How Suitable is Quantile Mapping For Postprocessing GCM Precipitation Forecasts?

Zhao, Tongtiegang; Bennett, James; Wang, QJ; Schepen, Andrew; Wood, A. W.; Robertson, D.E.; Ramos, Maria‐Helena

doi:10.1175/jcli-d-16-0652.1

Cited by 153 publications

(119 citation statements)

References 56 publications

Supporting

Mentioning

104

Contrasting

Unclassified

Order By: Relevance

“…Since RPP-S is built 5 upon the Bayesian joint probability (BJP) modelling approach, it explicitly models the correlation between the forecasts and observations, and thus takes into account model skill in the calibration. Our results add to the findings of Zhao et al (2017) who studied the post-processing of monthly rainfall forecasts from POAMA (Australia's GCM preceding ACCESS-S). While Zhao et al (2017) demonstrated that QM is very effective for bias correction, they did not consider accumulated totals.…”

Section: Discussionsupporting

confidence: 75%

“…Our results add to the findings of Zhao et al (2017) who studied the post-processing of monthly rainfall forecasts from POAMA (Australia's GCM preceding ACCESS-S). While Zhao et al (2017) demonstrated that QM is very effective for bias correction, they did not consider accumulated totals. In our study, we find that the RPP-S forecasts are less biased and more reliable than QM forecasts for accumulated totals.…”

Section: Discussionsupporting

confidence: 75%

“…The QM forecasts are supplied by the Bureau of Meteorology alongside the raw forecasts. QM matches the statistical distribution of past forecasts to the distribution of observations to reduce errors in the forecast mean and improve forecast 15 spread (Crochemore et al, 2016;Zhao et al, 2017). A post-processed forecast value is obtained by first working out the quantile fraction (cumulative probability) of the new forecast using the CDF of past forecasts, then inverting the quantile fraction using the CDF of observations.…”

Section: Comparison With Forecasts Post-processed By Using Quantile Mmentioning

confidence: 99%

“…For example, Crochemore et al (2016) recently evaluated linear scaling and quantile mapping for post-processing ECMWF System4 rainfall forecasts in France. Quantile mapping adjusts forecast means and ensemble spread but it is not a full calibration method because it does not account for the correlation between forecasts and observations (Zhao et al, 2017). It is useful for bias 10 correction of climate change projections where a full statistical calibration is inappropriate (Teutschbein and Seibert, 2012).…”

mentioning

confidence: 99%

See 3 more Smart Citations

A new method for post-processing daily sub-seasonal to seasonal rainfall forecasts from GCMs and evaluation for 12 Australian catchments

Schepen

Zhao

Wang

2017

Preprint

Self Cite

View full text Add to dashboard Cite

Abstract.Rainfall forecasts are an integral part of hydrological forecasting systems at sub-seasonal to seasonal time scales. In seasonal forecasting, global climate models (GCMs) are now the go-to source for rainfall forecasts. However, for hydrological 10 applications, GCM forecasts are often biased and unreliable in uncertainty spread, and therefore calibration is required before use. There are sophisticated statistical techniques for calibrating monthly and seasonal aggregations of the forecasts. However, calibration of seasonal forecasts at the daily time step typically uses very simple statistical methods or climate analogue methods. These methods generally lack the sophistication to achieve unbiased, reliable and coherent forecasts of daily amounts and seasonal accumulated totals. In this study, we propose and evaluate a Rainfall Post-Processing method for Seasonal 15 forecasts (RPP-S) based on the Bayesian joint probability approach for calibrating daily forecasts and the Schaake Shuffle approach for connecting the daily ensemble members of different lead times. We apply the method to post-process ACCESS-S forecasts for 12 perennial and ephemeral catchments across Australia and for 12 initialisation dates. RPP-S significantly reduces bias in raw forecasts and improves both skill and reliability. RPP-S forecasts are more skilful and reliable than forecasts derived from ACCESS-S forecasts that have been post-processed using quantile mapping, especially for monthly and seasonal 20 accumulations. Several opportunities to improve the robustness and skill of RPP-S are identified. The new RPP-S postprocessed forecasts will be used in ensemble sub-seasonal to seasonal streamflow applications.

