Temperature and precipitation seasonal forecasts over the Mediterranean region: added value compared to simple forecasting methods

This study considers a set of state-of-the-art seasonal forecasting systems (ECMWF, MF, UKMO, CMCC, DWD and the corresponding multi-model ensemble) and quantifies their added value (if any) in predicting seasonal and monthly temperature and precipitation anomalies over the Mediterranean region compared to a simple forecasting method based on the ERA5 climatology (CTRL) or the persistence of the ERA5 anomaly (PERS). This analysis considers two starting dates, May 1st and November 1st and the forecasts at lead times up to 6 months for each year in the period 1993–2014. Both deterministic and probabilistic metrics are employed to derive comprehensive information on the forecast quality in terms of association, reliability/resolution, discrimination, accuracy and sharpness. We find that temperature anomalies are better reproduced than precipitation anomalies with varying spatial patterns across different forecast systems. The Multi-Model Ensemble (MME) shows the best agreement in terms of anomaly correlation with ERA5 precipitation, while PERS provides the best results in terms of anomaly correlation with ERA5 temperature. Individual forecast systems and MME outperform CTRL in terms of accuracy of tercile-based forecasts up to lead time 5 months and in terms of discrimination up to lead time 2 months. All seasonal forecast systems also outperform elementary forecasts based on persistence in terms of accuracy and sharpness.

Attributing correlation skill of dynamical global precipitation forecasts to statistical ENSO teleconnection using a set theory based approach

Climate teleconnections are essential for the verification of valuable precipitation forecasts generated by global climate models (GCMs). This paper develops a novel approach to attributing correlation skill of dynamical GCM forecasts to statistical El Niño-Southern Oscillation (ENSO) teleconnection by using the coefficient of determination (R2). Specifically, observed precipitation is respectively regressed against GCM forecasts, Niño3.4 and both of them and then the intersection operation is implemented to quantify the overlapping R2 for GCM forecasts and Niño3.4. The significance of overlapping R2 and the sign of ENSO teleconnection facilitate three cases of attribution, i.e., significantly positive anomaly correlation attributable to positive ENSO teleconnection, attributable to negative ENSO teleconnection and not attributable to ENSO teleconnection. A case study is devised for the Climate Forecast System version 2 (CFSv2) seasonal forecasts of global precipitation. For grid cells around the world, the ratio of significantly positive anomaly correlation attributable to positive (negative) ENSO teleconnection is respectively 10.8 % (11.7 %) in December-January-February (DJF), 7.1 % (7.3 %) in March-April-May (MAM), 6.3 % (7.4 %) in June-July-August (JJA) and 7.0 % (14.3 %) in September-October-November (SON). The results not only confirm the prominent contributions of ENSO teleconnection to GCM forecasts, but also present spatial plots of regions where significantly positive anomaly correlation is subject to positive ENSO teleconnection, negative ENSO teleconnection and teleconnections other than ENSO. Overall, the proposed attribution approach can serve as an effective tool to investigate the source of predictability for GCM seasonal forecasts of global precipitation.

Assessment of seasonal forecast skills of temperature and precipitation: a comparison of 5 different models over the Mediterranean region

<p>Seasonal forecasts are increasingly employed as sources of information on the expected evolution of climate in the few months ahead by various end-users. This study provides an overall assessment of the skills of the main seasonal forecast systems available in the Copernicus Climate Data Store (C3S) in representing temperature and precipitation anomalies at the monthly time scale. The focus area is the Mediterranean, a densely populated region identified as a hotspot for climate change, where seasonal forecasts could be useful to a variety of economic sectors, including water management, hydropower production, agriculture.</p><p>In this study, seasonal forecast systems issued by 5 European institutions (ECMWF, Météo-France, UKMO, DWD, CMCC), together with two different Multi-Model Ensembles (MME) derived from them, have been analysed. The added value of these forecast systems with respect to simpler forecast approaches based on climatology and persistence has been investigated.</p><p>Different deterministic (Anomaly Correlation Coefficient) and probabilistic scores (Ranked Probability Score, Continuous Ranked Probability Score and Receiver Operating Characteristic Curve) have been employed to obtain an overall assessment of the quality of the forecasts (as of Murphy, 1993 and WMO, 2018), using ERA5 dataset as a reference. We performed the analysis using 6-month forecasts starting in May and November to reproduce the following summer and the winter seasons.</p><p>In general, temperature patterns and respective skill scores are better reproduced than those regarding precipitation. The anomaly correlation coefficients for MME reach the best agreement values for each season and variable except for winter temperature. Different behaviours are found for the different skill scores; their high spatial variability suggests that smaller regions could perform better for a single variable or starting date. Seasonal forecast systems, despite some limitations, show an added value with respect to simple forecast approaches based on the climatology or persistence.</p>

