WRF Simulations Research Articles

Simulating the characteristics of cut-off low rainfall over the Western Cape using WRF

Cut-off lows (COLs) cause hazardous weather but also play a crucial role in the annual rainfall of the Western Cape, especially during a drought year. However, there is a dearth of information on the capability of atmospheric models to reproduce the characteristics of COLs over this region. This study evaluates the capability of a regional climate model (WRF) in simulating COLs over the Western Cape, with emphasis on the drought periods. Observation, reanalysis and simulations datasets were analysed for the study. The simulated climatology and inter-annual variability of COL with the associated patterns were compared to the reanalysis and observation results. The Self Organising Map (SOM) was used to group the observed and simulated COL rainfall to similar patterns. The results of the analysis show that WRF captures the seasonal and annual climatology of COL and the associated rainfall, but the model struggles to simulate the inter-annual variability of the systems. The model reproduces all the COL rainfall patterns well, though it under-estimates the frequency of dry COLs. However, WRF simulation agrees with the reanalysis that wetter COLs over the Western Cape are associated with more transport of warm, moist air from the tropics. The results of the study have application in improving weather and seasonal forecasting of COLs over the Western Cape.

Move a Tropical Cyclone with 4D-Var and Vortex Dynamical Initialization in WRF Model

Previous studies showed that 4D-Var technique used for data assimilation could be modified for weather control. This study demonstrates the ability of 4D-Var to influence the future path of a tropical cyclone by calculating perturbations in WRF simulation. Given the background error covariance matrix, the initial field is improved by the vortex dynamic initialization technique. Our results show that 4D-Var can be applied to control the trajectory of simulated tropical cyclones by producing "optimal" perturbations. In the numerical simulation experiment of Typhoon Mitag in 2019, after this kind of weather control similar to data assimilation, the tropical cyclone moved obviously, and the damaging wind over the coastline weakened. The prediction results after the initial field modified by 4D-Var have a great change, and the position of the tropical cyclone moved about 0.5° southeastward after assimilation, which misses the southeast coast of China. Moreover, the damaging wind is also weakened. Since the 4D-Var is premised on the assumption that the model is perfect and does not consider the model error, then the research plan to consider model error and introduce new methods is discussed in the paper.

Accuracy assessment of heavy rainfall forecasting in the Western Caucasus with ICON-Eu regional atmospheric model for short-term flood forecasting

Heavy rainfalls and rain-induced floods regularly cause substantial damage in the Western Caucasus region. In this study, we assess the accuracy of heavy rainfall forecasting with ICON-Eu regional atmospheric model (developed by the weather service of Germany) and the possibility of its use as input data for rain flood prediction. The main conclusion is that the forecast accuracy is strongly determined by the nature of the rainfall in question (mainly convective or triggered by synoptic-scale processes) and the season of the year. The ICON-Eu model systematically underestimates (by 2-3 times) the precipitation amount in the warm season (April-September) and almost never reproduces local convective heavy rainfall events. Therefore, its forecasts for short-term prediction of summertime rain floods have low efficiency. On the other hand, in the cold season (October-March) the model adequately reproduces heavy precipitation events, with some underestimation of the maximum precipitation amount and overestimation of the coverage area. These forecasts can be used to improve the short-term prediction of flash floods on the rivers of the Black Sea coast of the Caucasus during the period from October to March. In addition, we have performed a detailed analysis of the heavy rainfall and a related flood event which occurred on August 17-18, 2019 by comparing hourly observed precipitation with ICON-Eu and Cosmo-Ru forecasts, as well as applying WRF simulation and rainfall-runoff modelling.

