1. Introduction
Throughout the last hundred years the National Meteorological Service of Mexico has operated a substantial number of climatological stations (reaching more than 4000 in 1982); however, the coverage in time and space of these stations is rather inhomogeneous. Available data from most stations dates back to the early 1950s. Since the purpose of this paper is to investigate the possible role of the Pacific decadal oscillation (PDO) in El Niño–Southern Oscillation (ENSO)-related Mexican climate anomalies, first we construct a reliable station-by-station climatological map of Mexico. That is, we choose only the stations that have at least 23 yr of monthly data with no daily gaps. The above results in approximately 1000 nonsimultaneous stations (Graef et al. 2000) for which an ad hoc temperature and precipitation climatology of Mexico was obtained [see Reyes et al. (2000) for the precipitation climatology]. The most prominent features of this climatology are as follows: 1) a rainy season over most of the country with a monsoonal behavior, starting in April at the Gulf of Mexico versant (GMV), extending northwestward throughout the summer, reverting by early September, and ending in November again at the GMV; 2) a region of Mediterranean climate over the northwestern region of the Baja California peninsula with a wintertime rainy season, in contrast to 1; 3) a region of dry climate in the northern half of Mexico; and 4) tropical and subtropical regimes, with mean annual temperature between 10° and 26°C for more than 90% of the country, with much higher annual temperature variability in the northern semiarid states than in the southern humid states. [For a more general discussion of Mexican climate the reader is referred to the work of Mosiño and García (1974).]
Several works addressing the effects of El Niño on Mexican climate and climate-related activities have been made in recent years (see, e.g., Galindo and Mosiño 1992; Magaña 1999; and the special issue of Geofísica Internacional 2003, Vol. 42, No. 3). Since during El Niño winters there is a southward shift of the intertropical convergence zone (ITCZ), the jet stream is displaced several hundred kilometers southward (allowing midlatitude cyclonic systems to reach farther south than during non–El Niño years). These phenomena have important climatic consequences for Mexico. El Niño winters show a relative increase in precipitation in the most northwestern and most northeastern regions of Mexico as well as in the Yucatan Peninsula, but a relative decrease in some parts of the southern states (see Fig. 2.5.a of Magaña 1999). During El Niño winters there is a generalized drop of temperature over central and northern Mexico. Only the Baja California Peninsula presents a considerable positive temperature anomaly, while in the south and southeast of the country the temperature anomalies do not show a clear pattern. La Niña winters exhibit precipitation and temperature anomalies almost opposite to the ones produced by El Niño, but somewhat less well defined.
The PDO, a lower-frequency oscillation compared to ENSO, is a decadal mode of the sea surface temperature spatial structure of tropical Pacific variability (see, among others, Trenberth 1990; Graham 1994; Latif and Barnett 1994; Minobe 1997; Zhang et al. 1997), and thus its associated index is based on sea surface temperature (Mantua et al. 1997). The North Pacific Oscillation (NPO), a higher-frequency oscillation compared to the PDO, is also a decadal mode of variability of the Pacific (see, among others, Bliss and Walker 19321; Rogers 1981), but its associated index is based on sea level pressure (Trenberth and Hurrell 1994). Therefore, both PDO and NPO are related to the variability of the Pacific coupled ocean–atmosphere system. In this work we select the lower-frequency PDO index as a tag to stratify ENSO years; however, using the NPO index makes almost no difference. Thus, the suggestion of Schneider and Cornuelle (2005) that the stratification of climate anomalies be based on the underlying indices of ENSO and the NPO rather than ENSO and PDO is not relevant in this case.
