Evapotranspiration and Crop Coefficient of Ratoon Rice Crop Determined by Water Depth Observation and Bayesian Inference

Abstract

Actual crop evapotranspiration (ET) and crop coefficient (Kc) of ratoon rice crop, which are necessary for irrigation planning, have been hardly reported. ET can be directly measured by lysimeter and eddy covariance but it is expensive, so it remains difficult to determine ET, especially in developing countries. The focus of this study was to evaluate the ET and Kc of ratoon cropping in a tropical region of Myanmar using a simplified method. Our method combined the manual observation of water depth in concrete paddy tanks and the ET model estimation using Bayesian parameter inference. The ET and Kc could be determined using this method with an incomplete observation dataset. The total ET of ratoon was 60–70% less than that of the main crop, but this difference was mainly attributed to climate conditions in each cultivation. The Kc regression curve between transplanted rice and ratoon crops was different because of the tillering traits. The results suggest that irrigation scheduling of ratoon cropping in the initial growth stage should take high crop water requirements into account. In addition, the productivity of ratoon crop is equivalent to transplanted rice, which was determined for cultivation in experiment conditions of small concrete tanks. Therefore, further study on ratoon in Myanmar is necessary for clarifying the viability of ratoon cropping.

1. Introduction

Global rice production is projected to increase by 116 million tons by 2035 to meet the growing demand of an increasing population [1]. However, cultivated rice land has declined in recent years because of urbanization, industrialization, and stagnant rice prices [2]. Multiple rice cropping, which includes double- or triple-cropping and requires frequent rice harvesting on the same land, is one strategy to solve land shortages [3]. However, labor shortages and rising labor costs are limiting factors [4]. Ratoon rice double-cropping (main crop (MC) and ratoon), which refers to producing a second rice crop from the residual stubble after the MC harvest, is useful as an alternate option to replace conventional double-cropping [5]. Although the annual yield of ratoon double-cropping is lower than conventional double-cropping, labor and production costs can be reduced. The annual net profit could be higher or equivalent to conventional double-cropping [6,7].

Moreover, the amount of irrigation water for ratoon crops (RCs) can be significantly reduced. The water utilization efficiency can be improved compared to the MC because the ratoon cropping has a short growth period and does not require irrigation for land preparation [8,9,10,11,12]. The growth period of ratoon is shorter by 40–60% than that of the MC [13,14,15]. The amount of irrigation water for land preparation has been observed at 100–500 mm [16,17] or 25% of the total water demand [18]. Therefore, there is a significant difference in the irrigation period and the amount of irrigation water required between ratoon and conventional double-cropping. Although there are many reports on evapotranspiration (ET) and crop coefficient (Kc) of single rice cropping [19,20,21,22], few studies have reported on ET estimation and Kc of ratoon double-cropping. Furthermore, complex and expensive observation systems such as weighing lysimeter and the eddy covariance system are commonly used for the estimation of ET and Kc.

In Myanmar, rice is a vital crop grown in 34% of the country’s total cultivated area [23]. Rice dominates the agricultural sector, which is the largest and most productive sector in Myanmar’s economy. However, the sector is crippled by labor [24] and water shortages, particularly in the dry season [25], driving the need for a water-saving cropping system. Hence, this study on water usage in ratoon double-cropping is critical in Myanmar, because it has the potential to reduce labor costs and improve WP.

Therefore, the objective of this study was to determine the ET and Kc of ratoon double-cropping in a tropical climate region of Myanmar using a simple method. In addition, we inferred the water productivity (WP) and viability of practicing ratoon cropping in Myanmar.

2. Materials and Methods

2.1. Study Site and Concrete Tank Paddy Experiment

The study site is in an experimental field at the Department of Agricultural Research (latitude 19.825, longitude 96.274), Ministry of Agriculture, Livestock and Irrigation, Myanmar, located in Zayarthiri Township in the Naypyidaw Union Territory at an elevation of 100 m above mean sea level (Figure 1a). The study experiment used 10 concrete tanks with an inside size of 1.62 m L × 0.84 m W × 0.4 m D; rice was planted in eight tanks for conventional transplanted rice and ratoon rice crops (four replications), and two tanks without plants to measure evaporation were used (two replications) (Figure 1b). The difference in daily pond water depth (WD) (mm) above the soil surface in a concrete tank was used to determine daily crop evapotranspiration (ET_tank) (mm day⁻¹) and evaporation (E_tank) (mm day⁻¹). Net radiation, air temperature, relative humidity, air pressure, and wind speed were measured at a paddy field near the concrete tank experiment.

