Abstract

Droughts and persistent wet spells over the United States and northwest Mexico have preferred regions of occurrence and persistence. Wet or dry conditions that persist more than 1 yr tend to occur over the interior United States west of 90°–95°W and northwest Mexico. In contrast, events over the eastern United States are less likely to occur and often last less than 6 months.

The long persistent drought and wet spells are often modulated by low-frequency sea surface temperature anomalies (SSTAs). The persistent dry or wet conditions over northwest Mexico and the Southwest are associated with decadal variability of SSTAs over the North Pacific. Persistent events over the northwestern mountains are associated with two decadal SSTA modes. One mode has loadings over three southern oceans and another one is an El Niño–Southern Oscillation (ENSO) like decadal mode. Wet and dry conditions over the Pacific Northwest and the Great Plains are often associated with ENSO.

The seasonal cycle of precipitation over the central-eastern United States, the East Coast, and the Ohio Valley is weak. Drought and wet spells over these regions are less persistent because the ENSO events have opposite impacts on precipitation for summer and winter.

1. Introduction

Droughts are recurrent climate events that have large societal impacts including human suffering and crop losses. The consequences of drought reach a wide range of sectors: agriculture, water resources management, and power supplies. Usually, droughts (wet spells) mean persistent below (above) normal rainfall over a long time period. During any time, there are some regions over the United States and Mexico under drought or wet spells.

Two indices commonly used to monitor drought are the standardized precipitation index (SPI) and the Palmer drought severity index (PDSI). The SPI is based on precipitation (P) alone (Hayes et al. 1999; McKee et al. 1993, 1995). It is relatively simple, but it does not take into account the water supply. The PDSI is based on the water balance between soil moisture supply and demand (Palmer 1965). The National Climatic Data Center (NCDC) has routinely produced the PDSI based on U.S. climate division data. There are many shortcomings to the PDSI (Alley 1984; Mo and Chelliah 2006). Because different indices have their advantages and shortcomings, more than one method or index is needed to study drought and floods. Droughts or wet spells can also be classified based on soil moisture. Soil moisture data from the North American Land Data Assimilation System depend on the forcing and model used (Robock et al. 2004). There are not enough observations to verify the outputs or to include in diagnostic studies.

Over the United States, there are preferred regions for persistent P events to occur. Based on the PDSI alone, Diaz (1983) and Karl (1983) concluded that drought and wet spells are more likely to occur over the interior western United States than other parts of the country. The above results were based on a single index and they did not discuss physical processes for the regional preference.

Long-term droughts are usually associated with low-frequency large-scale forcing. Dai et al. (1998) used the global PDSI to identify persistent events and linked them to El Niño–Southern Oscillation (ENSO). Later, Dai et al. (2004) used empirical orthogonal functions (EOFs) to identify two leading temporal and spatial global patterns of PDSI and linked them to a long-term upward trend. Rajagopalan et al. (2000) identified ENSO as a major forcing for summer droughts over the United States during the twentieth century. They also noticed that the relationships between droughts and sea surface temperatures (SSTs) change with time. Barlow et al. (2001) found that three SST patterns—the ENSO, the Pacific decadal mode, and the North Pacific SST mode—play important roles in maintaining long-term droughts over the United States. Soil moisture also is an important forcing for drought (Chang and Wallace 1987; Rind 1982). Many modeling studies have established that soil moisture is important to sustaining droughts, especially over the central United States (Oglesby and Erickson 1989; Koster and Suarez 2001; Koster et al. 2005). The model study by Schubert et al. (2004) found that both SSTs and soil moisture contribute to long-term drought over the Great Plains.

The purposes of this paper are 1) to identify regions where droughts and wet spells are most likely to occur and persist using different indices and methods, and 2) to understand the impact of low-frequency SSTAs and ENSO on persistent wet and dry conditions. Because we do not have a dataset of sufficient length to study the impact of soil moisture, we will concentrate on the SST forcing in this paper. Because of our data limitations, we will focus on the period from 1948 to the present. The datasets used are listed in section 2. The spatial patterns and regional preferences of droughts and persistent wet spells over the United States and Mexico are discussed in section 3. One important factor regulating the occurrence of droughts or wet spells is the seasonal cycle of precipitation; this is discussed in section 4. The relationships between SSTAs and persistent events are presented in section 5. Discussion and conclusions are presented in section 6.

2. Data

Both the SPI and the PDSI were used to identify droughts and wet spells. The PDSI was obtained from the NCDC from 1900 to 2004. This is the longest dataset available. The PDSI indices were derived from climate division data. There are 344 climate divisions over the United States but the data do not cover Mexico.

The SPI indices for 6 (SPI6), 12 (SPI12), and 60 months (SPI60) from 1948 to 2004 were computed using the gauge-based P analysis over the United States and Mexico. The method, data coverage, and quality of the gauge-based P data were discussed in Higgins et al. (2000). To obtain SPI6, a 6-month P mean P6 was calculated. The P6 (τ) value at time τ is the P averaged from the time τ − 5 to τ. A transform from a gamma distribution function to a normal distribution function was performed on P6. Then, SPI6 (τ) was determined based on the normal distribution of the transformed data. The SPI12 and the SPI60 were calculated the same way.

