Contributions of Mixed Physics versus Perturbed Initial/Lateral Boundary Conditions to Ensemble-Based Precipitation Forecast Skill

2008 ◽  
Vol 136 (6) ◽  
pp. 2140-2156 ◽  
Author(s):  
Adam J. Clark ◽  
William A. Gallus ◽  
Tsing-Chang Chen
Keyword(s):  
Lead Time ◽  
Ensemble Forecast ◽  
Forecast Skill ◽  
Precipitation Forecast ◽  
Forecast Errors ◽  
Lateral Boundary ◽  
Model Errors ◽  
Forecast Lead Time ◽  
Statistical Consistency ◽  
Made In

Abstract An experiment is described that is designed to examine the contributions of model, initial condition (IC), and lateral boundary condition (LBC) errors to the spread and skill of precipitation forecasts from two regional eight-member 15-km grid-spacing Weather Research and Forecasting (WRF) ensembles covering a 1575 km × 1800 km domain. It is widely recognized that a skillful ensemble [i.e., an ensemble with a probability distribution function (PDF) that generates forecast probabilities with high resolution and reliability] should account for both error sources. Previous work suggests that model errors make a larger contribution than IC and LBC errors to forecast uncertainty in the short range before synoptic-scale error growth becomes nonlinear. However, in a regional model with unperturbed LBCs, the infiltration of the lateral boundaries will negate increasing spread. To obtain a better understanding of the contributions to the forecast errors in precipitation and to examine the window of forecast lead time before unperturbed ICs and LBCs begin to cause degradation in ensemble forecast skill, the “perfect model” assumption is made in an ensemble that uses perturbed ICs and LBCs (PILB ensemble), and the “perfect analysis” assumption is made in another ensemble that uses mixed physics–dynamic cores (MP ensemble), thus isolating the error contributions. For the domain and time period used in this study, unperturbed ICs and LBCs in the MP ensemble begin to negate increasing spread around forecast hour 24, and ensemble forecast skill as measured by relative operating characteristic curves (ROC scores) becomes lower in the MP ensemble than in the PILB ensemble, with statistical significance beginning after forecast hour 69. However, degradation in forecast skill in the MP ensemble relative to the PILB ensemble is not observed in an analysis of deterministic forecasts calculated from each ensemble using the probability matching method. Both ensembles were found to lack statistical consistency (i.e., to be underdispersive), with the PILB ensemble (MP ensemble) exhibiting more (less) statistical consistency with respect to forecast lead time. Spread ratios in the PILB ensemble are greater than those in the MP ensemble at all forecast lead times and thresholds examined; however, ensemble variance in the MP ensemble is greater than that in the PILB ensemble during the first 24 h of the forecast. This discrepancy in spread measures likely results from greater bias in the MP ensemble leading to an increase in ensemble variance and decrease in spread ratio relative to the PILB ensemble.

scholarly journals Evaluating CFSv2 Subseasonal Forecast Skill with an Emphasis on Tropical Convection

2017 ◽  
Vol 145 (9) ◽  
pp. 3795-3815 ◽  
Author(s):  
Nicholas J. Weber ◽  
Clifford F. Mass
Keyword(s):  
Lead Time ◽  
Large Scale ◽  
Tropical Convection ◽  
Forecast Skill ◽  
Lead Times ◽  
Forecast Errors ◽  
Predictive Skill ◽  
Forecast Lead Time ◽  
Temporal Averaging ◽  
Total Variability

This study examines the subseasonal predictive skill of CFSv2, focusing on the spatial and temporal distributions of error for large-scale atmospheric variables and the realism of simulated tropical convection. Errors in a 4-member CFSv2 ensemble forecast saturate at lead times of approximately 3 weeks for 500-hPa geopotential height and 5 weeks for 200-hPa velocity potential. Forecast errors exceed those of climatology at lead times beyond 2 weeks. Sea surface temperature, which evolves more slowly than atmospheric fields, maintains skill over climatology through the first month. Spatial patterns of error are robust across lead times and temporal averaging periods, increasing in amplitude as lead time increases and temporal averaging period decreases. Several significant biases were found in the CFSv2 reforecasts, such as too little convection over tropical land and excessive convection over the ocean. The realism of simulated tropical convection and associated teleconnections degrades with forecast lead time. Large-scale tropical convection in CFSv2 is more stationary than observed. Forecast MJOs propagate eastward too slowly and those initiated over the Indian Ocean have trouble traversing beyond the Maritime Continent. The total variability of simulated propagating convection is concentrated at lower frequencies compared to observed convection, and is more fully described by a red spectrum, indicating weak representation of convectively coupled waves. These flaws in simulated tropical convection, which could be tied to problems with convective parameterization and associated mean state biases, affect atmospheric teleconnections and may degrade extended global forecast skill.

