Linear and nonlinear post-processing of numerically forecasted surface temperature

Casaioli, M.; Mantovani, Roberto; Scorzoni, F. Proietti; Puca, Silvia; Speranza, Antonio; Tirozzi, B.

doi:10.5194/npg-10-373-2003

Cited by 21 publications

(20 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Perhaps the most severe drawback of MLPs is the possibility that the learning procedure will end in a local minimum of the error function instead of reaching the global one. The seriousness of this problem can be reduced by means of repeated training or by application of a global minimum search technique (such as the one used by Casaioli et al, 2003), but usually at the cost of increased training time. There is also the possibility of overtraining, when the learning procedure runs for too long a time, and the network becomes overoptimized for the description of the cases from the learning set, thus losing its ability to generalize.…”

Section: Discussionmentioning

confidence: 99%

“…Several studies have been published devoted to downscaling or postprocessing of temperatures by nonlinear methods, mostly MLPs (Trigo and Palutikof, 1999;Schoof and Pryor, 2001;Marzban, 2003;Casaioli et al, 2003) or neural networks based on RBF functions (Weichert and Bürger, 1998). Here, downscaling of the gridded large-scale data was done for the predictand series of daily mean, minimum and maximum temperatures.…”

Section: Downscalingmentioning

confidence: 99%

See 1 more Smart Citation

Testing the performance of three nonlinear methods of time seriesanalysis for prediction and downscaling of European daily temperatures

Mikšovský

Raidl

2005

Nonlin. Processes Geophys.

View full text Add to dashboard Cite

Abstract.We investigated the usability of the method of local linear models (LLM), multilayer perceptron neural network (MLP NN) and radial basis function neural network (RBF NN) for the construction of temporal and spatial transfer functions between different meteorological quantities, and compared the obtained results both mutually and to the results of multiple linear regression (MLR). The tested methods were applied for the short-term prediction of daily mean temperatures and for the downscaling of NCEP/NCAR reanalysis data, using series of daily mean, minimum and maximum temperatures from 25 European stations as predictands. None of the tested nonlinear methods was recognized to be distinctly superior to the others, but all nonlinear techniques proved to be better than linear regression in the majority of the cases. It is also discussed that the most frequently used nonlinear method, the MLP neural network, may not be the best choice for processing the climatic time series -LLM method or RBF NNs can offer a comparable or slightly better performance and they do not suffer from some of the practical disadvantages of MLPs.Aside from comparing the performance of different methods, we paid attention to geographical and seasonal variations of the results. The forecasting results showed that the nonlinear character of relations between climate variables is well apparent over most of Europe, in contrast to rather weak nonlinearity in the Mediterranean and North Africa. No clear large-scale geographical structure of nonlinearity was identified in the case of downscaling. Nonlinearity also seems to be noticeably stronger in winter than in summer in most locations, for both forecasting and downscaling.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Downscalingmentioning

confidence: 99%

Testing the performance of three nonlinear methods of time seriesanalysis for prediction and downscaling of European daily temperatures

Mikšovský

Raidl

2005

Nonlin. Processes Geophys.

View full text Add to dashboard Cite

show abstract

“…To solve these difficulties different techniques have been applied by different authors, relating the predictions in grid points to real physical sites (Weichert and Bürger, 1998;Schoof and Pryor, 2001;Huth, 2002Huth, , 2004. Several methods were widely used to do the downscaling such as Perfect Prog, Model Output Statistics (MOS) (Wilks, 1995), Artificial Neural Networks (Hsieh et al, 1998) or even Multilinear Regression (Woodcock and Southern, 1983;Casaioli et al, 2003).…”

Section: Introductionmentioning

confidence: 99%

Determining temperature lapse rates over mountain slopes using vertically weighted regression: a case study from the Pyrenees

Pagès

Miró

2009

Meteorological Applications

View full text Add to dashboard Cite

This study took place in the Pyrenees Range, in the northeastern Iberian Peninsula. The Pyrenees extend longitudinally, separating the Iberian Peninsula from the rest of Europe, and high peaks around 3000 m arise from deep valleys. As a mountain range it creates a barrier to advection, in this case from the north and south, and typical meteorological phenomena of mountainous areas occur within it (inversions, Foehn effect, extreme wind-chill, snow storms). Thus, two specific valleys in Catalonia were considered, Val d'Aran and Cerdanya. In both valleys automatic weather stations (AWSs) are available at similar heights. Although these valleys are only 100 km apart, they have different climates. However, the main reason for developing the study was that Numerical Weather Prediction (NWP) has problems when forecasting temperatures in complex terrain areas, mainly in the valley floor in winter season.Firstly, different equations based on a multilinear regression were obtained for each weather station. Multilinear regression was considered in this case as the most suitable downscaling method and data used were provided by the AWSs and MM5 (PSU/NCAR mesoscale model) numerical weather prediction model outputs.These equations were obtained to set up a Geographically Weighted Regression (GWR) method, although this one was modified and changed to a Vertically Weighted Regression (VWR) in order to create vertical temperature profiles.

show abstract

“…The numerical output must usually undergo some post-processing treatment in order to fulfil the specific tasks of interest. Among the applications devised in the context of the project were, for example, the coupling of numerical model precipitation with river flow models for purposes of flood forecasting (Calenda et al, 2000) and the post-processing of surface temperature forecasts for agricultural applications (Casaioli et al, 2003). A first example of the capabilities of the 'sea section' of the system was presented by Bargagli et al (2002), who performed tidal propagation and sea-state hindcasting experiments.…”

Section: Objective Verification Of the System And Applicationsmentioning

confidence: 99%

SIMM: An integrated forecasting system for the Mediterranean area

Speranza¹,

Accadia²,

Mariani³

et al. 2007

Meteorological Applications

View full text Add to dashboard Cite

Many 'high-impact' meteorological, marine and hydrological events in the Mediterranean area are characterized by horizontal spatial scales of the order of 10-100 km. Such events are, sometimes, driven by complex dynamical processes involving planetary scale atmospheric flows. Several international programs (ALPEX, POEM, MAP, PYREX, MEDEX) have improved the understanding of some of these processes. However, because of the Mediterranean's geomorphological structure, characterized by mountain chains (e.g. the Alps), semi-enclosed sea basins and small river catchments, many problems remain. It is clear that such problems have to be faced in the context of analysis-prediction systems bridging the gap between global and local scales of motion. These systems should allow for an adequate representation of key dynamical processes at all the relevant scales of motion.The Hydro-Meteorological-Marine System ('Sistema Idro-Meteo-Mare', SIMM) is a first step in developing an integrated system, adequately covering all scales of motion from global to local. A short description of the system is presented, highlighting scientific concepts behind design choices. A summary of the results of verification tests is also illustrated, together with a general evaluation of the whole process in planning, developing and running SIMM in order to assist future updates of the system, currently under development.

show abstract

Linear and nonlinear post-processing of numerically forecasted surface temperature

Cited by 21 publications

References 15 publications

Testing the performance of three nonlinear methods of time seriesanalysis for prediction and downscaling of European daily temperatures

Testing the performance of three nonlinear methods of time seriesanalysis for prediction and downscaling of European daily temperatures

Determining temperature lapse rates over mountain slopes using vertically weighted regression: a case study from the Pyrenees

SIMM: An integrated forecasting system for the Mediterranean area

Contact Info

Product

Resources

About