The SST data are the monthly reconstructed SST data from Smith et al. (1996), covering the years from 1948 to 2004. The total soil water storage (SM) and the vertically integrated moisture fluxes (qu, qv) were obtained from the North American Regional Reanalysis (RR) from 1979 to 2006 (Mesinger et al. 2006). The study by Mo et al. (2005) indicated that the hydrological cycle over the United States and northern Mexico depicted by the RR is realistic except that the low-level jet from California is too strong. The monthly mean anomaly is defined as the departure from the monthly mean climatology for that month.

3. Regional preference of drought and wet spells

In this section, the regional preferences of wet and dry conditions are examined using the PDSI, SPI, and SM anomalies. Different methods and indices have their advantages and disadvantages, but important features related to droughts and wet spells should not depend on the methods or indices used.

a. Frequency of occurrence and persistence

The PDSI dataset is long enough to allow us to examine the frequency of occurrence at each climate division. Because the PDSI values between −2 and 2 are considered normal, a dry (wet) event is selected when the PDSI is below (above) or equal to −2 (2).

The ratio of dry or wet events to the record length (Figs. 1a and 1e) indicates that events are more likely to occur and persist for more than 1 yr over the interior United States west of roughly 90°–95°W, which is consistent with findings of Karl (1983) and Diaz (1983) (see Figs. 1d and 1h). The mountain areas including Montana, Wyoming, Nebraska, and Colorado have the largest number of persistent wet events. Over the Atlantic coast, droughts and wet spells are less likely to occur (less than 20%) and persist for less than 6 months (Figs. 1b and 1f). Over the central United States (85°–95°W), wet or dry events are mostly likely to persist from 6 months to 1 yr. Over the West Coast, events are likely to persist from 3 months to 1 yr.

Fig. 1.

(a) Ratio of the total number of months under wet spells (PDSI ≥ 2) to the record length from 1900 to 2004. Contour interval is 0.05. Values greater than 0.2 (0.3) are shaded light (dark). (b) Same as in (a) but for wet spells that persist from 1 to 5 months. Contour interval is 0.1. Values greater than 0.3 (0.4) are shaded light (dark). (c) Same as in (b) but for wet spells that persist from 6 to 11 months. (d) Same as in (b) but for events persisting 1 yr or longer. Values greater than 0.5 (0.6) are shaded light (dark). (e)–(h) Same as in (a)–(d) but for droughts (PDSI ≤ −2).

Fig. 1.

(a) Ratio of the total number of months under wet spells (PDSI ≥ 2) to the record length from 1900 to 2004. Contour interval is 0.05. Values greater than 0.2 (0.3) are shaded light (dark). (b) Same as in (a) but for wet spells that persist from 1 to 5 months. Contour interval is 0.1. Values greater than 0.3 (0.4) are shaded light (dark). (c) Same as in (b) but for wet spells that persist from 6 to 11 months. (d) Same as in (b) but for events persisting 1 yr or longer. Values greater than 0.5 (0.6) are shaded light (dark). (e)–(h) Same as in (a)–(d) but for droughts (PDSI ≤ −2).

Because the PDSI has shortcomings (Mo and Chelliah 2006; Alley 1984), SPI and SM are also used to examine the regional preference of wet or dry events. These datasets are not long enough to calculate the frequency of occurrence. Therefore, persistence is measured by the characteristic time T0.

The characteristic time T0 (Trenberth 1984) can be calculated from the autocorrelation R(i) at lag i month for i = 1–30,

 
formula

and N = 30. Values of T0 differ from one variable to another as expected. The important thing is that the patterns of T0 (Fig. 2) for PDSI, SPI6, SPI12, and SM from the RR results are similar. We calculated T0 for PDSI for the periods 1900–2004 (Fig. 2a) and 1950–2004 (not shown). The major features are similar. Large T0 values are located over the western interior states west of 90°–95°W and over northwest Mexico, consistent with the frequency of occurrence based on the PDSI (Fig. 1). One maximum is located over Montana, Wyoming, Nebraska, and a portion of Colorado with T0 longer than 20 months. Another maximum is located over northwest Mexico with T0 longer than 30 months. Events over the East Coast are less persistent when T0 is less than 1 yr. Over the central United States and the West Coast, T0 is between 9 and 15 months.

Fig. 2.

(a) Value of T0 for the PDSI calculated for the period 1900–2004. Contour interval is 4 months. Shading is indicated by the color bar. (b) Same as in (a) but for SPI6 for the period from 1950 to 2004. Contour interval is 3 months. (c) Same as in (b) but for SPI12. Contour interval is 6 months. (d) Same as in (a) but for total soil water storage SM from 1979 to 2006. Contour interval is 6 months.

Fig. 2.

(a) Value of T0 for the PDSI calculated for the period 1900–2004. Contour interval is 4 months. Shading is indicated by the color bar. (b) Same as in (a) but for SPI6 for the period from 1950 to 2004. Contour interval is 3 months. (c) Same as in (b) but for SPI12. Contour interval is 6 months. (d) Same as in (a) but for total soil water storage SM from 1979 to 2006. Contour interval is 6 months.

b. Spatial patterns

Seasonal mean precipitation often exhibits teleconnections between different areas. For example, the leading pattern for summer (June–September) P shows a phase reversal between the loadings of P over the northern plains and the Southwest (Higgins et al. 1997; Mo et al. 1997). During winter, the leading pattern for decadal, seasonal, and intraseasonal precipitation is a dipole between California and the Pacific Northwest (Mo and Higgins 1998; Dettinger et al. 1998). It is interesting to see whether droughts and wet events also have teleconnectivity.