Application of the WRF model to the coastal area at Ise Bay, Japan: evaluation of model output sensitivity to input data

ABSTRACT WRF simulations were conducted for Ise Bay, Japan for January and July 2016 to evaluate sensitivity of model input and output above sea surface to the replacement of three default input datasets with region-specific input datasets. For atmospheric input data, a final analysis created by the National Centers for Environmental Prediction (NCEP-FNL) was replaced with a mesoscale analysis created by the Japan Meteorological Agency (JMA). Topography and land use dataset released by the US Geological Survey were replaced with dataset released by the GeoSpatial Information authority of Japan. For sea surface temperature (SST) data, NCEP-FNL was replaced with an analysis created by JMA. Of the three region-specific datasets, replacement of atmospheric data results in the largest improvements in the accuracy of simulated wind speeds in January and July and of temperature in July 2016. Improvements in model output accuracy over sea surface can be seen near the coastline by replacing topography and land use data. Replacement of SST data results in the largest improvements in simulated temperature accuracy in January 2016. Replacing all three default input datasets results in the largest improvement, and expands on results from previous studies that focused on the effects of replacing only one input data.

Observation and Simulation of a Bifurcating Thunderstorm over Beijing

AbstractThe aim of this study was the analysis and simulation of the life cycle of a bifurcating thunderstorm that passed over Beijing, China, on 22 July 2015. Data from 150 surface weather sites and an S-band radar were used in conjunction with WRF simulations that used its multilevel Building Environment Parameterization (BEP) urbanization option. The Urban-case simulation used Beijing land-use information, and the NoUrban one replaced all urban areas by croplands. The Urban case correctly simulated both the observed weak 10-m winds over Beijing (<1.0 m s−1) and the weak 2-m urban heat island (<0.5°C). Observed radar and rain gauge data, as well as the Urban-case results, all showed precipitation bifurcation around Beijing, with maximum accumulations in convergent flow areas on either side of the city. The Urban case also reproduced the observed precipitation minima over the urban area and in a downwind rain shadow. The observations and Urban-case results both also showed bifurcated flow, even when the storm was still upwind of Beijing. The subsequent bifurcated precipitation areas thus each moved along a preexisting flow branch. Urban-case vertical sections showed downward motion in the divergence areas over the urban core and upward motions over the lateral convergence zones, both up to 6 km. Given that the NoUrban case showed none of these features, these differences demonstrate how the impact of cities can extend upward into deep local convection. Additional case-study simulations are needed to more fully understand urban storm bifurcation mechanisms in this and other storms for cities in a variety of climates.

An Examination of the Impact of Grid Spacing on WRF Simulations of Wintertime Precipitation in the Mid-Atlantic United States

AbstractA large set of deterministic and ensemble forecasts was produced to identify the optimal spacing for forecasting U.S. East Coast snowstorms. WRF forecasts were produced on cloud-allowing (~1-km grid spacing) and convection-allowing (3–4 km) grids, and compared against forecasts with parameterized convection (>~10 km). Performance diagrams were used to evaluate 19 deterministic forecasts from the winter of 2013–14. Ensemble forecasts of five disruptive snowstorms spanning the years 2015–18 were evaluated using various methods to evaluate probabilistic forecasts. While deterministic forecasts using cloud-allowing grids were not better than convection-allowing forecasts, both had lower bias and higher success ratios than forecasts with parameterized convection. All forecasts were underdispersive. Nevertheless, forecasts on the higher-resolution grids were more reliable than those with parameterized convection. Forecasts on the cloud-allowing grid were best able to discriminate areas that received heavy snow and those that did not, while the forecasts with parameterized convection were least able to do so. It is recommended to use convection-resolving and (if computationally possible) to use cloud-allowing forecast grids when predicting East Coast winter storms.

