3.2. Farmer-level adaptation
After the model was calibrated, the simulated anthesis day and maturity
day were exactly the same as the observed values. The simulated
yields of both the Sen Pidao and Phka Rumduol cultivars were nearly
equal to their measured values. The simulated tops weight and byproduct
weight were less than their corresponding observed weight;
the differences in the simulated tops weights for Sen Pidao and for
Phka Rumduol at anthesis were −0.53% and −13.50%, respectively.
The simulations of the by-products at maturity for both Sen Pidao and
Phka Rumduol were less than the observations, by −1.17% and
−23.73%, respectively.
The experimental data collected in 2010 (Sen Pidao and Phka
Rumduol) and 2013 (Phka Rumduol only) were used to validate the
CERES-Rice model. For the Sen Pidao cultivar, the agreement between
the predicted and observed values was good, with differences of
7 days for the anthesis day and 6 days for the maturity day. For Phka
Rumduol, the difference between the predicted and observed values
was 7 days for the anthesis day and 14 days for the maturity day in
2010. The difference between the simulated and observed timing of anthesis
for both rice cultivars was similar to that found by Yao et al.,
(2005), whose simulation was 10% off of the observed anthesis duration.
For 2013, the simulated panicle initiation day for Phka Rumduol
was only 2 days later than the observed day. However, the simulated
anthesis day and maturity daywere 15 and 13 days, respectively, longer
than observed values. Tongyai (1994) reported that the simulated number
of days to physiological maturitywas overestimated by 8 to 16 days
for three sites in Bangkok, Thailand. However, the simulated yields
agreed well with the observed values, with differences of 1.86% and
1.19% for Sen Pidao and Phka Rumduol, respectively, in 2010, and 3.1%
for Phka Rumduol in 2013. Timsina and Humphreys (2006) found that
the average difference between simulated and observed grain yields in
CERES-Rice and CERES-Wheat was approximately 23%.
The CERES-Rice model was used to recommend the best combination
of adaptation strategies for the grid cell (12.6 °N and 103.8 °E)
where irrigationwill most offset the negative impacts of climate change
on rice yields according to the GLAM-Ricemodel (Fig. 4). Fig. 5 displays
the changes in rice yields for the Sen Pidao and Phka Rumduol cultivars
under future climate conditions with different combinations of irrigation,
fertilizer, and adjusted planting date, relative to the baseline period.
To determine the planting dates for the greatest yields, simulations
were conducted that shifted the planting dates fromthe baseline period,
a shift of 50 days before and after, respectively, at an interval of 1 day.
Simulations by the field-scale crop model showed that the response of
rice yields to climate change will differ depending on cultivars, climate
scenarios, irrigation conditions, planting date adjustment, and fertilizer
application rates. The prices of the nitrogen fertilizer and the paddy rice
were approximately 0.65 USD (kg-N)−1 and 0.325 USD kg−1, respectively
(CARDI, 2012).We compared the changes in the costs and benefits
from the increases in the fertilizer rate based on the recommended
application rate (50 kg N ha−1). The increase in the fertilizer rate from
50 kg N ha−1 to 100 kg N ha−1 costs additionally about
32.5 USD ha−1, while the increase in the fertilizer rate increases rice
yield up to 350 kg ha−1. The benefit-cost ratio from this increase of
the fertilizer rate is approximately 3.5. However, the benefit-cost ratio
from the increase in the fertilizer rate from 50 kg N ha−1 to
150 kg N ha−1 is about 2.75. Less increase in the rice yield was found
from the increase in the fertilizer rate from 150 kg N ha−1 to
200 kg N ha−1 than from100 kg N ha−1 to 150 kg N ha−1. These results
show that 100 kg N ha−1 of fertilizer rate could be economically better.
We attempted to determine the best combination of adaptation strategies
in order to offset the negative impacts of climate change on rice
yields by examining the following scenarios:Case1. 50 kg N ha−1 and no water stress (denoted as NOAD)