Abstract
This study aims at estimating the inherent variability of microscale boundary-layer flows and its impact on air pollutant dispersion in urban environments. For this purpose, we present a methodology combining high-fidelity large-eddy simulation (LES) and a stationary bootstrap algorithm, to estimate the internal variability of time-averaged quantities over a given analysis period thanks to sub-average samples. A detailed validation of an LES microscale air pollutant dispersion model in the framework of the Mock Urban Setting Test (MUST) field-scale experiment is performed. We show that the LES results are in overall good agreement with the experimental measurements of wind velocity and tracer concentration, especially in terms of fluctuations and peaks of concentrations. We also show that both LES estimates and the MUST experimental measurements are subject to significant internal variability, which is therefore essential to take into account in the model validation. Moreover, we demonstrate that the LES model can accurately reproduce the observed internal variability.
Similar content being viewed by others
Notes
AVBP documentation, see https://www.cerfacs.fr/avbp7x/.
References
Antonioni G, Burkhart S, Burman J, Dejoan A, Fusco A, Gaasbeek R, Gjesdal T, Jäppinen A, Riikonen K, Morra P, Parmhed O, Santiago J (2012) Comparison of CFD and operational dispersion models in an urban-like environment. Atmos Environ 47:365–372. https://doi.org/10.1016/j.atmosenv.2011.10.053
Armand P, Duchenne C (2022) 3d multi-scale weather and dispersion models applied to assess the impact of industrial plants on human health and the environment. In: Air pollution modeling and its application xxviii. Springer, Berlin, pp 51–63. https://doi.org/10.1007/978-3-031-12786-1_7
Baklanov A (2000) Application of CFD methods for modelling in air pollution problems: possibilities and gaps. Environ Monit Assess 65(1):181–189. https://doi.org/10.1023/A:1006442514766
Biltoft C (2001) Customer report for Mock Urban Setting Test. West Desert Test Center, U.S. Army Dugway Proving Ground, Utah, USA, DPG Document No. WDTC-FR-01-121
Blocken B, Gualtieri C (2012) Ten iterative steps for model development and evaluation applied to computational fluid dynamics for environmental fluid mechanics. Environ Model Softw 33:1–22. https://doi.org/10.1016/j.envsoft.2012.02.001
Britter RE, Hanna SR (2003) Flow and dispersion in urban areas. Ann Rev Fluid Mech 35(1):469–496. https://doi.org/10.1146/annurev.fluid.35.101101.161147
Calaf M, Parlange MB, Meneveau C (2011) Large eddy simulation study of scalar transport in fully developed wind-turbine array boundary layers. Phys Fluids 23(12):126603. https://doi.org/10.1063/1.3663376
Camelli F, Lohner R, Hanna S (2005) VLES study of MUST experiment. In: 43rd AIAA aerospace sciences meeting and exhibit. https://doi.org/10.2514/6.2005-1279
Carlstein E (1986) The use of subseries values for estimating the variance of a general statistic from a stationary sequence. Ann Stat 14(3):1171–1179
Cassiani M, Bertagni MB, Marro M, Salizzoni P (2020) Concentration fluctuations from localized atmospheric releases. Boundary-Layer Meteorol 177(2):461–510. https://doi.org/10.1007/s10546-020-00547-4
Chan SC, Kendon EJ, Sè Berthou, Fosser G, Lewis E, Fowler HJ (2020) Europe-wide precipitation projections at convection permitting scale with the unified model. Clim Dyn 55(3):409–428. https://doi.org/10.1007/s00382-020-05192-8
Chang J, Hanna S (2004) Air quality model performance evaluation. Meteorol Atm Phys 87(1):167–196. https://doi.org/10.1007/s00703-003-0070-7
Chang JC, Hanna SR (2005) Technical descriptions and user’s guide for the boot statistical model evaluation software package, version 2.0. George Mason University and Harvard School of Public Health, Fairfax, VA, USA