Regional surface temperature simulations over the Iberian Peninsula: evaluation and climate projections

The realism of a specific configuration of the WRF Regional Climate Model (RCM) to represent the observed temperature evolution over the Iberian Peninsula (IP) in the 1971–2005 period has been analyzed. The E-OBS observational dataset was used for this purpose. Also, the textit{added value} of the WRF simulations with respect to the IPSL Earth System Model (ESM) used to drive the WRF RCM was evaluated. In general, WRF presents lower temperatures than in the observations (negative biases) over the IP. These biases are comparatively larger than those of the driving ESM. Once the biases are corrected, WRF provides an added value in terms of a higher spatial representation. WRF introduces more variability in some regions in comparison to gridded observation. Warming trends according to the observations are also well represented by the RCM. In the second part of this study, the projections of future climate performed with both the ESM and the RCM were evaluated for the RCP4.5 and RCP8.5 scenarios during the 21st century. Although both models simulate temperature increases, the RCM simulates a smaller warming than the ESM after the mid-21st century, except for winter. Using the WRF model, the maximum temperature increase reaches 6, ^circ hbox {C} and 3, ^circ hbox {C} for RCP8.5 and RCP4.5 in the south east of the Iberian Peninsula by the end of the 21st century, respectively.

IEA Wind Task 31: Design of a new comparison metrics simulation challenge for wind resource assessment in complex terrain Stage 1

In wind energy, the accuracy of the estimation of the wind resource has an enormous effect on the expected rate of return of a project. For a given project, the modeller is faced with a difficult choice of a wide range of simulation tools with varying accuracies and costs. In previous work, a new method for estimating the skill and cost scores of different wind modelling tools for a given project has been developed. Although this method worked well, it was shown that further studies are required for a wide range of input conditions and project types in order to develop project-specific transfer functions between the predicted and actual cost and skill scores. In this paper therefore, a new simulation challenge is designed with the goal of collecting comparison metrics data regarding the skill and cost scores of a range of different simulation tools for a complex terrain site, both before and after carrying out the simulations. The complex terrain site Perdigao is chosen for the challenge, due to the volume and quality of available measurement data and the complexity of the terrain. An initial data analysis and WAsP simulations allow mast 29 to be chosen for the input data and masts 7, 10, 20, 22, 25, 27, 34 and 37 for the validation data. The WAsP simulations are compared to previous WRF simulations and are found to capture the main features of the flow over the two ridges. The wind acceleration over the two ridges and the resulting maximum wind speeds at the peak of the ridges as well as the reduced velocity regions in the valley between the ridges are well captured. However, smaller-scale features such as small areas of separated flow seen in the WRF simulations cannot be captured by WAsP. This choice of mast for the challenge will therefore allow the capabilities of different tools for calculating flow in separated regions as well as on top of hills to be assessed.

Microscale CFD Simulations of a Wind Energy Test Site in the Swabian Alps with Mesoscale Based Inflow Data

The current study describes analyses of the WINSENT wind energy test site located in complex terrain in Southern Germany by highly resolved numerical simulations. The resolved atmospheric turbulence is simulated with Delayed Detached Eddy Simulations by the flow solver FLOWer without consideration of the research wind turbines.The mean inflow and wind direction of the analysed time period is provided by precursor simulations of project partners. The simulation model chain consists of three codes with different time scales and resolutions. The model chain provides a data transfer from mesoscale WRF simulations to OpenFOAM. As a next step OpenFOAM provides inflow data in the valley of the terrain site for the present FLOWer simulations, the code with the highest resolution in space and time. The mean velocity field provided by OpenFOAM is superimposed with fluctuations that are based on measurements to obtain the small turbulent scales within the FLOWer simulations, which the previous tools of the model chain can not resolve. Comparisons with the two already installed met masts clarify that the current FLOWer simulations provide an adequate agreement with measured data. The results are verified with the application of a second simulation, in which a homogeneous velocity profile is superimposed with turbulence. Thus, comparisons with measured data showed that the benefit of using the inflow data of this model chain is especially evident near the ground.

Can summer monsoon moisture invade the Jade Pass in Northwestern China?

