Forecast Datasets Research Articles

Improving global rainfall forecasting with a weather type approach in Japan

ABSTRACT An automated version of the weather type classification scheme was performed over Japan to characterize daily circulation conditions. A daily gridded field of mean sea-level pressure (MSLP) from the European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis dataset (ERA-interim) and the THORPEX Interactive Grand Global Ensemble (TIGGE) daily forecast dataset were used. The weather type is advantageous as it provides an opportunity to improve global rainfall prediction by refining statistical bias correction. We distinguished 11 weather types: anticyclone, cyclone, hybrid and eight purely wind directions. The results indicate that the main weather types contributing to the total volume of rainfall are cyclone, hybrid, purely westerly and northwest winds. A gamma-based bias correction decreases the global rainfall forecast root mean square by 10%, while specific weather type gamma bias correction accounts for 5–10% root mean square error reduction, with a total decrease of errors up to a maximum of 20%. Both global and weather type bias corrections improve the extreme dependency scores (EDS), but for different extreme rainfall thresholds. The study advocates the use of weather type bias-correction methods for extreme event rainfall intensity corrections higher than 100 mm/d. EDITOR A. Castellarin ASSOCIATE EDITOR A. Jain

A New Open-Access HUC-8 Based Downscaled CMIP-5 Climate Model Forecast Dataset for the Conterminous United States

Watershed-scale hydrologic simulation models generally require climate data inputs including precipitation and temperature. These climate inputs can be derived from downscaled global climate simulations which have the potential to drive runoff forecasts at the scale of local watersheds. While a simulation designed to drive a local watershed model would ideally be constructed at an appropriate scale, global climate simulations are, by definition, arbitrarily determined large rectangular spatial grids. This paper addresses the technical challenge of making climate simulation model results readily available in the form of downscaled datasets that can be used for watershed scale models. Specifically, we present the development and deployment of a new Coupled Model Intercomparison Project phase 5 (CMIP5) based database which has been prepared through a scaling and weighted averaging process for use at the level of U.S. Geological Survey (USGS) Hydrologic Unit Code (HUC)-8 watersheds. The resulting dataset includes 2,106 virtual observation sites (watershed centroids) each with 698 associated time series datasets representing average monthly temperature and precipitation between 1950 and 2099 based on 234 unique climate model simulations. The new dataset is deployed on a HydroServer and distributed using WaterOneFlow web services in the WaterML format. These methods can be adapted for downscaled General Circulation Model (GCM) results for specific drainage areas smaller than HUC-8. Two example use cases for the dataset also are presented.

Optimization of neural networks through grammatical evolution and a genetic algorithm

This paper proposes a hybrid neuro-evolutive algorithm (NEA) that uses a compact indirect encoding scheme (IES) for representing its genotypes (a set of ten production rules of a Lindenmayer System with memory), moreover has the ability to reuse the genotypes and automatically build modular, hierarchical and recurrent neural networks. A genetic algorithm (GA) evolves a Lindenmayer System (L-System) that is used to design the neural network’s architecture. This basic neural codification confers scalability and search space reduction in relation to other methods. Furthermore, the system uses a parallel genome scan engine that increases both the implicit parallelism and convergence of the GA. The fitness function of the NEA rewards economical artificial neural networks (ANNs) that are easily implemented. The NEA was tested on five real-world classification datasets and three well-known datasets for time series forecasting (TSF). The results are statistically compared against established state-of-the-art algorithms and various forecasting methods (ADANN, ARIMA, UCM, and Forecast Pro). In most cases, our NEA outperformed the other methods, delivering the most accurate classification and time series forecasting with the least computational effort. These superior results are attributed to the improved effectiveness and efficiency of NEA in the decision-making process. The result is an optimized neural network architecture for solving classification problems and simulating dynamical systems.