2.2. Ratoon Double-Cropping in Different Seasons

The rice (Oryza sativa L.) cultivar used was “Theehtetyin” (indica), which is a popular variety in Myanmar with normal crop duration of 110–115 days from sowing to harvesting. Forty-five plants (approximately 21-day-old seedlings, BBCH-scale = 14) were transplanted with 20cm × 20 cm spacing in nine rows and fine lines in the tank. A basal level of triple superphosphate fertilizer (18.5 kg P ha⁻¹) was applied before transplanting for the MC and two weeks before cutting the ratoon crop. Nitrogenous fertilizer (87 kg N ha⁻¹) and potash (31 kg K ha⁻¹) were applied equally to each tank at approximately 7, 30, and 50 days after transplanting and during the flowering period for the MC, and the ratoon crop was dosed at around 14, 30, and 50 days after cutting of the MC and at the flowering stage (BBCH-scale = 61).

The first trial of ratoon double-cropping planted in eight tanks, which are main crop1 (MC1) and ratoon1 (RC1), was from February 2019 to September 2019, and the second trial (MC2 and RC2) planted in eight tanks, was implemented from September 2019 to March 2020. In parallel with RC cropping, the transplanted rice (TC) planted in four tanks was prepared in each trial to compare plant growth and the ET (

Table 1. Cultivation information for transplanting and ratoon rice cropping.

Crop	Sowing Date	Transplanting Date	Harvesting Date
First trial
Main crop 1 (MC1)	15 February 2019	7 March 2019	1 June 2019
Ratoon 1 (RC1)			26 August 2019
Transplant 1 (TC1)	24 May 2019	14 June 2019	10 September 2019
Second trial
Main crop 2 (MC2)	6 September 2019	25 September 2019	12 December 2019
Ratoon 2 (RC2)			1 March 2020
Transplant 2 (TC2)	30 November 2019	19 December 2019	1 April 2020

and Figure 2). Water management for the tank experiments was performed by continuously flooding until around 14 days before harvesting. The stem cutting of MC for ratooning was implemented at a 5 cm height above the soil surface. The harvesting day at the physiological maturity stage (BBCH-scale = 87) was determined as approximately 25 days after the flowering day (BBCH-scale = 65). Myanmar has a typical tropical monsoon climate with three seasons: the hot and dry summer (i.e., March to May), the monsoon (i.e., June to September), and the cool and dry season (i.e., December to February).

2.3. Measurements of Rice Crop and Calculation of WP

We measured plant growth (i.e., height of the plant and number of tillers), number of spikelets per panicle, filled grain weight, 1000-grain weight, weight of dried straw, and biological yield from four adjacent hills, excluding the outer hills in each tank. Grain weight was adjusted to 14% moisture content. Biological yield (g m⁻²) was equal to the weight of dried straw and filled grains weight divided by the area of the four hills. Grain yield (g m⁻²) and number of panicles (g m⁻²) were determined on the basis of 21 harvested hills in seven rows and three lines. The WP of the rice crop regarding evapotranspiration (WP_ET) was defined as the ratio of the grain yield to the ET_tank (grain yield/ET_tank) (kg m⁻³).

2.4. Determination of E_tank, ET_tank, and Kc

The daily E_tank and ET_tank were determined as follows: (1) by observing pond WD in every concrete tank above the soil surface at 7 AM every day and calculating E_tank and ET_tank from the daily difference in WD; (2) using the modified E and ET model with Bayesian inference approach; (3) interpolating missing and error observation data by the model estimates for determination; and (4) by calculating Kc, which is defined as the ratio of the ET_tank to the E_tank (ET_tank/E_tank). The determination period of E_tank and ET_tank was the irrigation period, in other words, the time from transplanting or cutting the stem to 14 days before harvesting.

