Download - Peter J. Lamb 1, 3, Mostafa El Hamly 1, 2, Rachid Sebbari ...xtide.ldeo.columbia.edu/~visbeck/nao/poster/Lamb.pdfPeter J. Lamb 1, 3, Mostafa El Hamly 1, 2, Rachid Sebbari 1, 2, Diane

Transcript
  • Atlantic

    Spain

    Algeria

    Atlantic

    Spain

    Algeria

    Atlantic

    Spain

    Algeria

    Atlantic

    Spain

    Algeria

    Atlantic

    Spain

    Algeria

    Atlantic

    Spain

    Algeria

    NOV DEC

    JAN FEB

    MAR APRMed. Med.

    Med.

    Med.

    Med.

    Med.

    (b)EOF1 of Rogers' NAO

    1922-95

    EOF

    load

    ing

    Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr-1.0

    -0.5

    0.0

    0.5

    1.0

    -1.0

    -0.5

    0.0

    0.5

    1.0

    Total Explained Variance 18.4%

    -3.0-2.5-2.0-1.5-1.0-0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0

    -3.0-2.5-2.0-1.5-1.0-0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0

    30 40 50 60 70 80 90

    EOF

    tim

    e co

    effi

    cien

    t

    Year

    Prediction of the NAOThe NAO lacks the persistence of theSouthern Oscillation, making it more chal-lenging to use as a predictor.

    One prediction approach that we haveused is to try and eliminate the possibilityof an extreme mid-winter NAO event withreal-time monitoring of the NAO evolution.Fig. (a) documents the Aug-Jan evolutionof the NAO during seven seasons since1932 that culminated in the most extremepositive January values. A distinct Aug-Nov-Jan NAO oscillation is evident for sixof these seven seasons, and its inverse isalso partly characteristic of the evolution ofthe seven most extreme negative Jan NAOv a l u e s ( n o t s h o w n ) . E x t r e m epositive/negative January NAO values areassociated with Moroccan precipitation thatis below-average/above-average.

    The Aug-Nov-Jan NAO evolution that waspresent in Fig. (a) is also evident in EOF1of Rogers' NAO (Fig. (b)). Note that in thelate 1950s-1960s, an evolution that culmi-nated in a negative January was moreprevalent. This situation was reversed inthe 1980s-1990s.

    Historical analogue analyses of precipita-tion data for the 10 years that had the mostsimilar NAO evolutions were also used inthe seasonal prediction. These analyseswere updated monthly.

    1 1984 2 1990 3 1983 4 1989

    5 1986 6 1938 7 1974

    x

    x

    xx

    x

    x

    +

    +

    +

    ++ +

    x 1996-97

    Rog

    ers'

    NAO

    Inde

    x

    -5

    -4

    -3

    -2

    -1

    0

    1

    2

    3

    4

    5

    Aug Sep Oct Nov Dec Jan

    1

    2

    3

    4

    5

    67

    Nor

    th W

    est

    Sout

    h W

    est

    Atla

    s M

    ount

    ains

    2,5

    3

    46

    7,1 1,7

    2,6

    3

    45

    Fig. (a) shows the seasonal evolution of theNAO during seven seasons that culminatedwith the most extreme positive January values.Also shown are the NAO evolution curves forthe 1996-97 and 1997-98 seasons whose pre-diction verification will be presented in the nextpanel. The precipitation indices for the threeMoroccan regions that are most affected bythe NAO are also indicated for these sevenseasons.

    Seasonal Evolution of the NAO(a)

    1997-98+

    Reg

    iona

    l Pre

    cipi

    tatio

    n In

    dex

    in σ

    The month-to-month evolutions of the NAO during the (a)1996-97, (b) 1997-98,(c) 1998-1999, and (d) 1999-2000 winters and the precipitation anomalies for thenorthwest (NW), southwest (SW) and mountain (MT) regions of Morocco.