scholarly journals Limits to Flood Forecasting in the Colorado Front Range for Two Summer Convection Periods Using Radar Nowcasting and a Distributed Hydrologic Model

2013 ◽  
Vol 14 (4) ◽  
pp. 1075-1097 ◽  
Author(s):  
Hernan A. Moreno ◽  
Enrique R. Vivoni ◽  
David J. Gochis
Keyword(s):  
Lead Time ◽  
Flood Forecasting ◽  
Hydrologic Model ◽  
Colorado Front Range ◽  
Precipitation Forecast ◽  
Lead Times ◽  
Forecast Errors ◽  
Front Range ◽  
Distributed Hydrologic Model ◽  
Forecast Lead Time

Abstract Flood forecasting in mountain basins remains a challenge given the difficulty in accurately predicting rainfall and in representing hydrologic processes in complex terrain. This study identifies flood predictability patterns in mountain areas using quantitative precipitation forecasts for two summer events from radar nowcasting and a distributed hydrologic model. The authors focus on 11 mountain watersheds in the Colorado Front Range for two warm-season convective periods in 2004 and 2006. The effects of rainfall distribution, forecast lead time, and basin area on flood forecasting skill are quantified by means of regional verification of precipitation fields and analyses of the integrated and distributed basin responses. The authors postulate that rainfall and watershed characteristics are responsible for patterns that determine flood predictability at different catchment scales. Coupled simulations reveal that the largest decrease in precipitation forecast skill occurs between 15- and 45-min lead times that coincide with rapid development and movements of convective systems. Consistent with this, flood forecasting skill decreases with nowcasting lead time, but the functional relation depends on the interactions between watershed properties and rainfall characteristics. Across the majority of the basins, flood forecasting skill is reduced noticeably for nowcasting lead times greater than 30 min. The authors identified that intermediate basin areas [~(2–20) km2] exhibit the largest flood forecast errors with the largest differences across nowcasting ensemble members. The typical size of summer convective storms is found to coincide well with these maximum errors, while basin properties dictate the shape of the scale dependency of flood predictability for different lead times.

scholarly journals Calibrated Precipitation Forecasts for a Limited-Area Ensemble Forecast System Using Reforecasts

2010 ◽  
Vol 138 (1) ◽  
pp. 176-189 ◽  
Author(s):  
Felix Fundel ◽  
Andre Walser ◽  
Mark A. Liniger ◽  
Christoph Frei ◽  
Christof Appenzeller
Keyword(s):  
Return Period ◽  
Lead Time ◽  
Precipitation Forecast ◽  
Ensemble Prediction ◽  
Small Scale ◽  
Limited Area ◽  
Return Periods ◽  
Model Errors ◽  
Ensemble Prediction System ◽  
Forecast Lead Time

Abstract The calibration of numerical weather forecasts using reforecasts has been shown to increase the skill of weather predictions. Here, the precipitation forecasts from the Consortium for Small Scale Modeling Limited Area Ensemble Prediction System (COSMO-LEPS) are improved using a 30-yr-long set of reforecasts. The probabilistic forecasts are calibrated on the exceedance of return periods, independently from available observations. Besides correcting for systematic model errors, the spatial and temporal variability in the amplitude of rare precipitation events is implicitly captured when issuing forecasts of return periods. These forecast products are especially useful for issuing warnings of upcoming events. A way to visualize those calibrated ensemble forecasts conveniently for end users and to present verification results of the return period–based forecasts for Switzerland is proposed. It is presented that, depending on the lead time and return period, calibrating COSMO-LEPS with reforecasts increases the precipitation forecast skill substantially (about 1 day in forecast lead time). The largest improvements are achieved during winter months. The reasonable choice of the length of the reforecast climatology is estimated for an efficient use of this computational expensive calibration method.