The PDO–ENSO relationship has been recently studied (see, e.g., Newman et al. 2003; Yeh and Kirtman 2004; Schneider and Cornuelle 2005) addressing, most importantly, the ENSO forcing of the PDO. The influence of the NPO and/or PDO on ENSO was suggested by Gershunov and Barnett (1998), Gershunov et al. (1999), and Bove and O’Brien (2000). More specifically, McCabe and Dettinger (1999) investigated the decadal variations in the strength of ENSO teleconnections with precipitation in the western United States. And, by sorting ENSO events by warm phase (El Niño) and cold phase (La Niña) according to the prevailing high or low NPO phase, Gershunov et al. (1999) showed for the continental United States that El Niño years during the high phase of the NPO exhibit larger precipitation anomalies (using the mean heavy precipitation frequency anomalies) than during El Niño–only years. In other words, El Niño years in combination with the high phase of the NPO, constructively interfere to produce larger anomalies in rainfall, with more precipitation than normal in the western United States and less in the central and northeastern United States. They also showed that La Niña years during the low phase of the NPO produce larger anomalies than just La Niña years; that is, La Niña plus the low phase of the NPO seem to make a more pronounced effect on the climate of the United States than La Niña alone.
All these previous works constitute the basic motivation of this study. We shall try to elucidate if the tropical Pacific decadal variability (as measured by the PDO index) influences the way ENSO affects the climate of Mexico. Although Newman et al. (2003) question the validity of stratifying the extratropical response to ENSO by warm-phase or high PDO (HiPDO) and cold-phase or low PDO (LoPDO) events, especially during summer, we consider that since Mexico is mostly tropical and subtropical the questioning of Newman et al. (2003) might not be appropriate in this study. That is, by applying statistical techniques, we hope to be able to recognize quantitatively indices of ENSO, PDO, or ENSO–PDO combined that could be used for empirical prediction of Mexican seasonal climate anomalies.
2. Methodology
We use the PDO instead of the NPO but basically we attempt to follow the method of Gershunov et al. (1999) with the following differences. First, for all stations and for both precipitation and temperature we define standardized monthly anomalies, α = [x − μ(x)]/σ(x), where x is the mean temperature (or total rainfall) of the month, μ(x) is its climatological mean value, and σ(x) is its standard deviation. Second, for each station available year we construct winter and summer seasonal anomalies defined here as 3-month-averaged standardized monthly anomalies; that is, we average the anomalies of January, February, and March for winter and the anomalies of July, August, and September for summer. Third, we average the seasonal anomalies for the years of El Niño-HiPDO (years of El Niño during a high or warm PDO phase), La Niña-HiPDO, El Niño-LoPDO, La Niña-LoPDO (all of which are respectively defined in a similar fashion as above), El Niño, and La Niña. For example, considering a station having data for all the years listed in Table 1, we average 11 seasonal anomalies for El Niño-HiPDO, 6 seasonal anomalies for La Niña-HiPDO, 6 seasonal anomalies for El Niño-LoPDO, 10 seasonal anomalies for La Niña-LoPDO, 17 seasonal anomalies for El Niño, and 16 seasonal anomalies for La Niña. After these three steps we have for each station winter and summer averaged seasonal anomalies during the six cases: El Niño years, La Niña years, and the four combinations of ENSO and PDO phases. Fourth, for these six cases we compute the percentage of the ∼1000 stations whose averaged seasonal anomalies (winter and summer) go beyond arbitrary thresholds: ±0.50 for precipitation and ±0.75 for temperature (see Table 2, upper section).
a. Statistical significance test
To test the statistical significance of the percentage of stations whose seasonal anomalies exceed the defined thresholds we perform a Monte Carlo simulation. This is done as follows: for each station we shuffle all the seasonal anomalies and average them according to ENSO phase sample size as stated in Table 1. For example, consider a station that has seasonal anomalies for all the years listed in Table 1; for El Niño we would average 17 anomalies randomly chosen from all available anomalies and, similarly, for La Niña we would average 16 anomalies. We repeat this 100 times (since all stations have at least 23 yr of seasonal anomalies, the number of possible combinations of ENSO years is much larger than 100), and each time we compute the percentage of stations whose averaged seasonal anomalies go beyond the defined thresholds. Next, for each station we shuffle separately the El Niño and the La Niña years’ seasonal anomalies and average them according to the PDO phase sample size. For example, consider again a station that has seasonal anomalies for all the years listed in Table 1; for El Niño-HiPDO we would take 11 out of 17 El Niño years, for El Niño-LoPDO we would take 6 out of 17 El Niño years, for La Niña-HiPDO we would take 6 out of 16 La Niña years, and for La Niña-LoPDO we would take 10 out of 16 La Niña years. We repeat this also 100 times, excluding those stations with insufficient data such that the number of possible combinations of ENSO–PDO years is less than 100, each time computing the percentage of stations whose averaged seasonal anomalies go beyond the defined thresholds. Finally, we average the percentages obtained over the 100 realizations, calculate their standard error (see Table 2, lower section), and compare them with the true percentages first obtained with the original data (Table 2, upper section). A true percentage of stations would be considered statistically significant if it exceeds the Monte Carlo–averaged percentage plus its standard error (see Table 2).
