Regional trends of daily precipitation indices in ... - Wiley Online Library

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 115, D14111, doi:10.1029/2009JD013248, 2010

for

Full Article

Regional trends of daily precipitation indices in northwest Mexico and southwest United States Sarahí Arriaga‐Ramírez1 and Tereza Cavazos1 Received 22 September 2009; revised 2 March 2010; accepted 6 April 2010; published 24 July 2010.

[1] A regionalization based on a rotated principal component analysis (PCA) was used to produce six precipitation regions in northwest Mexico and the southwest United States. Monthly precipitation data from 184 meteorological stations for the 1960–1997 period were used in the PCA. The aim of this study was to estimate annual and seasonal trends of 10 daily precipitation indices in the six regions, including four indices related with extreme precipitation. The annual indices show a larger number of statistically significant trends than the seasonal indices, especially in Arizona‐New Mexico and in the monsoon region in northwest Mexico (MON). Significant positive trends common to these two contiguous regions are extreme precipitation exceeding the 95th (R95p) and 99th (R99p) percentiles. The analysis of summer (June–October) daily precipitation indices also reveals the occurrence of significant positive trends in R95p in MON, mainly due to tropical cyclone activity. With the exception of the trends in MON, the most important contribution to the annual trends comes from the winter indices. Four of the six regions in the study area show significant positive trends in extreme winter precipitation (R10mm, R95p or R99p) during the study period. The variability of the annual indices that show statistically significant trends in extreme precipitation are partially linked to natural variations resulting from the combined effects of El Niño/Southern Oscillation and the Pacific decadal oscillation (PDO) and, in most cases, the trends are explained by the PDO. Citation: Arriaga‐Ramírez, S., and T. Cavazos (2010), Regional trends of daily precipitation indices in northwest Mexico and southwest United States, J. Geophys. Res., 115, D14111, doi:10.1029/2009JD013248.

1. Introduction [2] Northwest Mexico (NW‐Mex) and the southwest United States (SW‐USA) are characterized by arid and semiarid climates (Figure 1) and high interannual climate variability. Precipitation fluctuations in the region have been associated with the El Niño/Southern Oscillation (ENSO) events at interannual timescales and/or with the Pacific Decadal Oscillation (PDO) at decadal timescales [e.g., Ropelewski and Halpert, 1986; Dettinger et al., 1998; Gershunov and Barnett, 1998; Cayan et al., 1999; Mason and Goddard, 2001; Gershunov and Cayan, 2003; Magaña et al., 2003; McCabe et al., 2004; Pavia et al., 2006; Goodrich, 2007; Higgins et al., 2007]. Stronger teleconnections usually occur during the winter season; a constructive combination of El Niño warm conditions and a warm PDO tends to favor above normal rainfall, and below normal rainfall tends to occur during La Niña and a cold PDO. Some studies have also shown that above (below) normal winter rainfall in the SW‐USA occurs during neutral ENSO events and warm (cold) phases of the PDO [Gershunov and Cayan, 2003; Goodrich, 2007]. 1 Departamento de Oceanografía Física, CICESE, Baja California, Mexico.

Copyright 2010 by the American Geophysical Union. 0148‐0227/10/2009JD013248

[3] Figure 2 shows the total annual precipitation derived from 184 stations in the study area (Figure 1). The time series show a relative dry period from 1960 to the mid‐ 1970s followed by a wetter period that lasts until the end of the1990s. In 1999 a strong La Niña event produced below normal precipitation in the region. In fact, a prolonged drought condition has persisted over the study area since the end of the 1990s and has been compared to the 1950s drought [e.g., Seager, 2007; Stahle et al., 2009]. These interdecadal fluctuations of annual precipitation are linked to the PDO shifts [Mantua et al., 1997], from cold to warm PDO in the mid‐1970s and from warm to a temporarily cold PDO at the end of 1990s, but in the 2000s the PDO has shown positive and negative fluctuations with no definite phase. Stahle et al. [2009] have suggested that the recent drought (or the 2000s drought) period, that will not be covered in this study due to insufficient daily data, is possibly associated with an aridity trend projected for the region, and for the subtropics in general, based on anthropogenic global warming and the associated poleward expansion of the Hadley circulation [Intergovernmental Panel on Climate Change (IPCC), 2007]. [4] In a study of climate change hot spots (highly vulnerable regions to climate change impacts) in the continental USA and northern Mexico, Diffenbaugh et al. [2008] found that areas of NW‐Mex and the SW‐USA are the most persistent hot spots in the 21st century, mainly associated with