    Verification of Prediction

    ·

    ·

    ·

    · ·

    ··

    ·

    ·

    ··

    ·

    ·

    · ·

    ·

    ·

    ·

    -4

    -2

    0

    2

    4

    -4

    -2

    0

    2

    4

    NWSWMT

    NWSW

    NWSWMT

    NWSWMT

    ·

    ··

    ··

    ··

    ··

    ·

    ·

    ··

    ··

    · ·

    ·

    -4

    -2

    0

    2

    4

    -4

    -2

    0

    2

    4

    NAO

    (c) 1998-99

    · · ·· ·

    ·

    · · ·· ·

    ·

    · ··

    · ·

    ·

    -4

    -2

    0

    2

    4

    -4

    -2

    0

    2

    4

    NAO

    (d) 1999-00

    Nov99

    Dec99

    Jan00

    Feb00

    Mar00

    Apr00

    Nov98

    Dec98

    Jan99

    Feb99

    Mar99

    Apr99

    NAO

    (a) 1996-97

    · ·

    · ·

    · ··

    · ··

    · ··

    ·

    ··

    ··

    -4

    -2

    0

    2

    4

    -4

    -2

    0

    2

    4

    NAO

    (b) 1997-98

    MT

    Nov96

    Dec96

    Jan97

    Feb97

    Mar97

    Apr97

    Nov97

    Dec97

    Jan98

    Feb98

    Mar98

    Apr98

    Peter J. Lamb 1, 3, Mostafa El Hamly 1, 2, Rachid Sebbari 1, 2, Diane H. Portis 1, M. Neil Ward 1, 4

    1Cooperative Institute for Mesoscale Meteorological Studies, The University of Oklahoma, Norman, Oklahoma 730192Moroccan Direction de la Météorologie Nationale (DMN), Casablanca, Morocco3School of Meteorology, The University of Oklahoma, Norman, Oklahoma 73019

    4International Research Institute for Climate Prediction, Lamont-Doherty Earth Observatory of Columbia University, Palisades, New York 10964

    NAO Control of Moroccan PrecipitationThe "storm track" during winter months characterized by nega-tive NAO has a west-east orientation across the central NorthAtlantic. This is very favorable for cyclonic and frontal systemsto affect the coast of northwest Africa. During positive NAOwinter months, in contrast, the southwest-northeast storm trackorientation steers weather systems away from the northwestAfrican coast.

    NAO control over Moroccan precipitation maximizes over thenorthwest and southwest regions of Morocco.

    NAO control over Moroccan precipitation steadily increasesfrom November through February, after which it decreases sig-nificantly during March and then becomes nonexistent in April.

    Monthly Correlation of NAO and Moroc-can Precipitation (1932-96)

    -0.1 -0.3 -0.5

    N.Am.

    Gr.Gr.

    Afr. Afr.

    N.Am.

    Strong + NAO Months Strong - NAO MonthsFrequency of Cyclones (After Rogers, 1990)

    Abstract

    An ongoing collaborative effort with the Kingdom of Morocco, that commenced in 1994, isattempting to increase our understanding of the interannual-to-decadal variability of Moroccan wintersemester (November-April) precipitation, and to use this knowledge to develop a seasonal precipita-tion prediction capability. This project was motivated by the predominance of extremely poor Moroc-can winter-precipitation seasons since the late 1970s. A major focus is on the role of North AtlanticOscillation (NAO) for the variability and predictability of Moroccan precipitation. The NAO control onMoroccan precipitation maximizes over the northwest and southwest regions of the country. Thatcontrol also increases steadily from October until January-February, after which it decreases signifi-cantly during March and then becomes non-existent in April.

    Since the NAO Index is much less persistent (on a month-to-month basis) than the SouthernOscillation Index, it is especially challenging to use the NAO as a tool for the seasonal prediction ofMoroccan precipitation. However, the evolution of the NAO in extreme years offers some promise inthat regard. A distinct August-November-January NAO oscillation is evident for six of the seven sea-sons since 1932 with the most extreme positive January NAO values, and its inverse is also partlycharacteristic of the evolution of the seven most extreme negative January NAO values. Extremepositive (negative) January NAO values tend to be associated with Moroccan precipitation that isbelow-average-to-very-deficient (above-average-to-abundant). The development/nondevelopment ofthe above oscillation or its inverse during a particular August-November period thus provides a basisfor the prediction/elimination of an extreme January NAO value (and its accompanying Moroccanprecipitation) with some level of confidence. The foregoing January characteristics also tend to per-sist into February.