Impacts of Noah model physics on catchment‐scale runoff simulations

AbstractNoah model physics options validated for the source region of the Yellow River (SRYR) are applied to investigate their ability in reproducing runoff at the catchment scale. Three sets of augmentations are implemented affecting descriptions of (i) turbulent and soil heat transport (Noah‐H), (ii) soil water flow (Noah‐W), and (iii) frozen ground processes (Noah‐F). Five numerical experiments are designed with the three augmented versions, a control run with default model physics and a run with all augmentations (Noah‐A). Each experiment is set up with vegetation and soil parameters from Weather Research and Forecasting data set, soil organic matter content from China Soil Database, 0.1° atmospheric forcing data from Institute of Tibetan Plateau Research (Chinese Academy of Sciences), and initial equilibrium model states achieved using a single‐year recurrent spin‐up. In situ heat flux, soil temperature (Ts), and soil moisture (θ) profile measurements are available for point‐scale assessment, whereas monthly streamflow is utilized for the catchment‐scale evaluation. The comparison with point measurements shows that the augmentations invoked with Noah‐H resolve issues with the heat flux and Ts simulation and Noah‐W mitigates deficiencies in the θ simulation, while Noah‐A yields improvements for both simulated surface energy and water budgets. In contrast, Noah‐F has a minor effect. Also, at catchment scale, the best model performance is found for Noah‐A leading to a base flow‐dominated runoff regime, whereby the surface runoff contribution remains significant. This study highlights the need for a complete description of vertical heat and water exchanges to correctly simulate the runoff in the seasonally frozen and high‐altitude SRYR at the catchment scale.

Expectation Errors in the Foreign Exchange Market

This paper argues that order flow can explain exchange rate forecasting errors. A unified theoretical model is developed showing that forecasting errors can be explained by both informational rigidities and portfolio shifts. This is applied to Brazilian data using a unique data set of daily consensus exchange rate forecasts managed by the Banco Central do Brasil along with order flow derived from the FX futures market. The results strongly support the theory.

Testing Reanalyses in Constraining Dynamical Downscaling

Reanalysis data sets have been widely used in regional climate dynamical downscaling studies. In this study, we test the use of various reanalysis data sets in constraining dynamical downscaling by assessing the reconstruction skill of the Yellow Sea coastal winds using the COSMO model in Climate Mode (CCLM) with 7-km resolution. Four reanalysis forcing data sets are used as lateral boundary conditions and internal large-scale constraints (spectral nudging): the National Centers for Environmental Prediction and National Center for Atmospheric Research (NCEP/NCAR) reanalysis data set (NCEP1) is downscaled to an intermediate domain with 55-km resolution (CCLM_55km), ERA-interim reanalysis data set (ERAint), NCEP climate forecast system reanalysis data set (CFSR), and Japanese 55-year reanalysis data set (JRA55). Several statistical analysis methods are employed to assess the modeled winds through comparison with observed offshore wind data from 2006, and it is found that the downscaled simulations yield good quality wind speed products. However, they all tend to overestimate observed low wind speeds and underestimate observed high wind speeds. Furthermore, the quality of the modeled wind direction is strongly associated with the wind speed intensities, exhibiting a much better reproduction of wind direction at strong wind speeds than at light wind speeds. The downscaling simulations driven by ERAint, JRA55, and CFSR are consistent with each other in the reproduction of local wind speed and direction; the simulations driven by ERAint and JRA55 are slightly better for strong winds and those driven by CFSR are better for light winds. All three simulations generate local wind estimates that are superior to those of the simulation driven by CCLM_55km. This superiority reflects the better quality of the CFSR, ERAint, and JRA55 reanalyses with regard to assimilated local observations compared with the CCLM_55km hindcast, which exploits only upper-air large scale NCEP1 wind fields.

Noisy Information and Expectation Formation in the Foreign Exchange Market

Market microstructure and the imperfect common knowledge literature in macroeconomics both analyze the effect of dispersed information on prices. This paper draws on both sources to understand exchange rate forecasting errors. A theoretical model is developed showing that forecasting errors depend on both forecast revisions as in the Woodford noisy information model and order flow as in the Evans-Lyons simultaneous trade model. This is applied to Brazilian data using a unique data set of daily consensus exchange rate forecasts managed by the Banco Central do Brasil along with order flow derived from the FX futures market. The results strongly support the theory.