D14111

1 of 10

D14111

ARRIAGA‐RAMÍREZ AND CAVAZOS: TRENDS OF DAILY PRECIPITATION

Figure 1. Mean total annual precipitation (mm) derived from the NARR data set for the 1979–2002 period. Dots indicate the location of the 184 selected meteorological stations from the GHCN‐Daily data set. high precipitation variability and higher temperatures. Most of the climate change scenarios predict that the study area will become drier in the 21st century [e.g., IPCC, 2007; Seager et al., 2007; Favre and Gershunov, 2008; Milly et al., 2008]. It has been suggested that the impact of climate change will be felt most strongly through changes in precipitation extremes rather than changes in mean precipitation [Allen and Ingram, 2002; Kharin and Zwiers, 2005; IPCC, 2007]. Similarly, Trenberth et al. [2005] argue that since the amount of moisture in the atmosphere is likely to rise much faster as a consequence of rising temperatures than the total precipitation, this should lead to an increase in the intensity of storms, offset by decreases in duration or frequency of events. [5] On the other hand, there is evidence that dry spell length has significantly increased in the SW‐USA [Groisman and Knight, 2008]. There is also evidence that the annual amount of heavy precipitation increased significantly in parts of NW‐Mex [Groisman et al., 2005; Alexander et al., 2006; Kunkel et al., 2008; Peterson et al., 2008] and the SW‐USA [Frich et al., 2002; Alexander et al., 2006] during the second half of the 20th century. However, it appears that these positive trends in heavy precipitation observed in the SW‐USA may have partially resulted from natural variability associated with ENSO events and the warm phase of the PDO after 1976 and before 2000 [e.g., Gershunov and Barnett, 1998; Cayan et al., 1999; Gershunov and Cayan, 2003; McCabe et al., 2004]. [6] Although heavy precipitation frequently results in flooding and landslides with negative impacts on society, it is also true that in arid and semiarid climates intense rainfall events play an important role on ecosystems [Knapp et al., 2008], and are a major source of water for dams and reservoirs. Moreover, trends in extreme daily precipitation have not been extensively documented for the region at seasonal timescales. Therefore, the objectives of this study are to estimate observed regional trends in annual and seasonal (summer and winter) daily precipitation indices, including several related with extreme precipitation, in

D14111

NW‐Mex and the SW‐USA during 1960–1997, and to explore the source of their annual variation through climatic teleconnections, such as ENSO and PDO. [7] One of the first studies to document global observed trends in climatic indices was carried out by Frich et al. [2002]. The importance of this and subsequent related studies motivated the World Meteorological Organization (WMO) through the Commission for Climatology (CCI) and Climate Variability (CLIVAR) Expert Team for Climate Change Detection Monitoring and Indices (ETCCDMI) to standardize a set of climate indices. The result was a group of 17 temperature and 10 precipitation indices that have been widely used in studies of climatic trends in observations and in climate change projections. In this analysis we obtained linear trends of the 10 precipitation indices at 184 selected meteorological stations in NW‐Mex and SW‐USA during 1960–1997 and estimated regional trends; we are particularly interested in the trends of four climatic indices that refer to extreme daily precipitation (R95p, R99p, R10mm, and R20mm), described in Table 1. [8] This paper is organized as follows: Sections 2 and 3 describe the data sets and methodology used to obtain regional trends of the 10 daily precipitation indices, respectively. Section 4 discusses the results of the regional trends at annual and seasonal (Jun‐Oct and Nov‐Mar) timescales. The last section presents a summary and conclusions of the main findings.

