A statistical method for forecasting extreme daily temperatures using ECMWF 2‐m temperatures and ground station measurements

Boi, Paolo

doi:10.1017/s1350482704001318

Cited by 14 publications

(23 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…• C. This is similar to results found by other authors using non-linear methods (Marzban, 2003, Casaioli et al, 2003, Boi, 2004. Nevertheless, it must be recognised that Atmos.…”

Section: Discussionsupporting

confidence: 88%

“…All the methods tested here clearly improve the raw NWP predictions. The best MOS approach (a nine-point RF) yields a mean average error (MAE) of 1.2 • C. This is similar to results found by other authors using non-linear methods (Marzban, 2003;Casaioli et al, 2003;Boi, 2004). Nevertheless, it must be recognised that the rather high correlation (r=0.95) between the DMO and measured values is itself a good starting point for designing a MOS procedure.…”

supporting

confidence: 83%

See 1 more Smart Citation

Prediction of minimum temperatures in an alpine region by linear and non-linear post-processing of meteorological models

Eccel

Ghielmi

Granitto

et al. 2007

Nonlin. Processes Geophys.

View full text Add to dashboard Cite

Abstract. Model Output Statistics (MOS) refers to a method of post-processing the direct outputs of numerical weather prediction (NWP) models in order to reduce the biases introduced by a coarse horizontal resolution. This technique is especially useful in orographically complex regions, where large differences can be found between the NWP elevation model and the true orography. This study carries out a comparison of linear and non-linear MOS methods, aimed at the prediction of minimum temperatures in a fruit-growing region of the Italian Alps, based on the output of two different NWPs (ECMWF T511-L60 and LAMI-3). Temperature, of course, is a particularly important NWP output; among other roles it drives the local frost forecast, which is of great interest to agriculture. The mechanisms of cold air drainage, a distinctive aspect of mountain environments, are often unsatisfactorily captured by global circulation models. The simplest post-processing technique applied in this work was a correction for the mean bias, assessed at individual model grid points. We also implemented a multivariate linear regression on the output at the grid points surrounding the target area, and two non-linear models based on machine learning techniques: Neural Networks and Random Forest. We compare the performance of all these techniques on four different NWP data sets. Downscaling the temperatures clearly improved the temperature forecasts with respect to the raw NWP output, and also with respect to the basic mean bias correction. Multivariate methods generally yielded better results, but the advantage of using non-linear algorithms was small if not negligible. RF, the best performing method, was implemented on ECMWF prognostic output at 06:00 UTC over the 9 grid points surrounding the target area. Mean absolute errors in the prediction of 2 m temperature at 06:00 UTC were approximately 1.2 • C, close to the natural variability inside the area itself.

show abstract

“…• C. This is similar to results found by other authors using non-linear methods (Marzban, 2003, Casaioli et al, 2003, Boi, 2004. Nevertheless, it must be recognised that Atmos.…”

Section: Discussionsupporting

confidence: 88%

supporting

confidence: 83%

Prediction of minimum temperatures in an alpine region by linear and non-linear post-processing of meteorological models

Eccel

Ghielmi

Granitto

et al. 2007

Nonlin. Processes Geophys.

View full text Add to dashboard Cite

show abstract

“…At noon, RAMS forecasts L v E ∼100 Wm −2 higher than WRF and also higher than the observations. Despite many models still suffer from a soil that reacts too slowly on the surface forcing [e.g., Zhang and Zheng , 2004; Boi , 2004; Rantamäki et al , 2005; Edwards et al , 2011], the current WRF simulation provides a satisfactory G , except at night when G is underestimated ∼20 Wm −2 (not shown). Note however, that a small error in G at night, might result in substantial errors in the estimated near surface temperature.…”

Section: Model Results For the Windy Period (Case I)mentioning

confidence: 94%

Confronting the WRF and RAMS mesoscale models with innovative observations in the Netherlands: Evaluating the boundary layer heat budget