Winter surface air temperature prediction over Japan using artificial neural networks

The machine learning technique, namely Artificial Neural Networks (ANN), is used to predict the surface air temperature (SAT) anomalies over Japan in the winter months of December, January and February for the period 1949/50 to 2019/20. The predictions are made for the four regions Hokkaido, North, Central and West of Japan. The inputs to the ANN model are derived from the anomaly correlation coefficients among the SAT anomalies over the regions of Japan and the global SAT and sea surface temperature anomalies. The results are validated using anomaly correlation coefficient (ACC) skill scores with the observation. It is found that the ANN predictions over Hokkaido have higher ACC skill scores compared to the ACC scores over the other three regions. The ANN predicted SAT anomalies are compared with that of ensemble mean of 8 of the North American Multi-Model Ensemble (NMME) models besides comparing them with the persistent anomalies. The ANN predictions over all the four regions have higher ACC skill scores compared to the NMME model skill scores in the common period of 1982/83 to 2018/19. The ANN predicted SAT anomalies also have higher Hit rate and lower False alarm rate compared to the NMME predicted SAT anomalies. All these indicate that the ANN model is a promising tool for predicting the winter SAT anomalies over Japan.

NMME-based Assessment of Prediction Skills of US Summertime Droughts and Pluvials: Role of Near-surface Temperature Prediction Skill

<p>This study used the North American Multi-Model Ensemble (NMME) system to understand the role of near surface temperature in the prediction skill for US climate extremes. In this study, the forecasting skill was measured by anomaly correlation coefficient (ACC) between the observed and forecasted precipitation (PREC) or 2-meter air temperature (T2m) over the contiguous United States (CONUS) during 1982–2012. The strength of the PREC-T2m coupling was measured by ACC between observed PREC and T2m or forecasted PREC and T2m over the CONUS. This study also assessed the NMME forecasting skill for the summers of 2004 (spatial anomaly correlation between PREC and T2m: 0.05), 2011 (-0.65), and 2012 (-0.60) when the PREC-T2m coupling is weaker or stronger than the 1982–2012 climatology (ACC:-0.34). This study found that most of the NMME models show stronger (negative) PREC-T2m coupling than the observed coupling, indicating that they fail to reproduce interannual variability of the observed PREC-T2m coupling. Some NMME models with skillful prediction for T2m show the skillful prediction of the precipitation anomalies and US droughts in 2011 and 2012 via strong PREC-T2m coupling despite the fact that the forecasting skill is year-dependent and model-dependent. Lastly, we explored how the forecasting skill for SSTs over north Pacific and Atlantic Oceans affects the forecasting skill for T2m and PREC over the US. The findings of this study suggest a need for the selective use of the current NMME seasonal forecasts for US droughts and pluvials.</p>

Significant spatial patterns from the GCM seasonal forecasts of global precipitation

Fully coupled global climate models (GCMs) generate a vast amount of high-dimensional forecast data of the global climate; therefore, interpreting and understanding the predictive performance is a critical issue in applying GCM forecasts. Spatial plotting is a powerful tool to identify where forecasts perform well and where forecasts are not satisfactory. Here we build upon the spatial plotting of anomaly correlation between forecast ensemble mean and observations to derive significant spatial patterns to illustrate the predictive performance. For the anomaly correlation derived from the 10 sets of forecasts archived in the North America Multi-Model Ensemble (NMME) experiment, the global and local Moran's I are calculated to associate anomaly correlations at neighbouring grid cells with one another. The global Moran's I associates anomaly correlation at the global scale and indicates that anomaly correlation at one grid cell relates significantly and positively to anomaly correlation at surrounding grid cells. The local Moran's I links anomaly correlation at one grid cell with its spatial lag and reveals clusters of grid cells with high, neutral, and low anomaly correlation. Overall, the forecasts produced by GCMs of similar settings and at the same climate centre exhibit similar clustering of anomaly correlation. In the meantime, the forecasts in NMME show complementary performances. About 80 % of grid cells across the globe fall into the cluster of high anomaly correlation under at least 1 of the 10 sets of forecasts. While anomaly correlation exhibits substantial spatial variability, the clustering approach serves as a filter of noise to identify spatial patterns and yields insights into the predictive performance of GCM seasonal forecasts of global precipitation.