The spatial patterns of droughts and persistent wet events are examined based on rotated EOF (REOF) analysis of seasonal mean SPI6 from 1948 to 2004 by pooling all seasons together. The number of EOFs entering the Varimax rotation is determined by the criterion established by O’Lenic and Livezey (1988). Eleven EOFs were entered into the Varimax rotation to obtain the REOFs (Fig. 3). Rotation was repeated based on 19 EOFs; the first 4 REOFs are the same, but the order of occurrence and percentage of variance explained differ slightly. The leading four REOFs also appear in the first six REOFs for SPI12 (not shown). The original SPI6 data were projected onto the REOF to obtain the corresponding principal components (RPCs). To confirm that the loadings represent the spatial scales of drought, the composites of the PDSI and the SPI6 were formed whenever the RPC is below negative one standard deviation. Figure 3 shows the leading REOFs and the explained total variance. The shading (Fig. 3) indicates areas that both the SPI6 and the PDSI composites identify as drought. This means that the composite of the PDSI is ≤−2 and the composite of the SPI6 ≤ −1.

Fig. 3.

(a) REOF1, (b) REOF2, (c) REOF3, and (d) REOF4. Contour interval is 2 nondimensional units. Shading indicates that composites for PDSI and SPI6 based on RPC all indicate droughts (PDSI ≤ −2 and SPI6 ≤ −1).

Fig. 3.

(a) REOF1, (b) REOF2, (c) REOF3, and (d) REOF4. Contour interval is 2 nondimensional units. Shading indicates that composites for PDSI and SPI6 based on RPC all indicate droughts (PDSI ≤ −2 and SPI6 ≤ −1).

The first four REOFs represent events over the Southwest, southern plains, northern Mexico, and the Pacific Northwest, respectively. REOF was also performed on the PDSI. The first four REOFs for PDSI have monopoles in northern Mexico, the Pacific Northwest, the Southeast, and the Southwest, respectively (not shown). The key message from these results is that the leading REOFs are monopoles. Unlike REOFs for seasonal P, there is no strong teleconnectivity among droughts or wet spells from one region to another. This suggests that the occurrence of droughts or wet conditions that persist for 6 months or longer is not only modulated by large-scale forcing like SSTs but is also influenced by regional factors like soil moisture or the seasonal cycle of the precipitation.

4. Seasonal cycle and moisture fluxes

a. Seasonal cycle

The seasonal cycle of P influences whether a drought or wet spell will last for a season or longer. The monthly mean climatology and standard deviation (STD) were computed for each calendar month using P data from 1948 to 2004. The difference between the maximum (Pmax) and minimum (Pmin) of the monthly mean P climatology measures the amplitude of the mean seasonal cycle. The STD indicates variability. The difference between (Pmax − STD) and (Pmin + STD) measures the strength of the seasonal cycle (Fig. 4b). The large positive values indicate a strong seasonal cycle with a clear P maximum and minimum. Negative values indicate a weak seasonal cycle.

Fig. 4.

(a) Seasonal cycle for P represented by the difference between the P maximum (Pmax) and P minimum (Pmin) of the monthly P climatology from 1948 to 2004. Contour interval is 1 mm day−1. Values greater than 3 (6) mm day−1 are shaded light (dark). (b) The difference between Pmax− Pmin and 2 std dev of the monthly mean climatology. Contour interval is 1 mm day−1. Zero contours are omitted. Contours −0.5 mm day−1 are added. Values greater 1 mm day−1 are shaded dark. (c) Monthly mean climatology for P averaged over northwest Mexico (24°–32°N, 103°–110°W) (solid line). The dark circles and solid lines indicate the mean value plus or minus 0.5 std dev. (d) Same as in (c) but for the East Coast (33°–42°N, 75°–85°W).

Fig. 4.

(a) Seasonal cycle for P represented by the difference between the P maximum (Pmax) and P minimum (Pmin) of the monthly P climatology from 1948 to 2004. Contour interval is 1 mm day−1. Values greater than 3 (6) mm day−1 are shaded light (dark). (b) The difference between Pmax− Pmin and 2 std dev of the monthly mean climatology. Contour interval is 1 mm day−1. Zero contours are omitted. Contours −0.5 mm day−1 are added. Values greater 1 mm day−1 are shaded dark. (c) Monthly mean climatology for P averaged over northwest Mexico (24°–32°N, 103°–110°W) (solid line). The dark circles and solid lines indicate the mean value plus or minus 0.5 std dev. (d) Same as in (c) but for the East Coast (33°–42°N, 75°–85°W).

Pronounced seasonal cycles with magnitudes greater than 3 mm day−1 are located over northwest Mexico, Florida, the West Coast from 36° to 48°N, and the central United States. The rainy season is clearly defined for these regions. For example, the monsoon season over northwest Mexico (24°–32°N, 103°–110°W) lasts from June to September and the area is dry for other months (Fig. 4c). Large contributions to the drought indices SPIs and PDSI are from summer monsoon months. If a drought occurs in late summer, there will be no rain to relieve the drought until the next monsoon season. This contributes to the persistence of drought.

