Forecasts of maximum and minimum air temperatures are essential to mitigate the damage of extreme weather events such as heat waves and tropical nights. The Numerical Weather Prediction (NWP) model has been widely used for forecasting air temperature, but generally it has a systematic bias due to its coarse grid resolution and lack of parametrizations. This study used random forest (RF), support vector regression (SVR), artificial neural network (ANN) and a multi‐model ensemble (MME) to correct the Local Data Assimilation and Prediction System (LDAPS; a local NWP model over Korea) model outputs of next‐day maximum and minimum air temperatures (Tmaxtþ1 and Tmintþ1) in Seoul, South Korea. A total of 14 LDAPS model forecast data, the daily maximum and minimum air temperatures of in‐situ observations, and five auxiliary data were used as input variables. The results showed that the LDAPSmodel had an R of 0.69, a bias of −0.85 °C and an RMSE of 2.08 °C for Tmaxtþ1 forecast, whereas the proposed models resulted in the improvement with R from 0.75 to 0.78, bias from −0.16 to −0.07 °C and RMSE from 1.55 to 1.66 °C by hindcast validation. For forecasting Tmintþ1 , the LDAPS model had an R 2 of 0.77, a bias of 0.51 °C and an RMSE of 1.43 °C by hindcast, while the bias correction models showed R values ranging from 0.86 to 0.87, biases from −0.03 to 0.03 °C, and RMSEs from 0.98 to 1.02 °C. The MMEmodel had better generalization performance than the three single machine learning models by hindcast validation and leave‐one‐station‐out cross‐validation.

/pdf/comparative-assessment-of-various-machine-learning-based-4d09twmos1.pdf

Comparative Assessment of Various Machine Learning-Based Bias Correction Methods for Numerical Weather Prediction Model Forecasts of Extreme Air Temperatures in Urban Areas

The conventional approach to the frequency analysis of extreme precipitation is complicated by non-stationarity resulting from climate variability and change. This study utilized a non-stationary frequency analysis to better understand the time-varying behavior of short-duration (1-, 6-, 12-, and 24-h) precipitation extremes at 65 weather stations scattered across South Korea. Trends in precipitation extremes were diagnosed with respect to both annual maximum precipitation (AMP) and peaks-over-threshold (POT) extremes. Non-stationary generalized extreme value (GEV) and generalized Pareto distribution (GPD) models with model parameters made a linear function of time were applied to AMP and POT respectively. Trends detected using the Mann–Kendall test revealed that the stations showing an increasing trend in AMP extremes were concentrated in the mountainous areas (the northeast and southwest regions) of South Korea. Trend tests on POT extremes provided fairly different results, with a significantly reduced number of stations showing an increasing trend and with some stations showing a decreasing trend. For most of stations showing a statistically significant trend, non-stationary GEV and GPD models significantly outperformed their stationary counterparts, particularly for precipitation extremes with shorter durations. Due to a significant-increasing trend in the POT frequency found at a considerable number of stations (about 10 stations for each rainfall duration), the performance of modeling POT extremes was further improved with a non-homogeneous Poisson model. The large differences in design storm estimates between stationary and non-stationary models (design storm estimates from stationary models were significantly lower than the estimates of non-stationary models) demonstrated the challenges in relying on the stationary assumption when planning the design and management of water facilities. This study also highlighted the need of caution when quantifying design storms from POT and AMP extremes by showing a large discrepancy between the estimates from those two approaches.

Non-stationary frequency analysis of extreme precipitation in South Korea using peaks-over-threshold and annual maxima

This study investigates the features of eastward extension of the South Asian high (SAH) and its connection with diabatic heating and rainfall over eastern Asia on subseasonal time scales. The causes of SAH’s eastward extension are examined by potential vorticity (PV) diagnosis with emphasis on the joint role of diabatic heating feedback and midlatitude wave train. The SAH’s eastward extension features eastward propagation of a wave train across Eurasia. Among the wave train, the migration of weak high from the western flank of the Tibetan Plateau (TP) to the east of TP contributes to the SAH’s eastward extension at the early stage. When the SAH approaches its easternmost position, a strong negative PV (positive geopotential height) center prevails to the east of the TP at 200 hPa. The associated anomalies in diabatic heating and rainfall include the anomalous heating and above-normal rainfall over the South China Sea (SCS) and subtropical western Pacific occurring 12 days before the SAH’s eastern...