Heavy precipitation events are increasingly concerned because their significant contribution to annual precipitation in the Northwestern China, which might be related to invasion of summer monsoon moisture. It is interest whether or not the same is Jade Pass as being outside the control of the Asian summer monsoon. In this work, six heavy precipitation events were selected based on the 95 percentiles of the daily precipitation at the 12 weather stations around the Jade Pass from 1970–2000, with consideration of the influences of elevation. The event on June 19th, 2013 was chosen for a detailed examination due to the fact that the day has a large-scale precipitation as revealed by a gridded precipitation dataset over a large region. Using a Weather Research and Forecasting Model (WRF) simulation with high spatiotemporal resolution and in situ isotopic tracing (δ18O, δD), under a large-scale heavy precipitation event, this study provides ambitious view at the synoptic scale. A dramatic decrease in the δ18O, δD and deuterium (d)-excess of precipitation, very high relative humidity (98%), and reduced air temperature indicate that the precipitation was a result of long-distance-transported monsoon vapor. In addition, the slope of the local water meteoric line (LWML) of the precipitation for this event was very close to that of the global meteoric water line (GWML), indicating the source of moisture was from the ocean. Meanwhile, the WRF simulation confirms that the precipitation at the Jade Pass was not caused by local convection, but by summer monsoon. Both WRF simulation and isotopic tracing support the view that the monsoon moisture could invade Jade Pass at the synoptic scale and impact on precipitation, which need be further investigated.

Highly resolved WRF-BEP/BEM simulations over Barcelona urban area with LCZ

This study evaluates the performance of urban schemes integrated in the Weather Research and Forecasting model (WRF) using Local Climate Zones (LCZ) as land use classification. We applied two multi-layer urban schemes: 1) Building Effect Parameterization (BEP) and 2) Building Energy Model coupled with BEP (BEP + BEM), over the Metropolitan Area of Barcelona (MAB) at 1km2 horizontal resolution for July 2016. These two simulations were compared with observations and a standard WRF simulation (BULK approach). Corine Land Cover 2012 provides background information for the entire simulation domain, while the LCZ covers MAB classifying the land cover into 10 classes according to urban morphology and thermal properties.BULK and multi-layer urban scheme experiments present a similar general error trend: overestimation of relative humidity and planetary boundary layer height and underestimation of temperature. Although BEP has the best correlation with observations, this is the scheme with the highest value of bias and RMSE for temperature and relative humidity, in particular during the night/morning. On the other hand, BEP + BEM performed with the minimum RMSE associated for temperature and relative humidity in the entire domain. BEP + BEM has shown to be more sensitive than the other schemes over locations where the land use in the model grid differs to the real one, which is a common consequent limitation of horizontal model resolution. This study also suggests that depending on the synoptic condition the scheme accuracy on determining PBLH might change considerably.

Electrical evolution of a rapidly developing MCS during its vigorous vertical growth phase

This study examined the coevolution of storm structure and electrical behavior of a rapidly developing mesoscale convective system (MCS) over Beijing metropolitan region during its vigorous growth phase based on observations and model simulations. When the isolated convection grew upscale and developed into a larger and organized MCS, it is found that the lightning production increased much faster than storm volume growth. Although the overall MCS went through rapid intensification, large differences exist in the radar reflectivity pattern of its convective cell members indicating their differential convective states. Several distribution peaks of lightning radiation pulses are also observed in the vertical direction, suggesting the overall complicated charge distribution inside the MCS. Given the decreased distances among cells during the upscale growth, it is hypothesized that such complicated charge distributions are more conducive to local high electric fields for lightning initiations and hence increased lightning production rates. The electrification-coupled WRF simulations support this proposed hypothesis, which clearly show that the MCS evolves from two layered charge regions to more complicated charge distributions. When some convective cells consisting of the MCS develop toward maturity, the precipitation particles descend to low levels and give rise to a lower positive charge region that is embedded inside a deep-layer negative charge region. In combination with enhanced interactions among the cells, a peak of electric field magnitude is generated at lower levels due to the approaching charge regions with opposite polarities, such that more lightning flashes are produced.