3. Results
The most noticeable result is that in general there is little effect of ENSO on Mexican climatological seasonal anomalies of precipitation and temperature. This is indicated by the low percentages of stations, between 0% and 3%, whose averaged seasonal anomalies exceed the thresholds when all El Niño and all La Niña years are considered (see the columns identified as ALL in Table 2). Only when ENSO years are classified according to the prevailing PDO phase do the results show particular PDO–ENSO effects on Mexican climate. This is indicated by the corresponding percentages, between 1% and 19%, which are relatively high compared to the above-mentioned percentages (see the columns identified as LoPDO and HiPDO in Table 2). In the latter case the average seasonal anomalies are taken over fewer years and thus more variability is expected; nevertheless, since the percentages are not always inversely proportional to the number of years averaged, this shows that the PDO affects the anomalies produced by ENSO in different manners.
a. Quantitative results
After sorting ENSO years by El Niño and La Niña events and prevailing PDO phase, HiPDO and LoPDO (see Table 2), it was found that
for precipitation during winters of HiPDO, where 14% of stations are wet versus 1% dry, and during summers of LoPDO, where 11% of stations are wet versus 1% dry, El Niño favors “wet conditions,” defined here as a statistically significant higher percentage of stations reporting values of its seasonal anomaly greater than 0.5, compared to the percentage of stations reporting values less than −0.5.
for mean temperature, “cooler conditions” (defined similarly as above, but with anomalies less than −0.75, compared with anomalies greater than 0.75) are favored during La Niña summers and during El Niño winters, regardless of the PDO phase (see Table 2). Indeed, in La Niña summers 9% of stations are cool versus less than 1% warm during LoPDO and 19% cool versus 5% warm during HiPDO, whereas in El Niño winters it is 6% cool versus 3% warm for LoPDO and 8% cool versus 3% warm for HiPDO. However “warmer conditions” (defined similarly as above, but with anomalies greater than 0.75, compared with anomalies less than −0.75) are favored by the HiPDO during summers of El Niño (10% warm versus 3% cool).
b. Regional aspects
The case El Niño-HiPDO corresponds to a constructive interference (Gershunov et al. 1999) that is particularly important in the northwestern part of the Baja California Peninsula (see Fig. 1), where the precipitation season is during the wintertime. Note also the wet conditions in the peninsula of Yucatan, as well as in most states south of 21°N. Furthermore the few negative anomalies in some stations of the southern states, although they are not statistically significant by our chosen arbitrary criteria, to some extent agree with the negative anomalies reported by Magaña (1999) during El Niño.
In the case of El Niño-LoPDO, the anomalously high precipitation during summers occurs mostly in stations south of 22°N, with high concentration of wet stations in the most southern states of Oaxaca and Chiapas (see Fig. 2).
During summers, for La Niña-LoPDO, most of the cool stations are in the central states between 18° and 22°N and in a localized region of the Sierra Madre Oriental around the triple boundary of the states of Nuevo Leon, Tamaulipas, and San Luis Potosi (around 23°N, 99°W), while for La Niña-HiPDO (where we found the highest percentage of all cases: 19%) the cool stations are all over the country. During winters, in the case of El Niño-LoPDO, most of the cool stations are north of the tropic of cancer, and we do not find a single warm station in the Baja California Peninsula and only one warm station in the Yucatan Peninsula; on the other hand, in the case of El Niño-HiPDO most of the cool stations are between 18° and 22°N and there are no cool stations in the Baja California Peninsula (maps not shown).
