تخمین فواصل پیش‌بینی در ریزمقیاس‌نمایی مدل‌ گردش عمومی جو بر پایه شبکه عصبی

نوع مقاله : مقاله پژوهشی

نویسندگان

1 استادیار/ گروه عمران آب، دانشکده مهندسی عمران، دانشگاه تبریز.

2 استاد/ گروه عمران آب، دانشکده مهندسی عمران، دانشگاه تبریز.

3 دانشجوی کارشناسی ارشد مهندسی و مدیریت منابع آب/ گروه عمران آب، دانشکده مهندسی عمران، دانشگاه تبریز.

چکیده

یکی از ابراز پرکاربرد در پیش‌بینی آب و هوا و بررسی تاثیرات تغییر اقلیم بر روی پارامترهای آب‌وهوا مدل گردش عمومی (GCM) می‌باشد. در این مطالعه ریزمقیاس‌نمایی GCMها با استفاده از شبکه عصبی مصنوعی انجام گرفته است. از آنجایی که روش کلاسیک شبکه عصبی مصنوعی (پیش بینی نقطه ای) هیچ اطلاعاتی درباره‌ی دقت پیش‌بینی نمی‌دهد، از فواصل پیش‌بینی برای کمیت‌سنجی دقت ریزمقیاس‌نمایی شبکه عصبی مصنوعی استفاده شده است. برای محاسبه فواصل پیش‌بینی از روش جدید حد بالا و پایین (LUBE) استفاده شده است، که در آن شبکه عصبی با دو خروجی برای تخمین حدود پیش‌بینی ساخته شده است. همچنین روش کلاسیک بوت‌استرپ، روشی برای ارزیابی عدم‌قطعیت پیش‌بینی مورد استفاده قرار گرفته و نتایج حاصل از دو روش مقایسه شده است. بنابراین دقت فواصل پیش‌بینی به وسیله دو معیار همگرایی فواصل پیش‌بینی و عرض فواصل، کمیت‌سنجی شده است. سه GCM،Can-ESM2 ,BNU- ESM ,INM-CM4 و ترکیب آنها، در چهار نقطه‌ی شبکه بر روی هر یک از دو ایستگاه تبریز و اردبیل در شمال غربی ایران، برای ارزیابی فواصل پیش‌بینی ریزمقیاس‌نمایی بارش ماهانه و دما استفاده شده است. مقایسه بین نتایج دو مدل نشان داده است که روش LUBE، قابلیت اطمینان بیشتری نسبت به بوت استرپ دارد. عرض فواصل پیش‌بینی و احتمال ‌همگرایی، به ترتیب 10% تا 40% کمتر و 2% تا 10% بیشتر از روش بوت‌استرپ برای GCM بوده است. ترکیب GCMها به نتایج دقیق‌تری منجر شده است و عرض فواصل پیش‌بینی و احتمال‌ همگرایی، به ترتیب 10% تا 60% کمتر و 2% تا 20% بیشتر از مدل‌های تکی GCM بوده است.

کلیدواژه‌ها


عنوان مقاله [English]

Estimation of Prediction Interval in ANN-based GCM Downscaling

نویسندگان [English]

  • Elnaz Sharghi 1
  • Vahid Nourani 2
  • Nardin Jabbarian Paknezhad 3
1 1*- Assistant Professor, Department of Water Resources Engineering, Faculty of Civil Engineering, University of Tabriz, Tabriz, Iran.
2 Professor, Department of Water Resources Engineering, Faculty of Civil Engineering, University of Tabriz, Tabriz, Iran.
3 M.Sc. Student in Water Resource Engineering, Department of Water Resources Engineering, Faculty of Civil Eng., University of Tabriz, Tabriz, Iran.
چکیده [English]