The areas along the Atlantic coast, the central Southeast, and Ohio Valley have the weakest seasonal cycles. The difference (PmaxPmin − 2STD) is less than −1 mm day−1. These are also the regions where the wet or dry conditions are least likely to occur and persist (Fig. 1). Based on climatological monthly means, P over the East Coast (33°–42°N, 75°–85°W over land) varies from a maximum of 3.4 mm day−1 in July to a minimum of 2.5 mm day−1 in October. The values of P from many seasons can contribute to annual mean precipitation and long-term measures such as the PDSI and the SPIs. For example, a winter drought may be relieved by heavy summer precipitation. Therefore, wet or dry conditions are less likely to persist.

b. Moisture transport

The P seasonal cycle is influenced by moisture transport. When moisture is transported into a region only for a limited period, then that region is likely to have a pronounced seasonal maximum. If moisture is available all year round, then the seasonal cycle is weak. Figure 5 shows the vertically integrated moisture flux climatology for four seasons: January–March (JFM), April–June (AMJ), July–September (JAS), and October–December (OND). The climatology was computed based on the RR from 1979 to 2006.

Fig. 5.

Vertical integrated moisture fluxes (qu, qv) (vectors) for the (a) JFM, (b) AMJ, (c) JAS, and (d) OND climatologies from 1979 to 2004. Data were taken from the regional reanalysis. Here, qv is contoured with intervals of 20 kg (m s)−1. Values greater than 40 (80) kg (m s)−1 are shaded light (dark). Moisture fluxes have the unit vector 150 kg (m s)−1

Fig. 5.

Vertical integrated moisture fluxes (qu, qv) (vectors) for the (a) JFM, (b) AMJ, (c) JAS, and (d) OND climatologies from 1979 to 2004. Data were taken from the regional reanalysis. Here, qv is contoured with intervals of 20 kg (m s)−1. Values greater than 40 (80) kg (m s)−1 are shaded light (dark). Moisture fluxes have the unit vector 150 kg (m s)−1

For the West Coast, P occurs in winter and fall when moisture fluxes bring needed moisture from the North Pacific to the West Coast. The seasonal P maxima (Pmax) for the Pacific Northwest and California occur in November and January, respectively. These regions are dry in summer.

The southern plains have a rainy season from April to September with a P maximum in May when moisture fluxes extend from the Gulf of Mexico to the southern plains while the Great Plains low-level jet (GPLLJ) strengthens. In spring, the maximum of the GPLLJ is located in Texas along the coast of the Gulf of Mexico. In June, the GPLLJ extends northward and, at the same time, P reaches a maximum over the northern plains.

The North American monsoon starts in June over northwest Mexico and in early July over the Southwest when the Gulf of California low-level jet strengthens. For the monsoon region, rainfall diminishes after September. For northwest Mexico, moisture is only available for summer. Therefore, that region has a strong seasonal cycle.

For the East Coast, the Gulf states, and the central eastern United States, the seasonal cycle of P is weak. There is no clear P maximum because moisture is brought into the region all year round. In spring and summer, moisture is transported by the zonal branch of the GPLLJ from the Gulf of Mexico. In summer (JAS), tropical storms can bring rain to the Atlantic coast and the Gulf coast. For winter and fall, the major moisture source comes from the Gulf of Mexico. A weaker branch of moisture from the Pacific through the central United States to the East or the Southeast also contributes to the rainfall totals.

5. Time scales of persistent wet or dry conditions

a. Decadal time scales

Multiyear droughts and wet spells are usually associated with low-frequency SST variability on both decadal and interannual time scales. To examine the decadal SST forcing and its linkages to droughts or wet spells, we turn to rotated EOF (REOF) analysis of low-pass-filtered SSTAs.

Monthly mean SSTA data were low-pass-filtered (LF) using a recursive filter to obtain fluctuations longer than 60 months. The data period covers from 1948 to 2004. To reduce the size of our matrix, the horizontal resolution was reduced to 6°. The EOF analysis was performed on the monthly mean LF SSTAs with all months pooled together. The Varimax rotations were performed using 13 EOFs. The procedure then was repeated using 18 modes. The first three leading rotated EOFs are the same. The LF SSTAs were projected onto the REOFs to obtain RPCs. RPCs then were regressed against SPI12. The statistical significance was assessed by assuming one degree of freedom per year. Areas where values are statistically significant at the 5% level are shaded in the figures.

The first REOF, which explains 26.1% of the total variance, shows warming over the southern Atlantic, the Indian Ocean, and the western Pacific (Fig. 6a). RPC1 shows positive trends overall. RPC1 increased rapidly from 1990 onward (Fig. 6d, solid line). The regression pattern against SPI12 indicates that the warning (cooling) of SSTAs is associated with persistent wet spells (droughts) extending from the interior western mountain region to the southern plains (Fig. 6e).

Fig. 6.

(a) REOF1, (b) REOF2, and (c) REOF3 for the LP SSTAs with all months pooled together. Contour interval is 0.5 nondimensional units. Zero contours are omitted. Positive (negative) values are shaded light (dark). (d) RPC1 (solid line) and RPC2 (dashed line). (e) The regression of SPI2 against RPC1. Contour interval is 1 psd. Areas where positive (negative) values are statistically significant at the 5% level and are shaded light (dark). (f) Same as in (e) but for RPC2 (Fig. 7d, dashed line). (g) Same as in (e) but for RPC3. (h) Same as in (d) but for RPC3.

Fig. 6.

