Adaptive Bias Correction for Improved Subseasonal Forecasting

Abstract

<p>Improving our ability to forecast the weather and climate is of interest to all sectors of the economy and government agencies from the local to the national level. In fact, weather forecasts 0-10 days ahead and climate forecasts seasons to decades ahead are currently used operationally in decision-making, and the accuracy and reliability of these forecasts has improved consistently in recent decades. However, many critical applications require subseasonal forecasts with lead times in between these two timescales. Subseasonal forecasting—predicting temperature and precipitation 2-6 weeks ahead—is indeed critical for effective water allocation, wildfire management, and drought and flood mitigation. Yet, accurate forecasts for the subseasonal regime are still lacking due to the chaotic nature of weather.</p><p>While short-term forecasting accuracy is largely sustained by physics-based dynamical models, these deterministic methods have limited subseasonal accuracy due to chaos. Indeed, subseasonal forecasting has long been considered a “predictability desert” due to its complex dependence on both local weather and global climate variables. Nevertheless, recent large-scale research efforts have advanced the subseasonal capabilities of operational physics-based models, while parallel efforts have demonstrated the value of machine learning and deep learning methods in improving subseasonal forecasting.</p><p>To counter the systematic errors of dynamical models at longer lead times, we introduce an <em>adaptive bias correction</em> (ABC) method that combines state-of-the-art dynamical forecasts with observations using machine learning. We evaluate our adaptive bias correction method in the contiguous U.S. over the years 2011-2020 and demonstrate consistent improvement over standard meteorological baselines, state-of-the-art learning models, and the leading subseasonal dynamical models, as measured by root mean squared error and uncentered anomaly correlation skill. When applied to the United States’ operational climate forecast system (CFSv2), ABC improves temperature forecasting skill by 20-47% and precipitation forecasting skill by 200-350%. When applied to the leading subseasonal model from the European Centre for Medium-Range Weather Forecasts (ECMWF), ABC improves temperature forecasting skill by 8-38% and precipitation forecasting skill by 40-80%.</p><p>Overall, we find that de-biasing dynamical forecasts with our learned adaptive bias correction method yields an effective and computationally inexpensive strategy for generating improved subseasonal forecasts and building the next generation of subseasonal forecasting benchmarks. To facilitate future subseasonal benchmarking and development, we release our model code through the subseasonal_toolkit Python package and our routinely updated SubseasonalClimateUSA dataset through the subseasonal_data Python package.</p>

A machine learning model that outperforms conventional global subseasonal forecast models

“…It is an important tool for various applications, such as agricultural planning, disaster preparedness and risk mitigation for extreme events like heatwaves, droughts, floods, and cold spells, as well as water resource management [2][3][4][5]. However, subseasonal forecasting is a challenging task due to its complex dependence on both local weather conditions and global climate system variables [6,7]. Furthermore, the subseasonal time scale has long been considered as a 'predictability desert' [8,9] as it suffers from the chaotic nature of the atmosphere [10] and the lack of predictive information from land beyond one month.…”

“…The FuXi-S2S 3 European Centre for Medium-Range Weather Forecasts (ECMWF) ensemble prediction system (EPS) [12] and the National Centers for Environmental Prediction (NCEP) Climate Forecast System version 2 (CFS v2) [13] are two leading operational subseasonal ensemble forecast systems, with 51 and 11 members, respectively. Nevertheless, these NWP forecasts exhibit considerable biases [7,14], particularly in predicting extreme events [15].…”

Deep learning for post-processing global probabilistic forecasts on sub-seasonal time-scales

“…One major challenge in statistical post-processing is to retain spatial and temporal relationships in the post-processed forecasts, as well as relationships between variables (Vannitsem et al, 2021). In particular, the few examples of S2S post-processing studies that we are aware of tend to separately operate on single grid-cells only, and thus are neither able to exploit spatial information in the raw forecasts, nor produce spatially homogeneous forecasts (e.g., Vigaud et al, 2017Vigaud et al, , 2019Mouatadid et al, 2021;Zhang et al, 2023). For short-to medium-range forecasts, convolutional neural network (CNN)-based architectures and model components have been used for a variety of post-processing applications (Dai and Hemri, 2021;Grönquist et al, 2021;Veldkamp et al, 2021;Chapman et al, 2022;Lerch and Polsterer, 2022;Li et al, 2022;Ben-Bouallegue et al, 2023;Hu et al, 2023).…”

An Extended Analysis of Temperature Prediction in Italy: From Sub-Seasonal to Seasonal Timescales

“…Moreover, in the sub-seasonal range, different techniques have been used to improve temperature forecasts, including machine learning [22,23], statistical models trained on dynamical models [24], and data-driven methods using random forest techniques [25]. Under this framework, in this current study, we extended a past investigation [13] where a benchmark analysis was only shown on data of one year for sub-seasonal and seasonal forecasts of temperatures by our meteorological model.…”

“…However, a decline in model performance was noticed between the 5th and 8th week of the prediction period. Moreover, in the sub-seasonal range, different techniques have been used to improve temperature forecasts, including machine learning [22,23], statistical models trained on dynamical models [24], and data-driven methods using random forest techniques [25].…”