Costes A, Rochoux MC, Lac C, Masson V (2021) Subgrid-scale fire front reconstruction for ensemble coupled atmosphere-fire simulations of the FireFlux I experiment. Fire Saf J 126(103):475. https://doi.org/10.1016/j.firesaf.2021.103475
Crippa M, Janssens-Maenhout G, Dentener F, Guizzardi D, Sindelarova K, Muntean M, Van Dingenen R, Granier C (2016) Forty years of improvements in European air quality: regional policy–industry interactions with global impacts. Atmos Chem Phys 16(6):3825–3841. https://doi.org/10.5194/acp-16-3825-2016
Dauxois T, Peacock T, Bauer P, Caulfield CP, Cenedese C, Gorlé C, Haller G, Ivey GN, Linden PF, Meiburg E, Pinardi N, Vriend NM, Woods AW (2021) Confronting grand challenges in environmental fluid mechanics. Phys Rev Fluids 6(020):501. https://doi.org/10.1103/PhysRevFluids.6.020501
Davison AC, Hinkley DV (1997) Bootstrap methods and their application. Cambridge Series in statistical and probabilistic mathematics. Cambridge University Press, Cambridge. https://doi.org/10.1017/CBO9780511802843
Defforge CL, Carissimo B, Bocquet M, Bresson R, Armand P (2021) Improving numerical dispersion modelling in built environments with data assimilation using the iterative ensemble Kalman smoother. Boundary-Layer Meteorol 179(2):209–240. https://doi.org/10.1007/s10546-020-00588-9
Dejoan A, Santiago J, Martilli A, Martin F, Pinelli A (2010) Comparison between large-eddy simulation and Reynolds-averaged Navier–Stokes computations for the MUST field experiment. Part II: Effects of incident wind angle deviation on the mean flow and plume dispersion. Boundary-Layer Meteorol 135(1):133–150. https://doi.org/10.1007/s10546-010-9467-2
Diffenbaugh NS, Singh D, Mankin JS, Horton DE, Swain DL, Touma D, Charland A, Liu Y, Haugen M, Tsiang M et al (2017) Quantifying the influence of global warming on unprecedented extreme climate events. Proc Natl Acad Sci USA 114(19):4881–4886. https://doi.org/10.1073/pnas.1618082114
Donnelly R, Lyons T, Flassak T (2009) Evaluation of results of a numerical simulation of dispersion in an idealised urban area for emergency response modelling. Atmos Environ 43(29):4416–4423. https://doi.org/10.1016/j.atmosenv.2009.05.038
Dumont Le Brazidec J, Bocquet M, Saunier O, Roustan Y (2023) Bayesian transdimensional inverse reconstruction of the Fukushima Daiichi caesium 137 release. Geosci Model Dev 16(3):1039–1052. https://doi.org/10.5194/gmd-16-1039-2023
EEA (2020) Air quality in Europe—2020 report. European Environment Agency
Efron B (1979) Bootstrap methods: another look at the jackknife. Ann Stat 7(1):1–26
Fernando HJS, Lee SM, Anderson J, Princevac M, Pardyjak E, Grossman-Clarke S (2001) Urban fluid mechanics: air circulation and contaminant dispersion in cities. Environ Fluid Mech 1(1):107–164. https://doi.org/10.1023/A:1011504001479
Franke J, Hellsten A, Schlunzen KH, Carissimo B (2011) The COST 732 best practice guideline for CFD simulation of flows in the urban environment: a summary. Int J Environ Pollut 44(1–4):419–427. https://doi.org/10.1504/IJEP.2011.038443
García-Sanchez C, Gorlé C (2018) Uncertainty quantification for microscale CFD simulations based on input from mesoscale codes. J Wind Eng Ind Aerodyn 176:87–97. https://doi.org/10.1016/j.jweia.2018.03.011
García-Sánchez C, Philips D, Gorlé C (2014) Quantifying inflow uncertainties for CFD simulations of the flow in downtown Oklahoma city. Build Environ 78:118–129. https://doi.org/10.1016/j.buildenv.2014.04.013
García-Sanchez C, van Beeck J, Gorlé C (2018) Predictive large eddy simulations for urban flows: challenges and opportunities. Build Environ 139:146–156
Gicquel LY, Gourdain N, Boussuge JF, Deniau H, Staffelbach G, Wolf P, Poinsot T (2011) High performance parallel computing of flows in complex geometries. Comptes Rendus Mécanique 339(2):104–124. https://doi.org/10.1016/j.crme.2010.11.006
