A new method for the characterization and verification of local spatial predictability for convective‐scale ensembles

Dey, Seonaid R. A.; Roberts, Nigel; Plant, Robert S.; Migliorini, Stefano

doi:10.1002/qj.2792

Cited by 32 publications

(79 citation statements)

References 63 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…First the rain rate values of f 1 and f 2 at grid point P (

f_{1 ij}^{0}

and

f_{2 ij}^{0}

, where the superscript ‘0’ indicates that we are comparing f 1 ij and f 2 ij at a single grid point, that is at a scale of 0) are compared by calculating the quantity

D_{ij}^{0} = ({, \begin{matrix} \frac{{(f_{1 ij}^{0} - f_{2 ij}^{0})}^{2}}{{(f_{1 ij}^{0})}^{2} + {(f_{2 ij}^{0})}^{2}} & if f_{1 ij}^{0} > 0 or f_{2 ij}^{0} > 0, \\ 1 & if f_{1 ij}^{0} = 0 and f_{2 ij}^{0} = 0 . \end{matrix})

f 1 ij and f 2 ij are considered to be suitably similar at this single grid point (a scale of zero) if

D_{ij}^{0} \leq D_{crit, ij}^{0}

, where

D_{crit, ij}^{0} = α

, a pre‐defined constant. Consistent with Dey et al () a value of α = 0.5 is used here. This choice means that a ratio

f_{1 ij}^{0}

f_{2 ij}^{0}

in the range 2

- \sqrt{3}

to 2+

\sqrt{3}

is considered suitably similar at the grid scale, so that the criterion is primarily dictated by whet...…”

Section: Analysis Methodssupporting

confidence: 91%

“…In particular, we use the average agreement scale between member–member pairs, denoted

S_{ij}^{normalA (\overset{mm}{false})}

, and the average agreement scale between member–radar pairs, denoted

S_{ij}^{normalA (\overset{mo}{false})}

. For ease of reference, we maintain the notation of Dey et al (). Thus, in

S_{ij}^{normalA (\overset{mm}{false})}

and

S_{ij}^{normalA (\overset{mo}{false})}

, we have S ij : A scale defining a square area (neighbourhood) centred upon grid point ij .…”

Section: Analysis Methodssupporting

confidence: 89%

“…This article measures the local spatial agreement between ensemble members, and between ensemble members and radar observations, using the methods of Dey et al (). In particular, we use the average agreement scale between member–member pairs, denoted

S_{ij}^{normalA (\overset{mm}{false})}

, and the average agreement scale between member–radar pairs, denoted

S_{ij}^{normalA (\overset{mo}{false})}

.…”

Section: Analysis Methodsmentioning

confidence: 99%

“…Similarly, for the calculation of

S_{ij}^{normalA (\overset{mo}{false})}

we have N ensemble member–radar pairs, N fields of

S_{ij}^{normalA (f_{1} f_{2})}

, and an average of these N values produces

S_{ij}^{normalA (\overset{mo}{false})}

. Although

S_{ij}^{normalA (\overset{mm}{false})}

and

S_{ij}^{normalA (\overset{mo}{false})}

are calculated by averaging over a different number of pairs, Dey et al () showed (using an idealised experiment) that they can be compared to diagnose the spatial spread–skill relationship of the ensemble.…”

Section: Analysis Methodsmentioning

confidence: 99%

“…The spatial spread–skill results are presented in the form of a binned scatter plot, as introduced in Dey et al (). The binned scatter plot allows the two fields of the

S_{ij}^{normalA (\overset{mm}{false})}

and the

S_{ij}^{normalA (\overset{mo}{false})}

to be compared (at a specified forecast time) while preserving location‐dependent information.…”

Section: Comparing With Observationsmentioning

confidence: 99%

See 4 more Smart Citations

Assessing spatial precipitation uncertainties in a convective‐scale ensemble

Dey

Plant

Roberts³

et al. 2016

Quart J Royal Meteoro Soc

Self Cite

View full text Add to dashboard Cite

New techniques have recently been developed to quantify the location‐dependent spatial agreement between ensemble members, and the spatial spread–skill relationship. In this article a summer of convection‐permitting ensemble forecasts are analysed to better understand the factors influencing location‐dependent spatial agreement of precipitation fields and the spatial spread–skill relationship over the UK. The aim is to further investigate the agreement scale method, and to highlight the information that could be extracted for a more long‐term routine model evaluation. Overall, for summer 2013, the UK 2.2 km grid spacing ensemble system was found to be reasonably well spread spatially, although there was a tendency for the ensemble to be overconfident in the location of precipitation. For the forecast lead times considered (up to 36 h), a diurnal cycle was seen in the spatial agreement and in the spatial spread–skill relationship: the forecast spread and error did not increase noticeably with forecast lead time. Both the spatial agreement and the spatial spread–skill were dependent on the fractional coverage and average intensity of precipitation. A poor spread–skill relationship was associated with a low fractional coverage of rain and low average rain rates. The times with a smaller fractional coverage, or lower intensity, of precipitation were found to have lower spatial agreement. The spatial agreement was found to be location dependant, with higher confidence in the location of precipitation to the northwest of the UK.

show abstract

“…First the rain rate values of f 1 and f 2 at grid point P (

f_{1 ij}^{0}

and

f_{2 ij}^{0}

, where the superscript ‘0’ indicates that we are comparing f 1 ij and f 2 ij at a single grid point, that is at a scale of 0) are compared by calculating the quantity

D_{ij}^{0} = ({, \begin{matrix} \frac{{(f_{1 ij}^{0} - f_{2 ij}^{0})}^{2}}{{(f_{1 ij}^{0})}^{2} + {(f_{2 ij}^{0})}^{2}} & if f_{1 ij}^{0} > 0 or f_{2 ij}^{0} > 0, \\ 1 & if f_{1 ij}^{0} = 0 and f_{2 ij}^{0} = 0 . \end{matrix})

f 1 ij and f 2 ij are considered to be suitably similar at this single grid point (a scale of zero) if

