## Probabilistic Quantitative Precipitation Forecasting using a Two-Stage Spatial Model (2008)

### Cached

### Download Links

Citations: | 1 - 0 self |

### BibTeX

@MISC{Berrocal08probabilisticquantitative,

author = {Veronica J. Berrocal and Adrian E. Raftery and Tilmann Gneiting},

title = {Probabilistic Quantitative Precipitation Forecasting using a Two-Stage Spatial Model},

year = {2008}

}

### OpenURL

### Abstract

Multidisciplinary University Research Initiative (MURI) program administered by the Office of

### Citations

246 |
Verification of forecasts expressed in terms of probability
- Brier
- 1950
(Show Context)
Citation Context ... probability score (11) is a proper scoring rule for a probabilistic forecast of a scalar quantity; for a deterministic forecast, it reduces to the absolute error. The Brier score or quadratic score (=-=Brier 1950-=-) for a probability forecast of a binary event is defined as bs(f, o) = (f − o) 2 , where f is the forecast probability for the event and o equals 1 if the event occurs and 0 otherwise. As the represe... |

233 |
Evaluating density forecasts with applications to financial risk management
- Diebold, Gunther, et al.
- 1998
(Show Context)
Citation Context ... of the predictive distributions, we use verification rank histograms (Anderson 1996; Talagrand et al. 1997; Hamill and Colucci 1997; Hamill 2001) and probability integral transform (PIT) histograms (=-=Diebold et al. 1998-=-; Gneiting et al. 2007). Verification rank histograms are used for ensemble forecasts when the number of members m is small. For each forecast case, the rank of the verifying observation is tallied wi... |

216 |
Estimation of Relationships for Limited Dependent Variables, Econometrica
- Tobin
- 1958
(Show Context)
Citation Context ...recipitation accumulation have been proposed in the literature. Stidd (1973), Bell (1987), Bardossy and Plate (1992), Hutchinson (1995), and Sansò and Guenni (1999, 2000, 2004) adapted a Tobit model (=-=Tobin 1958-=-; Chib 1992) to precipitation accumulation, working with a latent Gaussian process that relates to precipitation via a power transformation and a truncation. The power truncated normal (PTN) model off... |

202 | Geostatistics : Modeling Spatial Uncertainty - Chilès, Delfiner - 1999 |

91 | The SEM Algorithm: A Probabilistic Teacher Algorithm Derived From the EM Algorithm for the Mixture Problem - Celleux, Diebolt - 1985 |

71 | Using Bayesian Model Averaging to calibrate forecast ensembles’, Monthly Weather Review 133 - Raftery, Gneiting, et al. - 2005 |

55 | A description of the Fifth Generation - Grell, Dudhia, et al. |

53 |
A method for producing and evaluating probabilistic forecasts from ensemble model integrations
- Anderson
- 1996
(Show Context)
Citation Context ...ommodate precipitation occurrence and precipitation accumulation using a single latent spatial process. To assess the calibration of the predictive distributions, we use verification rank histograms (=-=Anderson 1996-=-; Talagrand et al. 1997; Hamill and Colucci 1997; Hamill 2001) and probability integral transform (PIT) histograms (Diebold et al. 1998; Gneiting et al. 2007). Verification rank histograms are used fo... |

51 |
The Use of Model Output Statistics (MOS) in Objective Weather Forecasting
- Glahn, Lowry
- 1972
(Show Context)
Citation Context ...hods have been developed to statistically postprocess numerical predictions of precipitation occurrence and produce quantitative probabilistic precipitation forecasts. They include linear regression (=-=Glahn and Lowry 1972-=-; Bermowitz 1975; Antolik 2000), quantile regression (Bremnes 2004; Friederichs and Hense 2007), logistic regression (Applequist et al. 2002; Hamill et al. 2004), neural networks (Koizumi 1999; Ramire... |

49 | Interpretation of rank histograms for verifying ensemble forecasts
- Hamill
(Show Context)
Citation Context ...n using a single latent spatial process. To assess the calibration of the predictive distributions, we use verification rank histograms (Anderson 1996; Talagrand et al. 1997; Hamill and Colucci 1997; =-=Hamill 2001-=-) and probability integral transform (PIT) histograms (Diebold et al. 1998; Gneiting et al. 2007). Verification rank histograms are used for ensemble forecasts when the number of members m is small. F... |

