S2S Model Research Articles

Clustering Anomalous Circulations and Their Implications on the Prediction of Minimum 2-m Air Temperature over Northeastern China during Winter

Abstract The ERA5 reanalysis during cold months (November–March) of 1979–2020 was used for determining four cluster centroids through the k-means for classifying regional anomalies of the daily geopotential height at 500 hPa (H500) over northeastern China. Empirical orthogonal function (EOF) was used to reduce dimensionality. Four clusters were linked to the EOF patterns with clear meteorological meanings, which are associated with the evolution of ridges and troughs over northeastern China. Those systems relate to warm and cold advections at 850 hPa. In each H500 cluster, the advection is the major contributor leading to temperature changes at 850 hPa, which significantly relates to the changes and anomalies of daily minimum air temperature at 2 m (T2min). Furthermore, the jet activities over Asia relate to more or less occurrence of specific H500 clusters in jet phases. This is because anomalous westerlies are generally in favor of positive anomalies of the vorticity tendency at 500 hPa. For the reforecasts during 2004–19 in the Chinese Meteorology Administration (CMA) S2S model, the hit rates above 50% for all the H500 clusters are within 9.5 days, which are in between those for the first two and the last two clusters. The correct prediction of H500 anomalies improves the T2min prediction up to 12 days, compared with 8 days for the incorrect one. The good prediction of the jet activities leads to a more accurate prediction of H500 anomalies. Therefore, improvement of the model prediction of jet activities and H500 anomalies will lead to better prediction of winter weather near the ground over northeastern China.

Sub-seasonal prediction skill: is the mean state a good model evaluation metric?

Evaluating forecast models encompasses assessing their ability to accurately depict observed climate states and predict future climate variables. Various evaluation methods, from computationally efficient measures like the anomaly correlation coefficient to more intricate approaches, have been formulated. While simpler methods provide limited information, climatology, due to its simplicity and immediate linkage to model performance, is a commonly utilized primary evaluation metric. In this study focusing on temperature and precipitation, we propose a novel metric based on the model’s mean state, integrating both climatology and the seasonal cycle for a more accurate assessment of the relationship between mean state performance and prediction skill on weather and sub-seasonal time scales compared to relying solely on climatology. This integrated metric reveals a robust correlation between temperature and precipitation across diverse geographical locations, with a more pronounced effect in tropical areas when considering the seasonal cycle. Additionally, we find that temperature exhibits higher prediction skill compared to precipitation. The discovered relationship serves as a potential early indicator for predicting the efficacy of Seasonal to Sub-seasonal (S2S) models and offers valuable insights for model development, emphasizing the significance of this integrated metric in enhancing S2S model performance and advancing climate prediction capabilities.

Sub-Seasonal Predictability of the Northeast China Cold Vortex in BCC and ECMWF S2S Model Forecasts for 2006–2021

Sub-Seasonal Predictability of the Northeast China Cold Vortex in BCC and ECMWF S2S Model Forecasts for 2006–2021

Probabilistic skill of statistically downscaled ECMWF S2S forecasts of maximum and minimum temperatures for weeks 1–4 over South Africa

AbstractThe probabilistic forecast skill level of statistically downscaled European Centre for Medium‐Range Weather Forecasts (ECMWF) subseasonal‐to‐seasonal (S2S) forecasts is determined in predicting maximum and minimum temperatures for weeks 1–4 lead times during 20‐year December–January–February (DJF) seasons from 2001 to 2020 over South Africa. Skilful S2S forecasts are vital in assisting decision‐makers in the development of contingency planning for any eventualities that may arise because of weather and climate phenomena. Extreme high‐ and low‐temperature events over a prolonged period can lead to hyperthermia and hypothermia, respectively, and can lead to loss of life. The results from the relative operating characteristic (ROC) and reliability diagrams indicate that the ECMWF S2S model has skill in predicting maximum temperature up to week 3 ahead, particularly over the central and eastern parts of South Africa. The ROC scores indicate that the model has skill in predicting minimum temperature up to week 4 ahead for the above‐normal category, particularly over the central and eastern parts of South Africa. Reliability diagrams indicate that the model has a tendency of overestimating the below‐normal category when predicting both maximum and minimum temperatures for weeks 1–4 lead times over South Africa. Furthermore, canonical correlation analysis (CCA) pattern analysis suggests that when there are anomalously positive and negative predicted 850‐hPa geopotential heights located over South Africa, there are anomalously hot and cold conditions during the DJF seasons over most parts of South Africa, respectively. These results suggests that statistical downscaling of model forecasts can improve forecast skill. Moreover, the results suggest that there is potential for S2S predictions in South Africa, and as such, S2S prediction system for maximum and minimum temperatures can be developed.