Characteristics and Mechanisms of the Subseasonal Eastward Extension of the South Asian High

In order to systematically and visually understand well-known but qualitative and complex relationships between synoptic fields and heavy rainfall events in Kyushu Islands, southwestern Japan, during the BAIU season, these synoptic fields were classified using the Self-Organizing Map (SOM), which can convert complex non-linear features into simple two-dimensional relationships. It was assumed that the synoptic field patterns could be simply expressed by the spatial distribution of (1) wind components at the 850 hPa level and (2) precipitable water (PW) defined by the water vapor amount contained in a vertical column of the atmosphere. By the SOM algorithm and the clustering techniques of the U-matrix and the K-means, the synoptic fields could be divided into eight kinds of patterns (clusters). One of the clusters has the notable spatial features represented by a large PW content accompanied by strong wind components known as low-level jet (LLJ). The features of this cluster indicate a typical synoptic field pattern that frequently causes heavy rainfall in Kyushu during the rainy season. In addition, an independent data set was used for validating the performance of the trained SOM. The results indicated that the SOM could successfully extract heavy rainfall events related to typical synoptic field patterns of the BAIU season. Interestingly, one specific SOM unit was closely related to the occurrence of disastrous heavy rainfall events observed during both training and validation periods. From these results, the trained SOM showed good performance for identifying synoptic fields causing heavy rainfall also in the validation period. We conclude that the SOM technique may be an effective tool for classifying complicated non-linear synoptic fields and identifying heavy rainfall events to some degree.

Identification of typical synoptic patterns causing heavy rainfall in the rainy season in Japan by a Self-Organizing Map

[1] Decadal changes in surface air temperature (SAT) variability and cold surge characteristics over Northeast Asia during late winter (January–March) are analyzed for the past three decades. Power spectrum analysis of SAT reveals that the low-frequency variabilities with a period longer than 10 days are significantly enhanced, while the high-frequency variabilities with a period shorter than 10 days are weakened in the 1980s and 2000s. Moreover, cold surges were stronger and lasted longer during the 1980s and 2000s compared to those that occurred in the 1990s. Here, we propose that large-scale atmospheric conditions manifested by a different phase of the Arctic Oscillation (AO) provide preconditioning for a cold surge event, which showed a prominent decadal fluctuation. The more (less) frequent strong and long-lasting cold surge occurrences in the 1980s and 2000s (1990s) are preceded by the more dominant negative (positive) phase of the AO. Lag-composite analyses for cold surge events categorized by the AO phases indicate that stronger and longer-lasting cold air advection dominates at the lower-level, when upper-level wave train and coastal trough are developed over East Asia under the strong negative AO phase. These results suggest that the decadal changes in SAT variability and cold surge characteristics are strongly associated with the decadal changes in the phase distribution of the AO.

https://agupubs.onlinelibrary.wiley.com/doi/pdf/10.1029/2011JD016929

Decadal changes in surface air temperature variability and cold surge characteristics over northeast Asia and their relation with the Arctic Oscillation for the past three decades (1979-2011)

The Korean peninsula has complex and diverse weather phenomena, and the Korea Meteorological Administration has been working on various numerical models to produce better forecasting data. The Unified Model Local Data Assimilation and Prediction System is a limited-area working model with a horizontal resolution of 1.5 km for estimating local-scale weather forecasts on the Korean peninsula. However, in order to numerically predict the detailed temperature characteristics of the urban space, in which surface characteristics change rapidly in a small spatial area, a city temperature prediction model with higher resolution spatial decomposition capabilities is required. As an alternative to this, a building-scale temperature model was developed, and a 25 m air temperature resolution was determined for the Seoul area. The spatial information was processed using statistical methods, such as linear regression models and machine learning. By comparing the accuracy of the estimated air temperatures with observational data during the summer, the machine learning was improved. In addition, horizontal and vertical characteristics of the urban space were better represented, and the air temperature was better resolved spatially. Air temperature information can be used to manage the response to heat-waves and tropical nights in administrative districts of urban areas.

Development of an Urban High-Resolution Air Temperature Forecast System for Local Weather Information Services Based on Statistical Downscaling

The first three principal modes of wintertime surface temperature variability in Seoul, South Korea (37.33°N, 126.59°E), are extracted from the 1979–2008 observed records via cyclostationary EOF (CSEOF) analysis. The first mode represents the seasonal cycle, the principle physical mechanism of which is associated with the continent–ocean sea level pressure contrast. The second mode mainly describes the overall wintertime warming or cooling. The third mode depicts subseasonal fluctuations of surface temperature. Sea level pressure anomalies to the west of South Korea (eastern China) and those with an opposite sign to the east of South Korea (Japan) are a major physical factor both for the second mode and the third mode. These sea level pressure anomalies with opposite signs alter the amount of warm air to the south of South Korea, which changes the surface temperature in South Korea. The PC time series of the seasonal cycle is significantly correlated with the East Asian winter monsoon index and e...

Physical Mechanisms of the Wintertime Surface Air Temperature Variability in South Korea and the near-7-Day Oscillations

Characteristics of localized heavy rainfall events (LHREs) in South Korea are analyzed using high-resolution precipitation reanalysis data from the Korea Meteorological Administration (KMA). The LHREs, defined as summertime heavy rainfall events with 30 mm/h or more, are detected using in-situ measurements, and their spatio-temporal structures are classified into the three groups using precipitation reanalysis data. The first group is mainly composed of band-type precipitation around the central regions of South Korea (CHREs). The second group predominantly consists of LHREs in the southern regions (SHREs). The third group contains diverse and isolated rainfall events (IHREs). Each group exhibits a different rainfall persistency, diurnal cycle, and occurrence frequency. In general, CHREs peak mostly in July and have shorter time scales compared to SHREs. SHREs are more nocturnal in nature with maximum frequencies in both July and August. IHREs, which consist of 47% of total LHREs, are composed of both nocturnal rainfall events in the southern coastal regions and daytime rainfall events in the inland regions. These results suggest that LHREs in South Korea have multiple characteristics and cannot be simply explained by one or two representative mechanisms.