49 |
Verification of Eta-RSM short-range ensemble forecasts
- Hamill, Colucci
- 1997
(Show Context)
Citation Context ...precipitation accumulation using a single latent spatial process. To assess the calibration of the predictive distributions, we use verification rank histograms (Anderson 1996; Talagrand et al. 1997; =-=Hamill and Colucci 1997-=-; Hamill 2001) and probability integral transform (PIT) histograms (Diebold et al. 1998; Gneiting et al. 2007). Verification rank histograms are used for ensemble forecasts when the number of members ... |

48 |
Simulation of stationary Gaussian processes
- Chan, Wood
- 1994
(Show Context)
Citation Context ...patial grid is of size approximately 10,000, so simulation from the required multivariate normal distributions is not a straightforward task. For doing this, we use the circulant embedding technique (=-=Wood and Chan 1994-=-; Dietrich and Newsam 1997; Gneiting et al. 2006) as implemented in the R package RandomFields (Schlather 2001). This is a very fast technique that can readily be used in real time. For verification p... |

44 |
Bayes Inference in the Tobit Censored Regression Model
- Chib
- 1992
(Show Context)
Citation Context ... accumulation have been proposed in the literature. Stidd (1973), Bell (1987), Bardossy and Plate (1992), Hutchinson (1995), and Sansò and Guenni (1999, 2000, 2004) adapted a Tobit model (Tobin 1958; =-=Chib 1992-=-) to precipitation accumulation, working with a latent Gaussian process that relates to precipitation via a power transformation and a truncation. The power truncated normal (PTN) model offers a unifi... |

43 |
Evaluation of probabilistic prediction systems
- Talagrand, Vautar, et al.
- 1997
(Show Context)
Citation Context ...itation occurrence and precipitation accumulation using a single latent spatial process. To assess the calibration of the predictive distributions, we use verification rank histograms (Anderson 1996; =-=Talagrand et al. 1997-=-; Hamill and Colucci 1997; Hamill 2001) and probability integral transform (PIT) histograms (Diebold et al. 1998; Gneiting et al. 2007). Verification rank histograms are used for ensemble forecasts wh... |

40 | Fast and exact simulation of stationary Gaussian processes through circulant embedding of the covariance matrix
- Dietrich, Newsam
- 1997
(Show Context)
Citation Context ...ze approximately 10,000, so simulation from the required multivariate normal distributions is not a straightforward task. For doing this, we use the circulant embedding technique (Wood and Chan 1994; =-=Dietrich and Newsam 1997-=-; Gneiting et al. 2006) as implemented in the R package RandomFields (Schlather 2001). This is a very fast technique that can readily be used in real time. For verification purposes, we need statistic... |

38 | Probabilistic forecasts, calibration, and sharpness
- Gneiting, Balabdaoui, et al.
- 2007
(Show Context)
Citation Context ...stage spatial yes yes no yes yes at several sites simultaneously. In our assessment, we are guided by the principle of maximizing the sharpness of the predictive distributions subject to calibration (=-=Gneiting et al. 2007-=-). In other words, we aim at predictive distributions that are as concentrated as possible while being statistically consistent with the observations. To provide summary measures of predictive perform... |

32 | Probabilistic quantitative precipitation forecasting using bayesian model averaging
- Sloughter, Raftery, et al.
(Show Context)
Citation Context ...tributions (Krzysztofowicz and Maranzano 2006), and two-stage models in which a Gamma density is employed to model precipitation accumulation (Wilks 1990; Hamill and Colucci 1998; Wilson et al. 1999; =-=Sloughter et al. 2007-=-). Common to all these methods is the underlying assumption that forecast errors at different locations are spatially independent. While this is not necessarily true, assuming conditional spatial inde... |

31 |
Scoring Rules for Continuous Probability Distributions
- Matheson, Winkler
- 1976
(Show Context)
Citation Context ... at individual sites as a function of the length M in days. To assess the quality of the predictive distributions for daily precipitation accumulation, we use the continuous ranked probability score (=-=Matheson and Winkler 1976-=-; Gneiting and Raftery 2007), which is a strictly proper scoring rule for the evaluation of probabilistic forecasts of a univariate quantity. It is negatively oriented, that is, the smaller the better... |

30 |
Initial results for a mesoscale short-range ensemble forecasting system over the Pacific Northwest
- Grimit, Mass
(Show Context)
Citation Context ...calendar year 2004, at a prediction horizon of 48 hours. We compare our method to competing prediction techniques, including an ensemble of NWP forecasts (University of Washington mesoscale ensemble; =-=Grimit and Mass 2002-=-; Eckel and Mass 2005), the Bayesian model averaging technique of Sloughter et al. (2007) and versions of the power truncated normal model of Bardossy and Plate (1992). In Section 4 we review other st... |