2. Data [9] Historical daily precipitation for NW‐Mex and SW‐ USA (Figure 1) in the period 1950–2000 was obtained from GHCN‐Daily (Global Historical Climatology Network‐ Daily, [Vose et al., 1992], http://www.ncdc.noaa.gov/oa/ climate/ghcn‐daily/index.php). The data set is freely available from ftp://ftp.ncdc.noaa.gov/pub/data/ghcn/daily/. It includes the climate records from meteorological stations from numerous sources, including the Mexican Servicio Meteorologico Nacional (SMN). The daily data was subjected to a suite of quality controls [Durre et al., 2008]. Although the climate data was available from 1950 to 2000, before 1960 and after 1997 the Mexican stations had insufficient continuous data coverage for this type of analysis.

Figure 2. Mean total annual precipitation (mm) over the study area derived from the 184 meteorological stations in Figure 1. The horizontal line indicates the mean annual precipitation during the period (380 mm).

2 of 10

D14111


D14111

Table 1. Ten Daily Precipitation Indices Used in This Study Index CDD CWD PRCPTOT R10mm R20mm R95p R99p RX1day RX5day SDII

Description Consecutive dry days: maximum number of consecutive days with daily precipitation < 1 mm. Consecutive wet days: maximum number of consecutive days with daily precipitation ≥ 1 mm. Total annual precipitation in wet days (≥ 1 mm). Number of days with precipitation ≥ 10 mm. Number of days with precipitation ≥ 20 mm. Annual contribution of total precipitation exceeding the 95th percentile. Annual contribution of total precipitation exceeding the 99th percentile. Maximum annual precipitation in one day. Maximum annual precipitation in 5 consecutive days. Simple precipitation intensity index: total annual precipitation divided by the total number of wet days (with precipitation ≥ 1 mm).

Thus, the effective period for this study is 1960–1997, and it does not cover the recent dry period (1999–2004) that has been documented for the study area [e.g., Seager, 2007; Stahle et al., 2009]. [10] The meteorological stations selected had at least 75% of daily data available at annual and seasonal (winter season: Nov‐Mar and summer season: Jun‐Oct) scales during 1960–1997. April and May were not included in the winter and summer seasons, respectively, because their contribution to the annual precipitation is small. In NW‐Mex 74 meteorological stations from SMN fulfilled the 75% requirement. In the SW‐USA the 110 stations from the United States Historical Climatology Network‐Daily (USHCN‐Daily), a subset of the GHCN‐Daily, had a high quality control and also satisfied the 75% requirement. The daily data were subjected to other quality controls, as described in the next section. [11] We also used the North American Regional Reanalysis (NARR) daily gridded precipitation [Mesinger et al., 2006] for the 1979–2002 period for a simple comparison with observed seasonal precipitation (Figure 3) and to show the spatial distribution of rainfall in the region (Figures 1, 4b and 4c). The NARR precipitation has a spatial resolution of 32 km; it is an assimilated product that used the Unified Climate Prediction Center (CPC) Daily Precipitation Analyses until 1998 [Mesinger et al., 2006]. However, the CPC

Units days days mm days days mm mm mm mm mm d−1

precipitation is known to have a dry bias in the monsoon region of NW‐Mex in the 1990s [Gutzler, 2004], due to coarser resolution in the daily Mexican station data. Figure 3 shows the seasonal (winter and summer) precipitation series of the entire region from observations and from the NARR data set. Although there is a relatively good agreement between the two series, the dry bias in the NARR precipitation is more obvious in the winter precipitation series. The seasonal rainfall series do not show any significant trends. [12] Monthly time series of the PDO Index was obtained from the University of Washington (http://jisao.washington. edu/pdo/PDO.latest) for the study period. Time series of the Oceanic Niño Index (ONI) was used to represent ENSO variability. ONI is defined as the three‐month running mean sea surface temperature (SST) departures in the Niño 3.4 region. Departures are based on a set of improved homogeneous historical SST analysis [Smith et al., 2008]. The ONI data was obtained from: http://www.cpc.noaa.gov/ products/analysis_monitoring/ensostuff/ensoyears.shtml).

