Main Title Is Maximum of 14 Words,The Title Is Composed of Result/Method+Significant+Object, Informational, And Representing The Entire Content of The Writing: Book Antiqua, Font size 16, Bold

Abstract

Put here your Indonesian version or Arabic Version of the Abstract. Abstracts made between 100-250 words contain problems, objectives, methods, and results. Abstract is written using Book Antiqua font size 10, space 1 and with text length between 100-250 words. The Indonesian version of the abstract is written using standard Indonesian or Arabic with improved spelling.

Figure 1.

HTTPS://JOURNALS.UMS.AC.ID/INDEX.PHP/FG/ISSN: 0852-0682 | E-ISSN: 2460-3945

Investigation of the Development of Tropical Storm Nicholas based on Global and Regional Climate Data

Intan Nuni Wahyuni ², Ayu Shabrina ², Fadhil Lobma ¹, Arnida L. Latifah ^1,2*

1 School of Computing, Telkom University, Bandung, Indonesia

Table 1.

Citation: Wahyuni, I. N., Shabrina, A., Lobma, F., & Latifah, A.L. (2023) Investigation of the Development of Tropical Storm Nicholas based on Global and Regional Climate Data. Forum Geografi. Vol. xx, No. x.Article history: Received: dateAccepted: datePublished: dateCopyright: © 2022 by the authors. Submitted for possible open access publication under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

2 Research Center for Computing, The National Research and Innovation Agency of Indonesia, Bogor, Indonesia

*) Correspondence: arnidalatifah@telkomuniversity.ac.id

This paper studies the simulation of tropical storm Nicholas that occurred close to the coastal area of Western Australia and fell in the mainland of Southwestern Australia. The simulation was conducted by a dynamical downscaling model, Weather Research and Forecasting (WRF), to obtain a higher resolution of the regional climate data. The model simulation is generated using a global reanalysis of climate data for the initial and lateral boundary conditions. We investigated the tropical storm response to the model regarding the track and intensity using a modified Kyklop method that seems more appropriate for a landfall cyclone. Our study suggests that the regional climate data computed by the model deviates from the storm track of the global climate data forcing field. In this study, the track of the simulated storm is parallel to the satellite data, but it is slightly shifted to the east, closer to the mainland. Nevertheless, the model simulation can perform the storm intensity as strongly as the observation, while the forcing data gives substantial underestimation.

Keywords

Cyclone, dynamical downscaling, simulation, WRF, hazard, coastal

Introduction

A tropical cyclone is a weather phenomenon caused by atmospheric conditions, humidity, and sea surface temperature (Hsu et al., 2019). Cyclones are storms with great strength and wide radius. This phenomenon generally occurs in tropical oceans with warm temperatures (Sobel et al., 2021). Even though they mainly occur in the oceans, tropical cyclones have a dangerous impact that can cause damage in the sea to the land, such as strong winds, heavy rain, and extreme waves that disrupt shipping and can potentially sink ships at sea (Bakkensen & Mendelsohn, 2019). While on land, strong winds are deadly and can destroy buildings, vehicles, and other places in their path to rubble. In addition, tropical cyclones also cause storm surges or sea level rises, such as high tides that come suddenly and are dangerous when they reach land (Wright et al., 2019).

Tropical cyclones frequently occur in Australia’s oceans and impact the coastal and mainland region. Tropical cyclones Jacob and George occurred on 2-12 March 2007: one year later, Nicholas cyclone on 11-20 February 2008, and Marcus cyclone on 14-27 March 2018 (Ningsih et al., 2020). The Nicholas cyclone brought an enormous impact as it ended by hitting Western Australia's land (van Keulen, 2018). The effect was not only in Australia, but the Indonesia region also felt the impact (Ningsih et al., 2020). These storms occurred relatively close to Indonesian territory and have contributed to the emergence of waves and sea level rise on the southern coast of Indonesia (Ningsih et al., 2020).

Moreover, the Nicholas cyclone, which occurred in Western Australia, also caused high wave phenomena and sea level anomalies on the western coast of Sumatra and the southern coast of Java and Bali (Ningsih et al., 2020). The cyclone came suddenly and was dangerous as it fell on the mainland (Mortlock et al., 2023). According to the Australian government summary, the tropical storm of Nicholas is a severe typhoon with the wind speed increased to about 10 to 13 km/h.

