Short-range Rainfall Research Articles

Error characterization of ARW model in Forecasting tropical cyclone rainfall over North Indian Ocean

The hydrological extremes due to the landfalling tropical cyclones (TCs) pose a severe threat to the coastal communities over the North Indian Ocean (NIO) region. Advanced Research Weather Research and Forecasting (ARW) model is obscure in rainfall prediction, while it is extensively evaluated for track and intensity prediction over the NIO. This study focuses on estimating the model rainfall errors based on a total of 280 TC forecast cases from 42 TCs from 2007 to 2018. The model rainfall errors are studied against rain gauge and Tropical Rainfall Measuring Mission (TRMM) data as a function of TC intensity stage and model forecast length.The short-range (24 h) rainfall guidance yields fewer errors than the long-range (48–96 h) forecast when the model is initialized at any TC intensity stage. The root mean square error (RMSE) and bias of ARW rainfall is higher when the model is initialized at weaker intensity (DD or CS) stages than initialized at stronger intensity (SCS and VSCS) stages. The inland rainfall errors increase with forecast lead. The model exhibited higher errors (~2 mm h−1) in the inner-core region (0–100 km) and lesser errors (~0.5 mm h−1) in the TC environment (200–400 km). The ARW model replicates the observed radial profiles of rainfall up to 400 km with 2–5 mm h−1 overestimation at different intensity stages.Rainfall error decomposition of contiguous rain area (CRA) analysis indicates that the pattern errors contribute the maximum (~50%) to the total error, followed by the displacement error (~35%). In comparison, the volume and rotational errors are less (10% and 2%, respectively). The mean CRA horizontal shift in rainfall decreases from weaker to stronger stage initialization. The radial-distance error of categorical rainfall distribution between the ARW model and TRMM is ~ 150–200 km. This error reduced to 20–50 km after correcting the model rainfall for CRA displacement errors. The RMSE of model-rainfall after CRA correction has reduced by 3–15% for any forecast length. This study helps to advance the regional hydrological model’s ability by improved model rainfall inputs at 72–96 h lead.

Spatial and Temporal Evaluation of Radar Rainfall Nowcasting Techniques on 1,533 Events

AbstractRadar rainfall nowcasting, the process of statistically extrapolating the most recent rainfall observation, is increasingly used for very short range rainfall forecasting (less than 6 hr ahead). We performed a large‐sample analysis of 1,533 events, systematically selected for 4 event durations and 12 lowland catchments (6.5–957 km2), to determine the predictive skill of nowcasting. Four algorithms are tested and compared with Eulerian Persistence: Rainymotion Sparse, Rainymotion DenseRotation, Pysteps deterministic, and Pysteps probabilistic with 20 ensemble members. We focus on the dependency of nowcast skill on event duration, season, catchment size, and location. Maximum skillful lead times increase for longer event durations, due to the more persistent character of these events. For all four event durations, Pysteps deterministic attains the longest average decorrelation times, with 25 min for 1‐hr durations, 40 min for 3 hr, 56 min for 6 hr, and 116 min for 24 hr. During winter, with more persistent stratiform precipitation, we find three times lower mean absolute errors than for convective summer precipitation. Higher skill is also found after spatially upscaling the forecast. Catchment location matters too: Given the prevailing storm movement, two times higher skillful lead times are found downwind than upwind toward the edge of the domain. In most cases, Pysteps algorithms outperform the Rainymotion benchmark algorithms. We speculate that most errors originate from growth and dissipation processes which are not or only partially (stochastically) accounted for.

Comparison of the Performance of Hybrid ETKF-3DVAR and 3DVAR Data Assimilation Systems on Short-Range Forecasts during Indian Summer Monsoon Season in a Limited-Area Model