scholarly journals Sensitivity of forecast skill to the parameterisation of moist convection in a limited-area ensemble forecast system

2018 ◽  
Author(s):  
Matteo Vasconi ◽  
Andrea Montani ◽  
Tiziana Paccagnella
Keyword(s):  
Forecast Skill ◽  
Precipitation Forecast ◽  
Limited Area ◽  
Forecast Errors ◽  
Total Precipitation ◽  
Moist Convection ◽  
Cosmo Model ◽  
Spatio Temporal ◽  
Convection Schemes ◽  
Ensemble Systems

Abstract. The parameterisation of convection in limited-area models is an important source of uncertainty as regards the spatio-temporal forecast of precipitation. As for the limited-area model COSMO, hitherto, only the Tiedtke convection scheme was available for the operational runs of the model in convection-parameterised mode. In addition to this the Bechtold scheme, implemented in ECMWF global model, has recently been adapted for COSMO applications. The development and implementation of ensemble systems in which different convection schemes are used, provides an opportunity to upgrade state-of-the-art probabilistic systems at the convection-parameterised scale. The sensitivity of the COSMO model forecast skill to the use of either the Tietdke or the Bechtold schemes is assessed by performing different sets of experiments. The performance of COSMO model run with the different schemes is investigated in ensemble mode with particular attention to the types of forecast errors (e.g. location, timing, intensity) provided by the different convection schemes in terms of total precipitation. A 10-member ensemble has been run for approximately 2 months with the Bechtold scheme, using the same initial and boundary conditions as members 1–10 of the operational COSMO-LEPS ensemble system (which has 20 members, all run with the Tiedtke scheme). The performance of these members is assessed and compared to that of the system made of members 1–10 of COSMO-LEPS in terms of total precipitation prediction. Finally, the performance of an experimental 20-member ensemble system (which has 10 members run with the Bechtold plus 10 members run with the Tiedtke scheme) is compared to that of operational COSMO-LEPS over the 2-month period. The new system turned out to have higher skill in terms of precipitation forecast with respect to COSMO-LEPS over the period. In this approach the use of the Bechtold scheme is proposed as a perturbation for the COSMO-LEPS ensemble, relatively to how uncertainties in the model representation of the cumulus convection can be described and quantified.

scholarly journals Introducing uncertainty of radar-rainfall estimates to the verification of mesoscale model precipitation forecasts

2008 ◽  
Vol 8 (3) ◽  
pp. 445-460 ◽  
Author(s):  
M. P. Mittermaier
Keyword(s):  
Weather Prediction ◽  
Forecast Skill ◽  
Forecast Errors ◽  
Intensity Scale ◽  
Radar Rainfall ◽  
Scale Method ◽  
Skill Scores ◽  
Forecast Lead Time ◽  
The Uk ◽  
Rainfall Estimates

Abstract. A simple measure of the uncertainty associated with using radar-derived rainfall estimates as "truth" has been introduced to the Numerical Weather Prediction (NWP) verification process to assess the effect on forecast skill and errors. Deterministic precipitation forecasts from the mesoscale version of the UK Met Office Unified Model for a two-day high-impact event and for a month were verified at the daily and six-hourly time scale using a spatially-based intensity-scale method and various traditional skill scores such as the Equitable Threat Score (ETS) and log-odds ratio. Radar-rainfall accumulations from the UK Nimrod radar-composite were used. The results show that the inclusion of uncertainty has some effect, shifting the forecast errors and skill. The study also allowed for the comparison of results from the intensity-scale method and traditional skill scores. It showed that the two methods complement each other, one detailing the scale and rainfall accumulation thresholds where the errors occur, the other showing how skillful the forecast is. It was also found that for the six-hourly forecasts the error distributions remain similar with forecast lead time but skill decreases. This highlights the difference between forecast error and forecast skill, and that they are not necessarily the same.

scholarly journals Assessing the predictability of WRF model for heavy rain by cold air associated with the easterly wind at high-level patterns over mid-central Vietnam

2018 ◽  
Vol 34 (1S) ◽  
Author(s):  
Nguyen Tien Toan ◽  
Cong Thanh ◽  
Pham Thi Phuong ◽  
Vu Tuan Anh
Keyword(s):  
Lead Time ◽  
Wrf Model ◽  
Heavy Rain ◽  
Rainfall Threshold ◽  
Precipitation Forecast ◽  
Easterly Wind ◽  
Cold Air ◽  
Central Vietnam ◽  
Forecast Lead Time ◽  
High Level