Previous studies have investigated tropical cyclones for various purposes, such as Quaill et al. (2019) observed the experiences of people with physical disabilities before, during, and after a cyclone in Australia to provide input into implementing the disability inclusive disaster risk reduction (DiDRR) policy. In addition, Lenzen et al. (2019) investigated the economy-wide repercussions of the biophysical damage due to Tropical Cyclone Debbie in Australia using multi-region economic input–output (MRIO) analysis to determine the storm's impact on supply chains. However, Bell et al. (2022) determines TC-wind speed ARIs from a series of tropical cyclone data sets to predict the likelihood of extreme winds. Then, Bruyère et al. (2019) created a physically based landfall tropical cyclone scenario based on the occurrence of severe tropical cyclone Debby in Australia in 2017 to support the risk assessment. Likewise, Anushka et al. (2018) has facilitated typhoon disaster preparedness in the Wet Tropics bioregion of Australia by evaluating the individual adaptive capacity and social capital.

Meanwhile, Parker et al. (2018) has made simulations with the Weather Research and Forecasting (WRF) model to determine the impact of climate change on the features of the tropical cyclones in northeastern Australia. WRF allows researchers to generate simulations that reflect actual data. In addition, WRF provides a relatively flexible and robust operating system for forecasting (Abualkishik, 2018; Andraju et al., 2019; Yang et al., 2021). Then, WRF is widely applicable for research and weather forecast. Understanding the characteristic response of cyclones to a downscaling model such as WRF can improve the prediction quality, which may estimate the cyclone impacts more accurately. Therefore, in line with the work of Parker et al. (2018), this study investigates the cyclone response against the WRF model by reconstructing the tropical cyclone of Nicholas 2018 in Australia. We compare the development of the Nicholas cyclone based on the global climate data and the regional climate data obtained from the WRF simulation. To evaluate the path of the cyclone, we compare both climate data and the satellite data, in term of the track and intensity of the cyclone.

Research Method

This section covers this study's methods, the study domain's information, and the dataset for the numerical simulation. To obtain the regional climate data over the ocean and coast of Western Australia, we downscale the global climate data using a numerical weather prediction model, Weather Research and Forecasting (WRF) model. To trace the development of the cyclone, we modify Kyklop method so that it can be applied to investigate the track and the intensity of a landfall cyclone. Then, we evaluate tropical storm Nicholas's development from global and regional climate data against satellite data. The framework of this study can be seen in Figure 1.

Figure . Framework of the study.

2.1. WRF model

Numerical Weather Prediction (NWP) is a system that support research on atmosphere and weather forecasts. WRF model is one of the NWP that has been widely used because of its ease of use and relatively accurate results (Gaur et al., 2021). The WRF model is a compressible Euler equation, including a non-hydrostatic equation on a regional scale. This study implemented the non-hydrostatic equation in the WRF version 3.8.1. (Skamarock et al., 2008) downloaded from the official MMM NCAR website. The WRF model has been developed from the collaboration of the National Center for Atmospheric Research (NCAR), National Center for Environmental Prediction (NCEP), Forecast System Laboratory of the NOAA (NOAA/FSL), and Air Force Weather Agency (AFWA). Further details of the WRF model about the dynamics, primitive equations, numeric, and physics were presented by Klemp et al. (2007).

In this study, the WRF model was applied to downscale tropical storm of Nicholas from a general circulation model (GCM), i.e., NCEP FNL reanalysis data. The model simulation is configured on a 10 km spatial grid size. It is limited to 27 levels in the vertical direction and dimensions of 400 x 400 grid points in the longitude and latitude directions. The model applied tropical physics suite with the cloud microphysics scheme of Lin (Chen and Sun, 2002), cumulus parameterization scheme of Kain-Fritsch (Kain, 2004), planetary boundary layer (PBL) parameterization scheme of Mellor-Yamada-Janjic TKE (Janjić, 1994), and the surface layer scheme of Eta Similarity Scheme (Skamarock et al., 2008).

