A lower bound for prediction uncertainty in nowcasting/forecasting models

J. Clayton Kerce; Francois Vandenberghe

doi:10.1117/12.740665

A lower bound for prediction uncertainty in nowcasting/forecasting models

J. Clayton Kerce, Francois Vandenberghe

Joint Center for Satellite Data Assimilation

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

We introduce a formalism for computing the Cramer-Rao lower bound (CRLB) for a general dynamical system, develop an approach to bounding the process noise for a general dynamical system, and discuss the application of this formalism in the context of a prototypical forecasting model. This model consists of a simple transport diffusion process with assimilation updates based on point source measurements. We investigate the use of Krylov subspace techniques for efficient computation of two point correlation functions, and the use of this technique in generating a coarse-grained state covariance.

Original language	English
Title of host publication	Assimilation of Remote Sensing and In Situ Data in Modern Numerical Weather and Environmental Prediction Models
DOIs	https://doi.org/10.1117/12.740665
State	Published - 2007
Event	Assimilation of Remote Sensing and In Situ Data in Modern Numerical Weather and Environmental Prediction Models - San Diego, CA, United States Duration: Aug 26 2007 → Aug 27 2007

Publication series

Name	Proceedings of SPIE - The International Society for Optical Engineering
Volume	6685
ISSN (Print)	0277-786X

Conference

Conference	Assimilation of Remote Sensing and In Situ Data in Modern Numerical Weather and Environmental Prediction Models
Country/Territory	United States
City	San Diego, CA
Period	08/26/07 → 08/27/07

Keywords

CRLB
Cramer Rao bound
Dynamical systems
Predictability
State estimation

Access to Document

10.1117/12.740665

Cite this

@inproceedings{15fc35bda1704eb594be847b89e3c756,

title = "A lower bound for prediction uncertainty in nowcasting/forecasting models",

abstract = "We introduce a formalism for computing the Cramer-Rao lower bound (CRLB) for a general dynamical system, develop an approach to bounding the process noise for a general dynamical system, and discuss the application of this formalism in the context of a prototypical forecasting model. This model consists of a simple transport diffusion process with assimilation updates based on point source measurements. We investigate the use of Krylov subspace techniques for efficient computation of two point correlation functions, and the use of this technique in generating a coarse-grained state covariance.",

keywords = "CRLB, Cramer Rao bound, Dynamical systems, Predictability, State estimation",

author = "Kerce, \{J. Clayton\} and Francois Vandenberghe",

year = "2007",

doi = "10.1117/12.740665",

language = "English",

isbn = "9780819468338",

series = "Proceedings of SPIE - The International Society for Optical Engineering",

booktitle = "Assimilation of Remote Sensing and In Situ Data in Modern Numerical Weather and Environmental Prediction Models",

note = "Assimilation of Remote Sensing and In Situ Data in Modern Numerical Weather and Environmental Prediction Models ; Conference date: 26-08-2007 Through 27-08-2007",

}

Kerce, JC & Vandenberghe, F 2007, A lower bound for prediction uncertainty in nowcasting/forecasting models. in Assimilation of Remote Sensing and In Situ Data in Modern Numerical Weather and Environmental Prediction Models., 668505, Proceedings of SPIE - The International Society for Optical Engineering, vol. 6685, Assimilation of Remote Sensing and In Situ Data in Modern Numerical Weather and Environmental Prediction Models, San Diego, CA, United States, 08/26/07. https://doi.org/10.1117/12.740665

A lower bound for prediction uncertainty in nowcasting/forecasting models. / Kerce, J. Clayton; Vandenberghe, Francois.
Assimilation of Remote Sensing and In Situ Data in Modern Numerical Weather and Environmental Prediction Models. 2007. 668505 (Proceedings of SPIE - The International Society for Optical Engineering; Vol. 6685).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - A lower bound for prediction uncertainty in nowcasting/forecasting models

AU - Kerce, J. Clayton

AU - Vandenberghe, Francois

PY - 2007

Y1 - 2007

N2 - We introduce a formalism for computing the Cramer-Rao lower bound (CRLB) for a general dynamical system, develop an approach to bounding the process noise for a general dynamical system, and discuss the application of this formalism in the context of a prototypical forecasting model. This model consists of a simple transport diffusion process with assimilation updates based on point source measurements. We investigate the use of Krylov subspace techniques for efficient computation of two point correlation functions, and the use of this technique in generating a coarse-grained state covariance.

AB - We introduce a formalism for computing the Cramer-Rao lower bound (CRLB) for a general dynamical system, develop an approach to bounding the process noise for a general dynamical system, and discuss the application of this formalism in the context of a prototypical forecasting model. This model consists of a simple transport diffusion process with assimilation updates based on point source measurements. We investigate the use of Krylov subspace techniques for efficient computation of two point correlation functions, and the use of this technique in generating a coarse-grained state covariance.

KW - CRLB

KW - Cramer Rao bound

KW - Dynamical systems

KW - Predictability

KW - State estimation

UR - https://www.scopus.com/pages/publications/42149179443

U2 - 10.1117/12.740665

DO - 10.1117/12.740665

M3 - Conference contribution

AN - SCOPUS:42149179443

SN - 9780819468338

T3 - Proceedings of SPIE - The International Society for Optical Engineering

BT - Assimilation of Remote Sensing and In Situ Data in Modern Numerical Weather and Environmental Prediction Models

T2 - Assimilation of Remote Sensing and In Situ Data in Modern Numerical Weather and Environmental Prediction Models

Y2 - 26 August 2007 through 27 August 2007

ER -

A lower bound for prediction uncertainty in nowcasting/forecasting models

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this