Gousseau P, Blocken B, Stathopoulos T, van Heijst G (2011) CFD simulation of near-field pollutant dispersion on a high-resolution grid: a case study by LES and RANS for a building group in downtown Montreal. Atmos Environ 45(2):428–438. https://doi.org/10.1016/j.atmosenv.2010.09.065
Grylls T, Cornec CML, Salizzoni P, Soulhac L, Stettler ME, van Reeuwijk M (2019) Evaluation of an operational air quality model using large-eddy simulation. Atmos Environ 3(100):041. https://doi.org/10.1016/j.aeaoa.2019.100041
Hanna SR (1989) Confidence limits for air quality model evaluations, as estimated by bootstrap and jackknife resampling methods. Atmos Environ 23(6):1385–1398. https://doi.org/10.1016/0004-6981(89)90161-3
Hanna SR, Hansen OR, Dharmavaram S (2004) FLACS CFD air quality model performance evaluation with Kit Fox, MUST, Prairie Grass, and EMU observations. Atmos Environ 38(28):4675–4687. https://doi.org/10.1016/j.atmosenv.2004.05.041
Harms F, Leitl B, Schatzmann M, Patnaik G (2011) Validating LES-based flow and dispersion models. J Wind Eng Ind Aerodyn 99(2):289–295. https://doi.org/10.1016/j.jweia.2011.01.007
Hayati AN, Stoll R, Kim JJ, Harman T, Nelson MA, Brown MJ, Pardyjak ER (2017) Comprehensive evaluation of fast-response, Reynolds-averaged Navier–Stokes, and large-eddy simulation methods against high-spatial-resolution wind-tunnel data in step-down street canyons. Boundary-Layer Meteorol 164(2):217–247. https://doi.org/10.1007/s10546-017-0245-2
Hertwig D, Soulhac L, Fuka V, Auerswald T, Carpentieri M, Hayden P, Robins A, Xie ZT, Coceal O (2018) Evaluation of fast atmospheric dispersion models in a regular street network. Environ Fluid Mech 18(4):1007–1044. https://doi.org/10.1007/s10652-018-9587-7
Hertwig D, Gough HL, Grimmond S, Barlow JF, Kent CW, Lin WE, Robins AG, Hayden P (2019) Wake characteristics of tall buildings in a realistic urban canopy. Boundary-Layer Meteorol 172(2):239–270. https://doi.org/10.1007/s10546-019-00450-7
Huybers P, McKinnon KA, Rhines A, Tingley M (2014) US daily temperatures: the meaning of extremes in the context of nonnormality. J Clim 27(19):7368–7384. https://doi.org/10.1175/JCLI-D-14-00216.1
Keating A, Piomelli U, Balaras E, Kaltenbach HJ (2004) A priori and a posteriori tests of inflow conditions for large-eddy simulation. Phys Fluids 16(12):4696–4712. https://doi.org/10.1063/1.1811672
Klein P, Leitl B, Schatzmann M (2007) Driving physical mechanisms of flow and dispersion in urban canopies. Int J Climatol 27(14):1887–1907. https://doi.org/10.1002/joc.1581
König M (2014) Large-eddy simulation modelling for urban scale, Ph.D. thesis. University of Leipzig
Koutsourakis N, Bartzis JG, Markatos NC (2012) Evaluation of Reynolds stress, k-\(\epsilon \) and RNG k-\(\epsilon \) turbulence models in street canyon flows using various experimental datasets. Environ Fluid Mech 12(4):379–403. https://doi.org/10.1007/s10652-012-9240-9
Kraichnan RH (1970) Diffusion by a random velocity field. Phys Fluids 13(1):22–31. https://doi.org/10.1063/1.1692799
Kumar P, Feiz AA, Ngae P, Singh SK, Issartel JP (2015) CFD simulation of short-range plume dispersion from a point release in an urban like environment. Atmos Environ 122:645–656. https://doi.org/10.1016/j.atmosenv.2015.10.027
Kunsch HR (1989) The jackknife and the bootstrap for general stationary observations. Ann Stat 66:1217–1241
Langmann B, Duncan B, Textor C, Trentmann J, van der Werf GR (2009) Vegetation fire emissions and their impact on air pollution and climate. Atmos Environ 43(1):107–116. https://doi.org/10.1016/j.atmosenv.2008.09.047
Larsson J, Kawai S, Bodart J, Bermejo-Moreno I (2016) Large eddy simulation with modeled wall-stress: recent progress and future directions. Mech Eng Rev 3(1):15-00418. https://doi.org/10.1299/mer.15-00418