Hybrid Deep Learning and S2S Model for Improved Sub-Seasonal Surface and Root-Zone Soil Moisture Forecasting

Surface soil moisture (SSM) and root-zone soil moisture (RZSM) are key hydrological variables for the agricultural water cycle and vegetation growth. Accurate SSM and RZSM forecasting at sub-seasonal scales would be valuable for agricultural water management and preparations. Currently, weather model-based soil moisture predictions are subject to large uncertainties due to inaccurate initial conditions and empirical parameterization schemes, while the data-driven machine learning methods have limitations in modeling long-term temporal dependences of SSM and RZSM because of the lack of considerations in the soil water process. Thus, here, we innovatively integrate the model-based soil moisture predictions from a sub-seasonal-to-seasonal (S2S) model into a data-driven stacked deep learning model to construct a hybrid SSM and RZSM forecasting framework. The hybrid forecasting model is evaluated over the Yangtze River Basin and parts of Europe from 1- to 46-day lead times and is compared with four baseline methods, including the support vector regression (SVR), random forest (RF), convolutional long short-term memory (ConvLSTM) and the S2S model. The results indicate substantial skill improvements in the hybrid model relative to baseline models over the two study areas spatiotemporally, in terms of the correlation coefficient, unbiased root mean square error (ubRMSE) and RMSE. The hybrid forecasting model benefits from the long-lead predictive skill from S2S and retains the advantages of data-driven soil moisture memory modeling at short-lead scales, which account for the superiority of hybrid forecasting. Overall, the developed hybrid model is promising for improved sub-seasonal SSM and RZSM forecasting over global and local areas.

Novel evaluation of sub‐seasonal precipitation ensemble forecasts for identifying high‐impact weather events associated with the Indian monsoon

AbstractWe assess the skill of the fully coupled lagged ensemble forecasts from GloSea5‐GC2, for the sub‐seasonal to seasonal (S2S) timescale up to 4 weeks, with the aim of understanding how these forecasts might be used in a Ready‐Set‐Go style decision‐making framework. Integrated Multi‐satellite Retrievals for Global Precipitation Measurement (IMERG‐GPM) are used to seamlessly verify these ensemble forecasts up to monthly timescales whereby forecast and observed precipitation fields are summed over a sequence of increasing lead time accumulation windows (LTAWs), from 1d1d up to 2w2w. Results show that model biases grow with increasing LTAW and with ensemble member age. The S2S model exhibits both wet and dry biases across different parts of the Indian domain. The S2S model error grows from around 10 mm for a 24‐h accumulation to 50 mm for the 2‐week LTAWs. The actual skill and potential skill of the ensemble forecasts reveal that the potential skill is not always greater than actual skill everywhere. The sensitivity to the number and age of ensemble members was tested, with potential skill showing more impact from the exclusion of older members at all LTAWs. We conclude that the older lagged members do not necessarily add value by being included in the short to medium range or even for the extended range forecasts. GloSea5‐GC2 shows some skill in detecting large accumulations, which are not always tied to locations where they are climatologically frequent.

An Application of Deep Learning Technique to Improve Subseasonal to Seasonal Rainfall Forecast over Java Island, Indonesia

Subseasonal to seasonal (S2S) rainfall forecast can benefit several sectors, such as water resources, hazard management, and agriculture. However, the forecast remains challenging due to its lack of skill. This study applies Convolutional AutoEncoders (ConvAE), a deep learning technique, to improve the quality of the S2S rainfall forecast. Seven S2S model output incorporated with Subseasonal Experiments Projects (SubX), including CCSM4, CFSv2, FIMr1p1, GEFS, GEOS_v2p1, GEPS6, and NESM, are corrected using the ConvAE approach. We combine 407 ground observations and the CHIRPS dataset using regression kriging methods producing gridded daily precipitation data with 0.05° spatial resolution. We utilize this dataset as a label to train ConvAE models and to perform bias corrections to all members of the SubX forecasts data. The results show that ConvAE is able to increase the quality of weekly S2S rainfall forecasts over Java, Indonesia. The Correlation Coefficient for 1 – 4 weeks lead time are improved from: 0.76, 0.715, 0.692 and 0.722 towards 0.809, 0.751, 0.719 and 0.74, respectively. Furthermore, the average CRPSS improves between 20 – 30% for all lead times.