2.4.1. Manual Observation of E_tank and ET_tank

The WD (mm) in every tank was observed every morning using a depth gauge by manual reading (Figure 1). The ET_tank and E_tank (mm day⁻¹) were determined by the daily difference of pond WD. The amount of water lost through percolation and surface runoff, and the amount of irrigation water applied were neglected because the concreate tank cannot be penetrated and there is no runoff and we observed the WD at two times before and after both irrigations. Moreover, when it rains, monitoring the amount of rainfall pouring into a tank is difficult, so that observed data with 5 mm daily precipitation or more were not used. On the other hand, precipitation less than 5 mm was regarded as negligible.

In addition, for inferring the differences in the microclimate between paddy fields and concrete paddy tanks, we compared the daily reference crop evapotranspiration by FAO-56 (ET₀) (mm day⁻¹) with the E_tank. The ET₀ expresses the evaporation power of the atmosphere at a specific location and time of the year, regardless of crop characteristics and soil factors [26], and it was calculated as:

$${\mathrm{ET}}_0=\frac{0.408\Delta R_n\left(R_n-G\right)+\gamma \frac{900}{T_a+273}{u}_2\left(e_s\left(T_a\right)-e\right)}{\Delta +\gamma \left(1+0.34u_2\right)}$$

(1)

where R_n is the net radiation (W m⁻²); G is the soil heat flux (W m⁻²) that was not used because of the daily analysis (G = 0); Δ is the slope of the saturation water vapor pressure curve at the mean air temperature (kPa °C⁻¹); γ is the psychrometric constant (approximately 0.066 kPa °C⁻¹); u₂ is the wind speed at 2 m height (m s⁻¹); e is the mean water vapor pressure (kPa) and e_s is the water vapor pressure at the evaporating surface at air temperature T_a (kPa).

2.4.2. Model Estimation of E_tank and ET_tank for Data Interpolation

Many missing and error data were generated due to inconsistencies in observation such as misreading or absence of reading. Therefore, using the Bayesian inference approach, a dataset was organized to interpolate the missing data with the model estimates by the modified Penman model (P) [27] for E_tank (mm day⁻¹) and the Penman–Monteith (PM) model [28] for ET_tank (mm day⁻¹). In the modified P and PM model, adjustment parameters θ_s, which present a₁, a₂, b₁, b₂, c₁, c₂, and d, were applied for correcting the energy balance and aerodynamic resistances under the condition of the concrete tank. The modified P model is expressed as

$${\mathrm{E}}_{\mathrm{tank}}=\frac{\Delta \left(R_n-G\right)a_1}{\left(\Delta +\gamma \right)\lambda }+\frac{\gamma }{\Delta +\gamma }f\left(u_2\right)\left(e_s\left(T_a\right)-e\right)$$

(2)

$$f\left(u_2\right)=b_1+b_2 u_2$$

(3)

where f(u₂) is a function of the horizontal wind velocity (unitless). The modified PM model is expressed as

$${\mathrm{ET}}_{\mathrm{tank}}=\frac{\mathsf{\Delta }\left(R_n-G\right)a_2+C_p\rho \left(e_s\left(T_a\right)-e\right)/r_a}{\lambda \left(\Delta +\gamma \left(1+r_s/r_a\right)\right)}$$

(4)

$$r_a=\frac{\ln \left(\frac{2.0-d}{Z_{0m}}\right)\ln \left(\frac{2.0-d}{Z_{0h}}\right)}{K^2{u}_2}$$

(5)

$$d=c_1 h_c,Z_{0m}=c_2 h_c\mathrm{and}{Z}_{0h}=0.1 Z_{0m}$$

(6)

$$r_s=\frac{d}{0.5\mathrm{LAI}}$$

(7)

where C_p is the specific heat capacity of dry air at constant pressure (approximately 1.01 kJ kg⁻¹ °C⁻¹); ρ is the mean air density (kg m⁻³); r_a is the aerodynamic resistance (s m⁻¹); r_s is the surface resistance (s m⁻¹); d is the zero-plane displacement height; h_c is the crop height (m) at observation day of WD; Z_0m is the roughness length governing momentum transfer (m), and Z_0h is the roughness length governing the transfer of heat and vapor (m); and K is the von Karman’s constant (0.41). The value of the leaf area index (LAI) was predicted from the measured crop height h_c using the following formula for typical values of field crops [29,30,31]:

LAI = LAI_max + 1.5 ln(h_c)

(8)

The maximum LAI (LAI_max) obtained from the literature was reported as 5.0 for rice [32,33].