show abstract

Section: Discussionsupporting

confidence: 75%

Section: Discussionsupporting

confidence: 75%

Section: Comparison With Forecasts Post-processed By Using Quantile Mmentioning

confidence: 99%

mentioning

confidence: 99%

See 2 more Smart Citations

A new method for post-processing daily sub-seasonal to seasonal rainfall forecasts from GCMs and evaluation for 12 Australian catchments

Schepen

Zhao

Wang

2017

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…There have been some criticisms raised lately regarding the 10 applicability of quantile mapping for bias adjusting seasonal data (e.g. Zhao et al, 2017). They point out that although quantile mapping approaches are effective at bias correction they cannot ensure reliability in forecast ensemble spread or guarantee coherence.…”

Section: Dynamical Modelling Ensemble (De)mentioning

confidence: 99%

The development and evaluation of a hydrological seasonal forecast system prototype for predicting spring flood volumes in Swedish rivers

Foster¹,

Uvo²,

Olsson³

2017

Preprint

View full text Add to dashboard Cite

Abstract. Hydropower makes up nearly half of Sweden's electrical energy production. However, the distribution of the water resources is not aligned with demand, most of the inflows to the reservoirs occur during the spring flood period. This means that carefully planned reservoir management is required to help redistribute the water resources to ensure optimal 10 production and accurate forecasts of the spring flood volume (SFV) is essential for this. The current operational SFV forecasts use a historical ensemble approach where the HBV model is forced with historical observations of precipitation and temperature. In this work we develop and test a multi-model prototype, building on previous work, and evaluate its ability to forecast the SFV in 84 sub-basins in northern Sweden. The testing is done using cross-validated hindcasts for the period 1981-2015 and the results are evaluated against both climatology and the current system to determine skill. Both the 15 considered multi-model methods considered showed skill over the reference forecasts, however the version that combined the historical modelling chain, dynamical modelling chain, and statistical modelling chain performed better than the other and was chosen for the prototype. The prototype was able to outperform the current operational system on average 65% of the time and reduce the error in the SFV by ~6% across all subbasins and forecast dates.

show abstract

Relating anomaly correlation to lead time: Clustering analysis of CFSv2 forecasts of summer precipitation in China

et al. 2017

Self Cite

View full text Add to dashboard Cite

Global climate model (GCM) forecasts are an integral part of long‐range hydroclimatic forecasting. We propose to use clustering to explore anomaly correlation, which indicates the performance of raw GCM forecasts, in the three‐dimensional space of latitude, longitude, and initialization time. Focusing on a certain period of the year, correlations for forecasts initialized at different preceding periods form a vector. The vectors of anomaly correlation across different GCM grid cells are clustered to reveal how GCM forecasts perform as time progresses. Through the case study of Climate Forecast System Version 2 (CFSv2) forecasts of summer precipitation in China, we observe that the correlation at a certain cell oscillates with lead time and can become negative. The use of clustering reveals two meaningful patterns that characterize the relationship between anomaly correlation and lead time. For some grid cells in Central and Southwest China, CFSv2 forecasts exhibit positive correlations with observations and they tend to improve as time progresses. This result suggests that CFSv2 forecasts tend to capture the summer precipitation induced by the East Asian monsoon and the South Asian monsoon. It also indicates that CFSv2 forecasts can potentially be applied to improving hydrological forecasts in these regions. For some other cells, the correlations are generally close to zero at different lead times. This outcome implies that CFSv2 forecasts still have plenty of room for further improvement. The robustness of the patterns has been tested using both hierarchical clustering and k‐means clustering and examined with the Silhouette score.

show abstract

How Suitable is Quantile Mapping For Postprocessing GCM Precipitation Forecasts?

Cited by 153 publications

References 56 publications

A new method for post-processing daily sub-seasonal to seasonal rainfall forecasts from GCMs and evaluation for 12 Australian catchments

A new method for post-processing daily sub-seasonal to seasonal rainfall forecasts from GCMs and evaluation for 12 Australian catchments

The development and evaluation of a hydrological seasonal forecast system prototype for predicting spring flood volumes in Swedish rivers

Relating anomaly correlation to lead time: Clustering analysis of CFSv2 forecasts of summer precipitation in China

Contact Info

Product

Resources

About