    The NAO control of Moroccan precipitation is weak at the beginning (November-December) andthe end (March-April) of the precipitation season. Results from Canonical Correlation Analysis (CCA)of historical data, and Global Climate Model (GCM) experimentation using the ECHAM4 model, indi-cate that tropical Pacific SSTs can be used to predict late-season Moroccan precipitation (March-April) especially for the western regions of Morocco. However, the large decline in December precipi-tation during recent decades has not yet been associated with other known factors in the climatesystem.

    In this presentation, we review the basis for and nature of the seasonal predictions issued for thewinter semesters of 1996-2000, and their subsequent evaluations. To date, this effort has used thetraditional fixed 2-point Rogers' NAO index. We also attempt an assessment of the utility of the pre-dictions for the management of Moroccan agricultural production, water resources, and public policy.

    It is clear that there was a strong negative NAO control onthe 1996-97 Moroccan precipitation season (Fig. (a)),which followed a pattern that has been especially charac-teristic of the 1990s. As we predicted, there was a largenegative Jan NAO value and above average (but notextreme) Jan precipitation. However, rather than peakingin Jan and persisting into Feb, as we had anticipated, thislarge negative NAO event maximized in Dec and then per-sisted with some weakening into Jan. By Feb the NAOhad reverted to extremely high positive values and wasaccompanied by substantial precipitation deficienciesacross the NW, SW and MT regions of Morocco.

    The negative NAO control on the 1997-98 precipitationseason (Fig. (b))was weaker than for 1996-97 and mostother winters during the 1990s. However, as we predicted,the early season (Nov-Dec) precipitation was not deficient,the extremely large negative Oct 1997 NAO Index value(see Fig. (a) in Prediction panel) did not persist throughDec-Feb, and precipitation was not abundant for extendedperiods during the season. Furthermore, with the excep-tion of ~10 days overlapping Jan and Feb 1998, the mid-Dec through Feb precipitation was consistent with our pre-diction of the below-average-to-very-low range for much ofthe upcoming precipitation season.

    The evolution of 1998-99 NAO (Fig. (c)) involved unusuallystrong persistence in the posit ive mode that, at aminimum, prevents Morocco from receiving abundantprecipitation. This evolution was largely consistent withour Prediction Statement #1 (Nov 1, 1998), subject to theDec 15, 1998 revision for Jan, 1999. The focused predic-tion statements were strongly validated by the precipitationoutcomes. While the late February precipitation reducedwhat would otherwise have been a severe deficiency forthat month, it apparently came too late in the month to beof significant assistance to agriculture.

    This 1999-2000 NAO evolution (panel (d)) was consistentwith Prediction Statement #1 (Nov 1, 1999) for Jan but notfor Feb. The validation of precipitation prediction state-ments was not as strong as for 1997-98 and 1998-99. Theprediction of a low probability of extremely deficient precip-itation was correct for Dec and Jan, but in these monthsthe moderately dry conditions were probably slightly drierthan anticipated. The almost total absence of precipitationin Feb ran strongly counter to our prediction.

    ··

    ·

    ·

    ·

    ··

    Northwest

    -0.6

    -0.4

    -0.2

    0.0

    0.2

    0.4

    0.6

    ··

    ·

    · · ·

    ·

    Southwest

    -0.6

    -0.4

    -0.2

    0.0

    0.2

    0.4

    0.6·

    ·

    ·

    · ··

    ·

    Mountain

    -0.6

    -0.4

    -0.2

    0.0

    0.2

    0.4

    0.6

    · ··

    ·

    ·

    ·

    ·

    Southeast

    -0.6

    -0.4

    -0.2

    0.0

    0.2

    0.4

    0.6

    ·

    ·

    ··

    ·

    ·

    ·

    Northeast

    -0.6

    -0.4

    -0.2

    0.0

    0.2

    0.4

    0.6

    Oct Nov Dec Jan Feb Mar Apr

    0

    5

    10

    15

    20

    25

    30

    0

    5

    10

    15

    20

    25

    30

    Prediction ChallengesThe recent series of very dry years, coveringapprox imate ly 1977-1978 to 1994-1995,appears rather unusual and may not be solelyattributable to the NAO.