2.4.3. Hierarchical Bayesian Approach

The θ_s values were estimated by hierarchical Bayesian inference, which can take into account the variation in observation condition such as seasonal fluctuation and plant growth stage for a dataset. These results can necessarily be generalized across different conditions and groups. The probabilistic model for estimating unknown adjustment parameters (i.e., a₁, a₂, b₁, b₂, c₁, c₂, d) accommodating group-level differences (i.e., different climate conditions and crop growth stages in a dataset) in the modified P and PM model are described as follows.

$$O_{ \left[c\right]}\left[t\right] ~ \mathrm{Normal}\left(S\left({\theta }_{s \left[c\right]}\right)\left[t\right], \sigma \right)$$

(9)

$${\theta }_{s \left[c\right]} ~ \mathrm{Normal}\left({\theta }_{s, mean}, {\sigma }_{{\theta }_s}\right)$$

(10)

where $O_{ \left[c\right]}\left[t\right]$ indicates the observed E_tank and ET_tank at day t in each observation period divided evenly into four periods c (1, 2, 3, 4) for E_tank and three periods c (1, 2, 3) for ET_tank (mm day⁻¹); S[t] is the estimated E_tank and ET_tank by the modified P and PM model at day t (mm day⁻¹); and σ is the standard deviance parameter that represents the measurement error variance for ET_tank and E_tank estimates (mm day⁻¹). Normal indicates a normal distribution from which ET estimates are generated. ${\theta }_{s \left[c\right]}$ represents a prior uncertainty in the parameters of group-level c and can be described as being stochastically generated from a normal distribution of ${\theta }_{s, mean}$ and ${\sigma }_{{\theta }_s}$; ${\theta }_{s, mean}$ is the overall mean distribution of each parameter in a dataset, while ${\sigma }_{{\theta }_s}$ is a random variable distribution that represents the difference in group level of the parameter. We assumed that the specific unknown parameters θ_s were distributed uniformly within the range of 0 ≤ a ≤ 2, 0 ≤ b ≤ 2, 0 ≤ c ≤ 1, and 30 ≤ d ≤ 150 we heuristically set.

In this study, all the simulations and calculations were performed in R version 4.0.2. For estimating the posterior distribution, we used RStan version 2.19.3 developed by the Stan Development Team [34], which employs a Markov Chain Monte Carlo (MCMC) technique to sample from the posterior distribution of a given model. We ran four MCMC chains with 20,000 iterations and monitored them to confirm that the MCMC chains converged to the target distributions. The model estimates of E_tank and ET_tank using the hierarchical Bayesian approach were evaluated by comparing with the observed using common criteria in model evaluation: the root mean squared percentage error (RMSPE) and the coefficient of determination (R²). The R² is a measure of the proportion of the total variance of observed data explained by the estimated data. Non-linear regression analysis for plant growth was analyzed by a generalized additive model explained in [35]. The differences between means were compared by analysis of variance (ANOVA) test at the 5% probability level.

3. Results

3.1. Crops Growth in Different Seasons

Plant growth was higher in the monsoon season than in the cool and dry season. The plant height of TC1 in the monsoon season was the highest, that of RC2 in the cool and dry season was the lowest, and the growth trends for plant height of TC and RC were similar (Figure 3a). In contrast, the number of tillers between TC and RC was substantially different. MC1 and TC1 tillers reached the maximum tillering stage at around 70–80 days after transplanting, while those of the RC1 reached 20 only 7 days after cutting the stem. Many tillers of RC emerged after stem cutting. In the RC2 in the cool and dry season, tillering was active until harvesting, with a large variation of the number of tillers (Figure 3b).

Figure 4 shows the comparison of the plant growth of transplanted rice (i.e., MC1, TC1, MC2, and TC2) and ratoon rice (i.e., RC1 and RC2) with non-linear regression curves and 95% confidence intervals. The accumulated air temperature was calculated from transplanting or cutting the stem to 14 days before harvesting.

The R² of the transplanted rice was higher than that of the ratoon in both plant height and number of tillers. A strong correlation with the accumulated temperature was observed in the number of tillers for the transplanted rice. The transplanted rice reached the maximum tillering at around 1500 °C accumulated temperature. Conversely, the plant growth of the ratoon did not depend on the accumulated temperature, and the R² was less than 0.5.