3. Methodology 3.1. Daily Precipitation Indices [13] The 10 daily precipitation indices analyzed in this study are described in Table 1. The 10 indices were calculated for each of the 184 meteorological stations over NW‐Mex

Figure 3. Mean total seasonal precipitation (mm) in the study area for (a) summer (Jun‐Oct) and (b) winter (Nov‐Mar) from the 184 meteorological stations (OBS) and from NARR derived precipitation over the continent. The NARR time series starts in 1979. 3 of 10

D14111


D14111

and SW‐USA for annual (Jan‐Dec), summer (Jun‐Oct), and winter (Nov‐Mar) timescales for the 1960–1997 period. The base period used to compute thresholds for the reference values in Table 1 was 1961–1990, consistent with similar studies [Alexander et al., 2006; Peterson et al., 2008]. [14] The daily precipitation indices were calculated with FClimDex, a Fortran‐based program of public domain (http://cccma.seos.uvic.ca/ETCCDMI/RClimDex/fclimdex.f) that only computes annual indices. It performs a quality control that replaces inconsistent daily precipitation (negative values) by ‐99.99, detects unusual frequencies of consecutive days with the same amount of precipitation, and identifies daily outliers. A threshold of 5 standard deviations above the climatological mean was used to detect daily outliers. Visual inspection, comparison with neighboring stations, and checks for the presence of storms and tropical cyclones were used to validate outliers and unusual frequency of consecutive days with the same amount of precipitation in each station. The FclimDex program was modified to also calculate seasonal indices.

Figure 4. (a) The six precipitation regions derived from a PCA are delimited by dashed lines and labeled with capitals. (b) Month of maximum precipitation (direction of the arrows) from the NARR data set for the 1979–2002 period. (c) Annual percent of summer (Jun‐Oct) precipitation from the NARR data set. Direction of arrows in Figure 4b is defined at the lower left. Shaded contours in Figure 4a represent the topography, with darker colors indicating higher elevations.

3.2. Regionalization of Monthly Precipitation [15] A regionalization based on a rotated principal component analysis (PCA) was used to produce precipitation regions in northwest Mexico and the southwest United States. The resultant components were used to estimate the observed trends of the 10 daily precipitation indices at regional scale. Other studies of climatic indices that have included the study area [e.g., Alexander et al., 2006; Peterson et al., 2008] have documented gridded trends, which help in the comparison of observed indices with those from global climate simulations. In the current study, the trends were evaluated in each of the PCA‐based regions, rather than on a regular grid, because there are wide areas that have insufficient long‐term data coverage, especially on the Mexican side (Figure 1). [16] An S‐mode PCA (multiple stations over time) was applied to the monthly (Jan ‐ Dec) precipitation of the 184 meteorological stations during the base period 1961–1990. Only for this analysis, monthly data unavailable at a particular station was replaced with its corresponding monthly climatology. To maximize the explained variance over specific regions an oblique rotation (oblimin), which tends to produce good climate regionalizations [White et al., 1991; Comrie and Glenn, 1998] was used. The g value (obliquity) suggested by Comrie and Glenn [1998] for the North American monsoon region is g = 0. The first five rotated components (regions) that explained 63.2% of the total precipitation variance were chosen based on a careful revision of scree plots and the eigenvalues (l) greater than one. Due to well‐known differences in climate features of the Sonora and Vizcaino deserts compared to Baja California Sur, this region was manually divided into two groups (DES and BCS, Figure 4a). [17] The criterion to delimit the six spatial regions obtained from the PCA, as shown in Figure 4a, was based on the precipitation stations (139 out of 184) with weights greater than 0.49 and lower than ‐0.49. In the next section we used this annual PCA‐based regionalization to obtain annual and seasonal (Jun‐Oct and Nov‐Mar) rainfall trends to determine which of the seasons contribute the most to the annual trends. However, it is important to notice that the

4 of 10

D14111


D14111

Table 2. Annual Trends of the 10 Daily Precipitation Indices Over the Six Regions in NW‐Mex and the SW‐USA During the 1960– 1997 Perioda Index/Regions

ANM (25)

TEX (18)

MON (47)

BCS (4)

CAL (31)

DES (14)