The impact of deploying a flow-dependent ensemble error covariance in Weather Research and Forecasting (WRF) three-dimensional variational (3DVAR) data assimilation (DA) system is examined for short-range rainfall forecasts during an Indian summer monsoon season. The flow-dependent background error covariance (BEC) is generated using a 50-member ensemble, which is further updated using the ensemble transform Kalman filter (ETKF). Assimilation is performed using a Hybrid variational-ensemble (“Hybrid”) and traditional 3DVAR DA system during the 4 weeks of July 2013. The forecasted wind, temperature, and rainfall from the assimilation experiments are verified against corresponding observations. The results indicate that the flow-dependent ensemble background error covariance in 3DVAR has systematically improved the forecasted wind and temperature when compared to the traditional 3DVAR. Similarly, rainfall forecast skill is superior in the Hybrid experiments relative to that of 3DVAR. Convection-permitting resolution rainfall forecast is validated against 746 telemetric rain gauge observations over the state of Karnataka. The Hybrid experiments show higher quantitative precipitation forecast skill than 3DVAR, particularly towards the later stages of data assimilation cycling. Spatially, the 3DVAR experiment shows a dry bias over the upper peninsular regions and a slight wet bias over the central and the northern Indian regions, while the magnitude of such wet and dry biases is smaller in forecasts from Hybrid analysis. Additionally, the westerly wind over the peninsular Indian landmass analyzed by 3DVAR is considerably weaker than that analyzed by the Hybrid experiments. This is proposed as a possible reason for the reduced dry bias in rainfall forecasts over the Indian landmass in Hybrid versus 3DVAR experiments.

Calibration Of Rainfall Ensemble Prediction Of ECMWF System 4 Using Bayesian Model Averaging

Bayesian model averaging (BMA) is a statistical post-processing method for producing probabilistic forecasts from ensembles in the form of predictive PDFs. It is known that BMA is able to improve the reliability of probabilistic forecast of short range and medium range rainfall forecast. This study aims to develop the application of BMA to calibrate seasonal forecast (long range) in order to improved quality of seasonal forecast in Indonesia. The seasonal forecast used is monthly rainfall from the output of the ensemble prediction system European Center for Medium-Range Weather Forecasts (ECMWF) system 4 model (ECS4) and it is calibrated against observational data at 26 stations of the Agency for Meteorology Climatology and Geophysics of Republic of Indonesia (BMKG) in Java Island in 1981 – 2018. BMA predictive PDFs is generated with Gamma distribution approach which is obtained based on sequential training windows (JTS) and conditionals training windows (JTC). BMA-JTS approach is done by selecting the width of the 30-month training window as the optimal training period while the BMA-JTC is carried out with a cross-validation scheme for each month. In general, Both of BMA-JTS and BMA-JTC better than RAW models. BMA-JTC calibration results are varying according to spatial and temporal, but in general the result is better in the dry season and during the El Nino phase. BMA is able to improve the distribution characteristics of the RAW model ECS4 prediction which is shown by: a smaller value of Continuous Rank Probability Score (CRPS), a larger value of the Continuous Rank Probability Skill Score (CRPSS) and more flat form of the Verification Rank Histogram (VRH) than the RAW model. BMA also increases the skill, esolution and reliability of prediction of probability Below Normal (BN) and Above Normal (AN), which is known from the increasing Brier Skill Score (BSS), and the increasing area under curve of Relative Operating Characteristics (ROC) compared to the RAW model. Furthermore, the reliability of BN and AN of BMA results also has the category of “still very useful” and “perfect” compared to RAW models that are in the “dangerous”, “not useful” and “marginally useful” categories. The reliability of BMA results with the category “still very useful” and “perfect” show that the probabilistic forecast of BN and AN events can be used in making decisions related to seasonal forecast.

Adding CO2 channel 16 to AHI data assimilation over land further improves short-range rainfall forecasts

The new generation of geostationary environmental operational satellite imager, the Himawari-8 Advanced Himawari Imager (AHI), adds two more water vapour channels and four more other channels than its predecessor, MTSAT-2. But except for the three water vapour channels, AHI channels are often not assimilated over land due to large uncertainty in surface parameters. Using the relative adjoint sensitivity analysis method, we show that the brightness temperature of AHI channel 16 is much more sensitive to the low-tropospheric atmosphere than to the surface emissivity, similar to those high-level water vapour channels. We thus assimilated AHI channel 16 brightness temperature observations together with AHI water vapour channels over land, and assessed the added benefits on short-range quantitative precipitation forecasts for several convection-induced rainfall cases. Results show that adding channel 16 over land to AHI data assimilation further improves short-range rainfall forecasts. Assimilation of AHI channel 16 improves the upstream near surface atmospheric temperature analysis and influences the development of downstream precipitating weather systems.