The General Circulation Model (GCM) is one of the most widely used tools for weather forecasting and assessing the impact of the climate change on the weather parameters. In this study, statistical downscaling of GCMs were applied via Artificial neural network (ANN). As the classic ANN method (point prediction), conveys no information about the accuracy of the prediction, so prediction intervals were used for quantifying the accuracy of downscaling via ANN. For calculating the prediction intervals, novel Lower Upper bound estimation method was used, in which an ANN with two outputs was used to estimate the bounds. Also, Bootstrap method as a classic technique for assessing uncertainty of ANN was used to further examine the proposed LUBE method. In this way, the accuracy of PIs was quantified by coverage and width criteria. Three GCMs, Can-ESM2, BNU-ESM, INM-CM4 and ensemble-GCM (ensemble of them) were used in four grid points around each of station for evaluating ANN-based downscaling of precipitation and temperature parameters. Comparison between the results of two methods indicated that LUBE method could lead to more reliable results than the Bootstrap method. PIs width and coverage probability were 10% to 40% lower and 2% to 10% higher than the Bootstrap method for different GCMs, respectively. Ensemble-GCM led to more accurate results so that computed PIs width and coverage probability were 10% to 60% lower and 2% to 20% higher than those for the single GCMs.

کلیدواژه‌ها [English]

