The 3-hour-interval prediction of ground-level temperature in South Korea using dynamic linear models

The 3-hour-interval prediction of ground-level temperature from 00 h out to 45 h in South Korea(38 stations) is performed using the DLM (dynamic linear model) in order to eliminate the systematic error of numerical model forecasts. Numerical model forecasts and observations are used as input values of the DLM. According to the comparison of the DLM forecasts to the KFM (Kalman filter model) forecasts with RMSE and bias, the DLM is useful to improve the accuracy of prediction.     

A Machine Learning-based Cloud Detection Algorithm for the Himawari-8 Spectral Image

Cloud Masking is one of the most essential products for satellite remote sensing and downstream applications. This study develops machine learning-based (ML-based) cloud detection algorithms using spectral observations for the Advanced Himawari Imager (AHI) onboard the Himawari-8 geostationary satellite. Collocated active observations from Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) are used to provide reference labels for model development and validation. We introduce both daytime and nighttime algorithms that differ according to whether solar band observations are included, and the artificial neural network (ANN) and random forest (RF) techniques are adopted for comparison. To eliminate the influences of surface conditions on cloud detection, we introduce three models with different treatments of the surface. Instead of developing independent ML-based algorithms, we add surface variables in a binary way that enhances the ML-based algorithm accuracy by ~5%. Validated against CALIOP observations, we find that our daytime RF-based algorithm outperforms the AHI operational algorithm by improving the accuracy of cloudy pixel detection by ~5%, while at the same time, reducing misjudgment by ~3%. The nighttime model with only infrared observations is also slightly better than the AHI operational product but may tend to overestimate cloudy pixels. Overall, our ML-based algorithms can serve as a reliable method to provide cloud mask results for both daytime and nighttime AHI observations. We furthermore suggest treating the surface with a set of independent variables for future ML-based algorithm development.     

Depositional mechanics and sequences of base surges, Songaksan tuff ring, Cheju Island, Korea

The Songaksan mount in the southwestern part of Cheju Island, Korea, is a Taalian tuff ring produced by phreatomagmatic explosions at an aquifer. A detailed analysis of proximal-to-distal facies changes reveals that the tuff ring sequence can be represented by 21 sedimentary facies; one lateral facies sequence (LFS) and three vertical facies sequences (VFS). The VFS 1 and 2 are representative of facies relationships in horizontal near-vent deposits. The VFS 1 comprises scour-fill bedded tuff, inversely graded tuff, massive tuff and laminated tuff from base to top. The VFS 2 is a variant of the VFS 1, replaced by an inversely graded lapilli tuff unit at the base. The sequences suggest traction carpet, suspension and minor traction sedimentation from a high-concentration near-vent base surge. The LFS 1 and the VFS 3 are distilled from outward-dipping flank deposits. Both sequences begin with disorganized lapilli tuff, followed successively by stratified (lapilli) tuff, dune-bedded (lapilli) tuff, very thinbedded tuff and accretionary lapilli. They are suggestive of waning base surge which decreases in particle concentration, suspended-load fall-out rate and flow regimes with an increase in traction and sorting. These facies sequences suggest that a base surge experiences flow transformation with its flow characters changing with time and space. A near-vent base surge is turbulent, uniformly mixed and highly concentrated and produces scour-fill bedded tuff. As capacity decreases, the surge transforms into a dense and laminar underflow and a dilute and turbulent upper part (gravity transformation), depositing inversely graded, massive and normally graded (lapilli) tuff. Ensuing loss of sediment load and mixing of ambient air result in flow dilution (surface transformation). Stratified and dune-bedded units are produced by tractional processes of turbulent and low-concentration surge. Further dilution causes deceleration and cooling and results in precipitation of moistened ash and accretionary lapilli from suspension.     

A New Infrared Atmospheric Sounding Interferometer Channel Selection and Assessment of Its Impact on Met Office NWP Forecasts

A new set of Infrared Atmospheric Sounding Interferometer(IASI) channels was re-selected from 314 EUMETSAT channels.In selecting channels,we calculated the impact of the individually added channel on the improvement in the analysis outputs from a one-dimensional variational analysis(1D-Var) for the Unified Model(UM) data assimilation system at the Met Office,using the channel score index(CSI) as a figure of merit.Then,200 channels were selected in order by counting each individual channel's CSI contribution.Compared with the operationally used 183 channels for the UM at the Met Office,the new set shares 149 channels,while the other 51 channels are new.Also examined is the selection from the entropy reduction method with the same 1D-Var approach.Results suggest that channel selection can be made in a more objective fashion using the proposed CSI method.This is because the most important channels can be selected across the whole IASI observation spectrum.In the experimental trial runs using the UM global assimilation system,the new channels had an overall neutral impact in terms of improvement in forecasts,as compared with results from the operational channels.However,upper-tropospheric moist biases shown in the control run with operational channels were significantly reduced in the experimental trial with the newly selected channels.The reduction of moist biases was mainly due to the additional water vapor channels,which are sensitive to the upper-tropospheric water vapor.     