24 | Space-time model for daily rainfall using atmospheric circulation patterns - Bardossy, Plate - 1992 |

24 |
Aspects of effective mesoscale, short-range ensemble forecasting Wea
- Eckel, Mass
- 2005
(Show Context)
Citation Context ... a prediction horizon of 48 hours. We compare our method to competing prediction techniques, including an ensemble of NWP forecasts (University of Washington mesoscale ensemble; Grimit and Mass 2002; =-=Eckel and Mass 2005-=-), the Bayesian model averaging technique of Sloughter et al. (2007) and versions of the power truncated normal model of Bardossy and Plate (1992). In Section 4 we review other statistical postprocess... |

21 |
The economic value of ensemble forecasts as a tool for risk assessment: From days to decades
- Palmer
- 2002
(Show Context)
Citation Context ... shortcoming in weather-related decision-making (National Research Council 2006). One approach to incorporate uncertainty information into weather forecasting is via ensembles of numerical forecasts (=-=Palmer 2002-=-; Gneiting and Raftery 2005). While this is a major advance, the use of statistical postprocessing techniques for numerical forecasts remains essential. Several methods have been developed to statisti... |

19 | Calibrated probabilistic mesoscale weather field forecasting: The geostatistical output perturbation method - Gel, Raftery, et al. |

16 | 2004: Ensemble reforecasting: improving medium-‐range forecast skill using retrospective forecasts
- Hamill, Whitaker, et al.
(Show Context)
Citation Context ...asts. They include linear regression (Glahn and Lowry 1972; Bermowitz 1975; Antolik 2000), quantile regression (Bremnes 2004; Friederichs and Hense 2007), logistic regression (Applequist et al. 2002; =-=Hamill et al. 2004-=-), neural networks (Koizumi 1999; Ramirez et al. 2005), binning techniques (Gahrs et al. 2003; Yussouf and Stensrud 2006), hierarchical models based on climatic prior distributions (Krzysztofowicz and... |

16 | Stochastic space-time weather models from ground-based data - Hutchinson - 1995 |

15 |
Simulation and analysis of random fields
- Schlather
(Show Context)
Citation Context ...not a straightforward task. For doing this, we use the circulant embedding technique (Wood and Chan 1994; Dietrich and Newsam 1997; Gneiting et al. 2006) as implemented in the R package RandomFields (=-=Schlather 2001-=-). This is a very fast technique that can readily be used in real time. For verification purposes, we need statistical forecast ensembles at observation sites, as opposed to the gridded forecasts in F... |

14 | L.: A space-time stochastic model of rainfall for satellite remote sensing studies - Bell - 1987 |

13 | Efficient Gibbs Sampling of Truncated Multivariate Normal with Application to Constrained Linear Regression, unpublished manuscript - Rodriguez-Yam, Davis, et al. - 2004 |

11 |
Studies in the history of probability and statistics XLIX On the Matern correlation family
- Guttorp, Gneiting
(Show Context)
Citation Context ...ccumulation are modeled as stationary and isotropic Gaussian processes with an exponential correlation function. More general covariance structures such as the Matérn covariance function (Stein 1999; =-=Guttorp and Gneiting 2006-=-) could be employed. In fitting the model, we used a sliding training window consisting of the recent past. This is a simple adaptive approach that yields good results. A different approach would be t... |

11 | Venezuelan rainfall data analysed by using a Bayesian space-time model - Sansó, Guenni - 1999 |

10 |
An overview of the National Weather Service’s centralized statistical quantitative precipitation forecasts
- ANTOLIK
- 2000
(Show Context)
Citation Context ...lly postprocess numerical predictions of precipitation occurrence and produce quantitative probabilistic precipitation forecasts. They include linear regression (Glahn and Lowry 1972; Bermowitz 1975; =-=Antolik 2000-=-), quantile regression (Bremnes 2004; Friederichs and Hense 2007), logistic regression (Applequist et al. 2002; Hamill et al. 2004), neural networks (Koizumi 1999; Ramirez et al. 2005), binning techni... |

9 | Combining spatial statistical and ensemble information in probabilistic weather forecasts - Berrocal, Raftery, et al. |

9 |
2004: Probabilistic forecasts of precipitation in terms of quantiles using NWP model output
- Bremnes
(Show Context)
Citation Context ...s of precipitation occurrence and produce quantitative probabilistic precipitation forecasts. They include linear regression (Glahn and Lowry 1972; Bermowitz 1975; Antolik 2000), quantile regression (=-=Bremnes 2004-=-; Friederichs and Hense 2007), logistic regression (Applequist et al. 2002; Hamill et al. 2004), neural networks (Koizumi 1999; Ramirez et al. 2005), binning techniques (Gahrs et al. 2003; Yussouf and... |