2.2. Kyklop method

To track the evolution of the cyclone, the recent method Kyklop is implemented with some adjustments. The main area of the tropical storm is detected where the wind speed is more extensive than 16.5 km/h, the temperature is higher than 298.15 K, and the surface pressure is less than 1000 Pa. These parameters are different from the parameter value used in Fuentes-Franco et al. (2017) and Latifah and Adytia (2019). The chosen parameters are based on the minimum error between the model result and the observation. The eye of the cyclone is an area with relatively low wind speed and cloudless located at the center of the cyclone. Unlike the Kyklop method (Fuentes-Franco et al., 2017) that detects the eye cyclone centered on the main area of the storm, we modified the method by excluding the eccentricity as it is less suitable for a landfall tropical cyclone. Instead, the eye cyclone is estimated by the maximum wind speed or the minimum sea level pressure.

2.3. Study Area

The study area is Western Australia which is located at latitudes 89°E to 130°E and longitudes 0°S to 38°S; see Figure 2. Australia has a varied climate with four seasons in most of its territory and a tropical climate in the north. Northern Australia is a tropical cyclone growth area that usually occurs between November and April. Australian region experiences an average of 11 cyclones a year, although only four to five cyclones make landfall that substantially impacts the environment, society, and economy (Dowdy, 2014). Most cyclones are formed in large cloudy areas because of the South Pacific Convergence Zone and the monsoon winds movement in southern Indonesia and northern Australia.

Figure 2. Domain for the WRF simulation.

2.4. Dataset

The geographical static data provided by Mesoscale and Microscale Meteorology (MMM) Laboratory, the National Center for Atmospheric Research (NCAR) was used for the WRF simulation, consisting of topography, land use and soil type. The atmospheric dataset from NCEP FNL, the National Centers for Environmental Prediction Final Operational Global Analysis is applied for the dynamic atmospheric data from 9-20 February 2008. The spatial resolution of the data is 10, with a time step of six hours for the lateral boundary condition. The atmospheric variables include air temperature, sea surface temperature, sea level and surface pressure, zonal and meridional winds, geopotential height, relative humidity, land characteristics, ice cover, vorticity, vertical motion, and ozone. For model validation, the simulation results are compared with the best track summary for Nicholas cyclone from Bureau of Meteorology in the Australian government. In addition, the spatial wind and sea level pressure pattern, the maximum wind, and the minimum sea level pressure are also compared with the dataset from KITAMOTO Asanobu, National Institute of Informatics (Kitamoto Laboratory, 2023).

Results and Discussion

3.1. Cyclone track

The tropical cyclone of Nicholas was a very slow-moving storm, averaging between 4 and 10 km/h on 16 February 2008, as summarized in the Australian Government report 2008. The report also explained that the wind speed increased to 10-13 km/h and the storm lasted for ten days. This is consistent with the cyclone reconstruction by the WRF model, which occurred from 11 to 20 February 2008. Figure 3 shows the track or path of the eye’s Nicholas cyclone movement. The formation of the Nicholas cyclone began on 11 February 2008 in the Indian sea in northeastern Australia. The storm’s path starts from coordinates of latitude 120° E and longitude 15°S, and then it moves towards the southwest along the outer coastline of Australia. It disappears in the land of Western Australia at latitude 115° E and longitude 25° S.

Figure 2.

Figure 3. The track of tropical storm Nicholas.

Besides comparing the cyclone track, we observe the spatial pattern changes of the wind speed and the sea level pressure during the storm on 16-19 February 2008 in Figure 4. From the satellite images in Figure 4, the storm is shown by moving large and thick clouds in a clockwise direction that is also shown in the wind direction in both WRF simulation and NCEP. The strong winds happened on 16-17 February 2008, then decreased on the next day. The storm moved downward to the southwest for three days and started dissipating on 19 February 2008 when the cloud started getting smaller. From the results of NCEP and WRF, the sea level pressure and wind speed form a vortex with a lower sea level pressure and a faster wind speed towards the eye of the storm. The WRF simulation shows more extreme wind speed over larger regions than NCEP. Moreover, the sea level pressure on the eye’s cyclone (indicated by the dark blue) from the WRF simulation is much lower than NCEP.

Figure 3.

Figure 4. Sea level pressure overlaid by the wind field of tropical storm Nicholas on 16-19 February 2008. From left to right: bi-linear interpolated NCEP, WRF simulation, satellite images.

