This study proposes a numerical method to estimate the variance in weather noise from monthly data within a specific season, and thereby the associated SPP. Applying it to the centennial long Tmaxand Tmin records in Taiwan, the results allow us to test the degree of fidelity of three assumptions made behind the usual analytical method in a monsoonal climate regime.
Firstly, the month-by-month variances in weather noise are not stationary and linear. The numerical method relaxes the linear assumption and results in a better coherence with the direct daily estimates.
Secondly, the stationary assumption of inter-monthly correlations is less realistic in JJA and SON for Tmax and in DJF and MAM for Tmin, which are qualitatively consistent with the direct daily data estimates. Finally, the correlations between two months which are one month apart, i.e. the 13, are not negligible in some cases. This non-zero 13 implies the existence of some low-frequency weather noise signals regularly visiting the same place at least for a considerable number of years. Indeed, the idea of fast annual cycle has been proposed (LinHo and Wang, 2002) to represent the climatological intraseasonal oscillation (CISO, Wang and Xu, 1997) resident in the East Asian summer monsoon regions. During the boreal summer the seasonal march of the East Asian subtropical rainfall band (Meiyu in China and Baiu in Japan; e.g. Hsu et al., 2014) is best described by the CISO activity, which takes about two months to advance northward from the northern part of the South China Sea in the middle of May to northern China by the end of July (e.g. Liu et al., 2008). The presence of a significant negative 13 value in some cases, which contributes to a decreased variance of weather noise and thus an add-on source of SPP, could reflect the passage of the alternative dry and wet phases of CISO over Taiwan. On the other hand, a significant positive 13 value likely signifies the prolonged residence of CISO activity. Put together, in a monsoonal environment the zero 13 assumption should be judged with cautions instead of an ad hoc decision.
The estimated weather noise componentsof Tmax (Tmin) are large for stations in the northwestern (western) portion of Taiwan in DJF but relatively small for stations to the southeast in SON (JJA). Such a spatiotemporal contrast is attributable to the combined effects of prevailing wintertime westerly disturbances, the barrier role of CMR and thermal adjustments (e.g. through the sea breeze circulation) from the adjacent Pacific Ocean during the summertime.
The proposed numerical approachincreases the estimates of Tmax (Tmin) in DJF (DJF and SON) but reduces them otherwise (in MAM and JJA). The resultant SPP changes in wintertime Tmax
then show a -6% reduction on average, which are in contrast with the widespread increases in other seasons (c.f. Table 4). Note the different response at Taipei in JJA. In response to the large decrease of , our new approach significantly increases the SPP of Tmax (from 37.7% to 56.2%), which is most welcome considering its implications for the potential socio-economic benefits. As for the Tmin, the new method generally reduces the SPP in cold DJF and SON seasons but raises it in warm MAM and JJA seasons. Such a tendency in reducing the SPP of Tmin in cold seasons is intentionally not what we
desire.
Annually, the numerical method shows relatively low seasonal-averaged SPP of Tmax at Taipei (33.5%) and Taichung (36.9%). Both cities are located in the northwestern portion of Taiwan. As well recognized, weather in this region is intermittently affected by the mid-latitude westerly disturbances almost throughout the year. The fast urbanization in both cities is another possible cause. As for the Tmin, the highest and the lowest SPP, 63.7% vs. 37.3%, appear at Penghu and Tainan, respectively. Although the geographical distance is short, Penghu is located in the middle of the Formosa Strait whereas Tainan is located in the wind shield as well as rain shadow (windward) region during the northeasterly (southwesterly) monsoon season. In Taiwan, the overall means of SPP in both Tmaxand Tminare 47.6%.
The results of trend effect examination showed that the coherency between a warming trend and an increased SPP is significantly higher in Tmin than in Tmax. The overall mean of trend-remained SPP for Tminin Taiwan is around 75%, a 27% increase compared to the de-trended SPP. On the other hand, the overall mean of trend-remained SPP for Tmax(48.5%) is roughly equal to that of the de-trended estimate (47.6%).Fig. 6 further examined the trend efficiency on the SPP, which is defined as the linear slope of changed SPP (SPP; trend-remained minus de-trended estimates) regressed on the linear trend values () of seasonal-mean temperature extremes of the employed stations. The estimated efficiency of all-season regression is 16.1% (2.0%) per 1.0℃/100-yr warming for Tmin(Tmax).
When the world is warming, the hydrological cycle is becoming more active. The role of the resultant cloud seems to play a different role from day to night in affecting the Tmax and Tmin, respectively. The increasing SPP of Tmin, which is in accordance with the increasing nighttime cloud amount (Weng, 2010), suggests the stabilized role (i.e. negative feedback) of hydrological cycle during the nighttime hours. While the surface cools down after sunset, the blanket effect of nighttime cloud tends to warm the lower troposphere thereby increasing the atmospheric stability and SPP. In SON, the above effect may reach its climax when the water vapor is profoundly supplied from the warm oceans.
Note that in western north Pacific, the SST reaches its maximum in September. On the other hand, the almost unchanged SPP of Tmax, which is also in accordance with the controversial change of daytime cloud amount in different stations and different seasons (Weng, 2010), indicates that the response of daytime hydrological cycle is not unified yet but tends to be case dependent.
