Evaluation of several different planetary boundary layer schemes within a
                        single model, a unified model and a multimodel superensemble

Krishnamurti, T. N.; Basu, Swati; Sanjay, J.; Gnanaseelan, C.

doi:10.1111/j.1600-0870.2007.00278.x

Cited by 7 publications

(7 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Here we utilized the planetary boundary layer formulations from the studies of Manobianco [], Kanamitsu [], Holtslag and Boville [], Hong and Pan [], and Stull []. These schemes have been described by Krishnamurti et al []. Each member model of a suite of models utilizes different planetary boundary layer schemes.…”

Section: Numerical Weather Predictionmentioning

confidence: 99%

“…Here these fluxes for training and forecast validation were estimated from the computations of the apparent heat source and the apparent moisture sink, following Yanai et al [], and using the 6 hourly reanalysis data sets for the summer of 2002. The details are provided in Krishnamurti et al []. Figure shows the predicted fluxes of moisture within the planetary boundary layer (PBL) for 1 day, 24 September 2002 at 12:00 UTC, over a monsoon domain.…”

Section: Numerical Weather Predictionmentioning

confidence: 99%

“…Geographical distribution of 24 h forecasts of latent heat flux valid at September 24 2002 at 12:00 UTC over the monsoon domain The eight panels include the observed Yannai fluxes, the results from the five‐member models that utilize a single scheme, the results for the unified model, and from the multimodel superensemble [ Krishnamurti et al , ].…”

Section: Numerical Weather Predictionmentioning

confidence: 99%

“…Following these experiments, a similar multimodel exercise was carried out by Krishnamurti et al [] to examine possible forecast impacts of various planetary boundary layer schemes within a single unified model (that includes several weighted schemes within one model) and for the construction of the multimodel superensemble. Five different planetary boundary formulations were included from large forecast experimentations.…”

Section: Numerical Weather Predictionmentioning

confidence: 99%

“…Five different planetary boundary formulations were included from large forecast experimentations. These schemes are (i) a local‐ K closure scheme [ Manobianco , ], in which the turbulent mixing is treated by a first‐order closure; (ii) a second local‐ K PBL scheme [ Kanamitsu , ; Basu et al , ], where the K is determined through mixing length considerations and using semi‐empirical stability dependent functions based on the bulk Richardson number; (iii) a nonlocal scheme [ Holtslag and Boville , ], which is based on a modified gradient (nonlocal) corrected approach; here, the fluxes are allowed to flow down the local gradient; an artificial gradient is added to the gradient during convective conditions; (iv) an operational version of the above local scheme [ Hong and Pan , ], which utilizes observation based empirical functions for the height divided by Monin Obukov length in the PBL that are similar to the surface similarity theory; and (v) a turbulent closure scheme [ Krishnamurti et al , ], which includes prognostic equations for the turbulent kinetic energy (TKE), the kinetic energy (up to a defined closure) K , and the turbulent energy dissipation E . The details of all these methods are provided in Krishnamurti et al [].…”

Section: Numerical Weather Predictionmentioning

confidence: 99%

See 4 more Smart Citations

A review of multimodel superensemble forecasting for weather, seasonal climate, and hurricanes

Krishnamurti

Kumar

Simon

et al. 2016

Reviews of Geophysics

Self Cite

View full text Add to dashboard Cite

This review provides a summary of work in the area of ensemble forecasts for weather, climate, oceans, and hurricanes. This includes a combination of multiple forecast model results that does not dwell on the ensemble mean but uses a unique collective bias reduction procedure. A theoretical framework for this procedure is provided, utilizing a suite of models that is constructed from the well-known Lorenz low-order nonlinear system. A tutorial that includes a walk-through table and illustrates the inner workings of the multimodel superensemble's principle is provided. Systematic errors in a single deterministic model arise from a host of features that range from the model's initial state (data assimilation), resolution, representation of physics, dynamics, and ocean processes, local aspects of orography, water bodies, and details of the land surface. Models, in their diversity of representation of such features, end up leaving unique signatures of systematic errors. The multimodel superensemble utilizes as many as 10 million weights to take into account the bias errors arising from these diverse features of multimodels. The design of a single deterministic forecast models that utilizes multiple features from the use of the large volume of weights is provided here. This has led to a better understanding of the error growths and the collective bias reductions for several of the physical parameterizations within diverse models, such as cumulus convection, planetary boundary layer physics, and radiative transfer. A number of examples for weather, seasonal climate, hurricanes and sub surface oceanic forecast skills of member models, the ensemble mean, and the superensemble are provided.A suite of models, each of which carry somewhat different representation of the above processes, can be combined to reduce the collective local biases in space, time, and for different variables from the different models. That is the theme of the multimodel superensemble. The multimodel superensemble (which is not an ensemble mean) utilizes as many as ten million weights toward the reduction of such systematic errors. This figure of 10 million comes from the products of the three-dimensional grid points, the number of dependent variables, the number of models, and the number of forecast intervals in time.The notion of the multimodel superensemble was first described in Krishnamurti et al. [1999]. This utilizes a training and a forecast phase. The training phase learns from the recent past performances of models and is used to determine statistical weights from a least square minimization via a simple multiple regression. That regression is carried out with respect to analyzed (assimilated) values. Given a number of grid locations, base variables, forecast intervals, and a suite of models, the number of statistical weights can be as high as 10 7 . That many coefficients are needed because of different responses to physical parameterizations of local features such as water bodies, local mountain features, and land surface details within d...

show abstract

Section: Numerical Weather Predictionmentioning

confidence: 99%

Section: Numerical Weather Predictionmentioning

confidence: 99%

Section: Numerical Weather Predictionmentioning

confidence: 99%

Section: Numerical Weather Predictionmentioning

confidence: 99%