Page 25 - APEC CLIMATE CENTER 2025 Annual Report

APEC CLIMATE CENTER 2025 ANNUAL REPORT

Highlighted The five improved impact indices go beyond simple meteorological statistics to provide Highlighted 6. Evaluation of Heatwave and Drought Predictability

practical decision-support information directly linked to productivity improvement and Using DCPP Prediction Models over East Asia
Achievements management stability in the agricultural field. Achievements
in 2025 First, it enables data-driven precision agriculture planning. By utilizing the standardized in 2025 ㉖ Ms. Daeun Jeong (downy@apcc21.org) Dr. Hyun-Ju Lee (asteria1104@apcc21.org)

'Growing Degree Days (GDD)' and the quantitatively calculated 'Growing Season Length
(GSL),' farmers can accurately forecast optimal timings for each growth stage, from germi-
Recently, East Asia has experienced a rapid increase in the frequency and intensity of ex-
nation to flowering and harvest. This allows farmers to determine the appropriate timing
treme hydrometeorological disasters, such as heatwaves, droughts, and floods, driven by
for fertilization and irrigation to enhance product quality and scientifically assess shifts
the acceleration of climate change. To proactively respond to the climate crisis and mini-
in suitable cultivation areas or the feasibility of double cropping due to climate change,
mize socio-economic damage, securing reliable mid-to-long-term prediction information
enabling preemptive preparation of future cropping systems.
1)
on a decadal scale is essential. In particular, the Decadal Climate Prediction Project (DCPP),
a core initiative of the World Climate Research Programme (WCRP), is an innovative system
Second, it contributes to energy efficiency and cost reduction for facility farmers. The
that projects the climate for the next decade by incorporating observed initial conditions
newly introduced 'Facility Heating Degree Days (HDD)' accurately predicts winter heating
into climate models. While its potential for global climate risk management is highly rec-
energy demand for greenhouses by considering the critical growth temperature for each
ognized, systematic diagnosis and verification of its performance regarding East Asian ex-
crop. This not only assists smart farms and facility horticulture farmers in significantly re-
tremes remain insufficient. Therefore, this study aims to precisely diagnose the character-
ducing energy costs by minimizing unnecessary heating but also serves as a crucial basis
istics and limitations of the DCPP prediction system. By providing essential baseline data
for quantitatively assessing the impact of greenhouse gas reduction in the agricultural
for enhancing forecast reliability and developing bias correction technologies, this research
sector at the national level.
seeks to contribute to establishing an effective framework for climate disaster response.
Third, it minimizes agricultural damage from meteorological disasters. 'Chill Days,'
All five DCPP prediction systems consistently exhibit positive biases (model values exceed-
which reflect daily meteorological data, provide sophisticated predictions for fruit tree
ing observations) across East Asia when predicting heatwave frequency indices of warm
dormancy break and flowering times, enabling preparation against spring frost damage
2)
3)
days (TX90p) and warm nights (TN90p). These indices show high sensitivity to warming,
caused by abnormal low temperatures. Additionally, the 'Livestock Heat Index (THI),'
with biases intensifying as forecast lead time increases. In contrast, the intensity indices—
which reflects thresholds for each livestock type (cattle, pigs, chickens), provides specific
4)
annual maximum of daily maximum temperature (TXx) and annual maximum of daily
risk information in 5 levels (Normal to Fatal). This significantly strengthens the risk man-
5)
minimum temperature (TNx)—generally show negative biases (model values lower than
agement capabilities of livestock farmers by mitigating livestock stress and preventing
observations). The upward trends in heatwave indices simulated by the models are over-
mass mortality during summer heatwaves.
estimated compared to observations, with particularly pronounced increases in TX90p
and TN90p after the 2000s. Predictability analysis indicates that frequency-based indices
exhibit higher skill than intensity-based indices; notably, MPI-ESM1-2-HR demonstrates high
Glossary correlations and low errors for heatwave frequency indices. In contrast, CMCC-CM2-SR5
shows low predictive skill across all heatwave indices, with negative correlations for TN90p.
1) DCPP:
Overall, models exhibiting relatively robust predictive performance may allow the direct use
Decadal Climate Prediction Project of uncorrected forecasts, whereas models characterized by strong warming trends and sub-
stantial biases would require bias correction based on observed trends.
2) TX90p (Warm days):
The drought prediction performance of the DCPP systems was rigorously evaluated
Number of days when daily maximum
through both deterministic and probabilistic verification, spanning lead times from in-
temperature is greater than the 90th
percentile dividual years to multi-year averages of three, five, and nine years. While most models
exhibited a positive precipitation bias, predictive performance improved significantly
when transitioning from single-year forecasts to multi-year average assessments. These
3) TN90p (Warm nights):
results empirically demonstrate that the DCPP systems possess a distinct strength in
Number of days when daily minimum
capturing the long-term decadal mean state rather than predicting the specific climate
temperature is greater than the 90th
variability of a single year. Furthermore, the study revealed clear discrepancies in perfor-
percentile
mance among the models depending on the verification framework utilized. Specifically,
MIROC6 demonstrated the highest skill in deterministic verification, whereas CanESM5
4) TXx:
excelled in probabilistic verification. These distinct performances are attributed to the
Annual maximum value of daily maxi- unique error characteristics and internal physical mechanisms inherent in each individ-
mum temperature
ual climate model. Consequently, the findings indicate that a uniform bias correction ap-
proach applied across all models is insufficient for fundamentally enhancing the accuracy
5) TNx: of DCPP-based drought information. Instead, the research emphasizes the necessity of
Annual maximum value of daily mini- implementing customized calibration methods that account for the specific prediction
mum temperature characteristics and error profiles of each model.
25
24 25
24