8 |
2002: Comparison of methodologies for probabilistic quantitative precipitation forecasting
- Applequist, Gahrs, et al.
(Show Context)
Citation Context ... overall steady improvement in the quality of numerical weather predictions, forecasts of precipitation accumulation are still not as accurate and reliable as those of other meteorological variables (=-=Applequist et al. 2002-=-; Stensrud and Yussouf 2007). Furthermore, quantitative precipitation forecasts obtained from a single NWP model are deterministic, and thus, do not convey any information about the degree of uncertai... |

8 |
2007: Reliable probabilistic quantitative precipitation forecasts from Short Range Ensemble Forecasting System
- Stensrud, Yussouf
(Show Context)
Citation Context ...ent in the quality of numerical weather predictions, forecasts of precipitation accumulation are still not as accurate and reliable as those of other meteorological variables (Applequist et al. 2002; =-=Stensrud and Yussouf 2007-=-). Furthermore, quantitative precipitation forecasts obtained from a single NWP model are deterministic, and thus, do not convey any information about the degree of uncertainty in the prediction, whic... |

7 | A Markov renewal model for rainfall occurrences
- Foufoula-Georgiou, Lettenmaier
- 1987
(Show Context)
Citation Context ...ccurrence. Common choices for the distribution of nonzero precipitation accumulation include exponential densities (Todorovic and Woolhiser 1975), mixtures of exponentials (Woolhiser and Pegram 1979; =-=Foufoula-Georgiou and Lettenmaier 1987-=-) and gamma densities (Stern and Coe 1984; Wilks 1989; Hamill and Colucci 1998; Wilson et al. 1999; Sloughter 3NWP forecast of precipitation accumulation valid January 5, 2004 4$ 44 46 48 50 !130 !1$... |

7 |
Statistical downscaling of extreme precipitation events using censored quantile regression. Monthly weather review
- P, Hense
(Show Context)
Citation Context ...tion occurrence and produce quantitative probabilistic precipitation forecasts. They include linear regression (Glahn and Lowry 1972; Bermowitz 1975; Antolik 2000), quantile regression (Bremnes 2004; =-=Friederichs and Hense 2007-=-), logistic regression (Applequist et al. 2002; Hamill et al. 2004), neural networks (Koizumi 1999; Ramirez et al. 2005), binning techniques (Gahrs et al. 2003; Yussouf and Stensrud 2006), hierarchica... |

6 |
Artificial neural network technique for rainfall forecasting applied to the Sa˜o Paulo region
- Rami´rez, Velho, et al.
- 2005
(Show Context)
Citation Context ...y 1972; Bermowitz 1975; Antolik 2000), quantile regression (Bremnes 2004; Friederichs and Hense 2007), logistic regression (Applequist et al. 2002; Hamill et al. 2004), neural networks (Koizumi 1999; =-=Ramirez et al. 2005-=-), binning techniques (Gahrs et al. 2003; Yussouf and Stensrud 2006), hierarchical models based on climatic prior distributions (Krzysztofowicz and Maranzano 2006), and two-stage models in which a Gam... |

6 | Extending the limits of ensemble forecast verification with the minimum spanning tree, Monthly Weather Rev
- Smith, Hansen
- 2004
(Show Context)
Citation Context ...is negatively oriented, that is, the smaller the better. To assess calibration for ensemble forecasts of multivariate weather quantities, we use j=1 19the minimum spanning tree (MST) rank histogram (=-=Smith and Hansen 2004-=-; Wilks 2004). If the ensemble has m members, the MST rank is found by tallying the length of the MST that connects the m ensemble members within the combined set of the m + 1 lengths of the ensemble-... |

5 | Fast and exact simulation of large Gaussian lattice systems - Sev˘cíková, Percival, et al. - 2006 |

5 | Generic probability distribution of rainfall in space: the bivariate model - Herr, Krzysztofowicz - 2005 |

5 |
A stochastic model of n-day precipitation
- Todorovic, Woolhiser
- 1975
(Show Context)
Citation Context ...ence first, and then model nonzero precipitation accumulation conditionally on its occurrence. Common choices for the distribution of nonzero precipitation accumulation include exponential densities (=-=Todorovic and Woolhiser 1975-=-), mixtures of exponentials (Woolhiser and Pegram 1979; Foufoula-Georgiou and Lettenmaier 1987) and gamma densities (Stern and Coe 1984; Wilks 1989; Hamill and Colucci 1998; Wilson et al. 1999; Slough... |