The Performance of S2S Models on Predicting the 21.7 Extreme Rainfall Event in Henan China

Extreme rainfall may cause meteorological disasters and has tremendous impact on societies and economics. Assessing the capability of current dynamic models for rainfall prediction, especially extreme rainfall event prediction, at sub-seasonal to seasonal (S2S) scale and diagnosing the probable reasons are quite important topics in the current climate study field. This study analyzes the formation mechanisms of the extreme rainfall event during 18–22 July 2021 in Henan Province and introduces the Tanimoto Coefficient (TC) to evaluate the prediction performance of S2S models. The results show that confrontation between low-latitude typhoon “In-Fa” and subtropical highs leads to sufficient water vapor transporting to Henan, and that remarkable upward air motion causes strong convergence of water vapor, thereby providing atmospheric conditions for this extreme rainfall event. Furthermore, five S2S models showed limited capability in predicting this extreme rainfall event 20 days in advance with the TCs of four models being below 0.1. Models could capture this event signal 6 days ahead with most TCs above 0.2. The performances of model prediction for this extreme rainfall event were closely related to the fact that the water vapor convergence, vertical movements, relative vorticity, and geopotential height predicted by the NCEP model 20 days ahead were close to the actual situation, in contrast to the other four models 6 days in advance. This study implies that S2S model predictions for this extreme rainfall event show obvious differences, and the application of S2S models in the prediction of extreme events needs to fully consider their prediction uncertainties. The capability of the models to properly reproduce local water vapor convergence and vertical motions is also shown to be crucial for correctly simulating the extreme event, which might provide some hints for the further amelioration of models.

Numerical simulation and conceptual design of an MW-grade space heat pipe radiator

In recent years, nuclear reactor systems have been widely used in the space. The design of a radiator is of great importance as its heat dissipation performance directly determines the upper limit of the overall output power of the space nuclear propulsion system. In the present paper, a space radiator is divided into heat dissipation units, among which the heat pipes and fins are numerically modeled and analyzed. Kalium was selected as the working medium in the heat pipe, and the size of the heat pipe was designed based on the limitations, with the length of the evaporation section and the condensation section as 150 mm and 250 mm, respectively. VOF- Lee model was applied to simulate the phase-change heat transfer process inside the heat pipe, by setting the parameters of the hot-end temperature, the cold-end heat flux, etc. In addition, S2S model was used to examine the influences of thickness and width of the fin on the heat dissipation performance, so as to determine the optimal fin size. Finally, the heat pipes and fins were assembled into a whole radiative cooling system, with a mass-area ratio below 7 kg/m2 and a total heat dissipation power at 3.58 MW.

Superensemble Approach to S2S Model for Predicting Surface Air Temperature in Summer in East Asia from 2016 to 2020

A multimodel superensemble technique was applied to improve the accuracy of a model predicting the subseasonal surface air temperatures in East Asia. Superensemble forecast data based on seven subseasonal-to-seasonal (S2S) forecast models were produced for the initial times from May to August in 2016–2020 and forecast days in weeks 1–4. Superensemble techniques based on the root-mean-square error ensemble mean (SUPR) and the linear regression coefficient ensemble mean (SUPL) were evaluated and compared with the forecast bias-removed ensemble mean (ENS) data. The magnitudes of the forecasting root-mean-square errors (RMSE) of the SUPR and SUPL methods were 4.2% and 13.8% lower, respectively, than that of the ENS method, and the most significant effect was in 2020. The RMSE improvements of SUPR and SUPL were high when the initial date was in May, June, and August, and the forecast date was within weeks 1–4. Applying multimodel superensemble methods effectively reduced the forecasting errors of the S2S models. If weight coefficients based on the characteristics of the models are appropriately applied, the forecasting performance can be significantly improved.