Improvement in the decadal prediction skill of the North Atlantic extratropical winter circulation through increased model resolution

In this study the latest version of the MiKlip decadal hindcast system is analyzed, and the effect of an increased horizontal and vertical resolution on the prediction skill of the extratropical winter circulation is assessed. Four different metrics – the storm track, blocking, cyclone and windstorm frequencies – are analyzed in the North Atlantic and European region. The model bias and the deterministic decadal hindcast skill are evaluated in ensembles of five members in a lower-resolution version (LR, atm: T63L47, ocean: 1.5∘ L40) and a higher-resolution version (HR, atm: T127L95, ocean: 0.4∘ L40) of the MiKlip system based on the Max Planck Institute Earth System model (MPI-ESM). The skill is assessed for the lead winters 2–5 in terms of the anomaly correlation of the quantities' winter averages using initializations between 1978 and 2012. The deterministic predictions are considered skillful if the anomaly correlation is positive and statistically significant. While the LR version shows common shortcomings of lower-resolution climate models, e.g., a storm track that is too zonal and southward displaced as well as a negative bias of blocking frequencies over the eastern North Atlantic and Europe, the HR version counteracts these biases. Cyclones, i.e., their frequencies and characteristics like strength and lifetime, are particularly better represented in HR. As a result, a chain of significantly improved decadal prediction skill between all four metrics is found with the increase in the spatial resolution. While the skill of the storm track is significantly improved primarily over the main source region of synoptic activity – the North Atlantic Current – the other extratropical quantities experience a significant improvement primarily downstream thereof, i.e., in regions where the synoptic systems typically intensify. Thus, the skill of the cyclone frequencies is significantly improved over the central North Atlantic and northern Europe, the skill of the blocking frequencies is significantly improved over the Mediterranean, Scandinavia and eastern Europe, and the skill of the windstorms is significantly improved over Newfoundland and central Europe. Not only is the skill improved with the increase in resolution, but the HR system itself also exhibits significant skill over large areas of the North Atlantic and European sector for all four circulation metrics. These results are particularly promising regarding the high socioeconomic impact of European winter windstorms and blocking situations.

Significant spatial patterns from the GCM seasonal forecasts of global precipitation

Fully-coupled global climate models (GCMs) generate a vast amount of high-dimensional forecast data of the global climate; therefore, interpreting and understanding the predictive performance is a critical issue in applying GCM forecasts. Spatial plotting is a powerful tool to identify where forecasts perform well and where forecasts are not satisfactory. Here we build upon the spatial plotting of anomaly correlation between forecast ensemble mean and observations and derive significant spatial patterns to illustrate the predictive performance. For the anomaly correlation derived from the ten sets of forecasts archived in the North America Multi-Model Ensemble (NMME) experiment, the global and local Moran's I are calculated to associate anomaly correlation at neighbouring grid cells to one another. The global Moran's I indicates that at the global scale anomaly correlation at one grid cell relates significantly and positively to anomaly correlation at surrounding grid cells, while the local Moran's I reveals clusters of grid cells with high, neutral, and low anomaly correlation. Overall, the forecasts produced by GCMs of similar settings and at the same climate center exhibit similar clustering of anomaly correlation. In the meantime, the forecasts in NMME show complementary performances. About 80 % of grid cells across the globe fall into the cluster of high anomaly correlation under at least one of the ten sets of forecasts. While anomaly correlation exhibits substantial spatial variability, the clustering approach serves as a filter of noise to identify spatial patterns and yields insights into the predictive performance of GCM seasonal forecasts of global precipitation.