Massey FJ (1951) The Kolmogorov–Smirnov test for goodness of fit. J Am Stat Assoc 46(253):68–78. https://doi.org/10.1080/01621459.1951.10500769
Merlier L, Jacob J, Sagaut P (2018) Lattice–Boltzmann Large-Eddy Simulation of pollutant dispersion in street canyons including tree planting effects. Atmos Environ 195:89–103. https://doi.org/10.1016/j.atmosenv.2018.09.040
Meroney RN, Leitl BM, Rafailidis S, Schatzmann M (1999) Wind-tunnel and numerical modeling of flow and dispersion about several building shapes. J Wind Eng Ind Aerodyn 81(1):333–345. https://doi.org/10.1016/S0167-6105(99)00028-8
Meyers J, Geurts B, Sagaut P (2008) Quality and reliability of large-eddy simulations, vol 12. Springer, Berlin. https://doi.org/10.1007/978-1-4020-8578-9
Milliez M, Carissimo B (2007) Numerical simulations of pollutant dispersion in an idealized urban area, for different meteorological conditions. Boundary-Layer Meteorol 122(2):321–342. https://doi.org/10.1007/s10546-006-9110-4
Montazeri H, Blocken B (2013) CFD simulation of wind-induced pressure coefficients on buildings with and without balconies: validation and sensitivity analysis. Build Environ 60:137–149. https://doi.org/10.1016/j.buildenv.2012.11.012
Muñoz-Esparza D, Kosović B, Mirocha J, van Beeck J (2014) Bridging the transition from mesoscale to microscale turbulence in numerical weather prediction models. Boundary-Layer Meteorol 153(3):409–440. https://doi.org/10.1007/s10546-014-9956-9
Munters W, Meneveau C, Meyers J (2016) Turbulent inflow precursor method with time-varying direction for large-eddy simulations and applications to wind farms. Boundary-Layer Meteorol 159(2):305–328. https://doi.org/10.1007/s10546-016-0127-z
Nagel T, Schoetter R, Masson V, Lac C, Carissimo B (2022) Numerical analysis of the atmospheric boundary-layer turbulence influence on microscale transport of pollutant in an idealized urban environment. Boundary-Layer Meteorol 184(1):113–141. https://doi.org/10.1007/s10546-022-00697-7
Nagel T, Schoetter R, Bourgin V, Masson V, Onofri E (2023) Drag coefficient and turbulence mixing length of local climate zone-based urban morphologies derived using obstacle-resolving modelling. Boundary-Layer Meteorol 186(3):737–769. https://doi.org/10.1007/s10546-022-00780-z
Nazarian N, Krayenhoff ES, Martilli A (2020) A one-dimensional model of turbulent flow through “urban” canopies (mlucm v2.0): updates based on large-eddy simulation. Geosci Model Dev 13(3):937–953. https://doi.org/10.5194/gmd-13-937-2020
Nicoud F, Ducros F (1999) Subgrid-scale stress modelling based on the square of the velocity gradient tensor. Flow Turbul Combust 62(3):183–200. https://doi.org/10.1023/A:1009995426001
Nony BX, Rochoux MC, Jaravel T, Lucor D (2023) Reduced-order modeling for parameterized large-eddy simulations of atmospheric pollutant dispersion. Stoch Environ Res Risk Assess 37(6):2117–2144. https://doi.org/10.1007/s00477-023-02383-7
Orsi M, Soulhac L, Feraco F, Marro M, Rosenberg D, Marino R, Boffadossi M, Salizzoni P (2021) Scalar mixing in homogeneous isotropic turbulence: a numerical study. Phys Rev Fluids 6(034):502. https://doi.org/10.1103/PhysRevFluids.6.034502
Paoli R, Poubeau A, Cariolle D (2020) Large-eddy simulations of a reactive solid rocket motor plume. AIAA J 58(4):1639–1656. https://doi.org/10.2514/1.J058601
Passot T, Pouquet A (1987) Numerical simulation of compressible homogeneous flows in the turbulent regime. J Fluid Mech 181:441–466. https://doi.org/10.1017/S0022112087002167
Patnaik G, Boris J, Young T, Grinstein F (2007) Large scale urban contaminant transport simulations with Miles. J Fluids Eng 129(12):1524–1532. https://doi.org/10.1115/1.2801368
Poinsot T, Lele S (1992) Boundary conditions for direct simulations of compressible viscous flows. J Comput Phys 101(1):104–129. https://doi.org/10.1016/0021-9991(92)90046-2