(a) REOF1, (b) REOF2, and (c) REOF3 for the LP SSTAs with all months pooled together. Contour interval is 0.5 nondimensional units. Zero contours are omitted. Positive (negative) values are shaded light (dark). (d) RPC1 (solid line) and RPC2 (dashed line). (e) The regression of SPI2 against RPC1. Contour interval is 1 psd. Areas where positive (negative) values are statistically significant at the 5% level and are shaded light (dark). (f) Same as in (e) but for RPC2 (Fig. 7d, dashed line). (g) Same as in (e) but for RPC3. (h) Same as in (d) but for RPC3.

The second REOF (REOF2) explains 22% of the total variance. It shows a broader ENSO-like pattern (Zhang et al. 1997) with positive loadings over the central and eastern Pacific. The corresponding RPC2 also shows positive trends (Fig. 6d, dashed line) with a sharp increase after 1976. This pattern is associated with the drought and persistent wet spells over the interior western region.

The large T0 values suggest that dry and wet events over the northwestern mountains often persist for more than 1 yr (Fig. 2). Over the western mountains, P has a pronounced seasonal cycle and most rainfall occurs from April to July. The contributions from other months are small (Fig. 7a).

Fig. 7.

(a) Monthly mean climatology for P averaged over the western mountain region (38°–45°N, 100°–110°W) (open circles). The dark circles and solid lines indicate the mean value plus or minus 0.5 std dev. (b) SPI12 (solid line) and SPI60 (crosses). (c) Time series of the 24-month running mean of SSTAs averaged over the western Pacific (equator–10°N, 120°–140°W). (d) Same as in (c) but averaged over the Indian Ocean (equator–10° N, 60°–80°W). (e) Correlation coefficient between monthly mean SPI12 averaged over the mountain region and SSTAs for AMJ. Contour interval is 0.1. Areas with positive (negative) values that are statistically significant at the 5% level are shaded dark (light).

Fig. 7.

(a) Monthly mean climatology for P averaged over the western mountain region (38°–45°N, 100°–110°W) (open circles). The dark circles and solid lines indicate the mean value plus or minus 0.5 std dev. (b) SPI12 (solid line) and SPI60 (crosses). (c) Time series of the 24-month running mean of SSTAs averaged over the western Pacific (equator–10°N, 120°–140°W). (d) Same as in (c) but averaged over the Indian Ocean (equator–10° N, 60°–80°W). (e) Correlation coefficient between monthly mean SPI12 averaged over the mountain region and SSTAs for AMJ. Contour interval is 0.1. Areas with positive (negative) values that are statistically significant at the 5% level are shaded dark (light).

The time series of the SPI12 and SPI60 (solid line and crosses, respectively, in Fig. 7b) indicate strong positive trends from 1950 onward with a rapid increase after 1990. The least square fit of SPI12 into a linear combination of two RPCs indicates

 
formula

This suggests that both REOFs contribute to the SPI12 variability, which is consistent with the regression maps (Figs. 6e and 6f). Because AMJ is the rainy season and the western mountain region has a strong seasonal cycle of P, we correlated SPI12 with SSTAs for AMJ. If we assume one degree of freedom per year, the positive (negative) correlations greater (less) than 0.25 (−0.25) are statistically significant at the 5% level. If we assume one degree of freedom per 2 yr, then the correlation magnitudes greater than 0.42 (0.32) are statistically significant at the 5% (10%) level.

The correlation shows warming over the South Atlantic, Indian Ocean, the western Pacific, and the eastern Pacific. The SSTAs averaged over the western Pacific and the Indian Ocean show similar positive trends as the SPIs. These features are consistent with the correlation maps (Figs. 6e and 6f). Again, this confirms that the SPI12 variability is associated with the first low-pass-filtered SST REOF1 and REOF2 modes.

The third REOF (REOF3) shows positive SSTAs in the North Pacific (Fig. 6c). RPC3 (Fig. 6g) indicates positive values from 1950 to 1965 and negative values from 1975 to 1985. RPC3 turns positive from 1995 onward. The associated SPI12 pattern shows that the positive SSTAs in the North Pacific are associated with negative SPI12 over northwest Mexico and the Southwest. It also shows negative SPI12 over Florida and positive SPI12 over the Ohio Valley.

The time series of SPI6 and SPI12 averaged over northwest Mexico (24°–32°N, 103°–110°W; see Figs. 8a and 8b) show low-frequency multidecadal changes similar to those of the SSTAs averaged over the North Pacific and the North Atlantic. The correlation maps shows that wet (dry) conditions over northwest Mexico (Figs. 8e and 8f) are associated with negative (positive) SSTAs in the North Pacific and the North Atlantic. This is consistent with the correlation between SPI12 and RPC3 (Fig. 6g).

Fig. 8.

(a) SPI6 averaged over northwest Mexico (24°–32°N, 103°–110°W). (b) Same as in (a) but for SPI12. (c) SSTAs averaged over the North Pacific (30°–40°N, 140°–160°W) (solid line) and 24-month running mean (crosses). (d) Same as in (c) but averaged over the North Atlantic. (e) Correlation coefficient between monthly mean SPI6 averaged over northwest Mexico and SSTAs for JAS. Contour interval is 0.1. Contours between −0.2 and 0.2 are omitted. Areas with positive (negative) values that are statistically significant at the 5% level are shaded dark (light). (f) Same as in (e) but for SPI12.

Fig. 8.