CDD CWD PRCPTOT R10mm R20mm R95p R99p RX1day RX5day SDII

−0.09 0.00 2.20(**) 0.09(**) 0.02 0.73(**) 0.23(*) −0.04(**) −0.06(**) 0.01

−0.19 0.01 1.43 0.06 0.02 0.46 0.04 −0.05(*) −0.07(*) 0.01

−0.24 −0.02 1.61 0.05 0.03 1.0(**) 0.41(*) 0.00 −0.03 0.03

−0.64 0.00 1.08 0.02 0.02 0.89(*) 0.04(**) 0.06 0.00 0.03

0.16 0.00 2.42 0.09 0.03 0.40 0.12 −0.06 NaN 0.02

−0.64 0.00 0.80 0.03 0.01 0.13 0.02 −0.10 −0.13 0.00

a Statistically significant trends are shown in bold and one (two) asterisks indicate statistical significance at the 90% (95%) levels. NaN indicates no data, when the data available in the region was less than 75%. See Table 1 for definition of indices and units, and Figure 4a for the geographical location of the six regions. The number of meteorological stations used for each region is shown in parentheses.

boundaries of the rainfall regions slightly differ from the ones shown in Figure 4a if the PCA analysis is performed separately for the winter and summer seasons, as our auxiliary material shows (Figure S1).1 The statistically significant rainfall trends shown in the auxiliary material (Figure S1) are, in general, consistent with the trend results discussed in Section 4.

the ENSO and PDO indices. In the presence of a significant relationship we analyzed the residuals; if their trend was not significantly different from zero (p < 0.1) we concluded that the observed trend in the precipitation index was due to natural variation.

3.3. Trends of Daily Precipitation Indices [18] After the 10 precipitation indices were calculated for each of the selected stations, annual time series anomalies of each index were obtained at annual and seasonal scales for the 1960–1997 period. The time series anomalies were averaged over the stations corresponding to each of the six precipitation regions (Figure 4a) to obtain the regional indices (see Table 2). The trend analysis was assessed over the regional anomalies (i.e., the regional indices). The slope of the trend was determined using Sen’s slope estimator [Sen, 1968]. This approach involves computing slopes for all the pairs of ordinal time points and then using the median of these slopes as an estimate of the overall slope. This estimator is a nonparametric alternative for estimating the slope for a univariate time series; it is robust to the effect of outliers in the series. It has been widely used to compute trends in hydrometeorological series [e.g., Alexander et al., 2006; Cavazos et al., 2008; Peterson et al., 2008]. The significance of the trend was calculated using the non‐ parametric Mann‐Kendall test [Mann, 1945; Kendall, 1955] by normal approximation; an adjustment was made for tied observations. This test is non‐parametric and therefore does not assume an underlying probability distribution of the data series. The statistical significance levels used for the trend analysis were 90% (p < 0.10), 95% (p < 0.05), and 99% (p < 0.01). Mainly significant trends are discussed in the following sections. [19] We explored the role of ENSO and PDO on the variability and trends of some of the annual precipitation indices that refer to extreme precipitation (R95p, R99p, R10mm, and R20mm). We assessed linear correlations for both original and detrended time series. The annual time series were detrended by subtracting a simple linear fit. We also linearly regressed the original precipitation indices onto

4.1. PCA‐Based Precipitation Regionalization [20] The basic characteristics of the seasonal precipitation and their main forcing mechanisms in each of the six PCA‐ based precipitation regions shown in Figure 4a are described in this section. [21] The monsoon region in NW‐Mex (MON) has a peak of precipitation in July, during the boreal summer (Figures 4b and 4c) associated with the North American monsoon system, tropical cyclones, gulf surges, and easterly waves [e.g., Vera et al., 2006; Douglas and Englehart, 2007]. This region also receives less than 30% of winter precipitation (Figure 4c) associated with the passage of cold fronts, enhanced by the southern displacement of the subtropical westerly jet stream, which is sometimes associated with El Niño events [e.g., Cavazos and Hastenrath, 1990; Mason and Goddard, 2001; Magaña et al., 2003]. Winter precipitation in the monsoon region is also known to be in phase with the PDO [e.g., Brito‐Castillo et al., 2002; Pavia et al., 2006]. [22] Precipitation in the southern tip of the Baja California Peninsula (BCS) is mainly due to tropical storms and hurricanes during the summer season [Farfán and Fogel, 2007]; it has a bimodal behavior, with peaks in September (Figure 4b) and July [Englehart and Douglas, 2001]. Meteorological stations located in the middle of the Baja California Peninsula, northern Sonora and southwestern Arizona (DES) are characterized by a desert climate with annual precipitation less than 100 mm yr−1 (Figure 1). Here precipitation is mainly due to tropical storms and hurricanes (or remnants) during the summer and the passage of cold fronts and troughs during winter [Cavazos, 2008]. Winter precipitation in BCS and DES regions is favored during neutral to El Niño conditions and the warm phase of the PDO [Pavia et al., 2006]. [23] The region over California and northwestern Baja California (CAL) has a Mediterranean climate with a peak of precipitation in Jan‐Feb (Figures 4b and 4c), when the