A New Algorithm for Sea Fog／Stratus Detection Using GMS-5 IR Data

A new algorithm for the detection of fog/stratus over the ocean from the GMS-5 infrared (IR) channel data is presented. The new algorithm uses a clear-sky radiance composite map (CSCM) to compare the hourly observations of the IR radiance. The feasibility of the simple comparison is justified by the theoretical simulations of the fog effect on the measured radiance using a radiative transfer model. The simulation results show that the presence of fog can be detected provided the visibility is worse than 1 km and the background clear-sky radiances are accurate enough with known uncertainties. For the current study, an accurate CSCM is constructed using a modified spatial and temporal coherence method, which takes advantage of the high temporal resolution of the GMS-5 observations. The new algorithm is applied for the period of 10-12 May 1999, when heavy sea fog formed near the southwest coast of the Korean Peninsula. Comparisons of the fog/stratus index, defined as the difference between the measured and clear-sky brightness temperature, from the new algorithm to the results from other methods, such as the dual channel difference of NOAA/AVHRR and the earth albedo method, show a good agreement. The fog/stratus index also compares favorably with the ground observations of visibility and relative humidity. The general characteristics of the fog/stratus index and visibility are relatively well matched, although the relationship among the absolute values, the fog/stratus index, visibility, and relative humidity, varies with time. This variation is thought to be due to the variation of the atmospheric conditions and the characteristics of fog/stratus, which affect the derived fog/stratus index.     

Two Types of Diurnal Variations in Heavy Rainfall during July over Korea※

This study examined the characteristics of the diurnal variations of heavy rainfall (≥110 mm in 12 hours) in Korea and the related atmospheric circulation for July from 1980?2020. During the analysis period, two dominant pattens of diurnal variation of the heavy rainfall emerged: all-day heavy rainfall (AD) and morning only heavy rainfall (MO) types. For the AD-type, the heavy rainfall is caused by abundant moisture content in conjunction with active convection in the morning (0000?1200, LST; LST = UTC + 9) and the afternoon hours (1200?2400 LST). These systems are related to the enhanced moisture inflow and upward motion induced by the strengthening of the western North Pacific subtropical high and upper-tropospheric jet. For the MO-type, heavy rainfall occurs mostly in the morning hours; the associated atmospheric patterns are similar to the climatology. We find that the atmospheric pattern related to severe heavy rainfalls in 2020 corresponds to a typical AD-type and resembles the 1991 heavy-rainfall system in its overall synoptic/mesoscale circulations. The present results imply that extremely heavy rainfall episodes in Korea during the 2020 summer may occur again in the future associated with the recurring atmospheric phenomenon related to the heavy rainfall.     

Characteristics of sea surface temperature retrieved from MTSAT-1R and in-situ data

Multi-channel sea surface temperature (MCSST) data were retrieved from the Japanese geostationary satellite, MTSAT-1R, for East Asia in western North Pacific. The coefficients used to calculate the MCSST data were estimated by assuming a linear relationship between the brightness temperatures obtained from the satellite and the in-situ buoy SST. It is important to remove cloud contamination pixels to retrieve meaningful information from infrared data. Therefore, the cloud detection algorithm was improved by using a 10-day maximum or minimum composite map for infrared and visible channels. The RMSE of the MCSST in comparison with the two-year buoy SST was about 0.89oC. The error was the largest at mid-latitudes in summer. Additionally, the error between the two SSTs showed that diurnal variation had a positive bias during daytime and a negative bias during nighttime. Furthermore, in 2007, both SSTs showed seasonal and spatial diurnal variation. The magnitude of the daily variation in the MCSST was two times larger than that in the buoy SST, and this was attributed to diurnal heating with a weak surface wind speed.     