  • general circulation models
  • Downscaling
  • Prediction interval
  • Artificial Neural Network
Afzali-Gorouh Z, Faridhosseini A and Bakhtiari B (2019) Investigation of climate change impact on 24-h probable maximum precipitation in a sub-humid climate. Journal of Iran-Water Resource Research 15(2) Available at: http://www.iwrr.ir/article_82591.html  (In Persian)
Ahmadalipour A, Rana A, Moradkhani H, and Sharma A (2017) Multi-criteria evaluation of CMIP5 GCMs for climate change impact analysis. Theoretical and Applied Climatology 128(1-2):71-87
Beecham S, Rashid M, and Chowdhury RK (2014) Statistical downscaling of multi‐site daily rainfall in a South Australian catchment using a Generalized Linear Model. International Journal of Climatology, Wiley Online Library 34(14):3654-3670
Chryssolouris G, Lee M, and Ramsey A (1996) Confidence interval prediction for neural network models. IEEE Transactions on Neural Networks IEEE 7(1):229-232
Danandeh Mehr A, Kahya E (2016) Grid-based performance evaluation of GCM-RCM combinations for rainfall
reproduction. Theoretical and Applied Climatology 129(1-2):47-57
Dhanesh Y and Sudheer KP (2010) Predictions in ungauged basins: can we use artificial neural networks. American Geophysical Union Joint Assembly, Foz doIguassu, Brazil, August 8-13
Duan Q and Phillips TJ (2010) Bayesian estimation of local signal and noise in multimodel simulations of climate change. Journal of Geophysical Research: Atmospheres, Wiley Online Library 115(D18)
Efron B and Tibshirani RJ (1994) An introduction to the bootstrap. CRC press
Farahani MA, Rasekhi A, Parmas B, and Keshvari A (2018) The effects of climate change on temperature, precipitation and drought in the the future Shadegan basin. Journal of Iran-Water Resource Research 14(3):125-13  (In Persian)
Feng J, Lee D-K, Fu C, Tang J, Sato Y, Kato H, Mcgregor JL, and Mabuchi K (2011) Comparison of four ensemble methods combining regional climate simulations over Asia. Meteorology and Atmospheric Physics Springer 111(1-2):41-53
Fyfe JC, Gillett NP, and Zwiers FW (2013) Overestimated global warming over the past 20 years. Nature Climate Change Nature Publishing Group 3(9):767
Gao Z, Gu B, and Lin J (2008) Monomodal image registration using mutual information based methods. Image and Vision Computing 26(2):164-173
Grant El, Leavenworth RS (1972) Statistical quality and control Table A P643. New York: McGraw-Hill
Hosseini Baghanam A, Nourani V, Keynejad M A, Taghipour H, & Alami M T (2018) Conjunction of wavelet-entropy and SOM clustering for multi-GCM statistical downscaling. Hydrology Reasearch Online Published doi:10.2166/nh.2018.169
Her Y, Yoo S-H, Cho J, Hwang S, Jeong J, and Seong C (2019) Uncertainty in hydrological analysis of climate change: multi-parameter vs. multi-GCM ensemble predictions. Scientific Reports Nature Publishing Group 9(1):4974
Kaboosi  K and Kordjazi M (2017) The effect of climate change on meteorological parameters and drought in Golestan province. Journal of Iran-Water Resource Research 13(3):205-213 (In Persian)
Kasiviswanathan KS and Sudheer KP (2016) Comparison of methods used for quantifying prediction interval in artificial neural network hydrologic models. Modeling Earth Systems and Environment Springer 2(1):22
Khosravi A, Nahavandi S, and Creighton D (2010) A prediction interval-based approach to determine optimal structures of neural network metamodels. Expert Systems With Applications Elsevier 37(3):2377-2387
Khosravi A, Nahavandi S, Creighton D, and Atiya AF (2011) Lower upper bound estimation method for construction of neural network-based prediction intervals. IEEE Transactions on Neural Networks United States 22(3):337-346
Lee J-K and Kim Y-O (2017) Selection of representative GCM scenarios preserving uncertainties. Journal of Water and Climate Change. International Water Association 8(4):641-651
Liu Y, Feng J, Shao Y, and Li J (2019) Identify optimal predictors of statistical downscaling of summer daily precipitation in China from three-dimensional large-scale variables. Atmospheric Research Elsevier
MacKay DJC (1992) A practical Bayesian framework for backpropagation networks. Neural computation MIT Press 4(3):448-472
Miao C, Duan Q, Yang L, and Borthwick AGL (2012) On the applicability of temperature and precipitation data from CMIP3 for China. PLoS One. Public Library of Science 7(9):e44659
Nam C, Bony S, Dufresne J, and Chepfer H (2012) The ‘too few, too bright’tropical low‐cloud problem in CMIP5 models. Geophysical Research Letters, Wiley Online Library 39(21)
Nourani V, Baghanam AH, and Gokcekus H (2018) Data-driven ensemble model to statistically downscale rainfall using nonlinear predictor screening approach. Journal of Hydrology, Elsevier 565:538-551
Nourani V, Khanghah TR, and Baghanam AH (2015) Application of entropy concept for input selection of wavelet-ANN based rainfall-runoff modeling. Journal of Environmental Informatics 26(1)
Osman YZ and Abdellatif ME (2016) Improving accuracy of downscaling rainfall by combining predictions of different statistical downscale models. Water Science, Elsevier 30(2):61-75
Quan H, Srinivasan D and Khosravi A (2014) Particle swarm optimization for construction of neural network-based prediction intervals. Neurocomputing, Elsevier 127:172-180
Sailor D J, Li X (1999) A semiempirical downscaling approach for predicting regional temperature impacts associated with climatic change. Journal of Climate 12(1):103-114
Samadi S, Wilson C, and Moradkhani H (2013) Uncertainty analysis of statistical downscaling models using Hadley Centre Coupled Model. Theoretical and Applied Climatology, Springer 114(3-4):673-690
Shannon CE (1948) A mathematical theory of communication. Bell System Technical Journal. Wiley Online Library 27(3):379-423
Shiravand H and Dostkamiyan M (2019) Analysis of temperature fluctuations in the South West of Iran based on general circulation model and neural network (Case study: plain and mountainous stations). Journal of Iran-Water Resource Research, Available at: http://www.iwrr.ir/article_90429.html (In Persian)
Srivastav RK, Sudheer KP, and Chaubey I (2007) A simplified approach to quantifying predictive and parametric uncertainty in artificial neural network hydrologic models. Water Resources Research, Wiley Online Library 43(10)
Wilby RL and Wigley TML (1997) Downscaling general circulation model output: a review of methods and limitations. Progress in physical geography. Sage Publications Sage CA: Thousand Oaks, CA 21(4):530-548
Yang HH, Van Vuuren S, Sharma S, and Hermansky H (2000) Relevance of time–frequency features for phonetic and speaker-channel classification. Speech communication, Elsevier 31(1):35-50
Zio E (2006) A study of the bootstrap method for estimating the accuracy of artificial neural networks in predicting nuclear transient processes. IEEE Transactions on Nuclear Science IEEE 53(3):1460-1478