Classification of Localized Heavy Rainfall Events in South Korea

[1] Three distinct physical mechanisms in the seasonal cycle of the 120 day (19 May to 15 September) summer precipitation in Korea (126°E–130°E, 33°N–38°N) were identified using the 1979–2008 observed precipitation records at 61 Korea Meteorological Administration stations. Detailed space-time structures of the physical mechanisms of precipitation variability were derived using the daily National Center for Environmental Prediction/National Center for Atmospheric Research reanalysis data over Asia (80°E–180°E, 0°–60°N). The seasonal cycle of summertime precipitation in Korea exhibits three principal temporal scales (seasonal, subseasonal, and high-frequency components) of variability, each with distinct physical mechanisms. The seasonal component represents the variability associated with the evolution of the Asian summer monsoon, specifically the East Asia summer monsoon, governed primarily by large-scale circulation as a result of changes in sea level pressure contrasts between the Asian continent and the surrounding oceans. The arrival and the duration of a monsoon front primarily shape the seasonal evolution of precipitation in Korea. The bimodal peaks are due to the low-level circulation change as a result of redistribution of temperature and, subsequently, of sea level pressure during summer. The subseasonal component has characteristic time scales of 10–30 days and is associated with eastward-moving upper-level disturbances at ∼40°N. The upper-level disturbances affect the meridional circulations, resulting in low-level convergence/divergence not only underneath but also to the south and to the north of the disturbance. From mid-July to mid-August, the subseasonal component is more clearly observable, and the period of oscillations is generally shorter, than during early or late summer. The high-frequency component with time scales of less than 10 days is associated with midlatitude baroclinic Rossby waves; synoptic-scale variations of upper-level geopotential height and low-level moisture convergence are well correlated with the high-frequency variability of precipitation in Korea. Positive precipitation anomalies over Korea correspond to an upper-level divergence with anticyclonic wind anomalies caused by baroclinic Rossby waves with axes around ∼40°N. The vertical structures of key physical variables are well explained in the context of baroclinic instability with the geopotential height anomaly field connecting the upper-level divergence and the lower-level convergence tilted westward with height and the temperature anomaly field slanting slightly in the opposite direction.

https://agupubs.onlinelibrary.wiley.com/doi/pdf/10.1029/2009JD013561

Physical mechanisms of the seasonal, subseasonal, and high-frequency variability in the seasonal cycle of summer precipitation in Korea

[1] Summertime precipitation variability in Korea has changed significantly in recent years. To understand how the physical mechanisms of summertime precipitation in Korea vary with climate change, we analyzed observed precipitation records for 1996–2008 in comparison with those for 1979–1995 at 61 Korea Meteorological Administration stations distributed over South Korea. We investigated detailed physical changes by extracting space-time structures of the physical mechanisms from the daily NCEP/NCAR reanalysis data over East Asia via cyclostationary EOF (CSEOF) analysis. Due to the change of low-level circulation patterns in the Asian region, the commencement, duration, and retreat of the East Asian monsoon front has varied significantly over recent years. Specifically, the first peak of the bimodal precipitation pattern in Korea has started earlier and significantly increased in intensity. The second peak has broadened in recent years and the typical seasonal period of decreased precipitation has weakened. The strength of the sub-seasonal component of precipitation has increased in recent years due to the strengthening of meridional circulation between the subtropics and the midlatitudes. A conspicuous change in the vertical structure of the sub-seasonal component is observed in recent records. Increased warm and moist advection from the south and decreased cold and dry advection from the north seem to be the primary reasons for such a change. The high-frequency component of summer precipitation with time scales shorter than 10 days is formed by the baroclinic instability observed in recent years. This component has increased by ∼25% due primarily to the strengthened dynamic and thermodynamic processes.

https://agupubs.onlinelibrary.wiley.com/doi/pdf/10.1029/2011JD017268

Joon-Woo Roh

Papers

Development of an Urban High-Resolution Air Temperature Forecast System for Local Weather Information Services Based on Statistical Downscaling

Physical Mechanisms of the Wintertime Surface Air Temperature Variability in South Korea and the near-7-Day Oscillations

Classification of Localized Heavy Rainfall Events in South Korea

Physical mechanisms of the seasonal, subseasonal, and high-frequency variability in the seasonal cycle of summer precipitation in Korea

Decadal changes in the physical mechanisms of the seasonal cycle of summertime precipitation variability in Korea