Design and Thermal Analysis of Temperature Control Experiment Box for Final Optic Assembly

The Final Optic Assembly (FOA) is one of the key components of the inertial confinement nuclear fusion device. The temperature change in the FOA is related to the laser frequency conversion efficiency of the frequency doubling crystal. In order to research the temperature control of FOA, this paper considers the internal atmosphere and internal structure of FOA, designed and fabricated a vacuum temperature control experiment box, and set up heat sources inside the box to simulate the internal heating of the FOA. This article uses different radiation heat transfer models to analysis the temperature field of the vacuum experiment box. Compare the thermal simulation results of the S2S model and the DO model, and compare them with the actual temperature of the vacuum experiment box under the same boundary conditions. The results show that the S2S model fits well with the experimental results and is more suitable for thermal analysis of FOA, which is of great significance to the follow-up FOA temperature control research.

Preliminary assessment for sub-seasonal to seasonal precipitation model on four specific conditions over western Indonesia

Preliminary assessment of sub-seasonal to seasonal reforecast precipitation model (S2S) was conducted to analyze the model's performance over western Indonesia on four conditions. The ECMWF S2S model was compared to quality controlled daily precipitation data from 645 observation points over the region. The control and perturbed model for the first three time steps and the last three were utilized to obtain the best performance comparison. The analysis was conducted in monthly period, MJO events, NCS events, and when both of them were active during period of November-December-January-February (NDJF) from 1998 to 2017. The results show that the first three time steps perform much better than the last one with a slightly higher correlation coefficient from the control model with relatively similar RMSE in Natuna Islands. Spatial analysis indicates that both of the control and perturbed models can catch the variation brought by the wet season in the NDJF period, by the MJO, show a hint of NCS effect, and the combination when MJO and NCS were active at the same time. The models can depict the precipitation pattern pretty well with the tendency to overestimate low rainfall intensity and underestimate the high one. The models relatively overestimate the intensity in Sumatra for the whole period. Meanwhile, consistently good spatial performance is shown by the models over Java, both in NDJF periods or MJO events.

Relationships between subseasonal‐to‐seasonal predictability and spatial scales in tropical rainfall

AbstractSubseasonal to seasonal (S2S) tropical rainfall predictability is assessed both from an analysis of the spatial scales of observed rainfall variability data, as well as from an S2S model reforecast skill. Observed spatial scales are quantified from gridded observed daily rainfall data, in terms of the size (area) of daily contiguous wet grid‐points (referred to as ‘wet patches’), as well as from the spatial autocorrelations of 7–91‐day running averages of rainfall. Model S2S reforecast skill is measured using the anomaly correlation coefficient between observed and simulated weekly and monthly rainfall from an 11‐member ensemble of European Centre for Medium‐Range Weather Forecasts (ECMWF) reforecasts (1998–2017). Both measures of S2S predictability are found to be systematically lower over land than sea, usually peaking at the start or end of the rainy season and decreasing during the core. Small spatial scales and low skill over equatorial/northern tropical Africa and western Amazonia coincide with small daily rainfall patch size and strong synoptic‐scale (≤7 days) variability there. Over most of South and SE Asia, daily wet patches are larger and strongly modulated by intraseasonal oscillations, boosting S2S rainfall predictability, while this is offset by large daily mean rainfall intensities that increase the noise. In consequence, S2S rainfall skill here generally remains low. Several land areas (as around Maritime Continent from the Philippines to Northern Australia, Eastern and Southern Africa, Eastern South America) exhibit larger spatial scales and skill, especially where the relative amplitude of SST‐forced interannual variations is strong. Most of the Maritime Continent illustrates such behaviour, but even here, the time‐averaged spatial scales and skill drop during the core of the rainy season.