The case El Niño-HiPDO during summers, which is considered warm, shows that most of the warm stations are south of the tropic of cancer (see Fig. 3), excluding the Yucatan Peninsula.
4. Discussion and conclusions
With the exception of the two El Niño-HiPDO cases mentioned in the previous section (Figs. 1 and 3), we do not find a general pattern of constructive interference (El Niño-HiPDO or La Niña-LoPDO) in our results. Although this study is merely statistical, this fact suggests that either the combined PDO-ENSO influence on the climate of Mexico is weak, or that the signals associated with these two phenomena are linearly independent. With the recent works on ENSO forcing of the PDO (Newman et al. 2003; Schneider and Cornuelle 2005) the latter case seems less likely than the former. Indeed we found the straightforward July to June annual averages of the Southern Oscillation and the PDO indices to be correlated above 0.6, which in a simpler way (compared to Newman et al. 2003 and Schneider and Cornuelle 2005) somewhat verifies the weak, but noticeable, PDO-ENSO influence on Mexican climate rather than PDO-ENSO linear independence.
Nonetheless, all of the cases presented in the previous section and highlighted in Table 3 are statistical significant, and the two constructive interference cases (Figs. 1 and 3) are important in the sense that they substantiate the speculation of Pavia and Graef (2002) about the heavy wintertime precipitation in a region of northwestern Mexico (Baja California) and the obvious assumption of warm summers during a warm ENSO event in a warm PDO phase.
The results reported in Magaña (1999) agree with our findings for winter anomalies during El Niño and La Niña years. We cannot say that this is also the case for the summer (despite the agreement in warm El Niño summers) because Magaña (1999) used the summer of the year before the winter, while we used the summer after the winter (as explained earlier). We decided to use the summer after the winter because we would like to use the strong ENSO signal (in its fully developed state) as a potential predictor of Mexican climate.
Finally, we believe that the combined effects of tropical Pacific variability and ENSO on the climate of Mexico have not been sufficiently investigated. There is a brief discussion of NPO and ENSO effects on Mexican climate in Castro et al. (2001); however, at this time we do not think that we can compare our results with those in other works. This study permits us only to suggest where, and where there is not, a statistical influence of the PDO on ENSO-related effects in the climate of Mexico (see Table 3 for a summary of our findings). Although there are only a couple of cases where a PDO–ENSO constructive interference in Mexican climate is identified (El Niño-HiPDO wet winters and warm summers), we trust that the complete set of our results will help improve ENSO-related empirical predictability of seasonal anomalies in the climate of Mexico.
Acknowledgments
We thank the Mexican water authority (Comisión Nacional del Agua) for providing us with the climatological data used in this study. Conversations with Sasha Gershunov, Niklas Schneider, and Jim O’Brien, as well as the opinions and suggestions of three anonymous reviewers, helped us to improve this paper notably. Our research is funded by the CONACYT-Mexico System.
REFERENCES
Bliss, E. W., and G. T. Walker, 1932: World Weather V. Mem. Roy. Meteor. Soc., 4 , 53–84.
Bove, M. C., and J. J. O’Brien, 2000: PDO modifications of U.S. ENSO climate impacts. The Florida State University, COAPS Tech. Rep. 2000-03, 103 pp.
Castro, C. L., T. B. McKee, and R. A. Pielke, 2001: The relationship of the North American monsoon to tropical and North Pacific sea surface temperatures as revealed by observational analyses. J. Climate, 14 , 4449–4473.
Galindo, I., and P. Mosiño, 1992: Precipitation patterns in Mexico associated with El Niño/Southern Oscillation (ENSO). Extended Abstracts, Paleo-ENSO Records Int. Symp., Lima, Peru, United Nations Development Programme, 111–115.
Gershunov, A., and T. Barnett, 1998: Interdecadal modulation of ENSO teleconnections. Bull. Amer. Meteor. Soc., 79 , 2715–2726.
Gershunov, A., T. Barnett, and D. Cayan, 1999: North Pacific interdecadal oscillation seen as a factor in ENSO-related North American climate anomalies. Eos, Trans. Amer. Geophys. Union, 80 , 25–30.