(a) SPI6 averaged over northwest Mexico (24°–32°N, 103°–110°W). (b) Same as in (a) but for SPI12. (c) SSTAs averaged over the North Pacific (30°–40°N, 140°–160°W) (solid line) and 24-month running mean (crosses). (d) Same as in (c) but averaged over the North Atlantic. (e) Correlation coefficient between monthly mean SPI6 averaged over northwest Mexico and SSTAs for JAS. Contour interval is 0.1. Contours between −0.2 and 0.2 are omitted. Areas with positive (negative) values that are statistically significant at the 5% level are shaded dark (light). (f) Same as in (e) but for SPI12.

b. Influence of ENSO

On interannual time scales, ENSO is a dominant forcing. The seasonal ENSO composites of P and PDSI between the warm and cold events for selected seasons are given in Figs. 9 and 10, respectively. The statistical significance is assessed by assuming a normal distribution and one degree of freedom per year. Areas where values are statistically significant at the 5% level are shaded. Information on the ENSO events is available online (http://www.cpc.ncep.noaa.gov). The following description is for warm events. The situation reverses for the cold ENSO events.

Fig. 9.

Composite difference of P anomalies between warm and cold ENSO events for (a) OND, (b) JFM, (c) AMJ, (d) JJA, (e) JAS, and (f) ASO. Contour interval is 0.3 mm day−1. Zero contours are omitted. Areas with positive (negative) values that are statistically significant at the 5% level are shaded dark (light).

Fig. 9.

Composite difference of P anomalies between warm and cold ENSO events for (a) OND, (b) JFM, (c) AMJ, (d) JJA, (e) JAS, and (f) ASO. Contour interval is 0.3 mm day−1. Zero contours are omitted. Areas with positive (negative) values that are statistically significant at the 5% level are shaded dark (light).

Fig. 10.

Same as in Fig. 9 but for the PDSI. Contour interval is 1.

Fig. 10.

Same as in Fig. 9 but for the PDSI. Contour interval is 1.

The responses to warm ENSO events for October–December (OND) and January–March (JFM) are negative P anomalies over the Pacific Northwest and northern California, with positive anomalies over the southern plains, the Southwest, northern Mexico, and the Gulf states. For winter (JFM), the influence also includes less P over the Ohio Valley and positive P anomalies over the East Coast. The ENSO influence decreases in spring. During AMJ, the composite map does not show any signal. Positive P anomalies appear in the north-central United States in JAS.

During summer (July–October), negative P anomalies over the East Coast are influenced by tropical storms. Warm ENSO events are often detrimental to the occurrence of tropical storms in the Atlantic (Gray 1984; Chelliah and Bell 2004). In ASO, positive P anomalies are located over the Great Plains and the western states including Utah, Colorado, Wyoming, and eastern New Mexico.

The composites of the PDSI show the cumulative impact of ENSO on droughts and wet spells. Overall, the cumulative impact shows negative PDSI over the Pacific Northwest from OND to AMJ and positive PDSI over the Southwest and the Gulf states. Over the central United States, cold ENSO is likely to enhance summer drought.

6. Conclusions

Based on the seasonal cycle of P (Fig. 4), the low-frequency influence of SSTAs (Fig. 6), and the ENSO composites (Figs. 9 and 10), the regional preference of droughts and wet spells can be explained as follows.

a. Northwest Mexico and the western mountain region

Droughts and wet spells are most likely to occur and to persist for more than 1 yr over northwest Mexico and the northwestern mountains over the United States (Figs. 1 and 2) because of the influence of low-frequency decadal SST variability. Warm (cold) SSTAs over the Indian Ocean and southern Atlantic (REOF1) and an ENSO-like warming (cooling) over the tropical Pacific (REOF2) (Figs. 6 and 7) are associated with the increase of wet (dry) events over the northwestern mountain region. The decadal variations of SSTA over the North Pacific are associated with events over northwest Mexico.

b. West Coast

Droughts and wet spells over the West Coast are less persistent than those over the interior western region. Persistent dry or wet events last from 3 months to a year (Fig. 1). There is no strong decadal SSTA signal. The largest influence is ENSO. The Pacific Northwest has a strong seasonal cycle of P and the rainy season lasts from fall to winter. The PDSI composites indicate that the ENSO modulation persists from late fall (OND) to spring (AMJ) (Fig. 10), which is less than 1 yr (Fig. 1).

The controlling factor of P over California is the precipitation pattern in the tropical eastern Pacific. If rainfall over the eastern tropical Pacific is suppressed, then California is likely to have a heavy rainy season (Mo and Higgins 1998; Cayan and Redmond 1994). The intraseasonal oscillations on the time scales of 36–40 and 22–25 days also can produce strong wet and dry episodes over California (Mo 1999). The cumulative impact is that droughts and wet spells over California are unlikely to persist beyond 1 yr.

c. Southwest

Over Arizona and western New Mexico, the P distribution is bimodal with one weaker maximum in winter (JFM) and another strong summer maximum associated with the monsoon. During winter, warm ENSO increases the chance of rain (Fig. 9). The SSTAs over the North Pacific play an important role in modulating low-frequency rainfall variations (Mo and Paegle 2000; Castro et al. 2001). The PDSI composite indicates that the accumulative ENSO influence is likely to last from OND to AMJ (Fig. 10). Because of the low-frequency decadal influence and ENSO, droughts and wet spells are likely to persist for more than 1 yr.