3.2. Rice Yield Components and Grain Yield

Table 2. Yield components and grain yield for concrete tanks.

Crop	No. of Panicles (m−2)	No. of Spikelets (Panicle−1)	Filled Grains (%)	1000-Grain Weight (g)	Biological Yield (g m−2)	Grain Yield (g m−2)
First trial
Main crop1 (MC1)	377 b ± 52	142 b ± 23	61 a ± 7	21 b ± 0.3	1572 a ± 150	555 a ± 133
Ratoon1 (RC1)	488 a ± 31	80 c ± 6	56 a,b ± 7	22 a ± 0.4	1259 b ± 253	439 a ± 54
Transplant1 (TC1)	262 c ± 17	174 a ± 28	50 b ± 3	22 a ± 0.7	1337 a,b ± 211	426 a ± 53
Second trial
Main crop2 (MC2)	329 b ± 21	115 b ± 9	62 a ± 5	22 a ± 0.7	1048 a ± 138	510 b ± 54
Ratoon2 (RC2)	501 a ± 66	74 c ± 8	30 b ± 10	19 b ± 0.6	699 b ± 103	197 c ± 57
Transplant2 (TC2)	310 b ± 6	158 a ± 35	66 a ± 11	20 b ± 0.5	1109 a ± 111	631 a ± 28

Values within one cropping and with the same lowercase letter are not significantly different at p = 0.05 level between crops of each trial according to Tukey’s honestly significant difference test.

presents the yield components and grain yield of each crop in the concrete tanks. The statistical comparison of the first and the second trial was not made due to the different seasons of the trials. For RCs, the number of panicles in both trials was significantly higher than that of the transplanted crops. The number of spikelets was significantly lower than that of the transplanted crops. The filled grains rate, the 1000-grain weight, the biological yield, and the grain yield of RC2 were significantly lower than those of the others. This was caused by the effect of low temperature in the reproductive stage (BBCH-scale = 51–69) of RCs. The spikelet fertility of the rice decreased, because the rapidly growing booting and reproductive tissues are highly sensitive to low temperatures [36]. The grain yield of RC1 was 20% lower than MC1 and identical with TC1 cultivated in parallel with RC1. However, the yield of RC2 was 39% of that for MC2, exhibiting a significant decrease. The highest grain yield among all crops was exhibited by TC2 together with the longest growth period.

3.3. E_tank and ET_tank Combined with the Observation and Model Estimates

3.3.1. Model Estimates of E_tank and ET_tank for Data Interpolation

For the E_tank, the mean value of the observation was 4.6 ± 1.4 mm (mean ± SD), and that of the model estimates was 4.6 ± 1.2 mm. Both mean values were almost identical, but variation in the dataset was different between the estimates and the observation, as shown in Figure 5a. The evaluation criterion for the model estimate was 36% for RMSPE and 0.59 for R². In addition, the values for ET₀ and for E_tank from the day of year (DOY) 66 to 280 were almost identical. However, approximately from DOY 280, when the air temperature began to drop gradually (Figure 2), the values for ET₀ were underestimated compared with the values for E_tank. We assume that the underestimated values were caused by the different surface condition at the paddy field and the concrete tanks after the harvest of the paddy fields. The paddy field, where the meteorological station was installed was harvested at the end of October (i.e., DOY 300). After harvesting, the field was drained and left fallow during the cool season. For the ET_tank, the RMSPE and R² explained that the model performance in the monsoon season (RC1 and TC1) was lower than other seasons, and the performance for RC was low compared to that of TC cultivated in parallel with RC (Figure 5b). The RC1 was the highest with RMSPE = 62% and lowest with R² = 0.278. In the TC1 and MC2 with an incomplete dataset, 60% and 40% data deficient, respectively, using the Bayesian parameter inference approach, the ET_tank for TC1 and MC2 could be estimated with a performance of RMSPE = 19.6% and 22%, and R² = 0.462 and 0.677, respectively. The MC1 and TP2 with sufficient datasets was with R² = 0.809 and 0.873, respectively, which indicated that the model explained more than 80% of the variability in the observed data (Figure 5b).

3.3.2. Determination of E_tank and ET_tank

Summary of the E_tank and ET_tank for the model estimates and for the corrected observations are shown in

Table 3. Summary of E_tank and ET_tank for the model estimates and for the corrected observations.

