Comparison and correction of IDW based wind speed interpolation methods in urbanized Shenzhen

Abstract

Based on the 2-min average wind speed observations at 100 automatic weather stations in Shenzhen from January 2008 to December 2018, this study tries to explore the ways to improve wind interpolation quality over the Shenzhen region. Three IDW based methods, i.e., traditional inverse distance weight (IDW), modified inverse distance weight (MIDW), and gradient inverse distance weight (GIDW) are used to interpolate the near surface wind field in Shenzhen. In addition, the gradient boosted regression trees (GBRT) model is used to correct the wind interpolation results based on the three IDW based methods. The results show that among the three methods, GIDW has better interpolation effects than the other two in the case of stratified sampling. The MSE and R² for the GIDW’s in different months are in the range of 1.096–1.605 m/s and 0.340–0.419, respectively. However, in the case of leave-one-group-out cross-validation, GIDW has no advantage over the other two methods. For the stratified sampling, GBRT effectively corrects the interpolated results by the three IDW based methods. MSE decreases to the range of 0.778–0.923 m/s, and R² increases to the range of 0.530–0.671. In the non-station area, the correction effect of GBRT is still robust, even though the elevation frequency distribution of the non-station area is different from that of the stations’ area. The correction performance of GBRT mainly comes from its consideration of the nonlinear relationship between wind speed and the elevation, and the combination of historical and current observation data.

Graphical abstract

Keywords

wind interpolation / Shenzhen / inverse distance weight / gradient boosted regression trees

References

[1] AguilarF J, AgüeraF, AguilarM A, CarvajalF. ( 2005). Effects of terrain morphology, sampling density, and interpolation methods on grid DEM accuracy. Photogramm Eng Remote Sensing, 71( 7): 805– 816

[2] AltmannA, ToloşiL, SanderO, LengauerT. ( 2010). Permutation importance: a corrected feature importance measure. Bioinformatics, 26( 10): 1340– 1347

[3] ArainM A, BlairR, FinkelsteinN, BrookJ R, SahsuvarogluT, BeckermanB, ZhangL, JerrettM. ( 2007). The use of wind fields in a land use regression model to predict air pollution concentrations for health exposure studies. Atmos Environ, 41( 16): 3453– 3464

[4] Bañuelos-RuedasF, Angeles-CamachoC, Rios-MarcuelloS. ( 2010). Analysis and validation of the methodology used in the extrapolation of wind speed data at different heights. Renew Sustain Energy Rev, 14( 8): 2383– 2391

[5] CasalicchioG MolnarC BischlB ( 2018). Visualizing the feature importance for black box models. In: Joint European Conference on Machine Learning and Knowledge Discovery in Databases. Cham: Springer

[6] ChangQ, LiS, WangY, WuJ, XieM. ( 2013). Spatial process of green infrastructure changes associated with rapid urbanization in Shenzhen, China. Chin Geogr Sci, 23( 1): 113– 128

[7] CoccoloS, MaureeD, NaboniE, KaempfJ, ScartezziniJ L. ( 2017). On the impact of the wind speed on the outdoor human comfort: a sensitivity analysis. Energy Procedia, 122: 481– 486

[8] CurryC L, van der KampD, MonahanA H. ( 2012). Statistical downscaling of historical monthly mean winds over a coastal region of complex terrain. I. Predicting wind speed. Clim Dyn, 38( 7−8): 1281– 1299

[9] DalyC, HalbleibM, SmithJ I, GibsonW P, DoggettM K, TaylorG H, CurtisJ, PasterisP P. ( 2008). Physiographically sensitive mapping of climatological temperature and precipitation across the conterminous United States. Intern J Climat, 28( 15): 2031– 2064

[10] ElithJ, LeathwickJ R, HastieT. ( 2008). A working guide to boosted regression trees. J Anim Ecol, 77( 4): 802– 813

[11] FriedmanJ H. ( 2001). Greedy function approximation: a gradient boosting machine. Ann Stat, 29( 5): 1189– 1232

[12] FriedmanJ H. ( 2002). Stochastic gradient boosting. Comput Stat Data Anal, 38( 4): 367– 378

[13] GhoshS P. ( 1958). A note on stratified random sampling with multiple characters. Calcutta Statistical Association Bulletin, 8( 2−3): 81– 90

[14] JarvisA ReuterH I NelsonA GuevaraE ( 2008). SRTM 90m Digital Elevation Database v4. 1

[15] JohnsonH K. ( 1999). Simple expressions for correcting wind speed data for elevation. Coast Eng, 36( 3): 263– 269

[16] KikumotoH, OokaR, SugawaraH, LimJ. ( 2017). Observational study of power-law approximation of wind profiles within an urban boundary layer for various wind conditions. J Wind Eng Ind Aerodyn, 164: 13– 21

[17] KleerekoperL, Van EschM, SalcedoT B. ( 2012). How to make a city climate-proof, addressing the urban heat island effect. Resour Conserv Recycling, 64: 30– 38