Politis DN, Romano JP (1994) The stationary bootstrap. J Am Stat Assoc 89(428):1303–1313. https://doi.org/10.1080/01621459.1994.10476870
Poubeau A, Paoli R, Cariolle D (2016) Evaluation of afterburning chemistry in solid-rocket motor jets using an off-line model. J Spacecr Rockets 53(2):380–388. https://doi.org/10.2514/1.A33311
Ramshaw J, O’Rourke P, Amsden A (1986) Acoustic damping for explicit calculations of fluid flow at low Mach number. Los Alamos National Laboratories, USA, Technical report no. LA-10641-MS
Richards P, Hoxey R (1993) Appropriate boundary conditions for computational wind engineering models using the \(k-\epsilon \) turbulence model. J Wind Eng Ind Aerodyn 46–47:145–153. https://doi.org/10.1016/0167-6105(93)90124-7
Risser MD, Paciorek CJ, Wehner MF, O’Brien TA, Collins WD (2019) A probabilistic gridded product for daily precipitation extremes over the united states. Clim Dyn 53(5):2517–2538. https://doi.org/10.1007/s00382-019-04636-0
Santiago JL, Dejoan A, Martilli A, Martin F, Pinelli A (2010) Comparison between large-eddy simulation and Reynolds-averaged Navier–Stokes computations for the MUST field experiment. Part I: Study of the flow for an incident wind directed perpendicularly to the front array of containers. Boundary-Layer Meteorol 135(1):109–132. https://doi.org/10.1007/s10546-010-9466-3
Schatzmann M, Leitl B (2011) Issues with validation of urban flow and dispersion CFD models. J Wind Eng Ind Aerodyn 99(4):169–186. https://doi.org/10.1016/j.jweia.2011.01.005
Schatzmann M, Olesen H, Franke J (2010) COST 732 model evaluation case studies: approach and results. University of Hamburg, Meteorological Institute, Tech rep
Scheiner SM, Gurevitch J (2001) Design and analysis of ecological experiments. Oxford University Press, Oxford. https://doi.org/10.1201/9781003059813
Schönfeld T, Rudgyard M (1999) Steady and unsteady flow simulations using the hybrid flow solver AVBP. AIAA J 37(11):1378–1385. https://doi.org/10.2514/2.636
Smirnov A, Shi S, Celik I (2001) Random flow generation technique for large eddy simulations and particle-dynamics modeling. J Fluids Eng 123(2):359–371. https://doi.org/10.1115/1.1369598
Sood I, Simon E, Vitsas A, Blockmans B, Larsen GC, Meyers J (2022) Comparison of large eddy simulations against measurements from the Lillgrund offshore wind farm. Wind Energy Sci 7(6):2469–2489. https://doi.org/10.5194/wes-7-2469-2022
Sousa J, Gorlé C (2019) Computational urban flow predictions with Bayesian inference: validation with field data. Build Environ 154:13–22. https://doi.org/10.1016/j.buildenv.2019.02.028
Tominaga Y, Stathopoulos T (2013) CFD simulation of near-field pollutant dispersion in the urban environment: a review of current modeling techniques. Atmos Environ 79:716–730. https://doi.org/10.1016/j.atmosenv.2013.07.028
Tominaga Y, Mochida A, Yoshie R, Kataoka H, Nozu T, Yoshikawa M, Shirasawa T (2008) AIJ guidelines for practical applications of CFD to pedestrian wind environment around buildings. J Wind Eng Ind Aerodyn 96(10):1749–1761. https://doi.org/10.1016/j.jweia.2008.02.058
Toparlar Y, Blocken B, Maiheu B, van Heijst G (2017) A review on the CFD analysis of urban microclimate. Renew Sust Energ Rev 80:1613–1640. https://doi.org/10.1016/j.rser.2017.05.248
Vasaturo R, Kalkman I, Blocken B, van Wesemael P (2018) Large eddy simulation of the neutral atmospheric boundary layer: performance evaluation of three inflow methods for terrains with different roughness. J Wind Eng Ind Aerodyn 173:241–261. https://doi.org/10.1016/j.jweia.2017.11.025
Vervecken L, Camps J, Meyers J (2015) Dynamic dose assessment by large eddy simulation of the near-range atmospheric dispersion. J Radiol Prot 35(1):165–178. https://doi.org/10.1088/0952-4746/35/1/165
Vervecken L, Camps J, Meyers J (2015) Stable reduced-order models for pollutant dispersion in the built environment. Build Environ 92:360–367. https://doi.org/10.1016/j.buildenv.2015.05.008