Crop	Total Period (Days)	Model Estimates				Corrected Observations
		Etank (mm)		ETtank (mm)		Etank (mm)		ETtank (mm)
		Total	Daily Mean	Total	Daily Mean	Total	Daily Mean	Total	Daily Mean
First trial
MC1	73	456	6.2	754	10	464	6.4	762	10
RC1	69	259	3.8	446	6.5	254	3.7	450	6.5
TC1	75	287	3.8	479	6.4	274	3.7	479	6.4
Second trial
MC2	65	276	4.2	483	7.4	289	4.5	483	7.4
RC2	67	237	3.5	363	5.4	237	3.5	356	5.3
TC2	91	370	4.1	606	6.7	364	4.0	606	6.7

MC, RC, and TC indicate the main crop, ratoon, and transplanted crop, respectively. The total period indicates the irrigated period in a tank for the observation of water depth from transplanting or cutting the stem to around 14 days before harvesting. The corrected observations represent the observed data with missing data interpolated by the model estimation.

. The total period represents the irrigated period in a tank for the observation of water depth from transplanting or cutting the stem to 14 days before harvesting. The model estimates for RC2 of E_tank, and TC1, MC2, and TC2 of ET_tank were identical with the corrected observations. For the daily mean of the E_tank and ET_tank on the corrected observations, the largest value was MC1 cultivated in the hot and dry summer, and the smallest was RC2 in the cool and dry season. Furthermore, the ET_tank of RC1 was 59% of MC1, and RC2 was 74% of MC2. Conversely, comparing TC cultivated in parallel with RC, the ET_tank of RC1 was 94% that of TC1, and RC2 was 59% that of TC2. The total ET_tank of the first and second ratoon double-cropping was 1212 mm and 839 mm, respectively.

3.4. Crop Coefficient

Figure 6 shows Kc values (ET_tank/E_tank) described with non-liner regression curves and bands corresponding to the 95% confidence intervals. The target period was from transplanting or cutting the stem of the MC to 14 days before harvesting. The Kc regression curve of all crops was approximately set in a range of 1 to 2. However, the variation of the Kc value of RC1 and TC1 cultivated in the monsoon was very large compared to others, resulting in R² = 0.100 and 0.211, respectively. The Kc regression curves of RC2 had a different shape from that of MC and TC. The regression curve of MC and TC showed a logistic curve, but that of RC2 formed a parabolic curve.

3.5. Water Productivtiy

Figure 7 presents the comparison of the ET_tank and grain yield as well as grain yield and WP_ET for each cropping. The log approximation between the grain yield and the ET_tank, and the WP_ET and the grain yield were moderately established (R² = 0.558 and R² = 0.579, respectively). The WP_ET in this study ranged from 0.55 to 1.06. The MC1 cultivated in the summer season exhibited the highest ET_tank together with a large yield. By contrast, the WP_ET of MC1 was lower than that of other crops except for RC2. Regarding the ratoon, the WP_ET of the RC1 of the first trail was higher than that for MC1 and TC1. The WP_ET of the RC2 was considerably lower than that of others because of the low yield induced by the low temperatures.

4. Discussion

4.1. Method for ET Dermination

This study combined manual observation of the WD in concrete paddy tanks and the ET model estimation using Bayesian inference; this enabled the ET_tank to be determined with an incomplete observation dataset, around 20–60% data deficit in total. Therefore, this method is useful when it is difficult to install a complex observation system and would be helpful when the observation error often occurs or the observation frequency needs to be reduced.

However, for the estimation of ET_tank for RC, the model performance was decreased compared to that of TC. We assumed that the regression formula of the LAI could not meet the actual plant growth of RC because the tillering trait of RC was significantly different from that of the TC. Moreover, the considerable variation in the number of tillers for RC made the model estimation difficult. Since ratoon ability is a complex trait dependent on many heritable and environmental factors [12], the relationship between regeneration traits and surface resistance in modeling the ET of ratoon crops needs to be further developed.