Steeneveld¹,

Tolk²,

Moene³

et al. 2011

J. Geophys. Res.

View full text Add to dashboard Cite

[1] The Weather Research and Forecasting Model (WRF) and the Regional Atmospheric Mesoscale Model System (RAMS) are frequently used for (regional) weather, climate and air quality studies. This paper covers an evaluation of these models for a windy and calm episode against Cabauw tower observations (Netherlands), with a special focus on the representation of the physical processes in the atmospheric boundary layer (ABL). In addition, area averaged sensible heat flux observations by scintillometry are utilized which enables evaluation of grid scale model fluxes and flux observations at the same horizontal scale. Also, novel ABL height observations by ceilometry and of the near surface longwave radiation divergence are utilized. It appears that WRF in its basic set-up shows satisfactory model results for nearly all atmospheric near surface variables compared to field observations, while RAMS needed refining of its ABL scheme. An important inconsistency was found regarding the ABL daytime heat budget: Both model versions are only able to correctly forecast the ABL thermodynamic structure when the modeled surface sensible heat flux is much larger than both the eddy-covariance and scintillometer observations indicate. In order to clarify this discrepancy, model results for each term of the heat budget equation is evaluated against field observations. Sensitivity studies and evaluation of radiative tendencies and entrainment reveal that possible errors in these variables cannot explain the overestimation of the sensible heat flux within the current model infrastructure.

show abstract

“…A detailed description of this procedure and the one year of forecast verifications at 58 locations are presented in Boi (2004); a comparison with other post-processing methods, such as the Kalman filter, MOS or Perfect Prog, confirms that this method performs well.…”

Section: Bias Correctionmentioning

confidence: 81%

“…This bias correction procedure has been applied successfully to forecast maximum and minimum temperatures at 58 stations on Sardinia (deterministic forecasts), by post-processing the TL511 ECMWF model. A detailed description of this procedure and the one year of forecast verifications at 58 locations are presented in Boi (2004); a comparison with other post-processing methods, such as the Kalman filter, MOS or Perfect Prog, confirms that this method performs well.…”

Section: Bias Correctionmentioning

confidence: 81%

Probabilistic temperature forecast by using ground station measurements and ECMWF ensemble prediction system

Boi¹

2004

Met. Apps

Self Cite

View full text Add to dashboard Cite

The ECMWF Ensemble Prediction System 2-metre temperature forecasts are affected by systematic errors due mainly to resolution inadequacies. Moreover, other errors sources are present: differences in height above sea level between the station and the corresponding grid point, boundary layer parameterisation, and description of the land surface. These errors are more marked in regions of complex orography. A recursive statistical procedure to adapt ECMWF EPS-2metre temperature fields to 58 meteorological stations on the Mediterranean island of Sardinia is presented. The correction has been made in three steps: (1) bias correction of systematic errors; (2) calibration to adapt the EPS temperature distribution to the station temperature distribution; and (3) doubling the ensemble size with the aim of taking into account the analysis errors. Two years of probabilistic forecasts of freezing are tested by Brier Score, reliability diagram, rank histogram and Brier Skill Score with respect to the climatological forecast. The score analysis shows much better performance in comparison with the climatological forecast and direct model output, for all forecast timse, even after the first step (bias correction). Further gains in skill are obtained by calibration and by doubling the ensemble size.

show abstract

A statistical method for forecasting extreme daily temperatures using ECMWF 2‐m temperatures and ground station measurements

Cited by 14 publications

References 7 publications

Prediction of minimum temperatures in an alpine region by linear and non-linear post-processing of meteorological models

Prediction of minimum temperatures in an alpine region by linear and non-linear post-processing of meteorological models

Confronting the WRF and RAMS mesoscale models with innovative observations in the Netherlands: Evaluating the boundary layer heat budget

Probabilistic temperature forecast by using ground station measurements and ECMWF ensemble prediction system

Contact Info

Product

Resources

About