Assessing Predictability of Seasonal Precipitation for May-June-July
in Kazakhstan
Tony Barnston, IRI, New York, US
Possible sources of seasonal climate predictability:
1. Tropical sea surface temperature (SST) anomalies such as El Nino and La Nina, or tropical Atlantic or Indian Ocean SST anomalies
2. Land surface anomalies (up to 1-2 months influence)
3. Persistent extratropical atmospheric circulation anomalies, such as the Arctic Oscillation
El Nino
ENSO-based Teleconnections: May-Jun-Jul El Nino
Probability of above normal precipitation
(uses CRU precipitation)
La Nina
ENSO-based Teleconnections: May-Jun-Jul La Nina
Probability of above normal precipitation
(uses CRU precipitation)
Seasonal precipitation forecastsfor May-June-July
for northern Kazakhstan
Using field of 500 hPa height as predictorof Kazakhstan rainfall in May-Jun-Jul
Lagged in time:
March-April 500 hPa is used to predict
May-Jun-Jul rainfall
x Distribution of Skill using Mar-Apr 500 hPa ht
Correlation of precip atpoint X with predictor 500 hPa ht
Using earlier (Mar-Apr) 500 hPa height as predictor for MJJ rain
Cross-validation: 5 years held out, middle one predicted
skill
Using observed tropical SST field as predictorof Kazakhstan rainfall in May-Jun-Jul
Lagged in time:
March-April SST is used to predict
May-Jun-Jul rainfall
Using earlier March SST as predictor
March SST Time Series MJJ Kaz precip
March SST Time Series MJJ Kaz precip
Mode 1 Mode 1
Mode 2 Mode 2
x Distribution of skill using March tropical SST
Cross-validation: 5 years held out, middle one predicted
skillMay-Jun-Jul
Current dynamical model climate predictionsfor May-June-July 2014
North American national multi-model ensemble forecastFor May-Jun-Jul 2014 rainfall
x
North American National Multi-model Ensemble Anomaly CorrelationPrecipitation May-June-July
x
skill
European national multi-model ensemble forecastFor May-Jun-Jul 2014 rainfall
x
North American national multi-model ensemble forecastFor May-Jun-Jul 2014 temperature
x
North American Multi-model Ensemble Anomaly CorrelationTemperature May-June-July
x
skill
European national multi-model ensemble forecastFor May-Jun-Jul 2014 temperature
x
Precipitation Skill IRI Forecasts 1998-2013 May-June-July 0.5-month lead
Heidke hit skill score
x
Using autocorrelations of precipitation
In the 3 states in northern part of Kazakhstan,autocorrelations for precipitation are generallyweak. However, autocorrelations of
July Augustare at least 0.3, and >0.4 at some stations.
Lag correlations of temperature precipitationare very weak during the growing season.
Global warming trend gives opportunity for some skill in seasonal temperature predictions:
With base period in the past, positive temperature anomalies are often a correct forecast.
Time series of monthly anomaly of maximum temperature at station 28698 (Omsk, Russia)
warming?
Warming trend is evident near Northern Kazakhstan
Time series of annual anomaly of maximum temperature at station 28698 (Omsk, Russia)
ConclusionsTropical SST anomalies during months earlier thanMay-June-July have almost no relationship with rainfallor temperature in northern Kazakhstan in May-June-July.
Upper air geopotential height (500 hPa) in preceding months is related only weakly to Kazakhstan precipitation and temperature in May-June-July. A connection with the Arctic Oscillation is weak.
Autocorrelation statistics for precipitation in northernKazakhstan show some July-to-August anomaly persistence.
Dynamical model predictions for Kazakstan show very slight skill for May-June-July precipitation. Fortemperature, skill is present due to warming trends. Anupward temperature trend exists in observations fornorthern Kazakhstan for the May-June-July season.