Record‐Breaking Meiyu Rainfall Around the Yangtze River in 2020 Regulated by the Subseasonal Phase Transition of the North Atlantic Oscillation

AbstractIn 2020, the Yangtze River (YR) suffered a long‐persisting Meiyu season. The accumulated rainfall broke its record since 1961 and caused severe flooding and death in China. Our results show the sequential warm and cold Meiyu front regulated by the North Atlantic Oscillation (NAO) was responsible for this unexpected extreme Meiyu event. From 11 to 25 June with the positive NAO, the interaction between the South Asian High (SAH) and the western Pacific subtropical high maintained a warm front to strengthen the rainband north of the YR. Afterward, the coupling between SAH and midlatitude Mongolian Cyclone induced a cold front, which retreated the rainband to the south of YR from 30 June to 13 July with the negative NAO. Although the ECMWF S2S model successfully predicted the warm‐front‐related Meiyu rainband, it failed to forecast the Meiyu rainband in the cold‐front period, suggesting a great challenge of S2S forecasting on Meiyu rainfall.

Experimental and Numerical Investigation of Car Cabin Temperature under Solar Load Condition using Phase Change Materials

Cars kept in open have the tendency to attain temperatures far higher than the one prevailing in the surroundings. This condition is especially severe during summer months, and can cause a lot of discomfort to passengers, who re-occupy the car after parking it in the open for a while. Employing a Phase Change Material (PCM) can be a possible solution to control the car cabin temperature to a certain extent. This paper is focused on experimental and numerical study on employing a suitable PCM to control the car cabin temperature. Four different locations inside the car cabin namely roof, head, bottom and feet were selected for monitoring the temperature using temperature sensors. The effect of 90 minutes of exposure to the solar load is simulated by employing the Surface to Surface (S2S) radiation model available in ANSYS FLUENT 18.2. Experimental and numerical results conveyed the reduction in cabin average air temperature by ∼ 8°C after Placing the phase change material inside the four wheeler cabin. Comparison of Numerical and experimental results was carried out and the absolute average deviation in temperature predicted by S2S model from the experimental data is 9.93 %.

Performance Evaluation of a Subseasonal to Seasonal Model in Predicting Rainfall Onset Over West Africa

AbstractThe present study evaluates the performance of a Subseasonal to Seasonal (S2S) model (called The China Meteorological Administration) in simulating rainfall onset dates (RODs) over West Africa. Using two ROD definitions, we compared the model's ROD at six lead‐time forecasts (10, 20, 30, 40, 50, and 60 days) with the observed ROD from satellite data sets, statistically quantified the model's capability to reproduce the interannual variability of RODs over the climatological zones in the subcontinent, and investigated how well the model links RODs with the dynamics of monsoon system over the subregion. The results show that the mean RODs follow a latitudinal progression and the dates increases northward from the coast. The performance of the S2S model in reproducing RODs largely depends on the definition used. For instance, the ability of the model to replicate the observed spatial pattern of RODs and all the essential features over the three zones in West Africa is stronger with DEF1 than DEF2. Regardless of the ROD definition used, the 20‐ to 60‐day forecasts produced more realistic simulation than the 10‐day forecast. We also found that the performance of the model in reproducing the interannual variability of RODs depends on the zones (Guinea, Savanna, and Sahel). Moreover, the model could reproduce the three main phases of the West African Monsoon (the onset, the peak, and the southward retreat of rainfall) and other dynamics. The results of the study have application in improving S2S forecasting over West Africa.

Predictability of the Strong Ural blocking Event in January 2012 in the Subseasonal to Seasonal Models of Europe and Canada

The occurrence of a Ural blocking (UB) event is an important precursor of severe cold air outbreaks in Siberia and East Asia, and thus is significant to accurately predict UB events. Using subseasonal to seasonal (S2S) models of the European Centre for Medium-Range Weather Forecasts (ECMWF) and the Environment and Climate Change Canada (ECCC), we evaluated the predictability of a persistent UB event on 18 to 26 January 2012. Results showed that the ECCC model was superior to the ECMWF model in predicting the development stage of the UB event ten days in advance, while the ECMWF model had better predictions than the ECCC model for more than ten days in advance and the decaying stage of the UB event. By comparing the dynamic and thermodynamic evolution of the UB event predicted by the two models via the geostrophic vorticity tendency equation and temperature tendency equation, we found that the ECCC model better predicted the vertical vorticity advection, ageostrophic vorticity tendency, the tilting effect, horizontal temperature advection, and adiabatic heating during the development stage, whereas the ECMWF model better predicted the three dynamic and the two thermodynamic terms during the decaying stage. In addition, during both the development and decaying stages, the two models were good (bad) at predicting the vortex stretching term (horizontal vorticity advection), with the PCC between both the predictions and the observations larger (smaller) than +0.70 (+0.10) Thus, we suggest that the prediction of the persistent UB event in the S2S model might be improved by the better prediction of the horizontal vorticity advection.