3.2. Cyclone intensity

Figure 5 shows the wind speed changes, and Figure 6 presents the sea level pressure of tropical cyclone Nicholas based on the best track, NCEP, and WRF model simulation. Despite no significant changes in wind speed and sea level pressure during the storm in NCEP, the WRF model simulation presents comparable results with the observation. The maximum wind speed (40 km/h) and the minimum sea level pressure (950 hPa) during the storm can be approximately reached by the WRF model. Both variables have two days delay from the observation. The increase of the wind speed and the decrease of the sea level pressure in the best track started on 14 February 2008, while those changes began on 16 February 2008 in the WRF model simulations. Before 14 February 2008, NCEP, WRF, and observations' wind speed values were 15-20 km/h. Then, the WRF value and the observations increased and peaked at 40 km/h on 17 February 2008, but NCEP remained the same. The wind speed from WRF shows a better comparison with observation data at the beginning and the end of the simulation. This also complies with the study of Quitián-Hernández et al. (2021). On 14 – 17 February 2008, WRF showed a negative bias toward observation data.

Meanwhile, the sea level pressure starts around 1000 hPa and then decreases at the storm’s peak to 950 hPa for observation and 960 hPa for WRF on 17 February 2008 in Figure 6. The WRF model simulation shows an underestimation except in the first two days, with the highest bias being 30 hPa on 15 February 2008. In the first and last three days, the simulation can capture the intensity of the observation data.

Figure 4.

Figure 5. The maximum wind speed during the storm with the data observation (OBS) was retrieved from Kitamoto Laboratory (2023).

3.3. Discussion

Observing the cyclone track, the cyclone generated by the WRF model is similar to the path of NCEP and observation, although the WRF track shifts towards the east, which is closer to the mainland. The deviation track of WRF also happened in the study of tropical cyclone tracks in the Western North Pacific by Lui et al. (2021), in which the WRF model also found an underestimation of the speed and intensity. In our study case, the different track starts at the initial stage, but the track gap between the model and the observation gets smaller in the final stage of the cyclone. This observation is also found in the work of Munsi et al. (2021) in the simulation of the cyclone track by WRF, which shows some deviation at the initial period of the genesis stage compared with IBTrACS observation.

The study of Lui et al. (2021) showed that the WRF model underpredicts the intensity and its trend. The same behaviour is also observed in our study. The underestimation could be related to the low intensity in the forcing field, resulting in a gap between the simulation tracking and the best tracks mentioned above. The drawbacks of the WRF model were also found in the study of Huang et al. (2022), Chutla et al. (2019), and Kashem et al. (2019), in which the model showed good accuracy of the cyclone track, but it overestimated the cyclone intensities.

Other research has looked at the accuracy of the WRF model in forecasting tropical storm trajectories. For example, one research examined the direct positioning errors (DPEs) in WRF model low storm track predictions across the Philippine area and found that the model had displacement errors (Moon et al., 2021). Another research evaluated the global and regional WRF models in track errors for four typhoons during the 2011 season and found that the regional WRF model had lower track errors (Yu, 2022). As also stated in the study of Munsi et al. (2021), even though the WRF model simulation can reproduce the maximum stages of the cyclone well, the initial track should be carefully handled so that the deviation can be minimized. Not only are the consequences of the forcing data, but the deviations might also come from the sensitivity parameters of the model itself (Zhang et al.,2019).

As presented in this study and previous studies that downscaling approaches such as the WRF model give more comprehensive climatic information, significant errors can still occur. Then, further research to improve the robustness of the WRF model in simulating cyclone tracks and intensities is still required. One study should deeply examine the sensitivity between the cyclone track and intensity, particularly in the early stages, with the forcing field (Aragon & Pura, 2016) or with the model parameterization. Another study is required to investigate strategies to enhance the accuracy of downscaling approaches.

Conclusion

Substantial errors may still occur among the added values of a downscaling method in giving more details of climate information. This study shows the cyclone track deviation during the Nicholas cyclone on 9-20 February 2008 reconstructed by the Weather Research and Forecasting (WRF) model, while the forcing data NCEP gives a good track. However, the physical changes in sea level pressure and wind speed cannot be captured by NCEP. Even though there is a weak intensity at the initial stage of the cyclone in the WRF model simulation, the WRF model can reflect the condition of the tropical cyclone by showing a significant change in sea level pressure and wind speed a bit later. This weak intensity in the initial stage may consequence an initial error of the cyclone track because the track is only shifted to the east. At the same time, the pattern follows the observation correctly. The dependence between the cyclone track and intensity, especially in the initial stage, should be investigated further in future studies. It may relate to the forcing field.