Evaluation of the Weather Research and Forecasting Model Forecasts for Indian Summer Monsoon Rainfall of 2014 Using Ground Based Observations

The present study focuses on evaluating the ability of short-range rainfall forecasts from the Weather Research and Forecasting (WRF) model against observed gridded rainfall data from the India Meteorological Department (IMD) available over the Indian landmass during Indian summer monsoon (ISM) viz. June to September for the year 2014. The spatial distribution of the WRF predicted rainfall matches well with the IMD observed rainfall and indicates systematic rainfall biases over the Indian landmass. In general, precipitation is underestimated in dry, low elevation areas and overestimated in wet, high elevation areas by the WRF model. Based on rainfall verification scores, it was found that low and moderate rainfall was forecasted reasonably well by the WRF model compared to heavy rainfall. Moreover, extremal dependency score (EDS) has been used to verify heavy rainfall forecasts, as the traditional threat scores saturates towards high rainfall thresholds. Further, the separate analysis was carried out over five different homogeneous rainfall zones of India. Results show that the WRF model was able to predict rainfall reasonably well with higher correlation coefficient, lower bias and root mean square deviation in most of the zones. Moreover, the WRF rainfall forecasts at high spatial resolution (5 km) were also examined over Karnataka, India using dense rain gauge network.

Skill of Predicting Heavy Rainfall Over India: Improvement in Recent Years Using UKMO Global Model

The quantitative precipitation forecast (QPF) performance for heavy rains is still a challenge, even for the most advanced state-of-art high-resolution Numerical Weather Prediction (NWP) modeling systems. This study aims to evaluate the performance of UK Met Office Unified Model (UKMO) over India for prediction of high rainfall amounts (>2 and >5 cm/day) during the monsoon period (JJAS) from 2007 to 2015 in short range forecast up to Day 3. Among the various modeling upgrades and improvements in the parameterizations during this period, the model horizontal resolution has seen an improvement from 40 km in 2007 to 17 km in 2015. Skill of short range rainfall forecast has improved in UKMO model in recent years mainly due to increased horizontal and vertical resolution along with improved physics schemes. Categorical verification carried out using the four verification metrics, namely, probability of detection (POD), false alarm ratio (FAR), frequency bias (Bias) and Critical Success Index, indicates that QPF has improved by >29 and >24% in case of POD and FAR. Additionally, verification scores like EDS (Extreme Dependency Score), EDI (Extremal Dependence Index) and SEDI (Symmetric EDI) are used with special emphasis on verification of extreme and rare rainfall events. These scores also show an improvement by 60% (EDS) and >34% (EDI and SEDI) during the period of study, suggesting an improved skill of predicting heavy rains.

The use of indigenous knowledge systems for short and long range rainfall prediction and farmers’ perceptions of science-based seasonal forecasts in Zimbabwe

Purpose This paper aims to document indigenous knowledge systems (IKS) used for short- and long-range rainfall prediction by small holder farmers in three communities of Guruve District, in north-eastern Zimbabwe. The study also investigated farmers’ perceptions of contemporary forecasts and the reliability of both IKS and contemporary forecasts. Design/methodology/approach Data were collected among small holder farmers in Guruve District using household interviews and focus group discussions in three wards in the district, grouped according to their agro-climate into high and low rainfall areas. To get an expert view of the issues, key informant interviews were held with key agricultural extension personnel and traditional leaders. Findings Results obtained showed show high dependence on IKS-based forecasts in the district. Over 80 per cent of the farmers used at least one form of IKS for short- and long-range forecasting, as they are easily understood and applicable to their local situations. Tree phenology, migration and behaviour of some bird species and insects, and observation of atmospheric phenomena were the common indicators used. Tree phenology was the most common with over 80 per cent of farmers using this indicator. While some respondents (60 per cent) viewed forecasts derived from IKS as more reliable than science-based forecasts, 69 per cent preferred an integration of the two methods. Originality/value The simplicity and location specificity of IKS-based forecasts makes them potentially useful to smallholder farmers, climate scientists and policymakers in tracking change in these areas for more effective climate change response strategies and policymaking.