Predicting Wind Power with Reforecasts

Abstract Energy traders and decision-makers need accurate wind power forecasts. For this purpose, numerical weather predictions (NWPs) are often statistically postprocessed to correct systematic errors. This requires a dataset of past forecasts and observations that is often limited by frequent NWP model enhancements that change the statistical model properties. Reforecasts that recompute past forecasts with a recent model provide considerably longer datasets but usually have weaker setups than operational models. This study tests the reforecasts from the National Oceanic and Atmospheric Administration (NOAA) and the European Centre for Medium-Range Weather Forecasts (ECMWF) for wind power predictions. The NOAA reforecast clearly performs worse than the ECMWF reforecast, the operational ECMWF deterministic and ensemble forecasts, and a limited-area model of the Austrian weather service [Zentralanstalt für Meteorologie und Geodynamik (ZAMG)]. On the contrary, the ECMWF reforecast has, of all tested models, the smallest squared errors and one of the highest financial values in an energy market.

Study of the tornado event in Greece on March 25, 2009: Synoptic analysis and numerical modeling using modified topography

Recent research revealed that western Greece and NW Peloponnese are regions that favor prefrontal tornadic incidence. On March 25, 2009 a tornado developed approximately at 10:30 UTC near Varda village (NW Peloponnese). Tornado intensity was T4–T5 (TORRO scale) and consequently caused an economic impact of 350,000 € over the local society. The goals of this study are: (i) to analyze synoptic and remote sensing features regarding the tornado event over NW Peloponnese and (ii) to investigate the role of topography in tornadogenesis triggered under strong synoptic scale forcing over that area.Synoptic analysis was based on the European Centre for Medium-Range Weather Forecasts (ECMWF) data sets. The analysis of daily anomaly of synoptic conditions with respect to 30 years' climatology (1981–2010), was based on the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis data sets. In addition, numerous remote sensing data sets were derived by the Hellenic National Meteorological Service (HNMS) weather station network in order to better interpret the examined tornado event.Finally, numerical modeling was performed using the non-hydrostatic Weather Research and Forecasting model (WRF), initialized by ECMWF gridded analyses, with telescoping nested grids that allow the representation of atmospheric circulations ranging from the synoptic scale down to the meso-scale. The two numerical experiments were performed on the basis of: (a) the presence and (b) the absence of topography (landscape), so as to determine whether the occurrence of a tornado – identified by diagnostic instability indices – could be indicated by modifying topography.The energy helicity index (EHI), the bulk Richardson number (BRN) shear, the storm-relative environmental helicity (SRH), and the maximum convective available potential energy (MCAPE, for parcels with maximum θe) were considered as principal diagnostic instability variables and employed in both numerical experiments. Furthermore, model verification was conducted, accompanied by analysis of the absolute vorticity budget.Synoptic analysis revealed that the synoptic weather conditions on March 25, 2009 are in agreement with the composite synoptic climatology for tornado days over western Greece. In addition, maximum daily anomalies at the barometric levels of 500, 700, 850 and 925hPa were found, compared to the climatology of composite mean anomalies for tornado days over western Greece. Numerical simulations revealed that the topography of NW Peloponnese did not constitute an important factor during the tornado event on March 25, 2009, based on EHI, SRH, BRN, and MCAPE analyses.

