The tenth IMSC, Beijing, China, 2007 - International Meetings on ...

egion can be explained by the random evolution of the stochastic models, i.e. about 45% of
the observed seasonal-mean rainfall variance lies outside of the regional chain-dependent
model’s capability (which is termed the “potentially predictable” variance). In addition, only a
small fraction of observed years (between 10% to 20% over a given sub-regime) contain
seasonal-mean precipitation anomalies that contribute to this potentially-predictable variance.
Next, the influences of the various component sub-processes – rainfall occurrence, rainfall
coverage, and rainfall intensity – upon the overall precipitation anomalies are analyzed. Study
results indicate that year-to-year variations of all three can produce rainfall anomalies in a
given year. However year-to-year variations in coverage have the greatest impact upon
seasonal rainfall anomalies in each region. Finally, the influence of regional rainfall variations
on precipitation anomalies at single station is analyzed. Study results indicate that the regional
chain-dependent models can capture more than 60% of the station-based variance in the
summertime precipitation. In addition, 70% of the seasonal precipitation variance at a given
station is accounted for by regional variability and only about 30% of the variance is
contributed by local processes.
A study of daily precipitation in North <strong>China</strong> using generalized linear models
Speaker: Yi Wang
Yi Wang
IAP
wangyi@mail.tea.ac.cn
Yan Zhongwei
IAP
Daily precipitation series for North <strong>China</strong> during 1961-2004 was analyzed with
generalized linear models (GLM). A two-stage approach was adopted. Firstly, the pattern of
wet and dry was fitted using logistic regression; secondly, the precipitation amount during wet
days was fitted under gamma distribution. <strong>The</strong> regional daily precipitation sequence was
simulated by incorporating significant large-scale climate indices and interactions between
predictors (e.g. seasonal effect, rainfall autocorrelation and persistence).
<strong>The</strong> results suggested most influential factors affecting precipitation in North <strong>China</strong>, as
listed below.
(1) North Pacific pattern (NP): a striking negative correlation exits between annual wet days
in North <strong>China</strong> and the NP index. <strong>The</strong> positive phase of the NP pattern is associated with less
annual wet day numbers in North <strong>China</strong>. <strong>The</strong> NP also influences the summer rainfall intensity
in North <strong>China</strong>. <strong>The</strong> positive (negative) phase of NP corresponds to deficient (excessive)
rainfall.
(2) <strong>The</strong> Arctic Oscillation (AO): this pattern is found to be significantly correlated with the
summer precipitation amount in North <strong>China</strong>. A larger AO index corresponds to weaker
precipitation.
89