| 15 Apr 2026
SEAS5-BCSD: A bias-corrected and downscaled global seasonal forecast reference dataset for 1981–2024
Abstract. Seasonal forecasts offer valuable information on upcoming conditions for the water, energy, and agricultural sectors. However, applications of raw data from global seasonal forecasts are limited, as they can show substantial biases and temporal drifts. In this study, we present a bias-corrected and downscaled global seasonal forecast reference dataset for precipitation and 2-m temperature for 1981 till 2024, provided at monthly resolution. We achieve this with the Bias Correction and Spatial Disaggregation (BCSD) method, combining ECMWF SEAS5 seasonal forecasts with ERA5 reanalysis data. The resulting post-processed product is a spatially refined and improved dataset for a wide range of seasonal applications in the water, energy and agricultural sectors. Unlike existing products, the dataset provides bias-corrected forecasts for all SEAS5 ensemble members over the full hindcast period (1981–2016) and even beyond (till 2024). The dataset spans all global land areas at 0.25° spatial resolution with a forecast lead time of up to seven months. It comprises 25 ensemble members for the period 1981–2016 and 51 ensemble members for 2017–2024. To assess probabilistic forecast quality, we conduct a comprehensive performance evaluation, using the Brier Skill Score (BSS) and the Continuous Ranked Probability Skill Score (CRPSS). The BCSD-corrected temperature forecasts outperform climatology across nearly all regions and lead times, with highest skill in flat and warm regions. Precipitation skill is highest in the tropics and humid regions. Semi-arid areas show solid skill during the rainy season but reduced performance in dry months. This skillful global seasonal forecast reference dataset can now be explored by the community for subsequent forecast evaluation, drought prediction studies, and water resource management applications. The BCSD-corrected seasonal forecast dataset is publicly available as NetCDF data under a Creative Commons Attribution 4.0 International License (CC BY 4.0) at the World Data Center for Climate (WDCC; DOI: 10.26050/WDCC/SEAS5-BCSD, Weber et al., 2026).
Competing interests: One of the co-authors, Christof Lorenz, is a member of the editorial board of Earth System Science Data.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.
This manuscript describes a global bias-corrected seasonal forecast dataset based on ECMWF SEAS5 and ERA5. While the dataset concept has merit, the manuscript suffers from critical methodological gaps, circular evaluation design, and insufficient rigor in several areas. I have several substantive concerns and recommendations that should be addressed.
1) A fundamental concern is that ERA5 is both the correction target and the verification benchmark. The BCSD maps SEAS5 quantiles onto ERA5 by design. Evaluating skill against the same ERA5 guarantees apparent improvement. This is methodologically trivial, not evidence of forecast skill. The authors must verify against independent observations (GPCC, CRU, station networks), not simply against ERA5. And the brief discussion in Appendix A1 does not resolve this fundamental problem.
2) Line 404 references "six variables" processed operationally, but only two variables (tp and t2m) are documented in this manuscript. What are the other four? The authors must provide systematic analysis for all six variables in the paper. Documenting only two variables is very far from sufficient for a dataset paper.
3) The authors mush provide systematic comparison with existing bias-corrected seasonal products (MSWX, C3S products, etc.). A comparison table (resolution, variables, ensemble size, methods, skill) is necessary to substantiate claims of novelty and complementarity. The authors should also conduct a direct skill comparison at least for selected regions and lead times between SEAS5-BCSD and one or two competing products using the same verification framework. The authors must demonstrate the dataset's added value over what is already publicly available.
4) The authors note the ensemble size inhomogeneity (25 members for 1981-2016 vs. 51 for 2017-2024) and state that "including the operational period in the statistical analysis would therefore introduce an inhomogeneity" (Lines 68-69). Yet the dataset spans both periods. How is the bias correction applied to the 2017-2024 operational forecasts? Are CDFs from 1981-2016 used to correct 2017-2024 data without updating?
5) The SD component is described in only two sentences (Lines 101-103). The authors must provide substantially more technical detail. What exactly are these "relative differences"? How is the "coarse-grid ERA5 climatology" constructed (long-term mean, monthly climatology, or daily climatology)? What interpolation method is used (bilinear, conservative)? How are the relative differences applied back to obtain the downscaled field? For precipitation, are multiplicative factors used?
6) The authors state that precipitation extremes are handled via "linear extrapolation/scaling" and temperature via an "additive delta approach" (Lines 124-125). These are described in one sentence each with no equations or further detail. The authors must add more technical details.
7) None of the skill scores (CRPSS, BSS) include confidence intervals or significance tests. The authors must provide bootstrapped confidence intervals on the global/regional mean skill scores, or apply a field significance test for the spatial maps. Without this, it is impossible to determine whether the reported positive skill at longer lead times is statistically significant.
8) The 1981-2016 period contains a strong warming trend and regional precipitation shifts. Empirical quantile mapping assumes stationary bias. The authors acknowledge this only tangentially (Line 496) and argue it away by assertion. The authors must show that the bias structure is stable over time, or acknowledge and quantify the resulting uncertainty.
9) The bias correction method employed here called pixel-wise Empirical Quantile Mapping (EQM) differs fundamentally from the calibration approaches widely adopted in the initialized prediction community (e.g., Doblas-Reyes et al., 2013, Nat. Commun., https://doi.org/10.1038/ncomms2704). Can the authors discuss the advantages and disadvantages of both approaches, and clarify why EQM is more suitable than mean-variance calibration for the intended applications of this dataset?
10) The current evaluation is limited to statistical skill metrics. Can the authors provide application case study (e.g., historical drought or extreme precipitation event) to demonstrate the dataset's practical utility?
11) Line 91: "a more complex seasonal climate (reference needed) and..." what's the reference?
12) Line 78: "ERA5 has its' limitations" grammatical error; should be "its limitations"
13) Redundant phrasing at Line 212: "while areas with lower skill..."