Polarization signatures and brightness temperatures caused by horizontally oriented snow particles at microwave bands: Effects of atmospheric absorption

AbstractThis study analyzes the effects of atmospheric absorption and emission on the polarization difference (PD) and brightness temperature (TB) generated by horizontally oriented snow particles. A three‐layer plane‐parallel atmosphere model is used in conjunction with a simplified radiative transfer (RT) scheme to illustrate the combined effects of dichroic and nondichroic media on microwave signatures observed by ground‐based and spaceborne sensors. Based on idealized scenarios which encompass a dichroic snow layer and adjacent nondichroic layers composed of supercooled liquid water (SCLW) droplets and water vapor, we demonstrate that the presence of atmospheric absorption/emission enhances TB and damps PD when observed from the ground. From a spaceborne perspective, however, TB can be reduced or enhanced by an absorbing/emitting layer above the snow layer, while a strong absorbing/emitting layer below the dichroic snow layer may even enhance PD. The induced PD and TB, which rely on snow microphysical assumptions, can vary up to 2 K and 10 K, respectively, due to the temperature‐dependent absorption of SCLW. RT calculations based on 223 snowfall profiles selected from European Centre for Medium‐Range Weather Forecasts data sets indicate that the existence of SCLW has a noticeable impact on PD and TB at three window frequencies (150 GHz, 243 GHz, and 664 GHz) during snowfall. Our results imply that while polarimetric channels at the three window channels have the potential for snowfall characterization, accurate information on liquid water is required to correctly interpret the polarimetric observations.

Impacts of adaptive collaboration on demand forecasting accuracy of different product categories throughout the product life cycle

Purpose – This paper aims to empirically analyze how adaptive collaboration in supply chain management impacts demand forecast accuracy in short life-cycle products, depending on collaboration intensity, product life-cycle stage, retailer type and product category. Design/methodology/approach – The authors assembled a data set of forecasts and sales of 169 still-camera models, made by the same manufacturer and sold by three different retailers in France over five years. Collaboration intensity, coded by collaborative planning forecasting and replenishment level, was used to analyze the main effects and specific interaction effects of all variables using ANOVA and ordered feature evaluation analysis (OFEA). Findings – The findings lend empirical support to the long-standing assumption that supply chain collaboration intensity increases demand forecast accuracy and that product maturation also increases forecast accuracy even in short life-cycle products. Furthermore, the findings show that it is particularly the lack of collaboration that causes negative effects on forecast accuracy, while positive interaction effects are only found for life cycle stage and product category. Practical implications – Investment in adaptive supply chain collaboration is shown to increase demand forecast accuracy. However, the choice of collaboration intensity should account for life cycle stage, retailer type and product category. Originality/value – This paper provides empirical support for the adaptive collaboration concept, exploring not only the actual benefits but also the way it is achieved in the context of innovative products with short life cycles. The authors used a real-world data set and pushed its statistical analysis to a new level of detail using OFEA.

Regional and seasonal characteristics of global horizontal irradiance forecasts obtained from the Japan Meteorological Agency mesoscale model

To obtain accurate forecasts of photovoltaic power generation, the use of forecast datasets of meteorological elements from numerical prediction models, specifically global horizontal irradiance (GHI), is necessary. This study seeks to validate, and therefore improve GHI forecasts. Ground-based data from Japan Meteorological Agency (JMA) stations are used in a JMA mesoscale model (MSM) during the time period from 2008 to 2012 and temporal and spatial characteristics of forecast errors are analyzed. Statistical monthly evaluations show that associated errors vary between seasons, with monthly GHI mean bias error values ranging from −60 to +45W/m2 and root mean square errors (RMSEs) ranging from 95 to 170W/m2. Mapping of forecast errors show that underestimation of GHI forecast values and large RMSE values are significant in the southern part of Japan (a subtropical region located along the Pacific Ocean), particularly during summers. In winter, overestimation of GHI forecasts is found throughout the entire Japanese archipelago. The frequency of different cloud type occurrences over the Japanese islands indicate that regional and seasonal variations in cloud types are related to relatively large GHI forecast errors. High-level cirrus clouds, mid-level altocumulus, and low-level stratus are often observed during summer, when forecasted values are underestimated, and during winter, when values are overestimated.

