Nao-Lisbon Model: Nao-Gibraltar Model

Nao-Lisbon Model: Nao-Gibraltar Model

S.R. Gámiz-Fortis(1), D. Pozo-Vázquez(2), M.J. Esteban-Parra(1), and Y. Castro-Díez(1) (1)Dept. Applied Physics, University of Granada, E-18071, Granada, Spain. (2)Dept. Physics, University of Jaén, E-23071, Jaén, Spain. University of Granada DATA METHODOLOGY RESULTS OBTAINED BY FOURIER ANALYSIS Ø Winter NAO index (December-January-February): Ø Analysis of the NAO index by the Singular Spectral T = 2-3 T = 4.8 T= 5.8 T = 7.8 T = 23.5 Trend Analysis (SSA) method and the Monte Carlo method (MC). (years) (years) (years) (years) (years) ü Hurrell et al. (1995): Difference of monthly pressure values between Lisbon and Iceland. Period 1865-1997. Ø Parameters: 5-6 NAO-LI 10-12 __ 9-11-13 2-3 7-8 1-4 ü Jones et al. (1997): Difference of monthly pressure values 1. Lagged-covariance matrix: Broomhead and King between Gibraltar and Iceland. Period 1865-1997. method. 14-15 2. Window length: M = 40 3-4 ü Barnston y Livezey (1987): Obtained by a PCA of SLP data. 3. Monte Carlo test basis: data Period: 1951-1997. 4. Number of red noise processes generated by MC: p = NAO-GI 7-10 11-12 8-9 1-2 __ 5-6 10000. 13-14 Ø Normalisation Period : 1951-1980. Ø Forecasting of the NAO index using models ARMA. Periods (in years) corresponding to different oscillatory modes represented by pairs of PCs 4 280 180 260 PC 1 PCs 1 & 2 3 160 240 PCs 2 & 3 Power espectrum for the PCs 3 & 4 220 PCs 5 & 6 2 PC 4 140 200 pairs of oscillatory PCs PC7 180 PCs 5 & 6 120 PC 8 1 obtained for the NAO- 160 PC 9 PCs 7 & 8 100 140 PC10 0 PCs 9&11&13 LISBON index 120 80 PCs 11 & 12 NAO PCs 10 & 12 Periodogram 100 PC 13 & 14 -1 60 periodogram 80 PCs 14 &15 PC15 60 Power espectrum for the 40 -2 40 pairs of oscillatory PCs 20 20 NAO-LISBON -3 0 NAO-GIBRALTAR obtained for the NAO- 0 0 0,1 0,2 0,3 0,4 0,5 0,6 NAO-BARNSTON 0 0,1 0,2 0,3 0,4 0,5 0,6 -4 GIBRALTAR index 1865 1885 1905 1925 1945 1965 1985 Frequency (cycles/year) Frequency (cycles/year) YEARS l FIRST MC TEST RESULTS. NH: Serie = AR(1) 5,5 Eigenvalues NAO-LISBON l FIRST MC TEST RESULTS. NH: Serie = Percentiles 97.5 & 2.5 Ø Significant EOFs at the M = 40 NH = AR(1) 4,5 4,0 AR(1) 97.5 level: 1, 2, 5-6. Eigenvalues NAO-GIBRALTAR 3,5 Percentiles 97.5 & 2.5 Ø Significant EOFs at the 97.5 3,5 3,0 M = 40 NH = AR(1) 2,5 Ø True confidence level to 2,5 level: 1-2, 3-4. 2,0 Power of EOF-k 1,5 reject this AR(1) noise null 1,5 Ø True confidence level to 1,0 0,5 Power of EOF-k hypothesis: 92% 0,5 reject this AR(1) noise null -0,5 ª NAO-LISBON: 0,0 0 5 10 15 20 25 30 35 40 -0,5 4 eigenvalue rank k 5 0,5 5,5 10,5 15,5 20,5 25,5 30,5 35,5 40,5 hypothesis: 91.5% Eigenvalues NAO-LISBON Eigenvalues NAO-GIBRALTAR Percentiles 97.5 & 2.5 eigenvalues rank k ü Percentiles 97.5 & 2.5 M = 40 NH = AR(1) 11 Reconstructed Components. ¨ EOF 1: F @ 0.019 years-1 (T @ 4 3 M = 40 NH = AR(1) 3 ü -1 It appears an oscillation with period 23 2 53y) ¨ EOFs 1-2: F @ 0.132 years (T 2 Power of EOF-k years (but it is not significant). @ 8y) Power of EOF-k - 1 ¨ EOFs 2-3: F @ 0.13years 1 1 ª -1 0 NAO-GIBRALTAR: ¨ @ ([email protected]) 0,0 0,1 0,2 0,3 0,4 0,5 EOFs 3-4: F 0.419 years 0 0,0 0,1 0,2 0,3 0,4 0,5 -1 Frq assoc with EOF-k (years ) ([email protected]) Frq assoc with EOF-k (years-1 ) ¨ EOFs 5-6: F @ 0.42years- ü 14 Reconstructed Components. 1 @ (T 2.4y) ü It appears a significant oscillation with l FOURTH MC TEST RESULTS.(Excluding EOFs 1-2,3-4,5-6,7-9- l FOURTH MC TEST RESULTS. (Excluding EOFs 1-4, 2-3, 5-6, 9-11- period 4.6 years, that it is not related with the 10, 8,11-12,13-14 ) oscillation with period 5.8 years in Lisbon. 