Deep Learning for Load Forecasting: Sequence to Sequence Recurrent Neural Networks With Attention

The biggest contributor to global warming is energy production and use. Moreover, a push for electrical vehicle and other economic developments are expected to further increase energy use. To combat these challenges, electrical load forecasting is essential as it supports energy production planning and scheduling, assists with budgeting, and helps identify saving opportunities. Machine learning approaches commonly used for energy forecasting such as feedforward neural networks and support vector regression encounter challenges with capturing time dependencies. Consequently, this paper proposes Sequence to Sequence Recurrent Neural Network (S2S RNN) with Attention for electrical load forecasting. The S2S architecture from language translation is adapted for load forecasting and a corresponding sample generation approach is designed. RNN enables capturing time dependencies present in the load data and S2S model further improves time modeling by combining two RNNs: encoder and decoder. The attention mechanism alleviates the burden of connecting encoder and decoder. The experiments evaluated attention mechanisms with different RNN cells (vanilla, LSTM, and GRU) and with varied time horizons. Results show that S2S with Bahdanau attention outperforms other models. Accuracy decreases as forecasting horizon increases; however, longer input sequences do not always increase accuracy.

Experimental and Numerical Investigation of Air Temperature Distribution inside a Car under Solar Load Condition

In this work both experimental and numerical analysis are carried out to investigate the effect of solar radiation on the cabin air temperature of Maruti Suzuki Celerio car parked for 90 min under solar load condition. The experimental and numerical analysis encompasses on temperature increment of air at various locations inside the vehicle cabin. The effect of 90 min exposure to the environment is simulated with the help of Discrete Ordinance (DO) and Surface to Surface (S2S) radiation models using ANSYS FLUENT 18.2. Moreover, the impacts of using different turbulence model on the accuracy of the simulation results and the comparison between steady state and transient state simulation results have also been studied. The results of the simulation are compared with the experimental data to contrast the model. The absolute average deviation in temperature predicted by DO and S2S model from the experimental data are 10.07 and 10.01%, respectively. In this work both experimental and numerical analysis are carried out to investigate the effect of solar radiation on the cabin air temperature of Maruti Suzuki Celerio car parked for 90 min under solar load condition. The experimental and numerical analysis encompasses on temperature increment of air at various locations inside the vehicle cabin. The effect of 90 min exposure to the environment is simulated with the help of Discrete Ordinance (DO) and Surface to Surface (S2S) radiation models using ANSYS FLUENT 18.2. Moreover, the impacts of using different turbulence model on the accuracy of the simulation results and the comparison between steady state and transient state simulation results have also been studied. The results of the simulation are compared with the experimental data to contrast the model. The absolute average deviation in temperature predicted by DO and S2S model from the experimental data are 10.07 and 10.01%, respectively.

Experimental and unsteady CFD analyses of the heating process of large size forgings in a gas-fired furnace

Comprehensive unsteady analysis of large size forged blocks heating characteristics in a gas-fired heat treatment furnace was carried out employing experimentally measured temperatures and computational fluid dynamics (CFD) simulations. Heat and fluid flow interactions, consisting of turbulence, radiation and combustion, were simultaneously considered using k-ε, DO and EDM models in a three dimensional CFD model of the furnace, respectively. Applicability of the S2S radiation model was also evaluated to quantify the effect of participating medium and radiation view factor in the radiation heat transfer. Temperature measurements at several locations of an instrumented large size forged block and the furnace chamber were performed. A maximum deviation of about 7% was obtained between the model predictions and the experimental measurements. Results showed that despite the unloaded furnace had uniform temperature distribution, after loading, product surfaces experienced different heating rates, resulting in temperature differences up to 200 K. Findings were correlated with furnace geometry, vortical structures and flow circulations around the workpiece. Besides, S2S model demonstrating reliable results highlighted the importance of radiation view factor optimization in this application. The study could directly be employed for optimization of the heat treatment process of large size steel components.