d. Southern plains

The seasonal cycle of P over the southern plains has its precipitation maximum in May, but the rainy season usually lasts from April to September (not shown). Wet and dry events are associated with SSTAs in the tropical Pacific associated with both ENSO and the low-frequency decadal variability (Fig. 6). The composites of PDSI (Fig. 10) indicate that the impact of ENSO can last for a whole year as long as SSTAs persist. These factors contribute to the persistence of droughts and wet spells.

e. North-central

Over the north-central United States, P has a strong seasonal cycle (Fig. 4) with a summer maximum. The largest contribution to annual P is from ENSO in late summer (ASO). Because of the strong seasonal cycle, the composites of PDSI indicate that ENSO’s influence can last for 1 yr (Fig. 10). Our results are consistent with the findings of Ting and Wang (1997).

f. East Coast

The P seasonal cycle for the East Coast (32°–37°N, 70°–82°W over land) is weak (Fig. 4d) and there is no influence of decadal SSTA variability. The interesting point to note is that ENSO’s influence on rainfall is seasonally dependent. The P composite (Fig. 9b) shows that warm (cold) ENSO is associated with positive (negative) P anomalies in winter. However, the warm (cold) ENSO decreases (increases) the possibility of tropical storm occurrence and the region receives less (more) tropical storm–related precipitation (Figs. 9e and 9f). If a cold ENSO event persists from winter to summer, then winter drought may be relieved by the increase of summer rainfall and vice versa. The persistence of droughts or wet spells is unlikely to last for more than one season (Fig. 10f).

For the other areas, like the Ohio Valley, the impact of ENSO only lasts for the winter season. Over AMJ, the impact is considerably weakened. The impacts of ENSO during winter and during summer are opposite in phase. All of these factors contribute to less persistence among the dry or wet events.

As indicated by Rajagopalan et al. (2000), the relationships between SSTAs and droughts can vary from one decade to another. The results presented here here are based on data from 1948 to 2004. There are no long-term observational soil moisture data available. Carefully designed modeling experiments are needed to address the feedback mechanisms between soil moisture and drought.

Acknowledgments

This work was supported by NCPO/CPPA Project GC06-012.