1 Auxiliary materials are available in the HTML. doi:10.1029/ 2009JD013248.

4. Results and Discussion

5 of 10

D14111


D14111

Pavia and Graef, 2002; Cavazos and Rivas, 2004; Pavia et al., 2006]. [24] The region in Texas State (TEX) has a peak of precipitation during the summer (Figures 4b and 4c) mainly due to tropical cyclones, mesoscale convective systems, and moisture from the Gulf of Mexico [Mock, 1996] and the Pacific. TEX has a secondary rainfall peak during winter (not shown) associated with the subtropical westerly jet stream and the passage of cold fronts. Winter precipitation over Arizona and New Mexico (ANM) is produced by cutoff lows associated with the southern displacement of the subtropical jet stream mainly during neutral and El Niño conditions. Summer precipitation (Figure 4c) in ANM is related to the monsoon system, remnants of tropical storms, and mesoscale convective systems [e.g., Mock, 1996; Higgins et al., 1997]. Mock [1996] suggests that the interaction between mesoscale convective systems and topography modulates the interannual precipitation on elevated terrain over TEX and ANM. McCabe et al. [2004] document that persistent dry conditions in all of the SW‐USA, including Texas, are favored during low phases of the PDO and warm phases of the Atlantic Multidecadal Oscillation (AMO); the opposite was found for wet conditions. Goodrich [2007] shows that dry winter conditions in the SW‐USA tend to occur during neutral to La Niña conditions and low phases of the PDO. In contrast, Castro et al. [2001] find that an in‐phase combination of La Niña conditions in the tropical Pacific and a low phase of the PDO are associated with a northward‐displaced monsoon anticyclone and therefore, wet summers in ANM.

Figure 5. (a) Statistically significant annual trends in each of the six precipitation regions. Annual time series anomalies and linear trends (dashed lines) of R95p in (b) ANM and (c) MON. One (two) asterisks in Figure 5a indicate statistical significance at the 90% (95%) level. semi‐permanent North Pacific anticyclone weakens allowing the entrance of convective and frontal systems into the region. Favre and Gershunov [2008] show that cyclones (anticyclones) are more frequent over the eastern North Pacific off California when the Aleutian low is strong (weak). These synoptic systems are in part modulated by ENSO and the PDO through teleconnection patterns that affect the winter climate of the entire study area [e.g., Gershunov and Barnett, 1998; Mason and Goddard, 2001;

4.2. Annual Trends of Daily Precipitation Indices [25] Table 2 shows the annual linear trends of the 10 daily precipitation indices for the six regions and the map in Figure 5 shows only the statistically significant trends in each region. The indices that do not show any significant trends are simple daily intensity index (SDII), days with precipitation greater than 20 mm (R20mm), and consecutive dry and consecutive wet days (CDD and CWD). Interestingly, Groisman and Knight [2008] find that CDD, with periods longer than one month, have significantly increased in the SW‐USA. However, their study covers the period 1967–2006, which may reflect the influence of the 2000s drought that our study does not include. [26] ANM is the region with the largest number of significant annual trends (Figure 5a and Table 2). It is characterized by a significant decrease in the maximum annual precipitation in 1 day (RX1day) and in 5 consecutive days (RX5day); TEX shows similar trends. ANM also shows positive trends in total annual precipitation (PRCPTOT), number of days with precipitation greater than 10 mm (R10mm), and total precipitation from heavy rainfall days (R95p and R99p), consistent with Alexander et al. [2006]. Frich et al. [2002] also document a positive trend in R10mm in the SW‐USA. To complement these results, Figure 5b shows the time series anomalies of the annual contribution of the precipitation exceeding the 95th percentile (R95p) in ANM; the linear trend shows a significant increase of 7.3 mm decade−1 in R95p during 1960–1997. A larger frequency of positive anomalies is seen after the end of the 1970s, possibly associated with the warm phase of the PDO and a larger number of El Niño warm events in this period

