Long-range Forecasting Models Research Articles

Objective and Probabilistic Long-Range Forecasts of Summertime Air Temperatures in South Korea Based on Gaussian Processes

Abstract We propose the objective long-range forecasting model based on Gaussian processes (OLRAF-GP), focusing on summertime near-surface air temperatures in June (1-month lead), July (2-month lead), and August (3-month lead). The predictors were objectively selected based on their relationships with the target variables, either from observations (GP-OBS) or from observations and dynamical climate model results from APEC Climate Center multimodel ensemble (APCC MME) for the period with no observed data (GP-MME). The performances of the OLRAF-GP models were compared with the model with predetermined predictors from observations (GP-PD). Both GP-MME and GP-OBS outperformed GP-PD in June [Heidke skill score (HSS); HSS = 0.46, 0.72, and 0.16 for mean temperature] and July (HSS = 0.53, 0.3, and 0.07 for mean temperature). Furthermore, GP-MME mostly outperformed GP-OBS and GP-PD in August (HSS = 0.52, 0.28, and 0.5, respectively, for mean temperature), implying larger contributions of the additional predictors from MME. OLRAF-GP models, especially GP-MME, are expected to better forecast summertime temperatures in regions where existing models have been struggling. We find that the physical processes associated with the notable predictors are aligned with those in previous studies, such as the attribution of the La Niña conditions in the previous winter, the related Indian Ocean capacitor effect, and the impacts of wintertime Polar/Eurasia pattern. These results imply that the mechanisms of the objectively selected predictors can be physically meaningful, and their inclusion can improve model performance and efficiency. Significance Statements This study aims to improve the long-range probabilistic forecasting of summertime near-surface temperatures for regions where the climate variability is not sufficiently explained by well-known key predictors. We propose objective and probabilistic forecasting models that use objectively selected predictors either from observations or from observations and results of the dynamical climate model. The overall skill scores of the proposed models (overall HSS = 0.33, 0.39) for the case study site of South Korea are higher than the model with predetermined predictors (overall HSS = 0.19). We also find that the mechanisms of the objectively selected predictors can be physically meaningful, and their inclusion can improve model performance and efficiency.

A long-range forecasting model for the thermosphere based on the intelligent optimized particle filtering

A long-range forecasting model for the thermosphere based on the intelligent optimized particle filtering

Quantifying the Predictive Skill in Long-Range Forecasting. Part II: Model Error in Coarse-Grained Markov Models with Application to Ocean-Circulation Regimes

AbstractAn information-theoretic framework is developed to assess the predictive skill and model error in imperfect climate models for long-range forecasting. Here, of key importance is a climate equilibrium consistency test for detecting false predictive skill, as well as an analogous criterion describing model error during relaxation to equilibrium. Climate equilibrium consistency enforces the requirement that long-range forecasting models should reproduce the climatology of prediction observables with high fidelity. If a model meets both climate consistency and the analogous criterion describing model error during relaxation to equilibrium, then relative entropy can be used as an unbiased superensemble measure of the model’s skill in long-range coarse-grained forecasts. As an application, the authors investigate the error in modeling regime transitions in a 1.5-layer ocean model as a Markov process and identify models that are strongly persistent but their predictive skill is false. The general techniques developed here are also useful for estimating predictive skill with model error for Markov models of low-frequency atmospheric regimes.

Long-Range Forecasting for a Consumer Durable in an International Market

There has been a substantial amount of interest recently in long-range planning. One necessary component of the long-range plan Is the long-range forecast. In contrast to the emphasis on the planning process, however, little attention has been given to forecasting. This study considers the problem of long-range forecasting in a situation which is of growing importance - forecasting sales for international markets.Many researchers appear to operate under the impression that causal models (i.e., models based on an analysis of underlying factors) lead to more accurate sales forecasts than those provided by naive models (i.e., projections based on historical sales data only). A survey of the research literature led to the conclusion that this confidence in causal models is virtually unsupported. One can hardly criticize firms, then, for relying primarily upon naive models for sales forecasting since these models are simpler and less expensive than causal model.This study was based on the hypothesis that causal models are superior to naive models in certain situations. The key element of these situations is that -there are 'large changes.' Long-range sales forecasting usually involves such large changes; and there are many reasons to expect that long-range forecasting for international markets is a situation in which substantial changes will occur (e.g., the Kennedy round tariff cuts and the formation of common markets.)A causal model was developed to provide long-range forecasts of the international market for still cameras. This model provided unconditional forecasts of unit camera sales by country for year t n on the basis of l) knowledge about camera sales in year t and 2) predicted changes in four causal variables from year t to t n. These four causal variables included, in order of importance, per capita income, price of cameras, number of potential buyers and quality of cameras.The predictive ability of the causal model was superior to that of a naive model purporting to represent current practice. Each model was used to provide backcasts of 1954 camera sales in 17 countries on the basis of data from 1967 to 1960 only. The mean absolute percentage error for the causal model was 23% while that for the naive model was 43%. This result was statistically significant (0? = .05); but, more importantly, it appeared to have great practical significance. An evaluation, based on very conservative subjective estimates, indicated that such an improvement in accuracy would have a present value worth in excess of one percent of a typical firm's yearly sales volume.Further support for the use of the causal model was obtained by noting that the standard errors of the estimated relationships were low (evidence of reliability), that the estimates of causal relationships from different measurement models were in rather close agreement (evidence of construct validity), and that the causal model performed well in another situation where predictions were provided for I960-65 camera sales in 11 'new' countries (evidence of concurrent validity).The causal relationships were initially specified by a subjective analysis. Various parts of the causal model were then updated by use of a number of measurement models including an analysis of differences among sales rates for 30 countries, of differences among changes in the sales rates from 1961 to 1965 for 21 countries, and of differences among six income categories from United States household survey data. This updating led to a modest, though valuable, gain as the mean absolute percentage error of the 1954 backcast was reduced from 30% to the 23% mentioned above.Additional benefits associated with the development of the causal model included the ability to evaluate large changes in the market; to estimate current sales where trade and production figures are inadequate; to evaluate alternative assumptions about the future rapidly and cheaply; and to identify markets which have not been fully exploited.In summary, the study argues that the development of better long-range forecasting models is an important problem; describes the development of causal models; and demonstrates the superiority of causal models over naive models in a case involving long-range forecasting for international markets.