Graef, F., E. G. Pavia, and J. Reyes, 2000: A new temperature and precipitation climatology of 1000 stations in Mexico. Eos, Trans. Amer. Geophys. Union, 81 , F744.
Graham, N. E., 1994: Decadal-scale climate variability in the 1970s and 1980s: Observations and model results. Climate Dyn., 10 , 135–159.
Latif, M., and T. P. Barnett, 1994: Causes of decadal climate variability over the North Pacific and North America. Science, 266 , 634–637.
Magaña, V., 1999: Los Impactos de El Niño en México. Dirección General de Protección Civil, Secretaría de Gobernación, México, 229 pp.
Mantua, N. J., S. R. Hare, Y. Zhang, J. M. Wallace, and R. C. Francis, 1997: A Pacific decadal climate oscillation with impacts on salmon. Bull. Amer. Meteor. Soc., 78 , 1069–1079.
McCabe, G. J., and M. D. Dettinger, 1999: Decadal variations in the strength of ENSO teleconnections with precipitation in the western United States. Int. J. Climatol., 19 , 1399–1410.
Minobe, S., 1997: A 50–70 year climatic oscillation over the North Pacific and North America. Geophys. Res. Lett., 24 , 683–686.
Mosiño, P., and E. García, 1974: The climate of Mexico. Climates of North America, R. A. Bryson and K. F. Hare, Eds., World Survey of Climatology, Vol. 11, Elsevier, 345–404.
Newman, M., G. P. Compo, and M. A. Alexander, 2003: ENSO-forced variability of the Pacific decadal oscillation. J. Climate, 16 , 3853–3857.
Pavia, E. G., and F. Graef, 2002: The recent rainfall climatology of the Mediterranean Californias. J. Climate, 15 , 2697–2701.
Reyes, J., E. G. Pavia, and F. Graef, 2000: Una nueva climatología de la precipitación pluvial en México. Comunicaciones Académicas, Serie Oceanografía Física CTOFT20001, CICESE, Ensenada, México, 12 pp.
Rogers, J. C., 1981: The North-Pacific Oscillation. J. Climatol., 1 , 39–57.
Schneider, N., and B. D. Cornuelle, 2005: The forcing of the PDO. J. Climate, 18 , 4355–4373.
Trenberth, K. E., 1990: Recent observed interdecadal climate changes in the Northern Hemisphere. Bull. Amer. Meteor. Soc., 71 , 988–993.
Trenberth, K. E., and J. W. Hurrell, 1994: Decadal atmosphere–ocean variations in the Pacific. Climate Dyn., 9 , 303–319.
Yeh, S-W., and B. Kirtman, 2004: The North Pacific Oscillation–ENSO and internal atmospheric variability. Geophys. Res. Lett., 31 .L13206, doi:10.1029/2004GL019983.
Zhang, Y., J. M. Wallace, and D. S. Battisti, 1997: ENSO-like interdecadal variability: 1900–93. J. Climate, 10 , 1004–1020.
Anomalies (+ positive, ○ negative) of winter precipitation during El Niño-HiPDO. Only positive anomalies are statistically significant.
Citation: Journal of Climate 19, 24; 10.1175/JCLI4045.1
Anomalies (+ positive, ○ negative) of summer precipitation during El Niño-LoPDO. Only positive anomalies are statistically significant.
Citation: Journal of Climate 19, 24; 10.1175/JCLI4045.1
Statistically significant anomalies (+ positive, ○ negative) of summer mean temperature during El Niño-HiPDO.
Citation: Journal of Climate 19, 24; 10.1175/JCLI4045.1
Classification of years (1934–99) based on ENSO and PDO phases.
Percentage of stations whose averaged seasonal anomaly exceeds an arbitrary threshold: ±0.50 (wet, dry) for precipitation and ±0.75 (warm, cool) for temperature. Real data and Monte Carlo simulated data.
Summary of the combined PDO–ENSO effects on the mean temperature and precipitation of Mexico. (Hi+LoPDO means HiPDO and LoPDO)
Although this is the usual form this paper is cited, the real order of the authors is Walker and Bliss.