Gargantuan Hail in Argentina

AbstractOn 8 February 2018, a supercell storm produced gargantuan (>15 cm or >6 in. in maximum dimension) hail as it moved over the heavily populated city of Villa Carlos Paz in Córdoba Province, Argentina. Observations of gargantuan hail are quite rare, but the large population density here yielded numerous witnesses and social media pictures and videos from this event that document multiple large hailstones. The storm was also sampled by the newly installed operational polarimetric C-band radar in Córdoba. During the RELAMPAGO campaign, the authors interviewed local residents about their accounts of the storm and uncovered additional social media video and photographs revealing extremely large hail at multiple locations in town. This article documents the case, including the meteorological conditions supporting the storm (with the aid of a high-resolution WRF simulation), the storm’s observed radar signatures, and three noteworthy hailstones observed by residents. These hailstones include a freezer-preserved 4.48-in. (11.38 cm) maximum dimension stone that was scanned with a 3D infrared laser scanner, a 7.1-in. (18 cm) maximum dimension stone, and a hailstone photogrammetrically estimated to be between 7.4 and 9.3 in. (18.8–23.7 cm) in maximum dimension, which is close to or exceeds the world record for maximum dimension. Such a well-observed case is an important step forward in understanding environments and storms that produce gargantuan hail, and ultimately how to anticipate and detect such extreme events.

Realistic Simulation of Tropical Atmospheric Gravity Waves Using Radar-Observed Precipitation Rate and Echo Top Height.

Gravity waves (GWs) generated by tropical convection are important for the simulation of large‐scale atmospheric circulations, for example, the quasi‐biennial oscillation (QBO), and small‐scale phenomena like clear‐air turbulence. However, the simulation of these waves still poses a challenge due to the inaccurate representation of convection, and the high computational costs of global, cloud‐resolving models. Methods combining models with observations are needed to gain the necessary knowledge on GW generation, propagation, and dissipation so that we may encode this knowledge into fast parameterized physics for global weather and climate simulation or turbulence forecasting. We present a new method suitable for rapid simulation of realistic convective GWs. Here, we associate the profile of latent heating with two parameters: precipitation rate and cloud top height. Full‐physics cloud‐resolving WRF simulations are used to develop a lookup table for converting instantaneous radar precipitation rates and echo top measurements into a high‐resolution, time‐dependent latent heating field. The heating field from these simulations is then used to force an idealized dry version of the WRF model. We validate the method by comparing simulated precipitation rates and cloud tops with scanning radar observations and by comparing the GW field in the idealized simulations to satellite measurements. Our results suggest that including variable cloud top height in the derivation of the latent heating profiles leads to better representation of the GWs compared to using only the precipitation rate. The improvement is especially noticeable with respect to wave amplitudes. This improved representation also affects the forcing of GWs on large‐scale circulation.

Observations and numerical simulations of the October 25th, 2015 super-cell thunderstorm over Central Israel

An exceptionally rare super-cell thunderstorm occurred on the morning hours of October 25th 2015 in the Eastern Mediterranean. The storm developed off the Egyptian coastline near Alexandria and moved north-west, crossing the Israeli coast just north of Tel-Aviv at 0700 UTC. Rapidly developing, deep convective cells, with thunderclouds exhibiting cloud top temperatures colder than −70 °C (17 km) and radar reflectivity cores >65 dBz at 12 km, were accompanied by intensive lightning activity, severe hail, downbursts and massive rain rates. Between 0500 and 0900 UTC, more than 28,440 cloud-to-ground lightning strokes were registered by the lightning detection system of the Israeli Electrical Corporation. Additionally, weak strokes detected with peak currents lower the ±5 kA were classified as intracloud, and constituted additional 29,878 strokes; the storm's total for the 4-h period was 58,318 strokes, exceeding the total annual amount for the entire country. The maximum total stroke rate between 0700 and 0800 UTC exceeded 500 strokes per minute. Synoptic analysis and WRF simulations suggest that electrification processes were enhanced through a combination of dynamic factors and dust induced invigoration of microphysical processes, leading to the prodigious amount of lightning.