We have noticed that the trend efficiency in JJA drops to 4.0% (increases up to 21.4%) per 1.0℃
/100-yr warming for Tmin (Tmax), mainly attributable to the unusual behaviors at Taidong and Taipei (Penghu and Henchun). Note that in JJA Taidong and Taipei rank at the top two stations of de-trended SPP of Tmin (c.f. Table 5) whereas Penghu and Henchun have the lowest de-trended SPP of Tmaxamong stations. It seems that the high (low) de-trended SPP at former (later) two stations leaves no (plenty) room for the warming trend to increase the de-trended SPP further and our question becomes: why are the de-trended SPP of Tmin(Tmax) at Taidong and Taipei (Penghu and Henchun) so high (low). Taidong is located in the exit region of the East Rift Valley of Taiwan. On summer nights the local weather is controlled by the downslope winds from the CMR to the west and frequent foehn events induced by the
prevailing southwestlies. Taipei is located in a basin bounded by the Yangming Mountain to the north/northwest andthe Xueshan Mountain Range to the south/southeast. The downdraft associated with the prevailing easterlies stabilizes the nighttime air aloft. On the other hand, Penghu and Henchun are two islandic/rural cities deeply affected by the adjacent open oceans. Local weather on summer days is controlled by the sea breeze circulation. Both cities have relatively low mean seasonal-mean Tmax but high standard deviation in seasonal-mean Tmax. Put together, it suggests that the interaction between local topography and monsoon winds could be an influential factor in the SPP of temperature extremes.
Results of the 60-yr sliding time window analysis show the existence of long-term variability in de-trended SPP time series for both Tmaxand Tmin. The SPP series in both Tmaxand Tmin are more stable in DJF. Since the linear trend has been removed, such stability may be in response to the dominance of the warming trend over other factors in supporting the SPP during boreal winters. Notwithstanding, because it is only the linear part of trend removed, the eccentric de-trended SPP series such as the existence of long term tendency and recent increasing/decreasing tendency found at different stations and in different seasons may in parts signify the nonlinearity of warming/cooling trends in temperature extremes (Weng 2010). On the other hand, the de-trended SPP series at some stations in some seasons show a flip-flop between the increasing and decreasing tendency, or vice versa. This implies that factors other than the long term trend, such as the phase of atmospheric low-frequency modes (e.g. Arctic Oscillation; c.f. Lorenz, 1951, Thompson and Wallace, 1998; Pacific Decadal Oscillation; c.f. Yu et al., 2015), multi-decadal ENSO variability (e.g. Wang, 1995; Wu and Wang, 2002) and variability of CISO possibly interacting with the ENSO and atmospheric low-frequency modes (e.g. Ding, 2007), could contribute to the long term variability of SPP. For example, the CISO promotes the summertime SPP in Taiwan, and its variabilitywould disturb its promising signal. For practical prediction needs, it would be of interest to understand thesepossible sources of SPP and how the variations in them can affect the SPP of local climates. Using the proposed numerical method, the above tasks are underway and the results will be reported in future papers.
Acknowledgment
The author wishes to thank Mr. Chen-Dau Yang for the figures preparation. This study was sponsored by the Ministry of Science and Technology of Taiwan under Grants MOST106-2621-M-865-001-.
References
Barzilai, J., and Borwein, J. (1988). Two-point step size gradient methods, IMA Journal of Numerical
Analysis, 8: 141-148.
Brown, L. (2011). World on the edge: how to prevent environmental and economic collapse. W. W.
Norton, New York.
Buishand, T.A., and Beersma, J.J. (1993). Jackknife tests for differences in autocorrelation between climate time series. Journal of Climate, 6: 2490-2495.
Ding, Y.H. (2007). The variability of the Asian summer monsoon. Journal of the Meteorology Society of
Japan, 85: 21-54.
Feng, X., Delsole, T., and Houser, P. (2014). Comparison of seasonal potential predictability of precipitation. Journal of Climate, 27: 4094-4110.
Frederiksen, C.S., and Zheng, X. (2007). Variability of seasonal-mean fields arising from intraseasonal variability. Part 3: Application to SH winter and summer circulations. Climate Dynamics, 28:
849-866.
Grainger, S., Frederiksen, C.S., and Zheng, X. (2009). Estimating the potential predictability of Australian surface maximum and minimum temperature. Climate Dynamics, 32: 443-455.
Hsu, H.H., Zhou, T., and Matsumoto, J. (2014). East Asian, Indochina and western north Pacific summer monsoon: An update. Asia-Pacific Journal of Atmospheric Sciences, 50: 45-68.
Hung, C.W. (2009). Temperature discontinuity caused by relocation of meteorological stations in Taiwan. Terrestrial, Atmospheric and Oceanic Sciences, 20: 607-617.
Kumar, A., and Hoerling, M.P. (2000). Analysis of a conceptual model of seasonal climate variability and implications for seasonal prediction, Bulletinof American Meteorological Society, 81: 255-264.
Leith, C.E. (1973). The standard error of time-average estimates of climatic means. Journal of Applied
Meteorology, 12: 1066-1069.
LinHo, and Wang, B. (2002). The time-space structure of the Asian-Pacific summer monsoon: A fast annual cycle view, Journal of Climate, 15: 2001-2019.
Liu, J., Wang, B., and Yang, J. (2008). Forced and internal modes of variability of the East Asian summer monsoon. Climate of the past Discussions, 4: 225-233.
Lorenz, E.N. (1951). Seasonal and irregular variations of the Northern Hemisphere sea-level pressure profile. Journal of Meteorology, 8: 52-59.
Lorenz, E.N. (1963). Deterministic nonperiodic flow. Journal of Atmospheric Sciences, 20: 130-141.
Madden, R.A. (1976). Estimates of natural variability of time averaged sea level pressure. Monthly
Weather Review, 104: 942-952.
Meehl, G.A., and Tebaldi, C. (2004). More intense, more frequent, and longer lasting heat waves in the 21stcentury. Science, 305: 994-997.
Mercer, J.B. (2003). Cold: an underrated risk factor for health. Environmental Research, 92: 8-13.
Narapusetty, B., Delsole, T., and Tippett, M.K. (2009). Optimal estimation of the climatological mean.