A nonlinear statistical ensemble model for short-range rainfall prediction

Following the practice of the numerical weather ensemble prediction, a nonlinear statistical ensemble prediction model has been developed based on a neural network technique with a Particle Swarm Optimization (PSO) algorithm. The model is validated by short-range climate forecasts of monthly mean rainfall at 37 stations in Guangxi, China during the first rainy season (April, May, and June). Independent prediction results show that the Particle Swarm Optimization Neural Network ensemble prediction model is clearly better than the traditional linear statistical method, such as the multiple regression method and the stepwise regression method. It is also suggested that by applying multiple ensemble members with each member objectively determined by the PSO algorithm, the generalization capacity of the ensemble prediction model is enhanced, demonstrating a vast range of possibilities for operational short-range climate prediction.

Doppler radar data assimilation with a local SVD-En3DVar method

An observation localization scheme is introduced into an ensemble-based three-dimensional variational (3DVar) assimilation method based on the singular value decomposition technique (SVD-En3DVar) to improve assimilation skill. A point-by-point analysis technique is adopted in which the weight of each observation decreases with increasing distance between the analysis point and the observation point. A set of numerical experiments, in which simulated Doppler radar data are assimilated into the Weather Research and Forecasting (WRF) model, is designed to test the scheme. The results are compared with those obtained using the original global and local patch schemes in SVD-En3DVar, neither of which includes this type of observation localization. The observation localization scheme not only eliminates spurious analysis increments in areas of missing data, but also avoids the discontinuous analysis fields that arise from the local patch scheme. The new scheme provides better analysis fields and a more reasonable short-range rainfall forecast than the original schemes. Additional forecast experiments that assimilate real data from 10 radars indicate that the short-term precipitation forecast skill can be improved by assimilating radar data and the observation localization scheme provides a better forecast than the other two schemes.

Uncertainty analysis of hydrological ensemble forecasts in a distributed model utilising short-range rainfall prediction

Abstract. Advances in mesoscale numerical weather predication make it possible to provide rainfall forecasts along with many other data fields at increasingly higher spatial resolutions. It is currently possible to incorporate high-resolution NWPs directly into flood forecasting systems in order to obtain an extended lead time. It is recognised, however, that direct application of rainfall outputs from the NWP model can contribute considerable uncertainty to the final river flow forecasts as the uncertainties inherent in the NWP are propagated into hydrological domains and can also be magnified by the scaling process. As the ensemble weather forecast has become operationally available, it is of particular interest to the hydrologist to investigate both the potential and implication of ensemble rainfall inputs to the hydrological modelling systems in terms of uncertainty propagation. In this paper, we employ a distributed hydrological model to analyse the performance of the ensemble flow forecasts based on the ensemble rainfall inputs from a short-range high-resolution mesoscale weather model. The results show that: (1) The hydrological model driven by QPF can produce forecasts comparable with those from a raingauge-driven one; (2) The ensemble hydrological forecast is able to disseminate abundant information with regard to the nature of the weather system and the confidence of the forecast itself; and (3) the uncertainties as well as systematic biases are sometimes significant and, as such, extra effort needs to be made to improve the quality of such a system.

A short-range quantitative precipitation forecast algorithm using back-propagation neural network approach

A back-propagation neural network (BPNN) was used to establish relationships between the short-range (0-3-h) rainfall and the predictors ranging from extrapolative forecasts of radar reflectivity, satellite-estimated cloud-top temperature, lightning strike rates, and Nested Grid Model (NGM) outputs. Quantitative precipitation forecasts (QPF) and the probabilities of categorical precipitation were obtained. Results of the BPNN algorithm were compared to the results obtained from the multiple linear regression algorithm for an independent dataset from the 1999 warm season over the continental United States. A sample forecast was made over the southeastern United States. Results showed that the BPNN categorical rainfall forecasts agreed well with Stage III observations in terms of the size and shape of the area of rainfall. The BPNN tended to over-forecast the spatial extent of heavier rainfall amounts, but the positioning of the areas with rainfall ⩾25.4 mm was still generally accurate. It appeared that the BPNN and linear regression approaches produce forecasts of very similar quality, although in some respects BPNN slightly outperformed the regression.