13, 10-12) CNH3:Serie = Oscillation (T=8y) + Oscillation (T=2.4y) + Trend CNH3: Serie =Trend (>50y) + Oscillation (T=8y) + Oscillation (T=2.4y) + ª The filtered index has got the same phase (>50y) + Oscillation (T=2.3y) + Oscillation (T=5.6y) + Oscillation Oscillation (T=5.8y) + oscillation (T=2.2y) + AR(1) than the non filtered index for the most cases, (T=4.5y) + Oscillation (T=2.9y)4,0 + AR(1) 3,5 Eigenvalues NAO-GIBRALTAR Percentiles 97.5 & 2.5 5,5 for both series. 3,0 M = 40 CNH3 Eigenvalues NAO-LISBON 2,5 4,5 Percentiles 97.5 & 2.5 2,0 M = 40 CNH3 ª Filtered indices represent very well the 1,5 3,5 1,0 extreme values series, specially in the minimum Power of EOF-k 2,5 0,5 case. 0,0 1,5 Power of EOF-k -0,5 0 5 10 15 20 25 30 35 40 eigenvalue rank k 0,5 ª The using two different and quasi Ø -0,5 independent data sets Þ a major confidence in Significant EOFs at the 97.5 level: 19 0 5 10 15 20 25 30 35 40 eigenvalue rank k the results obtained. Ø True confidence level to reject this composite null Ø Significant EOFs at the 97.5 level: NONE hypothes: 71.4% Ø True confidence level to reject this composite null hypothes: 70% METHOD CNH3 is accepted § NAO-GIBRALTAR MODEL: CNH3 is accepted •Modelling of the winter-annual NAO index § NAO-LISBON MODEL: based in models ARMA. ¯An oscillation of period 7.8 years,(EOFs 1-2). ¯A non linear trend, that contains the variability with •Adjustment period of the model: 1865-1985 ¯Three oscillations of periods 2.4, 2.3 and 2.9 years periods of more than 50 years and indicated by EOFs 1 (EOFs 3-4, 7-9-10 and 13-14 respectively), (QBO). and 4. •Checking period: 1986-1997 ¯A no linear trend, that contains the variability with ¯ An oscillation of period 7.8 years (EOFs 2-3) •Forecasting period: 1998-2010 periods of more than 50 years, indicated by the EOFs 5- ¯Two oscillations of periods 2.4 year (EOFs 5-6) and 2.23 6. years (EOFs 10-12), (QBO). ¯Two oscillations of periods 4.5 and 5.6 years (indicated Y ¯An oscillation of period 5.8 years (EOFs 9-11-13), that NAO LISBON : by the EOFs 11-12 and 8 respectively), (ENSO). can be associated with the influence of the ENSO on the © 5 5 Model: ARMA(6, 2) ¯A red noise process AR(1).. NAO-GIBRALTAR RCs 1 & 4 4 NAO. RCS 5-6 4 RC1-13 ©Akaike information criterion (AIC): 223 RCS 1-14 NAO-LISBON 3 ¯3A red NAO-BARNSTONnoise process AR(1). NAO-BARNSTON © 2 2 Absolute mean error (period 1865-1985):0.49 1 1 ©Explained variance : 50% 0 0 Y -1 -1 NAO GIBRALTAR : -2 -2 ©Model: ARMA(9, 5) -3 -3 1865 1885 1905 1925 1945 1965 1985 1865 1885 1905 1925 1945 1965 1985 Years Years ©AIC: 190 RCs for NAO-GI RCs for NAO-LI ©Absolute mean error (period 1865-1985): 0.38 index index 4 3 ©Explained variance: 45% OBSERVED VALUE LISBON ONE STEP FORECASTING 3 2 1986-2010 FORECASTING 2 1 1 0 0 -1 Index Index CONCLUSION -1 -2 -3 ª OBSERVED VALUES GIBRALTAR -2 Quantitative forecasts can’t be done using ARMA lineal models. ONE STEP FORECASTING -4 FORECASTIN 1986-2010 FORECASTING FORECASTIN -3 ª Qualitative forecasts can be done about the oscillation phase a year -5 1960 1970 1980 1990 2000G 2010 1960 1970 1980 1990 G 2000 2010 Time (years) early. Time (years).

View Full Text

Details

  • File Type
    pdf
  • Upload Time
    -
  • Content Languages
    English
  • Upload User
    Anonymous/Not logged-in
  • File Pages
    1 Page
  • File Size
    -

Download

Channel Download Status
Express Download Enable

Copyright

We respect the copyrights and intellectual property rights of all users. All uploaded documents are either original works of the uploader or authorized works of the rightful owners.

  • Not to be reproduced or distributed without explicit permission.
  • Not used for commercial purposes outside of approved use cases.
  • Not used to infringe on the rights of the original creators.
  • If you believe any content infringes your copyright, please contact us immediately.

Support

For help with questions, suggestions, or problems, please contact us