REFERENCES

REFERENCES
Alley
,
W. M.
,
1984
:
The Palmer drought severity index: Limitations and assumptions.
J. Climate Appl. Meteor.
,
23
,
1100
1109
.
Barlow
,
M.
,
S.
Nigam
, and
E. H.
Berbery
,
2001
:
ENSO, Pacific decadal variability, and the U.S. summertime precipitation drought and streamflow.
J. Climate
,
14
,
2105
2128
.
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 temperature as revealed by observational analyses.
J. Climate
,
14
,
4449
4473
.
Cayan
,
D. R.
, and
K. T.
Redmond
,
1994
:
ENSO influences on atmospheric circulation and precipitation in the western United States. Proc. 10th Annual Pacific Climate (PACLIM) Workshop, Pacific Grove, CA, California Department of Water Resources, 5–26
.
Chang
,
F.
, and
J.
Wallace
,
1987
:
Meteorological conditions during heat waves and droughts in the United States Great Plains.
Mon. Wea. Rev.
,
115
,
1253
1269
.
Chelliah
,
M.
, and
G. D.
Bell
,
2004
:
Tropical multidecadal and interannual climate variability in the NCEP–NCAR reanalysis.
J. Climate
,
17
,
1777
1803
.
Dai
,
A.
,
K. E.
Trenberth
, and
T. R.
Karl
,
1998
:
Global variations in droughts and wet spells 1900–1995.
Geophys. Res. Lett.
,
25
,
3367
3370
.
Dai
,
A.
,
K. E.
Trenberth
, and
T.
Quin
,
2004
:
A global dataset of Palmer drought severity index for 1987–2002: Relationships with soil moisture and effects of surface warming.
J. Hydrometeor.
,
5
,
1117
1130
.
Dettinger
,
M. D.
,
D. R.
Cayan
,
H. F.
Diaz
, and
D. M.
Meko
,
1998
:
North–south precipitation patterns in western North America on interannual-to-decadal timescales.
J. Climate
,
11
,
3095
3111
.
Diaz
,
H. P.
,
1983
:
Some aspect of major dry and wet periods in the contiguous United States.
J. Climate Appl. Meteor.
,
22
,
3
16
.
Gray
,
W. M.
,
1984
:
Atlantic seasonal hurricane frequency. Part I: El Niño and 30 mb quasi-biennial oscillation influences.
Mon. Wea. Rev.
,
112
,
1649
1668
.
Hayes
,
M. J.
,
M. D.
Svoboda
,
D. A.
Wilhite
, and
O. V.
Vanyarkho
,
1999
:
Monitoring the 1996 drought using the standardized precipitation index.
Bull. Amer. Meteor. Soc.
,
80
,
429
438
.
Higgins
,
R. W.
,
Y.
Yao
,
E. S.
Yarosh
,
J. E.
Janowiak
, and
K. C.
Mo
,
1997
:
Influence of the Great Plains low-level jet on summertime precipitation and moisture transport over the central United States.
J. Climate
,
10
,
481
507
.
Higgins
,
R. W.
,
W.
Shi
,
E.
Yarosh
, and
R.
Joyce
,
2000
:
Improved United States precipitation quality control system and analysis. NCEP/CPC Atlas 7, Climate Prediction Center, Camp Springs, MD
.
Karl
,
T. R.
,
1983
:
Some spatial characteristics of drought duration in the United States.
J. Climate Appl. Meteor.
,
22
,
1356
1366
.
Koster
,
R. D.
, and
M. J.
Suarez
,
2001
:
Soil moisture memory in climate models.
J. Hydrometeor.
,
2
,
558
570
.
Koster
,
R. D.
, and
Coauthors
,
2005
:
GLACE: The Global Land–Atmosphere Coupling Experiment.
J. Hydrometeor.
,
7
,
611
625
.
McKee
,
T. B.
,
N. J.
Doesken
, and
J.
Kleist
,
1993
:
The relationship of drought frequency and duration to time scales. Preprints.
Eighth Conf. on Applied Climatology, Anaheim, CA, Amer. Meteor. Soc., 179–184
.
McKee
,
T. B.
,
N. J.
Doesken
, and
J.
Kleist
,
1995
:
Drought monitoring with multiple time scales. Preprints.
Ninth Conf. on Applied Climatology, Dallas, TX, Amer. Meteor. Soc., 233–236
.
Mesinger
,
F.
, and
Coauthors
,
2006
:
North American Regional Reanalysis.
Bull. Amer. Meteor. Soc.
,
87
,
343
360
.
Mo
,
K. C.
,
1999
:
Alternating wet and dry episodes over California and intraseasonal oscillations.
Mon. Wea. Rev.
,
127
,
2759
2776
.
Mo
,
K. C.
, and
R. W.
Higgins
,
1998
:
Tropical influence on California precipitation.
J. Climate
,
11
,
412
430
.
Mo
,
K. C.
, and
J. N.
Paegle
,
2000
:
Influence of sea surface temperature anomalies on the precipitation regimes over the southwest United States.
J. Climate
,
13
,
3588
3598
.
Mo
,
K. C.
, and
M.
Chelliah
,
2006
:
The modified Palmer drought severity index based on the NCEP North American Regional Reanalysis.
J. Appl. Meteor. Climatol.
,
45
,
1362
1375
.
Mo
,
K. C.
,
J.
Nogues Paegle
, and
R. W.
Higgins
,
1997
:
Atmospheric processes associated with summer floods and droughts in the central United States.
J. Climate
,
10
,
3028
3046
.
Mo
,
K. C.
,
M.
Chelliah
,
M. L.
Carrera
,
R. W.
Higgins
, and
W.
Ebisuzaki
,
2005
:
Atmospheric moisture transport over the United States and Mexico as evaluated in the NCEP regional reanalysis.
J. Hydrometeor.
,
6
,
710
728
.
Oglesby
,
R. J.
, and
D. J.
Erickson
,
1989
:
Soil moisture and the persistence of North American drought.
J. Climate
,
2
,
1362
1380
.
O’Lenic
,
E. A.
, and
R. E.
Livezey
,
1988
:
Practical considerations in the use of rotated principal component analysis in diagnostic studies of upper air fields.
Mon. Wea. Rev.
,
116
,
1682
1698
.
Palmer
,
W. C.
,
1965
:
Meteorological drought. U.S. Weather Bureau Research Paper 45, 58 pp
.
Rajagopalan
,
B.
,
E.
Cook
,
U.
Lall
, and
B. K.
Ray
,
2000
:
Spatiotemporal variability of ENSO and SST teleconnections to summer drought over the United States during the twentieth century.
J. Climate
,
13
,
4244
4254
.
Rind
,
D.
,
1982
:
The influence of ground soil moisture conditions in North America on summer climate as modeled in the GISS GCM.
Mon. Wea. Rev.
,
110
,
1487
1494
.
Robock
,
A.
, and
Coauthors
,
2004
:
Evaluation of the North American Land Data Assimilation System (NLDAS) over the southern Great Plains during the warm season.
J. Geophys. Res.
,
109
.
8843, doi:10.1029/2002JD003246
.
Schubert
,
S. D.
,
M. J.
Suarez
,
P.
Pegion
,
R. D.
Koster
, and
J. T.
Bacmeister
,
2004
:
Causes of long-term drought in the U.S. Great Plains.
J. Climate
,
17
,
485
503
.
Smith
,
T. M.
,
R. W.
Reynolds
,
R. E.
Livezey
, and
D. C.
Stokes
,
1996
:
Reconstruction of historical sea surface temperatures using empirical orthogonal functions.
J. Climate
,
9
,
1403
1420
.
Ting
,
M.
, and
H.
Wang
,
1997
:
Summertime U.S. precipitation variability and the relation to Pacific sea surface temperature.
J. Climate
,
10
,
1853
1873
.
Trenberth
,
K. E.
,
1984
:
Some effects of finite sample size and persistence on meteorological statistics. Part II: Potential predictability.
Mon. Wea. Rev.
,
112
,
2369
2379
.
Zhang
,
Y.
,
J. M.
Wallace
, and
D. S.
Battisti
,
1997
:
ENSO-like interdecadal variability: 1900–93.
J. Climate
,
10
,
1004
1020
.

Footnotes

Corresponding author address: Kingtse C. Mo, NOAA/NWS/Climate Analysis Center, 5200 Auth Rd., Camp Springs, MD 20746. Email: kingtse.mo@noaa.gov