Implementing a physical initialization procedure in a regional spectral model: impact on the short-range rainfall forecasting over South America

Implementing a physical initialization procedure in a regional spectral model: impact on the short-range rainfall forecasting over South America

Implementing a physical initialization procedure in a regional spectral model: impact on the short-range rainfall forecasting over South America

In this paper we describe the implementation of a physical initialization procedure in the recently developed Florida State University nested regional spectral model, and its impact on short-term forecasting over South America. Because the regional model forecasts perturbations to the global model results, we seek to determine the impact of the boundary condition on the regional model assimilation. The regional model is able to assimilate the satellite-derived rain rates rather well, regardless of whether the global model, providing the boundary conditions, has been physically initialized. The regional model is able to assimilate higher-resolution precipitation data, and also the global model assimilates coarser resolution data as reported in earlier studies. Ten experiments are performed over the northern part of South America during the tropical rainy season of January and February 1999. The initial correlation coefficients of the rain rates exceed 0.9 for all physically initialized cases. The subsequent 24-h forecast is also improved, as measured by spatial correlation coefficients and equitable threat scores.

GANDOLF: a system for generating automated nowcasts of convective precipitation

AbstractDuring the past decade hydrologists have become increasingly aware of the problems of fluvial flood prediction during periods of intense convection, particularly in urbanised catchments whose rainfall‐runoff responses tend to be rapid. Existing approaches to deterministic, short‐range rainfall prediction are often deficient in their treatment of convective precipitation because they cannot resolve individual convective clouds or effectively model their evolution. In 1994 the UK Met. Office established a joint R&D programme with the Environment Agency (responsible for flood prediction in England and Wales) to explore the benefits of an Object‐Oriented conceptual Model (OOM) of convection in the nowcasting of fluvial floods. This involved the development of an automated nowcasting system (GANDOLF) designed to run the OOM during episodes of air mass convection. This paper describes the structure and function of the GANDOLF system and compares the performance of the OOM with that of two other precipitation models routinely used by Thames Region of the Agency. Copyright © 2000 Royal Meteorological Society

Prediction of a Flash Flood in Complex Terrain. Part I: A Comparison of Rainfall Estimates from Radar, and Very Short Range Rainfall Simulations from a Dynamic Model and an Automated Algorithmic System

Operational prediction of flash floods caused by convective rainfall in mountainous areas requires accurate estimates or predictions of the rainfall distribution in space and time. The details of the spatial distribution are especially critical in complex terrain because the watersheds generally are small in size, and position errors in the placement of the rainfall can distribute the rain over the wrong watershed. In addition to the need for good rainfall estimates, accurate flood prediction requires a surface-hydrologic model that is capable of predicting stream or river discharge based on the rainfall-rate input data. In part 1 of this study, different techniques for the estimation and prediction of convective rainfall are applied to the Buffalo Creek, Colorado, flash flood of July 1996, during which over 75 mm of rain from a thunderstorm fell on the watershed in less than 1 h. The hydrologic impact of the rainfall was exacerbated by the fact that a considerable fraction of the watershed experienced a wildfire approximately two months prior to the rain event. Precipitation estimates from the National Weather Service Weather Surveillance Radar-1988 Doppler and the National Center for Atmospheric Research S-band, dual-polarization radar, collocated east of Denver, Colorado, were compared. Very short range simulations from a convection-resolving dynamic model that was initialized variationally using the radar reflectivity and Doppler winds were compared with simulations from an automated algorithmic forecast system that also employs the radar data. The radar estimates of rain rate and the two forecasting systems that employ the radar data have degraded accuracy by virtue of the fact that they are applied in complex terrain. Nevertheless, the dynamic model and automated algorithms both produce simulations that could be useful operationally for input to surface-hydrologic models employed for flood warning. Part 2 of this study, reported in a companion paper, describes experiments in which these radar-based precipitation estimates and dynamic model‐ and automated algorithm‐based precipitation simulations are used as input to a surface-hydrologic model for simulation of the stream discharge associated with the flood.