Figure 5.

Figure 6. The sea level pressure during the storm. Data observation (OBS) was retrieved from Kitamoto Laboratory (2023).

Table 2.

AcknowledgementsThe simulation in this study had been done using the high-performance computing facilities of MAHAMERU, the National Research and Innovation Agency of Indonesia. We also acknowledge the provider of the dataset used in this paper, NCEP-FNL and KITAMOTO Asanobu, National Institute of Informatics.

References

Abualkishik, A.Z. (2018). A comparative study on the software architecture of wrf and other numerical weather prediction models. Journal of Theoretical and Applied Information Technology 96, 8244–8254.

Andraju, P., Kanth, A.L., Kumari, K.V., Vijaya Bhaskara Rao, S. (2019). Performance optimization of operational wrf model configured for Indian monsoon region. Earth Systems and Environment 3, 231–239.

Anushka, S., Amy, D., Georgina, G., Tristam, R. (2018). Perceptions of cyclone preparedness: Assessing the role of individual adaptive capacity and social capital in the wet tropics Australia. Sustainability 10, 1165.

Aragon, L. G. B., & Pura, A. G. (2016). Analysis of the displacement error of the WRF–ARW model in predicting tropical cyclone tracks over the Philippines. Meteorological Ap-plications, 23(3), 401-408.

Bakkensen, L. A., & Mendelsohn, R. O. (2019). Global tropical cyclone damages and fatalities under climate change: An updated assessment. Hurricane risk, 179-197.

Bell, S. S., Dowdy, A. J., Ramsay, H. A., Chand, S. S., Su, C. H., & Ye, H. (2022). Using historical tropical cyclone climate datasets to examine wind speed recurrence for coastal Australia. Scientific Reports, 12(1), 11612.

Bruyère, C.L., Done, J.M., Jaye, A.B., Holland, G.J., Buckley, B., Henderson, D.J., Leplastrier, M., Chan, P. (2019). Physically based landfalling tropical cyclone scenarios in support of risk assessment. Weather and Climate Extremes 26, 100229.

Chen, S.H., Sun, W.Y. (2002). A One-dimensional Time Dependent Cloud Model. Journal of the Meteorological Society of Japan. Ser. II 80, 99–118. doi:10.2151/jmsj.80.99.

Chutla, L., Pathak, B., Parottil, Ajay adn Bhuyan, P.K. (2019). Impact of microphysics parameterizations and horizontal resolutions on simulation of “mora” tropical cyclone over bay of bengal using numerical weather prediction model. Meteorology and Atmospheric Physics 131, 1483–1495.

Fuentes-Franco, R., Giorgi, F., Coppola, E., Zimmermann, K. (2017). Sensitivity of tropical cyclones to resolution, convection scheme and ocean flux parameterization over eastern tropical pacific and tropical north atlantic oceans in the regcm4 model. Climate dynamics 49, 547–561.

Gaur, A., Lacasse, M., Armstrong, M., Lu, H., Shu, C., Fields, A., Palou, F.S., Zhang, Y. (2021). Effects of using different urban parametrization schemes and land-cover datasets on the accuracy of wrf model over the city of ottawa. Urban Climate 35, 100737.

Hsu, W. C., Patricola, C. M., & Chang, P. (2019). The impact of climate model sea surface temperature biases on tropical cyclone simulations. Climate Dynamics, 53, 173-192.

Huang, C.Y., Jia-Yang, L., Kuo, H.C., Chen, D.S., Hong, J.S., Hsiao, L.F., Chen, S.Y. (2022). A numerical study for tropical cyclone atsani (2020) past offshore of southern taiwan under topographic influences. Atmosphere 13, 618.

Kashem, M., Ahmed, M.K., Akhter, M., Masud-Ul-Alam, M., Nahian, S., Loodh, R. (2019). Simulation of tropical cyclone viyaru using weather research and forecasting (wrf-arw) mesoscale model

Kitamoto Laboratory (2023). Digital Typhoon: Cyclone 200811 (Nicholas) - Detailed Track Information. Retrieved October 18, 2023, from http://agora.ex.nii.ac.jp/digital-typhoon/summary/wsp/l/200811.html.en

Latifah, A.L., Adytia, D. (2019). Effect of dynamical downscaling to cyclone simulation: a study case for haiyan typhoon, in: Journal of Physics: Conference Series, IOP Publishing. p. 012060.