Application of a long-range forecasting model to earthquakes in the Japan mainland testing region

The Every Earthquake a Precursor According to Scale (EEPAS) model is a long-range forecasting method which has been previously applied to a number of regions, including Japan. The Collaboratory for the Study of Earthquake Predictability (CSEP) forecasting experiment in Japan provides an opportunity to test the model at lower magnitudes than previously and to compare it with other competing models. The model sums contributions to the rate density from past earthquakes based on predictive scaling relations derived from the precursory scale increase phenomenon. Two features of the earthquake catalogue in the Japan mainland region create difficulties in applying the model, namely magnitude-dependence in the proportion of aftershocks and in the Gutenberg-Richter b-value. To accommodate these features, the model was fitted separately to earthquakes in three different target magnitude classes over the period 2000–2009. There are some substantial unexplained differences in parameters between classes, but the time and magnitude distributions of the individual earthquake contributions are such that the model is suitable for three-month testing at M ≥ 4 and for one-year testing at M ≥ 5. In retrospective analyses, the mean probability gain of the EEPAS model over a spatially smoothed seismicity model increases with magnitude. The same trend is expected in prospective testing. The Proximity to Past Earthquakes (PPE) model has been submitted to the same testing classes as the EEPAS model. Its role is that of a spatially-smoothed reference model, against which the performance of time-varying models can be compared.

Long-range earthquake forecasting allowing for aftershocks

SUMMARY An earthquake occurrence model aimed at forecasting main shocks can be adjusted to allow for the occurrence of aftershocks conforming to well-known empirical relations describing their temporal decay, magnitude distribution and spatial extent. Modifications are proposed to the Every Earthquake a Precursor According to Scale (EEPAS) long-range forecasting model to allow for the occurrence of aftershocks of predicted events. Using earthquake catalogues of California and the Kanto region, central Japan, versions of the modified and original EEPAS model are fitted to a period of about 10 yr and independently tested on a later period of about 10 yr of each catalogue. For the testing period, the increase in the log-likelihood per earthquake of the modified models over their original counterparts is about 0.1 on average. This gives preliminary confirmation of the efficacy of the modifications. Versions of the modified model will be submitted to regional testing centres of the Collaboratory for the Study of Earthquake Predictability for further independent performance testing in real time.

A Study of Six Operational Sea Surface Temperature Analyses

This study results from recommendations made by a 1984 WMO Expert Committee on Ocean-Atmosphere Interaction Relevant to Long-Range Forecasting. The committee suggested that comparisons be carried out between monthly sea surface temperature (SST) analyses routinely made in several different countries in near real time. Emphasis was placed on the improvement of such analyses for use in operational long-range forecasting, especially for initializing dynamical long-range forecasting models. Six different monthly averaged SST analyses have been compared. The extent to which the analyses agree on several space scales and for regions covering the global oceans is shown, together with estimates of the magnitude of various types of errors. Independent estimates of SST obtained from expendable bathythermogmphs indicate that the monthly mean Meteorological Office (UKMO), Climate Analysis Center (CAC) in situ, and CAC blended analyses showed small differences (biases) from the expendable bathythermograph data. The differences were near to or below the margins of statistical significance over the Northern Hemisphere and the Southern Hemisphere tropics. Apparent negative biases in the analyses were noted, however, in the extratropical Southern Hemisphere. The authors finish with a discussion of recent improvements to the accuracy and scope of SST analyses for both long-range forecasting and climate studies. These improvements include an integrated analysis of ice limit, in situ and satellite SST data, and the developing use of optimum interpolation as a method of SST analysis.

Predicting doctorate production in the U.S.A.: Some lessons for long-range forecasters

The nine major independently conducted forecasts of doctoral degree production for the 1970s shared a consistently high bias, although one was quite accurate. In this paper we compare the forecasting models based upon results of these studies and discuss their implications for long-range forecasting. We show that the bias was mainly due to a common core assumption (exemplifying ‘assumption drag’) that students would pursue degrees at close to the existing rates in spite of decreasing traditional demand for graduates. The results also indicate the value of simple models and eclectic forecasting methods, the desirability of robust causal models and the dangers of extrapolative methods (especially with insufficient historical data) when structural changes are anticipated, the usefulness of current expert opinion, the inappropriateness of intentions surveys for time horizons longer than the lead time needed for significant change, and the importance of physical and economic limitations and principles (such as substitution effects and interactions between supply and demand) in developing long-range forecasting models.