    The NAO does contain a strong upward wintertrend, but this trend is centered on the month ofMarch. From the upper panel, we see that thisupward trend explains 25% of the March NAOvariance over the 1960-94 period.

    Along the northwest and southwest coasts,where the NAO explains the most precipitationvariance, the precipitation downtrends are cen-tered on December (see bottom panel), forwhich there has been no pronounced NAO trend(top panel) . Farther in land, in the At lasMoun ta ins , the December p rec ip i ta t i ondecreased by 64% during this period. In thesoutheast, there has been a dramatic increasein February precipitation.

    Currently, we are focusing on the origin and pre-dictability of the fraction of Moroccan precipita-tion variance that is not explained by the NAO --particularly in the early winter.

    Prediction-related references:Ward, M.N., P.J. Lamb, M. El Hamly, R. Sebbari, and D.H.

    Portis, 1999: Climate Variability in Northern Africa: Under-standing Droughts in the Sahel and the Maghreb. In:Beyond El Niño: Decadal Variability in the Climate System,Anton io Navarra , ed i tor , Spr inger Ver lag, Ber l in ,Heidelberg, New York, 119-140.

    El Hamly, M., R. Sebbari, P.J. Lamb, M.N.Ward and D. Portis,1999: Towards the Seasonal Prediction of Moroccan Pre-cipitation and its Implications for Water ResourcesManagement. Proceedings of Abidjan '98: InternationalConference on Water Resources Variability in Africa duringthe XXth Century, Abidjan, Côte d'Ivoire, 16-19 November1998, International Association of Hydrological Sciences,Oxfordshire, UK, No. 252, 79-87.

    Calendar monthly average of normalized precipitation departures for all five Moroccan regionsfor 1979-94, with reference to the 1932-98 base period.

    The first coupled mode identified by the Canonical CorrelationAnalysis of November-January tropical Pacific SST and Febru-ary-April Moroccan precipitation (1951-1995). The tropicalPacific SST was represented by its first eleven EOFs (84.5% ofthe total variance) and Moroccan precipitaiton was representedby five EOFs (~100% of the total variance).

    Other Climatic ControlsTropical Pacific SSTs can be used to pre-dict late-season Moroccan precipitation(February, and especially March-April)several months in advance, with a level ofskill well above that achievable by chance.There is a negative association of latewinter precipitation in the western regionsof Morocco with midwinter central-to-eastern tropical Pacific SSTs.

    Specifically, this exploited the tendency forlate season Moroccan precipitation to bemore/less plentiful in years when LaNiña/El Niño conditions prevail.

    Perc

    ent V

    aria

    nce

    in L

    inea

    r Tre

    nd

    Northwest PrecipitationNAO

    Period:1960-1994

    Oct Nov Dec Jan Feb Mar Apr

    Oct Nov Dec Jan Feb Mar Apr

    Oct Nov Dec Jan Feb Mar Apr

    Oct Nov Dec Jan Feb Mar Apr

    Oct Nov Dec Jan Feb Mar Apr

    -0.7

    Utility of Prediction Statements: Water Resource Management Example

    AGU Chapman Conference on "The North Atlantic Oscillation", November 28 - December 1, 2000, Orense, Galicia, Spain.

    During the last 30 years, Morocco has significantly improved its water resource management throughthe construction of ~90 large dams (with 14 X 10^9 cubic meters storage capacity) that impound runofffrom Atlas and Rif Mountain precipitation.

    Since these dams are drawn down during the mid- and late-dry seasons, the Moroccan DirectionGénérale de l'Hydraulique is faced with especially crucial water resource management decisions(irrigation, energy, domestic consumption) before the start of each precipitation (and agriculturalproduction) season in Nov.

    Additionally, the recently developed capability to transfer water from previously hydrologically wet to dryregions further complicates the water management decision process.

    Since an imbalance between the initial dam levels and the subsequent precipitation can lead to severeflooding or water shortages, the potential value of seasonal predictions of Moroccan precipitation of thetype discussed above is very high. Recent international surveys have identified Morocco as a countrythat will need continued water resource development to meet the increasing needs of future genera-tions (e.g., Water International, vol. 20, 1995, p. 177).