the proportion of soil surface that is not covered by vegetation (set as 20%); n₁ = total of rainy days; E = actual evapotranspiration (mm month⁻¹).

River discharges: (5)${\displaystyle Q_{river}=\left(Q_{total}xA\right)/t}$ (6)${\displaystyle Q_{total}{=Q}_{base+}{Q}_{direct}{+Q}_{storm},}$ where, Q_river = discharges of a river (m³s⁻¹), Q_total = total runoff (mm month⁻¹), A = watershed area (Ha), t = time (second) Q_base=baseflow (mm month⁻¹), Q_direct = direct runoff (mm month⁻¹), and Q_storm = storm runoff (mm month⁻¹).

Baseflow: (7)${\displaystyle Q_{base}{=inf-G.STOR}_t{+G.STOR}_{\left(t-1\right)}}$ (8)${\displaystyle inf=WSxIF}$ (9)${\displaystyle {WS=ISM+R}_e-E-SMS}$ (10)${\displaystyle {SMS=ISM+R}_e-E}$ (11)${\displaystyle {G.STOR}_t{=G.STOR}_{\left(t-1\right)}xRc+0.5\left(1+Rc\right)xinf,}$ where, inf = infiltration (mm month⁻¹); G. STOR_t = groundwater storage at the beginning of the month (mm month⁻¹); G. STOR_(t−1) = groundwater storage at the end of the month (mm month⁻¹); IF = infiltration factor (set as 0.4); WS = water surplus (mm month⁻¹); ISM = initial soil moisture (set as 200 mm month⁻¹); R_e = monthly rainfall (mm month⁻¹); SMS = soil moisture storage (mm month⁻¹); Rc = flow reduction coefficient (set as 0.6).

Direct runoff: (12)${\displaystyle Q_{direct}=Wsx\left(1-IF\right),}$ where Ws = water surplus (mm)

Storm runoff: (13)${\displaystyle Q_{storm}=RexPF,}$ where, PF = precipitation factor (%).

Water demand

Two kinds of water needs were considered to calculate the water demand. First, water demand for irrigation in the Krueng Jrue sub-watershed between 2008–2017 was projected based on the area of irrigated land according to irrigation water needs, which was calculated as follows: (14)${\displaystyle DR=\frac{NFR}{e\times 8.64},}$ where, DR = diversion requirement (l s⁻¹ha⁻¹); NFR = net water requirement in paddy field (l s⁻¹ha⁻¹); e = irrigation efficiency; 1/8.64 = conversion value from (mm day⁻¹) to (l s⁻¹ha⁻¹).

Second, the water demand for households in the Krueng Jrue sub-watershed during 2008–2017 was calculated using the assumed population growth (1.4% year⁻¹) and standard water demand per capita (0.06 m³day⁻¹).

Model validation

Validation was performed by comparing the observation discharges (Qo) with the Mock model discharges (Qc) using linear regression. The statistical parameters were determined using the coefficient of determination (R²) and value of the regression coefficient (r). They should be near 1 in order to ensure that the Mock model is valid and suitable to predict discharges for this region. The correlation coefficient (r) can be calculated using the following equation (Ward & Trimble, 2003): (15)${\displaystyle r=\frac{\mathrm{n}\left(\sum QoQc\right)-\left(\sum Qo\right)\left(\sum Qc\right)}{\sqrt{n\sum Q{o}^2+\sum Q{o}^2}.\sqrt{n{\left(\sum Qc\right)}^2-\sum Q{c}^2}}.}$

Model validation

Model validation was conducted using regression analysis to observe the correlation between the observed discharges and calculated discharges using the Mock model. Statistically, the model showed a very significant regression coefficient (r = 0.95**) and coefficient of determination (R² = 0.90**). Those values prove the validation of the Mock model to predict discharges for this region. The correlation between observed discharges and calculated discharges is shown in Fig. 2.

Monthly rainfall

The average monthly rainfall in the Krueng Jrue sub-watershed (2008–2017) is shown in Fig. 3. Indonesia has two seasons: the rainy season (October–March) and the dry season (April–September). The highest monthly average rainfall occurred in December, November, and January at 325.7, 324.4, and 269.9 mm, respectively. The lowest monthly average rainfall occurred in February (98.3 mm) and July (68.7 mm). February is included in the rainy season, but the rainfall observed in February (98.3 mm) was less than the average rainfall observed for all of the months (190.3 mm). However, April and May are included in the dry season, but had average rainfalls of 247.4 and 219.6, respectively, which was above the average rainfall for all of the months.