Model sensitivities in the case of high-resolution Eulerian simulations of local tephra transport and deposition

Tephra transport modelling has predominantly been carried out over synoptic-scale resolutions, resolving only fundamental atmospheric circulations. Consequently, model sensitivity studies have focused on these resolutions. In recent years, improved eruption source parameter (ESP) estimation methods and computational efficiency have progressively led to the use of higher resolutions to increase the fidelity of tephra transport simulations. This is a new computational regime whose model sensitivities remain largely unexplored. Here, high-resolution simulations using a 300 m horizontal grid spacing were carried out using WRF and FALL3D to assess sensitivity to: meteorological input data (using both reanalysis and forecast data), boundary layer parametrisation, plume model, settling velocity, total grain-size distribution (TGSD), and aggregation, leading to a total of 1280 possible configurations. An eruption that occurred on 28 July 2019 at Sakurajima volcano, Japan, is used as a case study, as the northwards dispersal of a 3.8 km-high plume led to the temporary closure of Kagoshima airport. Sedimentation was observed at three optical disdrometers on the volcano, providing insight on the deposition timing and particle size distribution. Results reveal that meteorological input data can have a significant impact over the fidelity of the simulations, with errors changing over one order of magnitude depending on the initialisation dataset. This was found to be the result of local bias in the simulated wind fields over the region of dispersal, which was not represented in the domain-wide assessment of the WRF simulations. Ensemble simulations using different meteorological input data were found to mitigate this, significantly increasing the fidelity of the final ensemble product. Use of a plume model and the correct estimation of the TGSD were seen to significantly improve results. Finally, even with detailed data to inform the choice, use of an empirical aggregation scheme was seen to significantly degrade the accuracy of the simulations.

Investigating Future Urbanization’s Impact on Local Climate under Different Climate Change Scenarios in MEGA-urban Regions: A Case Study of the Pearl River Delta, China

Urbanization is one of the most significant contributing factors to anthropogenic climate change. However, a lack of projected city land use data has posed significant challenges to factoring urbanization into climate change modeling. Thus, the results from current models may contain considerable errors in estimating future climate scenarios. The Pearl River Delta region was selected as a case study to provide insight into how large-scale urbanization and different climate change scenarios impact the local climate. This study adopts projected land use data from freely available satellite imagery and applies dynamic simulation land use results to the Weather Research and Forecasting Model (WRF). The simulation periods cover the summer periods in 2010 and 2029–2031, the latter of which is averaged to represent the year 2030. The WRF simulation used the observed local climate conditions in 2010 to represent the current scenario and the projected local climate changes for 2030 as the future scenario. Under all three future climate change scenarios, the warming trend is prominent (around 1–2 °C increase), with a widespread reduction in wind speed in inland areas (1–2 ms−1). The vulnerability of human health to thermal stress was evaluated by adopting the wet-bulb globe temperature (WBGT). The results from the future scenarios suggest a high public health risk due to rising temperatures in the future. This study provides a methodology for a more comprehensive understanding of future urbanization and its impact on regional climate by using freely available satellite images and WRF simulation tools. The simulated temperature and WBGT results can serve local governments and stakeholders in city planning and the creation of action plans that will reduce the potential vulnerability of human health to excessive heat.

