Adapting to climate variability for rice cultivation paddies in the lowland coastal regions of Kien Giang Province, Vietnam

Kien Giang is a southwestern coastal province, stretching from 9°23′50″ N–104°26′40″ E to 10°32′30″ N–105°32′40″ E (Figure 1), with the territory varying from 0.2 to 1.0 m above average sea level (IMHEN, Ca Mau Peoples Committee, and Kien Giang Peoples Committee, 2011; Dang et al., 2021). Rice production is one of the strengths of the province, with an estimated total area of agricultural land of up to 350,000 ha (Trang, 2016; Dang et al., 2021). Rice cultivation paddies are supplied with freshwater via irrigation channels and local precipitation (CGIAR Research Program on Climate Change, 2016; Dang et al., 2021).

Figure 1.

The area is dominated by two monsoon circulations including the southwest monsoon that lasts from May to October and the northeast monsoon activities from November to April (IMHEN, Ca Mau Peoples Committee, and Kien Giang Peoples Committee, 2011; Dang et al., 2021). The tropical climate has an average temperature of approximately 27.5°C, humidity of 82%, and rainfall ranging from 2200 to 2400 mm (Figure 2). These climatic conditions are highly suitable for agricultural activities (Kontgis et al., 2019; Nguyen et al., 2022).

Figure 2.

In CCRs of Kien Giang Province, farmers sow single or double crops each year, except in a small part of the U Minh Thuong region and Vinh Thuan districts that are triple cropped (Kontgis et al., 2019; Nguyen et al., 2022). Sowing and harvesting schedules are based on local weather conditions and information recommended by the local extension agency (Dang et al., 2021). CCSs of winter–spring (WS) and summer–fall (SF) crops begin around the third week of November and third week of March (Table 1).

To shorten the cultivation period as well as optimize economic efficiency, farmers adapt cultivation practices including limiting the number of seeds per hectare and using highquality rice varieties (e.g., OM18, OM4900, OM5451, GKG1, and GKG9) and short rice varieties with a life cycle varying from 85 to 95 days. These varieties have the advantages of resistance to brown planthopper, salt tolerance, and a high yield, ranging from 5 to 9 tons/ha. To reduce production costs, farmers have applied the “three reductions and three increases” program to reduce the number of seeds, fertilizers, and pesticides while increasing the yield productivity, quality, and profit. Information about seeds sown, fertilizer rates, and irrigation water is presented in Table 1. Soil samples were obtained from rice paddies at the Ha Tien, Kien Luong, and Rach Gia stations, and the physicochemical properties of the soil were analyzed in the laboratory using standard procedures (Table 2). The physicochemical properties of the soil consisting of textural class, bulk density (BD), field capacity (FC), permanent wilting points (PWP), total available moisture (TAM), ions such as magnesium (Mg²⁺), sodium (Na⁺), potassium (K⁺), and calcium (Ca²⁺), and soil pH in the soil profiles are presented in Table 2.

Temperature and rainfall data and other climate variables for the period 2000–2021 at Kien Luong station (not show Ha Tien and Rach Gia stations) were collected from the Southern Regional Hydrometeorological Center, Vietnam (Figure 3). Specifically, model validation was conducted during the period 2000–2010 and model calibration was performed during the period 2011–2021.

Figure 3.

AquaCrop is a crop model that was developed by Food and Agriculture Organization to calculate the grain and biomass yields of several crops under different weather conditions (Shrestha et al., 2016; Dang et al., 2021). An advantage of the AquaCrop model is that it is easy to use based on the link between climate, soil, and management modules (Dang et al., 2021). The AquaCrop model requires a small number of input variables for the simulation, and its performance has been validated as both efficient and accurate (Greaves and Wang, 2016; Shrestha et al., 2016).