Wiersema DJ, Lundquist KA, Chow FK (2020) Mesoscale to microscale simulations over complex terrain with the immersed boundary method in the weather research and forecasting model. Mon Weather Rev 148(2):577–595. https://doi.org/10.1175/MWR-D-19-0071.1
Wise D, Boppana V, Li K, Poh H (2018) Effects of minor changes in the mean inlet wind direction on urban flow simulations. Sustain Cities Soc 37:492–500. https://doi.org/10.1016/j.scs.2017.11.041
Yee E, Biltoft CA (2004) Concentration fluctuation measurements in a plume dispersing through a regular array of obstacles. Boundary-Layer Meteorol 111(3):363–415. https://doi.org/10.1023/B:BOUN.0000016496.83909.ee
Acknowledgements
The authors acknowledge Tim Nagel (CNRM/Météo-France), Bertrand Carissimo (CEREA/École Nationale des Ponts et Chaussées), Julien Boé (CECI/CNRS-CERFACS), Bastien Nony (CECI/CNRS-CERFACS), and Eric Pardyjak (University of Utah) for helpful discussions as well as the Defense Threat Reduction Agency (DTRA) for providing access to the MUST dataset. They also acknowledge Laëtitia Thouron and Géraldine Rea for their contribution to previous related works as well as the technical help of the Computer Support Group at CERFACS.
Author information
Authors and Affiliations
Contributions
Eliott Lumet, Thomas Jaravel, Mélanie Rochoux and Simon Lacroix collectively developed and planned the study. M.R., T.J., and S.L. took care of the funding acquisition and the project administration. Olivier Vermorel and T.J. carried out specific development of the model used in this study. The presented simulation and analyses were performed by E.L. and T.J. The first draft of the manuscript was written by E.L. with significant contributions from M.R. and T.J. All authors read and approved the final manuscript.
Corresponding author
Ethics declarations
Ethical Approval
Ethical approval does not apply to the content of this study.
Competing Interests
The authors have no competing interests to declare that are relevant to the content of this article.
Funding
This work received financial support from the FCS-STAE foundation jointly with IRT Saint-Exupéry (PPM project, 2018-2020), which enhanced research on environmental CFD at CERFACS. Eliott Lumet’s Ph.D. thesis was funded by the Université Fédérale Toulouse Midi-Pyrénées jointly with Région Occitanie (AtmoDrones project, 2020-2023). This work was granted access to the HPC resources from GENCI-TGCC/CINES (A0062A10822 project, 2020-2022).
Availability of Data and Materials
The datasets generated during the current study, and the analysis codes developed by the authors are available from the corresponding author on reasonable request. The LES solver AVBP, https://www.cerfacs.fr/avbp7x/, as well as the Antares post-processing library, https://cerfacs.fr/antares/, are under license but can be provided under agreement for academic purposes.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix I: Validation of the Stationary Bootstrap Approach
Appendix I: Validation of the Stationary Bootstrap Approach
In this appendix, several convergence tests of the stationary bootstrap approach applied to the MUST trial are shown to verify the robustness and plausibility of the confidence intervals predicted by bootstrap as a complement to Sect. 4.
1.1 Convergence with the Number of Bootstrap Replicates
With bootstrap methods, such as the stationary bootstrap, bootstrap replicates of one original sample are used to compute Monte Carlo estimates of several statistics of the physical quantities of interest (time-averages or fluctuations), such as their variance (Eq. 12) or confidence intervals. The convergence of the estimator error is therefore in \(\mathcal {O}(1/\sqrt{B})\) with B the number of bootstrap replicates.