Internal variability versus multi‐physics uncertainty in a regional climate model

AbstractIn a recent study, Coppola et al. assessed the ability of an ensemble of convection‐permitting models (CPM) to simulate deep convection using three case studies. The ensemble exhibited strong discrepancies between models, which were attributed to various factors. In order to shed some light on the issue, we quantify in this article the uncertainty associated to different physical parameterizations from that of using different initial conditions, often referred to as the internal variability. For this purpose, we establish a framework to quantify both signals and we compare them for upper atmospheric circulation and near‐surface variables. The analysis is carried out in the context of the CORDEX Flagship Pilot Study on Convective phenomena at high resolution over Europe and the Mediterranean, in which the intermediate RCM WRF simulations that serve to drive the CPM are run several times with different parameterizations. For atmospheric circulation (geopotential height), the sensitivity induced by multi‐physics and the internal variability show comparable magnitudes and a similar spatial distribution pattern. For 2‐m temperature and 10‐m wind, the simulations with different parameterizations show larger differences than those launched with different initial conditions. The systematic effect over 1 year shows distinct patterns for the multi‐physics and the internal variability. Therefore, the general lesson of this study is that internal variability should be analysed in order to properly distinguish the impact of other sources of uncertainty, especially for short‐term sensitivity simulations.

Orographic effects on the propagation and rainfall modification associated with the 2007–08 Madden–Julian oscillation (MJO) past the New Guinea Highlands

Based on the tropical rainfall measuring mission (TRMM)-measured rainfall and estimated outgoing longwave radiation (OLR) fields, it is found that 2007–08 Madden–Julian Oscillation (MJO07-08) went through blocking, splitting, and merging stages when it passed over the New Guinea Highlands (NGH). The TRMM-estimated OLR fields fail to capture detailed TRMM rainfall field and thus is not suitable to serve as proxy for rainfall, as also found in previous studies. The mechanism of orographic blocking is explained by strong orographic blocking on the incoming, low-Froude number, and moist flow, which belonged to the flow-around regime. This is evidenced by estimating the Froude number by upstream soundings. The strong blocking forced the flow to go around the mountains on NGH, leading to the splitting of flow and MJO precipitating system and the merging at the southeast tip of New Guinea. Orographic, MJO, and cyclone clouds were shown in both observed and model-simulated results. The major differences of the model-simulated and TRMM-measured precipitation are as follows: (a) the model-simulated rainfall area is much larger than that covered by the observed rainfall and (b) even though they both show comparable maximum rainfall rate, the rainfall estimated by TRMM reveals more localized rainfall spots, which is unexpected since the WRF simulation uses a relatively fine resolution (5 km). In summary, during the blocking stage, the mountains have slowed down the MJO propagation and increased the rainfall amount upstream of the local mountains, while during the splitting and merging stages, the mountains have made significant impacts on the MJO rainfall distribution.

Improving wind speed forecasts from the Weather Research and Forecasting model at a wind farm in the semiarid Coquimbo region in central Chile

AbstractAccurate predictions of the wind field are key for better wind power forecasts. Wind speed forecasts from numerical weather models present differences with observations, especially in places with complex topography, such as the north of Chile. The present study has two goals: (a) to find the WRF model boundary layer (PBL) scheme that best reproduces the observations at the Totoral Wind Farm, located in the semiarid Coquimbo region in north‐central Chile, and (b) to use an artificial neural network (ANN) to postprocess wind speed forecasts from different model domains to analyze the sensitivity to horizontal resolution. The WRF model was run with three different PBL schemes (MYNN, MYNN3, and QNSE) for 2013. The WRF simulation with the QNSE scheme showed the best agreement with observations at the wind farm, and its outputs were postprocessed using two ANNs with two algorithms: backpropagation (BP) and particle swarm optimization (PSO). These two ANNs were applied to the innermost WRF domains with 3‐km (d03) and 1‐km (d04) horizontal resolutions. The root‐mean‐square errors (RMSEs) between raw WRF forecasts and observations for d03 and d04 were 2.7 and 2.4 ms−1, respectively. When both ANN models (BP and PSO) were applied to Domains d03 and d04, the RMSE decreased to values lower than 1.7 ms−1, and they showed similar performances, supporting the use of an ANN to postprocess a three‐nested WRF domain configuration to provide more accurate forecasts in advance for the region.