D_{ij}^{0} \leq D_{crit, ij}^{0}

, where

D_{crit, ij}^{0} = α

, a pre‐defined constant. Consistent with Dey et al () a value of α = 0.5 is used here. This choice means that a ratio

f_{1 ij}^{0}

f_{2 ij}^{0}

in the range 2

- \sqrt{3}

to 2+

\sqrt{3}

is considered suitably similar at the grid scale, so that the criterion is primarily dictated by whet...…”

Section: Analysis Methodssupporting

confidence: 91%

“…In particular, we use the average agreement scale between member–member pairs, denoted

S_{ij}^{normalA (\overset{mm}{false})}

, and the average agreement scale between member–radar pairs, denoted

S_{ij}^{normalA (\overset{mo}{false})}

. For ease of reference, we maintain the notation of Dey et al (). Thus, in

S_{ij}^{normalA (\overset{mm}{false})}

and

S_{ij}^{normalA (\overset{mo}{false})}

, we have S ij : A scale defining a square area (neighbourhood) centred upon grid point ij .…”

Section: Analysis Methodssupporting

confidence: 89%

S_{ij}^{normalA (\overset{mm}{false})}

, and the average agreement scale between member–radar pairs, denoted

S_{ij}^{normalA (\overset{mo}{false})}

.…”

Section: Analysis Methodsmentioning

confidence: 99%

“…Similarly, for the calculation of

S_{ij}^{normalA (\overset{mo}{false})}

we have N ensemble member–radar pairs, N fields of

S_{ij}^{normalA (f_{1} f_{2})}

, and an average of these N values produces

S_{ij}^{normalA (\overset{mo}{false})}

. Although

S_{ij}^{normalA (\overset{mm}{false})}

and

S_{ij}^{normalA (\overset{mo}{false})}

Section: Analysis Methodsmentioning

confidence: 99%

“…The spatial spread–skill results are presented in the form of a binned scatter plot, as introduced in Dey et al (). The binned scatter plot allows the two fields of the

S_{ij}^{normalA (\overset{mm}{false})}

and the

S_{ij}^{normalA (\overset{mo}{false})}

to be compared (at a specified forecast time) while preserving location‐dependent information.…”

Section: Comparing With Observationsmentioning

confidence: 99%

See 3 more Smart Citations

Assessing spatial precipitation uncertainties in a convective‐scale ensemble

Dey

Plant

Roberts³

et al. 2016

Quart J Royal Meteoro Soc

Self Cite

View full text Add to dashboard Cite

show abstract

A percentile‐based approach to rainfall scenario construction for surface‐water flood forecasts

Böing

Birch

Rabb

et al. 2020

Meteorological Applications

View full text Add to dashboard Cite

A novel technique to produce reasonable worst-case rainfall scenarios from ensemble forecasts is presented. This type of scenario is relevant for predicting the risk of localized, intense rainfall events with a duration between 15 min and several hours. Such rainfall events can cause surface-water (pluvial) flooding. Producing useful forecasts of these events at lead times of more than a few hours is challenging due to the precision and accuracy in rainfall intensity, duration and location that is required. The technique described here addresses these challenges by constructing appropriate scenarios using a neighbourhood technique in combination with ensemble forecasting. It is similar to the distance-dependent depth-duration analysis described in earlier studies, but it introduces an additional post-processing step based on probability distribution functions of rainfall accumulation near a location of interest. This additional step makes the reasonable worst-case scenarios less dependent on grid-scale behaviour, and helps to generate scenarios with a consistent interpretation. The method is used to compare forecasts with a lead time of 6-36 hr to radar data for several case studies that occurred in Yorkshire. These comparisons also introduce new techniques to present maps of the reasonable worst-case rainfall accumulation at each location.

show abstract

The Italian open data meteorological portal: MISTRAL

Bottazzi

Scipione

Marras

et al. 2021

Meteorological Applications

View full text Add to dashboard Cite

At the national level, in Italy, observational and forecast data are collected by various public bodies and are often kept in various small, heterogeneous and non‐interoperable repositories, released under different licenses, thus limiting the usability for external users. In this context, MISTRAL (the Meteo Italian SupercompuTing PoRtAL) was launched as the first Italian meteorological open data portal, with the aim of promoting the reuse of meteorological data sets available at national level coverage. The MISTRAL portal provides (and archives) meteorological data from various observation networks, both public and private, and forecast data that are generated and post‐processed within the Consortium for Small‐scale Modeling‐Limited Area Model Italia (COSMO‐LAMI) agreement using high performance computing (HPC) facilities. Also incorporated is the Italy Flash Flood use case, implemented with the collaboration of European Centre for Medium‐Range Weather Forecasts (ECMWF), which exploits cutting edge advances in HPC‐based post‐processing of ensemble precipitation forecasts, for different model resolutions, and applies those to deliver novel blended‐resolution forecasts specifically for Italy. Finally, in addition to providing architectures for the acquisition and display of observational data, MISTRAL also delivers an interactive system for visualizing forecast data of different resolutions as superimposed multi‐layer maps.

show abstract

A new method for the characterization and verification of local spatial predictability for convective‐scale ensembles

Cited by 32 publications

References 63 publications

Assessing spatial precipitation uncertainties in a convective‐scale ensemble

Assessing spatial precipitation uncertainties in a convective‐scale ensemble

A percentile‐based approach to rainfall scenario construction for surface‐water flood forecasts

The Italian open data meteorological portal: MISTRAL

Contact Info

Product

Resources

About