We assess the convergence for the 2.5th and 97.5th percentiles as it requires more bootstrap replicates than for bias or variance estimation (Davison and Hinkley 1997). Table 5 shows the evolution of the 2.5th percentile of the mean concentration at tower B at \(z=2\,\)m and of the model validation metrics evaluated according to the bootstrap procedure (Eq. 14) for different values of B. The bootstrap estimations of the 2.5th percentiles show some variability for very low numbers of bootstrap replicates (between 100 and 500), but then quickly converge for all the considered quantities. The same analysis was carried out for the 97.5th percentile and gave similar results. We conclude that B = 5000 bootstrap samples are more than sufficient to achieve convergence. This result is in line with the literature, which recommends between 1000 and 10,000 replicates (Davison and Hinkley 1997; Chang and Hanna 2005).
1.2 Effect of the Sub-Averaging Period
Since both simulated and measured time series are well sampled in time, it is possible to change the sub-averaging period \(\delta _t\) to adjust the number of sub-average samples \(N_t\) (Eq. 10). To assess the effect of the sub-averaging period on the internal variability estimation, the estimated percentiles obtained with a stationary bootstrap of sub-averages over \(\delta _t =10\,\textrm{s}\) and \(\delta _t =5\,\textrm{s}\) are compared. By reducing the sub-averaging period, the samples are getting more dependent. It is therefore mandatory to adapt the mean block length parameter \(\ell \) of the stationary bootstrap method. For \(\delta _t =5\,\textrm{s}\), it results to new values of \(\ell _{sim}=1.38\) and \(\ell _{obs}=2.62\) for the time-averaged concentrations. This is consistent since it means that, for more dependent data, the blocks should be larger than the ones used for \(\delta _t =10\,\textrm{s}\) (Table 2, page 21). Table 6 shows the 2.5th percentile estimates for the main quantities of interest for the two different values of \(\delta _t\). Results indicate that changing the sub-averaging period has a very limited impact on the stationary bootstrap estimations. This is because changing the sub-averaging period only changes the division of the original sample (Eq. 10) and so does not provide any additional information on the underlying distribution of the time-averaged quantities.
1.3 Convergence with the Number of Sub-average Samples
As mentioned in Sect. 4.4.3, it is essential to have a sufficient number of sub-average samples \(N_t\) in the original sample. In particular, too few samples may result in internal variability underestimation. To increase \(N_t\), the LES simulation acquisition time is increased from 200 s to 400 s, and then 600 s (Fig. 4). With \(\delta _t = 10\) s the resulting number of sub-averages is 40, and 60 respectively, against 20 for the reference sample. In any case, the bootstrap replicates are obtained by resampling only 20 sub-averages over the \(N_t\) available, even if \(N_t = 60\). Indeed, the objective is still to quantify the variability over the 200-s analysis period and not over 600 s. Two additional realizations of 200-s averages (in cyan and magenta in Fig. 4) are used for validation purposes.
The distributions of the air quality metrics estimated by stationary bootstrap with resampling of 20 sub-averages among \(Nt=20\), 40 and 60 are shown in Fig. 13. The bootstrap ensemble averages slightly change because the time-averaged quantities over 200, 400 and 600 s are different. Nevertheless, increasing the number of samples for stationary bootstrap gives similar estimations of the metrics dispersion. For FAC2, MG and VG metrics, which are nonlinear, the tails of the distributions seem more dependent on the number of samples. In all cases, the three bootstrap estimates cover the two independent realizations, once again supporting the validity of the stationary bootstrap.
As the orders of magnitude of the three estimates are overall consistent with each other, we conclude that \(N_t = 20\) samples of sub-averages are sufficient for the stationary bootstrap method to converge. This implies that it is not required to run longer simulations to capture internal variability. The convergence with the number of samples is similarly verified for the envelopes of tracer concentration and wind velocity statistics presented in Sect. 4.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Lumet, E., Jaravel, T., Rochoux, M.C. et al. Assessing the Internal Variability of Large-Eddy Simulations for Microscale Pollutant Dispersion Prediction in an Idealized Urban Environment. Boundary-Layer Meteorol 190, 9 (2024). https://doi.org/10.1007/s10546-023-00853-7
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s10546-023-00853-7