5 |
Conditioning stochastic daily precipitation models on total monthly precipitation
- Wilks
- 1989
(Show Context)
Citation Context ...ion include exponential densities (Todorovic and Woolhiser 1975), mixtures of exponentials (Woolhiser and Pegram 1979; Foufoula-Georgiou and Lettenmaier 1987) and gamma densities (Stern and Coe 1984; =-=Wilks 1989-=-; Hamill and Colucci 1998; Wilson et al. 1999; Sloughter 3NWP forecast of precipitation accumulation valid January 5, 2004 4$ 44 46 48 50 !130 !1$5 !1$0 !115 0 $0 40 60 80 100 (a) NWP forecast of pre... |

5 |
Maximum likelihood estimation of Fourier coefficients to describe seasonal variation of parameters in stochastic daily precipitation models
- Woolhiser, Pegram
- 1979
(Show Context)
Citation Context ...tion conditionally on its occurrence. Common choices for the distribution of nonzero precipitation accumulation include exponential densities (Todorovic and Woolhiser 1975), mixtures of exponentials (=-=Woolhiser and Pegram 1979-=-; Foufoula-Georgiou and Lettenmaier 1987) and gamma densities (Stern and Coe 1984; Wilks 1989; Hamill and Colucci 1998; Wilson et al. 1999; Sloughter 3NWP forecast of precipitation accumulation valid... |

4 |
An application of model output statistics to forecasting quantitative precipitation
- Bermowitz
- 1975
(Show Context)
Citation Context ...ed to statistically postprocess numerical predictions of precipitation occurrence and produce quantitative probabilistic precipitation forecasts. They include linear regression (Glahn and Lowry 1972; =-=Bermowitz 1975-=-; Antolik 2000), quantile regression (Bremnes 2004; Friederichs and Hense 2007), logistic regression (Applequist et al. 2002; Hamill et al. 2004), neural networks (Koizumi 1999; Ramirez et al. 2005), ... |

4 |
2005: Weather forecasting using ensemble methods
- Raftery
(Show Context)
Citation Context ...in weather-related decision-making (National Research Council 2006). One approach to incorporate uncertainty information into weather forecasting is via ensembles of numerical forecasts (Palmer 2002; =-=Gneiting and Raftery 2005-=-). While this is a major advance, the use of statistical postprocessing techniques for numerical forecasts remains essential. Several methods have been developed to statistically postprocess numerical... |

4 |
An objective method to modify numerical model forecasts with newly given weather data using an artificial neural network
- Koizumi
- 1999
(Show Context)
Citation Context ...Glahn and Lowry 1972; Bermowitz 1975; Antolik 2000), quantile regression (Bremnes 2004; Friederichs and Hense 2007), logistic regression (Applequist et al. 2002; Hamill et al. 2004), neural networks (=-=Koizumi 1999-=-; Ramirez et al. 2005), binning techniques (Gahrs et al. 2003; Yussouf and Stensrud 2006), hierarchical models based on climatic prior distributions (Krzysztofowicz and Maranzano 2006), and two-stage ... |

4 | Simulation of precipitation fields from probabilistic quantitative precipitation forecast - SEO, PERICA, et al. - 2000 |

4 |
A strategy for verifying weather element forecasts from an ensemble prediction system
- WILSON, BURROWS, et al.
- 1999
(Show Context)
Citation Context ...n climatic prior distributions (Krzysztofowicz and Maranzano 2006), and two-stage models in which a Gamma density is employed to model precipitation accumulation (Wilks 1990; Hamill and Colucci 1998; =-=Wilson et al. 1999-=-; Sloughter et al. 2007). Common to all these methods is the underlying assumption that forecast errors at different locations are spatially independent. While this is not necessarily true, assuming c... |

3 |
Improved results for probabilistic quantitative precipitation forecasting. Weather and Forecasting
- GAHRS, APPLEQUIST, et al.
- 2003
(Show Context)
Citation Context ...ntile regression (Bremnes 2004; Friederichs and Hense 2007), logistic regression (Applequist et al. 2002; Hamill et al. 2004), neural networks (Koizumi 1999; Ramirez et al. 2005), binning techniques (=-=Gahrs et al. 2003-=-; Yussouf and Stensrud 2006), hierarchical models based on climatic prior distributions (Krzysztofowicz and Maranzano 2006), and two-stage models in which a Gamma density is employed to model precipit... |