Argo data assimilation and its effect on climate state estimation and forecasting in the western North Pacific using a coupled model

AbstractIn this study, we investigated the effects of ocean subsurface data (Argo data) on climate state estimation and forecasting, focusing on the reproduction of North Pacific subtropical mode water (STMW) using a four‐dimensional variational data assimilation system with a coupled model. We produced two reanalysis plus forecast data sets for the ocean and atmosphere in 2010 using a 3 month assimilation period: the first including Argo data (Argo case) and the second did not include Argo data (control case). In the control case, the Kuroshio, Kuroshio Extension front, and recirculation gyres along the front were not adequately reproduced. Consequently, there were large biases in temperature and salinity in the western North Pacific. The assimilation of Argo data effectively corrected these biases and significantly improved reproduction of the Kuroshio fronts and recirculation gyres, resulting in a more realistic reproduction of the winter mixed layer and STMW. The correction of these biases is critical to the 1–3 year predictions of the STMW core properties, and the assimilation of Argo data enables prediction of these properties for more than a year. We showed that assimilation of Argo data affects the surface atmospheric temperature above the STMW formation region.

Impact of hindcast length on estimates of seasonal climate predictability.

It has recently been argued that single-model seasonal forecast ensembles are overdispersive, implying that the real world is more predictable than indicated by estimates of so-called perfect model predictability, particularly over the North Atlantic. However, such estimates are based on relatively short forecast data sets comprising just 20 years of seasonal predictions. Here we study longer 40 year seasonal forecast data sets from multimodel seasonal forecast ensemble projects and show that sampling uncertainty due to the length of the hindcast periods is large. The skill of forecasting the North Atlantic Oscillation during winter varies within the 40 year data sets with high levels of skill found for some subperiods. It is demonstrated that while 20 year estimates of seasonal reliability can show evidence of overdispersive behavior, the 40 year estimates are more stable and show no evidence of overdispersion. Instead, the predominant feature on these longer time scales is underdispersion, particularly in the tropics.Key PointsPredictions can appear overdispersive due to hindcast length sampling errorLonger hindcasts are more robust and underdispersive, especially in the tropicsTwenty hindcasts are an inadequate sample size to assess seasonal forecast skill

Operational monitoring and forecasting of bathing water quality through exploiting satellite Earth observation and models: The AlgaRisk demonstration service

Coastal zones and shelf-seas are important for tourism, commercial fishing and aquaculture. As a result the importance of good water quality within these regions to support life is recognised worldwide and a number of international directives for monitoring them now exist. This paper describes the AlgaRisk water quality monitoring demonstration service that was developed and operated for the UK Environment Agency in response to the microbiological monitoring needs within the revised European Union Bathing Waters Directive. The AlgaRisk approach used satellite Earth observation to provide a near-real time monitoring of microbiological water quality and a series of nested operational models (atmospheric and hydrodynamic-ecosystem) provided a forecast capability. For the period of the demonstration service (2008–2013) all monitoring and forecast datasets were processed in near-real time on a daily basis and disseminated through a dedicated web portal, with extracted data automatically emailed to agency staff. Near-real time data processing was achieved using a series of supercomputers and an Open Grid approach. The novel web portal and java-based viewer enabled users to visualise and interrogate current and historical data. The system description, the algorithms employed and example results focussing on a case study of an incidence of the harmful algal bloom Karenia mikimotoi are presented. Recommendations and the potential exploitation of web services for future water quality monitoring services are discussed.

The GEFS-Based Daily Reference Evapotranspiration (ETo) Forecast and Its Implication for Water Management in the Southeastern United States