[18] LiJ, TianY, ZhuY, ZhouT, LiJ, DingK, LiJ. ( 2020). A multicenter random forest model for effective prognosis prediction in collaborative clinical research network. Artif Intell Med, 103: 101814

[19] LiL. ( 2019). Geographically weighted machine learning and downscaling for high-resolution spatiotemporal estimations of wind speed. Remote Sens, 11( 11): 1378

[20] LiL, QianJ, OuC Q, ZhouY X, GuoC, GuoY. ( 2014). Spatial and temporal analysis of air pollution index and its timescale-dependent relationship with meteorological factors in Guangzhou, China, 2001–2011. Environ Pollut, 190: 75– 81

[21] LiuC, LiQ, ZhaoW, WangY, AliR, HuangD, LuX, ZhengH, WeiX. ( 2020). Spatiotemporal characteristics of near-surface wind in Shenzhen. Sustainability, 12( 2): 739

[22] LouppeG, WehenkelL, SuteraA, GeurtsP. ( 2013). Understanding variable importances in forests of randomized trees. In: Proceedings of the 26th International Conference on Neural Information Processing Systems, 431– 439

[23] MasseranN, RazaliA M, IbrahimK. ( 2012). An analysis of wind power density derived from several wind speed density functions: the regional assessment on wind power in Malaysia. Renew Sustain Energy Rev, 16( 8): 6476– 6487

[24] McCutchanM H, FoxD G. ( 1986). Effect of elevation and aspect on wind, temperature and humidity. J Clim Appl Meteorol, 25( 12): 1996– 2013

[25] MitasL MitasovaH ( 1999). Spatial interpolation. In: Longley P A, Goodchild M F, Maguire D J, Rhind D W, eds. Geographical Information Systems: Principles, Techniques, Management and Applications. New York: Wiley

[26] MorrisH WardleM MainelliM ( 2020). The global financial centres index 28. Long Finan Financ Centr Future. 1− 69

[27] NalderI A, WeinR W. ( 1998). Spatial interpolation of climatic normals: test of a new method in the Canadian boreal forest. Agric Meteorol, 92( 4): 211– 225

[28] OzelkanE, ChenG, UstundagB B. ( 2016). Spatial estimation of wind speed: a new integrative model using inverse distance weighting and power law. Int J Digit Earth, 9( 8): 733– 747

[29] PalutikofJ P, KellyP M, DaviesT D, HallidayJ A. ( 1987). Impacts of spatial and temporal windspeed variability on wind energy output. J Clim Appl Meteorol, 26( 9): 1124– 1133

[30] PengS ( 2017). Optimized study on spatial interpolation methods for meteorological element. Geospat Inform, 15( 7): 86− 89 (in Chinese)

[31] ToblerW R. ( 1970). A computer movie simulating urban growth in the Detroit region. Econom Geogra, 46( sup1): 234– 240

[32] WangJ, HuJ, MaK. ( 2016a). Wind speed probability distribution estimation and wind energy assessment. Renew Sustain Energy Rev, 60: 881– 899

[33] WangP, DouY, XinY. ( 2016b). The analysis and design of the job recommendation model based on GBRT and time factors. In: 2016 IEEE International Conference on Knowledge Engineering and Applications (ICKEA). IEEE, 29– 35

[34] WangY TangY (2019). A recommendation algorithm based on item genres preference and GBRT. In: Journal of Physics Conference Series. IOP Publishing, 2019, 1229( 1): 012053

[35] WanninkhofR. ( 2014). Relationship between wind speed and gas exchange over the ocean revisited. Limnol Oceanogr Methods, 12( 6): 351– 362

[36] WinstralA, JonasT, HelbigN. ( 2017). Statistical downscaling of gridded wind speed data using local topography. J Hydrometeorol, 18( 2): 335– 348

[37] XuY Wan X FuC LiuC ( 2012). Wind speed interpolation under complex terrain conditions: a case study of Jilin Province. Yunnan Geogr Environ Res, 24 ( 4), 78− 81 (in Chinese)

[38] YangJ, ShiB, ShiY, MarvinS, ZhengY, XiaG. ( 2020). Air pollution dispersal in high density urban areas: research on the triadic relation of wind, air pollution, and urban form. Sustain Cities Soc, 54: 101941

[39] YuanW, XuB, ChenZ, XiaJ, XuW, ChenY, WuX, FuY. ( 2015). Validation of China-wide interpolated daily climate variables from 1960 to 2011. Theor Appl Climatol, 119( 3−4): 689– 700

[40] ZhangZ, YangW, WushourS. ( 2020). Traffic accident prediction based on LSTM-GBRT model. J Contr Sci Eng, 2020: 4206919

[41] ZhiS, QianG, LuoJ. ( 2001). The variation of coastal wind with height in Guangdong Province. Trop Geogr, 21( 2): 131– 134

Acknowledgments

This study was supported by the Science and Technology Department of Guangdong Province (No. 2019B111101002) and the Innovation of Science and Technology Commission of Shenzhen Municipality Ministry (No. JCYJ 20210324101006016).

RIGHTS & PERMISSIONS

2022 Higher Education Press