Lenzen, M., Malik, A., Kenway, S., Daniels, P., Lam, K. L., & Geschke, A. (2019). Economic damage and spillovers from a tropical cyclone. Natural Hazards and Earth System Sciences, 19(1), 137-151.

Lui, Y.S., Tse, L.K.S., Tam, C.Y., Lau, K.H., Chen, J. (2021). Performance of mpas-a and wrf in predicting and simulating western north pacific tropical cyclone tracks and intensities. Theoretical and Applied Climatology 143, 505–520.

NCEP, F. (2000). National centers for environmental prediction/national weather service/noaa/us department of commerce. 2000, updated daily. ncep fnl operational model global tropospheric analyses, continuing from july 1999. research data archive at the national center for atmospheric research, computational and information systems laboratory.

Moon, J., Park, J., Cha, D. H., & Moon, Y. (2021). Five-day track forecast skills of WRF model for the western North Pacific tropical cyclones. Weather and Forecasting, 36(4), 1491-1503.

Mortlock, T. R., Nott, J., Crompton, R., & Koschatzky, V. (2023). A long-term view of tropical cyclone risk in Australia. Natural Hazards, 118(1), 571-588.

Munsi, A., Kesarkar, A., Bhate, J., Panchal, A., Singh, K., Kutty, G., & Giri, R. (2021). Rapidly intensified, long duration North Indian Ocean tropical cyclones: Mesoscale downscaling and validation. Atmospheric Research, 259(105678).

Ningsih, N.S., Hanifah, F., Tanjung, T.S., Yani, L.F., Azhar, M.A. (2020). The effect of tropical cyclone nicholas (11–20 february 2008) on sea level anomalies in indonesian waters. Journal of Marine Science and Engineering 8, 948.

Parker, C.L., Bruyère, C.L., Mooney, P.A., Lynch, A.H. (2018). The response of land-falling tropical cyclone characteristics to projected climate change in northeast australia. Climate dynamics 51, 3467–3485.

Quaill, J., Barker, R. N., & West, C. (2019). Experiences of people with physical disabilities before, during, and after tropical cyclones in Queensland, Australia. International journal of disaster risk reduction, 39, 101122.

Quitián-Hernández, L., Bolgiani, P., Santos-Muñoz, D., Sastre, M., Díaz-Fernández, J., González-Alemán, J.J., Farrán, J., Lopez, L., Valero, F., Martín, M. (2021). Analysis of the october 2014 subtropical cyclone using the wrf and the harmonie-arome numerical models: Assessment against observations. Atmospheric Research 260, 105697.

Sobel, A. H., Wing, A. A., Camargo, S. J., Patricola, C. M., Vecchi, G. A., Lee, C. Y., & Tippett, M. K. (2021). Tropical cyclone frequency. Earth's Future, 9(12), e2021EF002275.

van Keulen, M. (2018). Multiple climate impacts on seagrass dynamics: Amphibolis antarctica patches at Ningaloo Reef, Western Australia. Pacific Conservation Biology, 25(2), 211-212.

Vigh, J.L., Schubert, W.H., 2009. Rapid development of the tropical cyclone warm core. Journal of the Atmospheric Sciences 66, 3335 – 3350. doi:10.1175/2009JAS3092.1.

Wright, L. D., Resio, D. T., & Nichols, C. R. (2019). Causes and impacts of coastal inundation. Tomorrow's coasts: Complex and impermanent, 103-118.

Yang, Q., Yu, Z., Wei, J., Yang, C., Gu, H., Xiao, M., Shang, S., Dong, N., Gao, L., Arnault, J., et al. (2021). Performance of the wrf model in simulating intense precipitation events over the hanjiang river basin, china–a multi-physics ensemble approach. Atmospheric Research 248, 105206

Yu, J. (2022). Numerical tests for tropical cyclone track prediction by the global WRF model. Tropical Cyclone Research and Review, 11(4), 252-264.

Zhang, M., Perrie, W., & Long, Z. (2019). Sensitivity study of North Atlantic summer cyclone activity in dynamical downscaled simulations. Journal of Geophysical Research: Atmospheres,124, 7599–7616