Abstract: In this study, the authors assessed the possibility of predicting precipitation due to cold air associated with the easterly wind at the high level in the WRF model with 2 days lead time for the Mid-Central Vietnam region. The results show that in the 24-hour forecasts lead time should use medium rainfall threshold (16-50 mm/day) and heavy rain (50-100 mm/day) to referent for quantitative precipitation forecast and rainfall area; for 48-hour forecast lead time should choose moderate rainfall threshold. The threshold of over 100 mm, the results of all of forecast lead time is not good, almost unpredictable. The results of this study can help forecasters have more information for forecasting rain due to cold air associated easterly wind at the high level for the Mid-Central Vietnam region.

Multi-model Subseasonal Precipitation Forecasts over the Contiguous United States: Skill Assessment and Statistical Postprocessing

2021 ◽  
Author(s):  
Yanzhong Li ◽  
Di Tian ◽  
Hanoi Medina
Keyword(s):  
United States ◽  
Central Region ◽  
Lead Time ◽  
Forecast Skill ◽  
Generalized Extreme Value Distribution ◽  
Precipitation Forecast ◽  
North Central Region ◽  
Model Ensemble ◽  
North Central ◽  
Individual Model

AbstractThis study assessed multi-model subseasonal precipitation forecasts (SPFs) from eight subseasonal experiment (SubX) models over the contiguous United States (CONUS) and explored the generalized extreme value distribution (GEV)-based ensemble model output statistics (EMOS) framework for postprocessing multi-model ensemble SPF. The results showed that the SubX SPF skill varied by location and season, and the skill were relatively high in the western coastal region, north-central region, and Florida peninsula. The forecast skill was higher during winter than summer seasons, especially for lead week 3 in the northwest region. While no individual model consistently outperformed the others, the simple multi-model ensemble (MME) demonstrated a higher skill than any individual model. The GEV-based EMOS approach dramatically improved the MME subseasonal precipitation forecast skill at long lead times. The continuous ranked probability score (CRPS) was improved by approximately 20% in week 3 and 43% in lead week 4; the 5-mm Brier skill score (BSS) was improved by 59.2% in lead week 3 and 50.9% in lead week 4, with the largest improvements occurring in the northwestern, north-central, and southeastern CONUS. Regarding the relative contributions of the individual SubX model to the predictive skill, the NCEP model was given the highest weight at the shortest lead time, but the weight decreased dramatically with the increase in lead time, while the CESM, EMC, NCEP, and GMAO models were given approximately equal weights for lead weeks 2-4. The presence of active MJO conditions notably increased the forecast skill in the north-central region during weeks 3-4, while the ENSO phases influenced the skill mostly in the southern regions.

scholarly journals Prediction of the 14 June 2010 Oklahoma City Extreme Precipitation and Flooding Event in a Multiphysics Multi-Initial-Conditions Storm-Scale Ensemble Forecasting System

2016 ◽  
Vol 31 (4) ◽  
pp. 1215-1246 ◽  
Author(s):  
Nathan Dahl ◽  
Ming Xue
Keyword(s):  
Initial Conditions ◽  
Stage Iv ◽  
Radar Data ◽  
Ensemble Forecast ◽  
Forecast Skill ◽  
Oklahoma City ◽  
Precipitation Forecast ◽  
Cold Pool ◽  
Statistical Correlations ◽  
Outflow Boundary