Guidance on the Choice of Threshold for Binary Forecast Modeling

This paper proposes useful guidance on the choice of threshold for binary forecasts. In weather forecast systems, the probabilistic forecast cannot be used directly when estimated too smoothly. In this case, the binary forecast, whether a meteorological event will occur or not, is preferable to the probabilistic forecast. A threshold is needed to generate a binary forecast, and the guidance in this paper encompasses the use of skill scores for the choice of threshold according to the forecast pattern. The forecast pattern consists of distribution modes of estimated probabilities, occurrence rates of observations, and variation modes. This study is performed via Monte-Carlo simulation, with 48 forecast patterns considered. Estimated probabilities are generated by random variate sampling from five distributions separately. Varying the threshold from 0 to 1, binary forecasts are generated by threshold. For the assessment of binary forecast models, a 2×2 contingency table is used and four skill scores （Heidke skill score, hit rate, true skill statistic, and threat score） are compared for each forecast pattern. As a result, guidance on the choice of skill score to find the optimal threshold is proposed.     

The Ilchulbong tuff cone, Cheju Island, South Korea

The Ilchulbong mount of Cheju Island, South Korea, is an emergent tuff cone of middle Pleistocene age formed by eruption of a vesiculating basaltic magma into shallow seawater. A sedimentological study reveals that the cone sequence can be represented by nine sedimentary facies that are grouped into four facies associations. Facies association I represents steep strata near the crater rim composed mostly of crudely and evenly bedded lapilli tuff and minor inversely graded lapilli tuff. These facies suggest fall-out from tephra finger jets and occasional grain flows, respectively. Facies association II represents flank or base-of-slope deposits composed of lenticular and hummocky beds of massive or backset-stacked deposits intercalated between crudely to thinly stratified lapilli tuffs. They suggest occasional resedimentation of tephra by debris flows and slides during the eruption. Facies association III comprises thin, gently dipping marginal strata, composed of thinly stratified lapilli tuff and tuff. This association results from pyroclastic surges and cosurge falls associated with occasional large-scale jets. Facies association IV comprises a reworked sequence of massive, inversely graded and cross-bedded (gravelly) sandstones. These facies represent post-eruptive reworking of tephra by debris and stream flows. The facies associations suggest that the Ilchulbong tuff cone grew by an alternation of vertical and lateral accumulation. The vertical buildup was accomplished by plastering of wet tephra finger jets. This resulted in oversteepening and periodic failure of the deposits, in which resedimentation contributed to the lateral growth. After the eruption ceased, the cone underwent subaerial erosion and faulting of intracrater deposits. A volcaniclastic apron accumulated with erosion of the original tuff cone; the faulting was caused by subsidence of the subvolcanic basement within the crater.     

Depositional processes of the Suwolbong tuff ring, Cheju Island (Korea)

The Suwolbong pyroclastic sequence in the western part of Cheju Island, Korea, comprises partly preserved rim beds of a Quaternary basaltic tuff ring whose vent lies about 1 km seaward of the present shoreline. The sequence consists of breccia, lapillistone, lapilli tuff and tuff. Eighteen sedimentary facies are established and organized into six lateral facies sequences (LFS) and seven vertical facies sequences (VFS). The LFS 1, 4 and 5 begin with massive lapilli tuff which transforms downcurrent into either planar-bedded (LFS 1), undulatory-bedded (LFS 4) or climbing dune-bedded (LFS 5) (lapilli) tuff units. They are representative of relatively ‘dry’ base surge whose particle concentration decreases downcurrent with a progressive increase in both tractional processes and sorting. The LFS 2 begins with disorganized and massive lapilli tuff and transforms into crudely stratified units downcurrent. It results from relatively ‘wet’ base surge in which sorting is poor due to the cohesion of damp ash. The LFS 3 comprises well-sorted lapilli tuff and stratified tuff further downcurrent, suggestive of deposition from combined fall and surge of relatively ‘dry’ hydroclastic eruption. All seven vertical facies sequences generally comprise two facies units of coarse-grained fines-depleted lapilli tuff and an overlying fine-grained tuff. These sequences are suggestive of deposition from base surge that consists of a turbulent head and a low-concentration tail. Depositional processes in the Suwolbong tuff ring were dominated by a relatively ‘dry’ base surge. The base surge comprises turbulent and high-concentration suspension near the vent whose deposits are generally unstratified due to the lack of tractional transport. As the base surge becomes diluted downcurrent through fallout of clasts and mixing of ambient air, it develops large-scale turbulent eddies and is segregated into coarse-grained bedload and overlying fine-grained suspension forming thinly stratified units. Further downcurrent, the base surge may be either cooled and deflated or pushed up into the air, depending on its temperature. The Suwolbong tuff ring comprises an overall wet-to-dry cycle with several dry-to-wet cycles in it, suggestive of overall decrease in abundance of external water and fluctuation in the rate of magma rise.