Improving Precipitation Forecasts over the Global Tropical Belt

This study explores the nowcasting and short-range forecasting (up to 3 days) skills of rainfall over the tropics using a high resolution global model. Since the model-predicted rainfall is very sensitive to model parameters, four key model parameters were first selected. They are the Asselin filter coefficient, the fourth order horizontal diffusion coefficient, the surface moisture flux coefficient, and the vertical diffusion coefficient. The optimal values were defined as those which contributed to the best one day rainfall forecasts in the present study. In order to demonstrate and improve the precipitation forecast skill, several numerical experiments were designed using the 14-level Florida State University Global Spectral Model (FSUGSM) at a resolution of T106. Comparisons were also made of the short-range forecasts obtained from a control experiment subjected to normal mode initialization (NMI) versus experiments based on physical initialization (PI). The latter experiments were integrated using the original FSUGSM and a modified version. This modified FSUGSM was developed here by applying a reverse cumulus parameterization alorithm to the regular forecast model, which restructures the vertical humidity distribution and constrains the large-scale model’s moisture error growth during the model integration. An improved short-range rainfall prediction skill was achieved from the modified FSUGSM in this study. The results showed a better agreement between model-based and observed rainfall intensity and pattern.

Tangent linear and adjoint of “on-off” processes and their feasibility for use in 4-dimensional variational data assimilation

Problems in using the adjoint of moist physical processes that include “on-off” switches are analyzed via using both analytical examples and a numerical model. The analytic examples allow us to access such problems accurately since there exists the truth for both the model perturbation solution and the gradient of the cost function. The numerical model, which is the Penn State/NCAR nonhydrostatic mesoscale model version 5 (MM5), is used to test the “on-off” problems in a real-data environment. The SESAME 10—11 April 1979 storm case is chosen for the MM5 simulation since the forecast is strongly affected by moist diabatic processes. Analysis based on idealized continuous examples shows that if (i) the moist physics contains “on-off” switches, (ii) these switches are retained in the assimilation model, (iii) the tangent linear model, (TLM) and adjoint model follow the same route as in the NLM for the basic state around which the linearization is carried out (a way of constructing the TLM and adjoint model of a discretized numerical model), a sufficient condition for such developed TLM and adjoint model is that the perturbation solution of the NLM or the tendency equation is continuous at switching points. Otherwise, the accuracy of the TLM and adjoint model depends on the relative magnitude of the jump caused by “on-off” processes with respect to other terms in the tendency equations. However, numerical experiments using a discretized model including cumulus parameterization show that such errors, if any, do not seem to impact the validity of the TLM and the adjoint model for use in rainfall assimilation. The presence of “on-off” switches increased the degree of the nonlinearity more than that of the loss of accuracy in TLM and adjoint model, the former makes the TLM solution different from the true perturbation solution after a few hours of model integration. Three 4DVAR experiments are carried out in order to test the performance of the minimization procedure using numerical models which possess different “on-off” properties. The 4DVAR experiments are carried out over a 3-h window in which the initial precipitation associated with the Wichita Falls tornado outbreak took place. Assimilation of the 3-h rainfall observation produces a close fit to the observed 3-h rainfall during the assimilation window. Combination of the 3-h rainfall with the wind, temperature, surface moisture, and precipitable water produces a significant improvement on the short-range rainfall prediction. The presence of the “bad behavior” of the Kuo scheme with convection being turned on and off every other time step does not cause the minimization to fail, and similar results are obtained as far as the convergence rate and improvement to the model forecast are concerned.

Analysis of radar images for rainfall forecasting using neural networks

This paper describes a new approach to the analysis of weather radar data for short-range rainfall forecasting based on a neural network model. This approach consists in extracting synthetic information from radar images using the approximation capabilities of multilayer neural networks. Each image in a sequence is approximated using a modified radial basis function network trained by a competitive mechanism. Prediction of the rain field evolution is performed by analysing and extrapolating the time series of weight values. This method has been compared to the conventional cross-correlation technique and the persistence method for three different rainfall events, showing significant improvement in 30 and 60 min ahead forecast accuracy.

Short-range rainfall prediction model based on a physical method using three-dimensionally scanning Doppler radar

A physically based model which predicts rainfall distribution using three-dimensionally scanning Doppler radar is proposed.Most of short-range rainfall prediction models using three-dimensionally scanning radar are based on time-extrapolating method. Since physical processes of rainfall phenomenon are not included i those models, the accuracy of prediction is difficult to be more improved.So we offer the method which takes into account the water balance and its phase change in the atomosphere as well as the wind components estimated by VAD analysis using three-dimensionally scanning Doppler radar. The results from a numerical experiment indicate that this method is more effective for short-range rainfall prediction than time-extrapolating method.