Abstract NOAA’s second-generation retrospective forecast (reforecast) dataset was created using the currently operational Global Ensemble Forecast System (GEFS). It has the potential to accurately forecast daily reference evapotranspiration ETo and can be useful for water management. This study was conducted to evaluate daily ETo forecasts using the GEFS reforecasts in the southeastern United States (SEUS) and to incorporate the ETo forecasts into irrigation scheduling to explore the usefulness of the forecasts for water management. ETo was estimated using the Penman–Monteith equation, and ensemble forecasts were downscaled and bias corrected using a forecast analog approach. The overall forecast skill was evaluated using the linear error in probability space skill score, and the forecast in five categories (terciles and 10th and 90th percentiles) was evaluated using the Brier skill score, relative operating characteristic, and reliability diagrams. Irrigation scheduling was evaluated by water deficit WD forecasts, which were determined based on the agricultural reference index for drought (ARID) model driven by the GEFS-based ETo forecasts. All forecast skill was generally positive up to lead day 7 throughout the year, with higher skill in cooler months compared to warmer months. The GEFS reforecast improved ETo forecast skill for all lead days over the SEUS compared to the first-generation reforecast. The WD forecasts driven by the ETo forecasts showed higher accuracy and less uncertainty than the forecasts driven by climatology, indicating their usefulness for irrigation scheduling, hydrological forecasting, and water demand forecasting in the SEUS.

Observations of Riparian Tree Geometry and Bank Characteristics along a Bank Stability Gradient at the South River, VA

AbstractIn this study, we evaluate the use of riparian tree‐geometry and bank‐characteristic data in developing riverbank stability forecasts at four study sites along a mercury‐contaminated reach of the South River, Virginia. Our forecast data set includes measurements of tree diameter, lean angle and percent root‐plate undercut, coupled with bank characteristics included in the Pfankuch Index. Stepwise comparisons and statistical analyses of our tree‐geometry data reveal that lean angle and percent root‐plate undercut are effective measures for discriminating between riparian tree populations at eroding and stable banks, with the strongest discriminant power from the percent root‐plate undercut. Our stability forecasts, using both tree and bank characteristics, agree with long‐term observations, underscoring the value of rarely used tree‐geometry data in such evaluations. The application of these results is important in long study reaches where forecasts of erosion potential from contaminated banks are required to develop mitigation strategies, but direct measurements of bank stability are not feasible. Copyright © 2014 John Wiley & Sons, Ltd.

Assessing the Macroeconomic Forecasting Performance of Boosting: Evidence for the United States, the Euro Area and Germany

ABSTRACTThe use of large datasets for macroeconomic forecasting has received a great deal of interest recently. Boosting is one possible method of using high‐dimensional data for this purpose. It is a stage‐wise additive modelling procedure, which, in a linear specification, becomes a variable selection device that iteratively adds the predictors with the largest contribution to the fit. Using data for the United States, the euro area and Germany, we assess the performance of boosting when forecasting a wide range of macroeconomic variables. Moreover, we analyse to what extent its forecasting accuracy depends on the method used for determining its key regularization parameter: the number of iterations. We find that boosting mostly outperforms the autoregressive benchmark, and that K‐fold cross‐validation works much better as stopping criterion than the commonly used information criteria. Copyright © 2014 John Wiley & Sons, Ltd.

Improved weather-based late blight risk management: comparing models with a ten year forecast archive

SUMMARYAgroecosystem decision support systems typically rely on some types of weather data. Although many new digital weather and forecast datasets are gridded data, the current authors feel that evaluating previous methods with data of increased archive length is critical in aiding the transition to new datasets that lack extensive archives. To that end, the present paper reviews the improvements made to an artificial neural network for forecasting weather-based potato late blight (Phytophthora infestans) risk at 26 locations in the Great Lakes region. Accuracies of predictions made using an early model, developed in 2007, are compared with accuracies of predictions made using a new 10-year hourly optimized model. In nearly every comparison by month, forecast lead time and spatial region, the newly optimized model is more accurate, especially when the weather is conducive to high disease levels.

The Efficiency of Multivariate Macroeconomic Forecasts

We examine the efficiency of multivariate macroeconomic forecasts by estimating a vector autoregressive model on the forecast revisions of four variables (GDP, inflation, unemployment and wages). Using a data set of professional forecasts for the G7 countries, we find evidence of cross‐series revision dynamics. Specifically, forecasts revisions are conditionally correlated to the lagged forecast revisions of other macroeconomic variables, and the sign of the correlation is as predicted by conventional economic theory. This indicates that forecasters are slow to incorporate news across variables. We show that this finding can be explained by forecast underreaction.