4.2. ET Characteristic of RC

The total ET_tank of RC in the two different seasons was 450 mm (RC1) and 356 mm (RC2), 59% and 74% that of MC, respectively. This reduction ratio of RC was similar to previous studies [10,11,12]. However, in this study, the growth period for MC and RC was nearly the same. Additionally, the daily mean of ET_tank for RC1 (6.5 mm) was equivalent to TC1 (6.4 mm) cultivated in parallel with RC1. In other words, it indicates that the difference in the amount of ET_tank between MC and RC was mainly attributed to the climate conditions in each cultivation. In most previous studies [13,14,15,37], the growth period of RC was reduced to 40–60% of MC, contributing to lowering irrigation water supply for RC, which was different from the results of this study. The factors in the growth period of RC included the effect of varieties, environmental conditions, and cultivation practices [38]. Therefore, for understanding the ET characteristics of ratoon cropping, the impact of different varieties and cultivation management and environment on the plant growth of RC should be investigated.

Although there are many studies on the Kc of single rice cropping, few studies have reported on the Kc of ratoon cropping. In this study, the Kc value was determined from the evaporation loss from a ponding tank and the crop evapotranspiration was determined from a paddy tank. This result showed the different trend of Kc values between TC2 and RC2. Since the initial tillering of RC2 was more vigorous than that of TC2, the increase rate of Kc at the initial stage within one month for RC2 (83%) was higher than that for TC2 (20%). It is suggested that an irrigation scheduling of RC in the initial stage should take high crop water requirements into account. However, since the study on Kc of ratoon is limited, it is essential to study the Kc of RC under various conditions in the future.

4.3. Viablility of Practice for Ratoon Cropping

The ratoon rice exhibited approximately a 40–60% reduction in grain yield and growth period compared with the main crop [13,14,15,39,40]. As a result of the well-established linear relationship between plant biomass and transpiration [41], the fundamental factor causing the low yield of the ratoon crop is likely due to the shorter growth period. This leads to a decrease in total transpiration and a decrease in plant biomass, which affects the sink and source capacity of the yield performance. However, in this study, the growth period of ratoon crops (RC1 and RC2) was similar to that of the transplanted rice. For this reason, the observed grain yield for RC1 was similar to that of TC1 cultivated in parallel with RC1. Moreover, the WP_ET in this study ranged from 0.55 to 1.06, which constitutes a larger range than those reported in previous studies (e.g., 0.19–0.29 [42]; 0.22–0.39 [43]; 0.23–0.27 [44]; 0.24–0.70 [45]; 0.30–0.33 [46]). This indicates that the rice variety and environment applied in this study might be suitable for ratoon cropping. However, because the results of this study are based on small tank cultivation, careful evaluation of plant growth and yield performance of ratoon crop in fields is necessary. Further study on ratoon in Myanmar is necessary for generating reliable recommendations for farmers.

5. Conclusions

Few studies on the ET and Kc of ratoon rice double-cropping have been reported, and it remains difficult to observe ET in developing countries. Therefore, the ET and Kc for ratoon cropping during 2019–2020 in the tropical region of Myanmar were determined using a simple method. The conclusions are described as follows:

• To determine ET in ratoon cropping, we combined manual observation of WD in concrete paddy tanks and the ET model estimation using the Bayesian inference approach. The ET and Kc could be determined using this method with an incomplete observation dataset. The complex regeneration traits of ratoon affected the prediction of LAI and ET. The relationship between regeneration traits and surface resistance in modeling the ET of ratoon crops needs to be further developed.
• The total amount of ET for the RC1 (450 mm) was reduced by around 60% compared to the MC1 (762 mm). However, the growth period (82 d) and the mean ET value (6.5 mm day⁻¹) of ratoon were similar to transplanted rice (89 d and 6.4 mm day⁻¹) cultivated in parallel with ratoon. Thus, the difference in the ET was mainly attributed to the difference in climate conditions in each cropping period.
• The Kc regression curve between transplanted rice and RCs was different because of the tillering traits, and the increase rate of Kc at the initial stage for RC2 (83%) was higher than that for TC2 (20%). It is suggested that the irrigation scheduling of ratoon cropping in the initial growth stage should take high crop water requirements into account.
• The yield (439 g m⁻²) and WP_ET (0.98 kg m⁻³) of ratoon crop equivalent to the yield (426 g m⁻²) and WP_ET (0.89 kg m⁻³) of transplanted rice was observed for crops cultivated in concrete tanks. Further study on ratoon in Myanmar is essential for clarifying the viability of ratooning and for generating reliable recommendations for farmers.