In the model, actual grain yield (Ya) is defined based on the relationship between maximum grain yield (Ym) and water stress, which is described through the maximum evapotranspiration (ETx) and actual evapotranspiration (ET). Specifically, Ym responses to irrigation water are calculated using Eq. (1): (1) Ym−YaYm=KyETx−ETaETx \left( {{{{Y_m} - {Y_a}} \over {{Y_m}}}} \right) = {K_y}\left( {{{E{T_x} - E{T_a}} \over {E{T_x}}}} \right) where ETa is calculated by Eq. (2): (2) ETa=E+Tr E{T_a} = E + Tr In Eq. (2), E and Tr are the soil evaporation and crop transpiration, respectively. Tr in Eq. (3) is defined based on the reference evaporation (ET_o) with crop transpiration coefficient (KcTr), the effect of water (Ks), and temperature stresses (KsTr). (3) Tr=Ks KsTr KcTrETo Tr = \left( {Ks\;K{s_{Tr}}\;K{c_{Tr}}} \right)E{T_o} ET_o in Eq. (3) is used for simulating the model based on the FAO Penman–Monteith equation as given in Eq. (4): (4) ETO=0.408 ΔRn−G+γ900T+273u2es−eaΔ+γ1+0.34u2 E{T_O} = {{0.408\;\Delta \left( {{R_n} - G} \right) + \gamma {{900} \over {T + 273}}{u_2}\left( {{e_s} - {e_a}} \right)} \over {\Delta + \gamma \left( {1 + 0.34{u_2}} \right)}} where R_n is radiation at the soil surface, G is soil heat flux density, T is average daily temperature, u₂ is wind speed at 2.0 m height, e_s is saturation vapor pressure, e_a is the actual vapor pressure, Δ is the slope of the vapor pressure curve, and γ is a psychrometric constant.

Finally, grain yield (Y) (Eq. 5) is defined based on multiplying the harvest index (HI) with aboveground biomass production (B) and the stresses (f_HI). (5) Y=fHIHI*B Y = {f_{HI}}{\rm{HI*B}} Normally, the economic efficiency of using water irrigation efficiently (WIE) (Eq. 6) is commonly defined through the link between the grain yield produced and water evapotranspiration (Steduto et al., 2012). (6) WIE=yield producedwater evapotranspired {\rm{WIE}} = {{{\rm{yield}}\;{\rm{produced}}} \over {{\rm{water}}\;{\rm{evapotranspired}}}} Accordingly, the economic efficiency of using irrigation water is considered an indicator to assess the performance of an applied cultivation system.

To save time and increase the performance of the model, a sensitivity analysis was conducted in model simulations. Determining the sensitivity parameters for the simulation model is a necessary procedure (Dang et al., 2021). In this study, a sensitivity analysis was conducted to define the optimum values for simulating CCSs in Kien Giang Province. The optimal values of the main variables are presented in Table 3.

Model validation and calibration were conducted by comparing the simulated results and observed rice grain yield based on the index of agreement (d), the coefficient of determination (R²), and the root mean square error (RMSE) during the period 2000–2021 (Table 4).

Validation was based on comparison of the simulated results with observed yields for the WS and SF crops during the period 2000–2010, while calibration was based on comparison of the simulated results with observed yields for the WS and SF crops during the period 2011–2021 (Figure 4).

Figure 4.

The validation procedure found high correlations between observed and simulated yields according to d, R², RMSE, and mean absolute error, varying from 0.76 to 0.83, 0.77 to 0.83, 0.17 to 0.21, and 14.9% to 17.6%, respectively, while the calibration procedure demonstrated high reliability, with d = 0.78–0.86, R² = 0.79–0.85, RMSE = 0.11–0.16, and MAE = 12.6%–13.4% (Table 5).

In recent years, CCRs have been facing challenges caused by climate change and sea level rise. RCPs of Kien Giang Province are facing the challenges of saline intrusion, drought, lack of irrigation water, and sea level rise (Dang et al., 2021). According to Vo et al. (2021), climate change is expected to affect CCRs and, especially, lowland cultivation paddies along the coast of Ken Giang province (Dang et al., 2021). Analysis of CCSs for the WS and SF vegetation seasons based on consideration of input temperature and rainfall pointed out that the rice grain yield of both the WS and SF crops could increase if CCSs were shifted to adapt to changing climate conditions (Figure 5).

Figure 5.

For the WS crop, the results showed that bringing CCS forward by around 20 days will increase the rice grain yield by 3.5% to 7.8%, while delaying sowing of the SF vegetation season by 10–20 days will increase the rice grain yield by 2.1% to 5.6% (Table 6). Based on the simulated rice grain yields, a shift in CCSs from baseline would be appropriate when taking into consideration changes in climate variables (e.g., temperature, precipitation, and evapotranspiration).

According to the local agricultural agency, the “three reductions and three increases” program combined with changing CCS could reduce the amount of seed sown by 30–50 kg/ha, the amount of fertilizers of all kinds by 10–45 kg/ha, and the production costs by around 4,195,000 VND/ha.