Monthly discharges

Table 2 shows the discharges of the Krueng Jrue sub-watershed generated using the Mock model. The higher the rainfall, the higher the water discharge generated by the Mock model. The average monthly discharges of the Krueng Jrue sub-watershed from 2008–2017 ranged from 3.16–31.18 m³s¹. The highest monthly average discharges during the rainy season (October–March) occurred in November (31.18 m³s⁻¹) and December (28.02 m³s⁻¹), while the lowest monthly discharge in the dry season (April–September) occurred in July (3.16 m³s⁻¹), and June (3.36 m³s⁻¹). Further analysis provided information for the entire observation year. February, which is included in the rainy season, had a meager monthly discharge value (1.78–3.76 m³s⁻¹), that was below the average monthly discharge, except in 2012 and 2013.

Table 2 shows that the average monthly discharge was 12.86 m³ s⁻¹. Average monthly discharges for 2010, 2012, 2013, 2015, and 2017 were above the average monthly discharges for the whole year. On the contrary, the average discharges for 2008, 2019, 2011, and 2016 were below the average discharges. Based on the average discharges for each month, August had an average discharge below the average monthly discharge throughout the year. In the rainy season (October–March), the average monthly discharge should have been above the average monthly discharge for the whole year, but the average discharges for October, February, and March were below the average monthly discharge for the whole year. Conversely, for the dry season (April–September), the average discharge should have been below the monthly average discharge for the entire year, but the results of the average discharge for April were above the monthly average discharge for the entire year.

Z-score values for precipitation and discharge

The Z-score for precipitation (ZSP) and discharge (ZSD) from 2008 to 2017 are shown in Tables 3 and 4. Positive ZSP or ZSD values indicate greater than median precipitation or discharges and negative values indicate lower than median precipitation or discharges (Tsakiris & Vangelis, 2004). This region had drought indices for ZSP: 90 normal (N) events (75%), six moderate wet (MW) events (5%), seven very wet (VW) events (5.8%), six extreme wet (EW) events (5%), and 11 moderate drought (MD) events (9.2%). Drought indices for ZSP with severe drought and extreme drought were not found in this region. Furthermore, this region had drought indices for ZSD: 89 normal (N) events (74.2%), 11 moderate wet (MW) events (9.2%), four very wet (VW) events (3.3%), six extreme wet (EW) events (5%), nine moderate drought (MD) events (7.5%), and one severe drought (SD) event (0.8%). Drought indices for ZSD with extreme drought were not found in this region. The ZSP and ZSD indices in the dry season (April–September) tended to be negative, indicating that rainfall or discharge was under the average index. However, they tended to be positive for the rainy season (October–March).

The agreement between ZSP and ZSD indices reached 85.8%, with hydrological drought analyzed based on water discharge (ZSD), and meteorological drought analyzed based on rainfall (ZSP). Both indices tended to be negative during the dry season (April–September) indicating that the rainfall or discharge was below the average index. However, they were above the average index during the rainy season (October–March) (Table 5).

Water availability and demand

Surplus and water deficits in the Krueng Jrue sub-watershed were calculated using the difference between water availability (discharges) and water demand (water irrigation + water for households). A positive difference between water availability and water demand indicates a surplus. On the contrary, a negative difference between water availability and water demand indicates a deficit (Table 6 and Fig. 4). Generally, there is a surplus of water availability during the rainy season (October–March) and a water deficit during the dry season (April–September).

The monthly average water needs for irrigation and households were 5 m³ s⁻¹ and 0.041 m³ s⁻¹, respectively. There was a total water need of 5.041 m³ s⁻¹ per month, and the average water availability (Table 2) from 2008 to 2017 in June (3.36 m³ s⁻¹), July (3.16 m³ s⁻¹), and August (4.19 m³ s⁻¹) was unable to meet the water needs. A more detailed analysis of water availability for each year was conducted (Table 6 and Fig. 4). In 2009, there were five consecutive months during the gadu planting season (June, July, August, and September) with water availability lower than 3 m³ s⁻¹, as well as from 2011 to 2012 (May to August) and from 2013 to 2017 (June to September). On the other hand, during the rendeng planting season (rainy season) in October 2009, 2013, 2016, and 2017, and February 2008-2011 and 2014–2017, the water availability was not enough to meet irrigation and household needs. Water deficit and surplus also occurred in both the rainy and dry seasons.