Abstract Prolonged heavy rainfall produced widespread flooding in the Oklahoma City area early on 14 June 2010. This event was poorly predicted by operational models; however, it was skillfully predicted by the Storm-Scale Ensemble Forecast produced by the Center for Analysis and Prediction of Storms as part of the Hazardous Weather Testbed 2010 Spring Experiment. In this study, the quantitative precipitation forecast skill of ensemble members is assessed and ranked using a neighborhood-based threat score calculated against the stage IV precipitation data, and Oklahoma Mesonet observations are used to evaluate the forecast skill for surface conditions. Statistical correlations between skill metrics and qualitative comparisons of relevant features for higher- and lower-ranked members are used to identify important processes. The results demonstrate that the development of a cold pool from previous convection and the movement and orientation of the associated outflow boundary played dominant roles in the event. Without assimilated radar data from this earlier convection, the modeled cold pool was too weak and too slow to develop. Furthermore, forecast skill was sensitive to the choice of microphysics parameterization; members that used the Thompson scheme produced initial cold pools that propagated too slowly, substantially increasing errors in the timing and placement of later precipitation. The results also suggest important roles played by finescale, transient features in the period of outflow boundary stalling and reorientation associated with the heaviest rainfall. The unlikelihood of a deterministic forecast reliably predicting these features highlights the benefit of using convection-allowing/convection-resolving ensemble forecast methods for events of this kind.

scholarly journals Error Propagation of Radar Rainfall Nowcasting Fields through a Fully Distributed Flood Forecasting Model

2007 ◽  
Vol 46 (6) ◽  
pp. 932-940 ◽  
Author(s):  
Enrique R. Vivoni ◽  
Dara Entekhabi ◽  
Ross N. Hoffman
Keyword(s):  
Error Propagation ◽  
Lead Time ◽  
Hydrologic Model ◽  
Forecast Errors ◽  
Flood Forecast ◽  
Spatiotemporal Resolution ◽  
Radar Rainfall ◽  
Forecast Lead Time ◽  
Flood Simulations ◽  
Storm Characteristics

Abstract This study presents a first attempt to address the propagation of radar rainfall nowcasting errors to flood forecasts in the context of distributed hydrological simulations over a range of catchment sizes or scales. Rainfall forecasts with high spatiotemporal resolution generated from observed radar fields are used as forcing to a fully distributed hydrologic model to issue flood forecasts in a set of nested subbasins. Radar nowcasting introduces errors into the rainfall field evolution that result from spatial and temporal changes of storm features that are not captured in the forecast algorithm. The accuracy of radar rainfall and flood forecasts relative to observed radar precipitation fields and calibrated flood simulations is assessed. The study quantifies how increases in nowcasting errors with lead time result in higher flood forecast errors at the basin outlet. For small, interior basins, rainfall forecast errors can be simultaneously amplified or dampened in different flood forecast locations depending on the forecast lead time and storm characteristics. Interior differences in error propagation are shown to be effectively averaged out for larger catchment scales.

Improvement of Medium-Range Forecasts Using the Analog-Dynamical Method

2014 ◽  
Vol 142 (4) ◽  
pp. 1570-1587 ◽  
Author(s):  
Haipeng Yu ◽  
Jianping Huang ◽  
Jifan Chou
Keyword(s):  
Lead Time ◽  
Forecast Skill ◽  
Forecast Errors ◽  
Weather Forecasts ◽  
Planetary Scale ◽  
Anomaly Correlation Coefficient ◽  
Systematic Correction ◽  
Medium Range ◽  
Dynamical Method ◽  
Height Fields

Abstract This study further develops the analog-dynamical method and applies it to medium-range weather forecasts. By regarding the forecast field as a small disturbance superimposed on historical analog fields, historical analog errors can be used to estimate and correct forecast errors. This method is applied to 10-day forecasts from the Global and Regional Assimilation and Prediction System (GRAPES). Both the distribution of atmospheric circulation and the pattern of sea surface temperature (SST) are considered in choosing the analog samples from a historical dataset for 2001–10 based on NCEP Final (FNL) data. The results demonstrate that the analog-dynamical method greatly reduces forecast errors and extends the period of validity of the global 500-hPa height field by 0.8 days, which is superior to results obtained using systematic correction. The correction effect at 500 hPa is increasingly significant when the lead time increases. Although the analogs are selected using 500-hPa height fields, the forecast skill at all vertical levels is improved. The average increase of the anomaly correlation coefficient (ACC) is 0.07, and the root-mean-square error (RMSE) is decreased by 10 gpm on average at a lead time of 10 days. The magnitude of errors for most forecast fields, such as height, temperature, and kinetic energy is decreased considerably by inverse correction. The model improvement is primarily a result of improvement for planetary-scale waves, while the correction for synoptic-scale waves does not affect model forecast skill. As this method is easy to operate and transport to other sophisticated models, it could be appropriate for operational use.

Sign in / Sign up

Export Citation Format

Share Document