6 of 10

D14111


D14111

2009a]. To test the relationship between the annual R10mm and R95p time series with the annual PDO and the Oceanic El Niño Index (ONI), linear correlations were obtained for both raw and detrended time series. There is a significant positive correlation (p < 0.01) between PDO and R10mm for both raw (r = 0.47) and detrended (r = 0.43) annual time series. Similar results were obtained between ONI and R95p. PRCPTOT in ANM is also significantly correlated with PDO and ONI. After removal of the PDO and ENSO signals (by linear regression) the trends in the residuals of R10mm and PRCPTOT are not statistically significant, indicating that the observed annual trends are mainly associated to PDO and ENSO natural variations. However, it is important to notice that the residuals show large interannual variations. [27] On the Mexican side, stations located in the monsoon region (MON) and in BCS show significant positive trends in R95p and R99p (Figure 5a). As in ANM, extreme precipitation derived from R95p has also increased significantly (10 mm decade−1) in MON and in BCS (8.9 mm decade−1), as seen in Figure 5c and Table 2. The large annual peak in 1968 in MON (Figure 5c) comes from monsoon and tropical cyclone‐derived rainfall during the summer (Figure 6b). The significant positive annual trends in R95p in ANM and MON are consistent with the results documented by Alexander et al. [2006]. [28] BCS presents significant positive trends in R95p and R99p at annual timescales. Interestingly, BCS does not show any significant trend at seasonal timescales (Figures 6a and 7a). It was found that the major contribution of R95p and R99p comes from the summer season, mainly associated with tropical storms and hurricanes [e.g., Englehart and Douglas, 2001]. But, in some cases of strong El Niño events, the winter contribution of R95p and R99p can be also important. Thus, the combined contribution of the seasonal indices in BCS results in significant positive trends at annual timescales. A linear regression analysis shows that both PDO and ENSO are significantly correlated with R95p and R99p in BCS, and the trends are partially due to natural variations. In contrast, CAL and DES regions do not show any statistically significant annual trends; however, as will be shown later, these two regions have some significant trends at seasonal timescale (Figures 6 and 7).

Figure 6. (a) Statistically significant summer (Jun‐Oct) trends in each of the six precipitation regions. Jun‐Oct time series anomalies and linear trends (dashed lines) of (b) R95p in MON and (c) SDII in CAL. One (two) asterisks in Figure 6a indicate statistical significance at the 90% (95%) level. than in 1960–1976. The large peak in 1965 comes from the contribution of winter rains (Figure 7b) during the 1965–1966 El Niño event, that occurred during a cold phase of the PDO. The PDO has a strong positive trend during the study period mainly due to its natural oscillation; it has been also postulated that the transition to the warm PDO phase in the mid‐1970s is partially related to external forcing from increases in anthropogenic greenhouse gases [Meehl et al.,

4.3. Summer Trends of Daily Precipitation Indices [29] In contrast to the annual results (Figure 5), significant trends during the summer season (Jun‐Oct) are few, as illustrated in the map of Figure 6. The monsoon region in Mexico (MON) shows a significant increase in extreme summer precipitation (8 mm decade−1) exceeding the 95th (R95p) percentile mainly associated with tropical cyclone activity, as documented by Cavazos et al. [2008], who found a significant increase in the intensity of tropical cyclone‐derived heavy rainfall (R95p), but the frequency of the events did not show any significant trend. The interannual variability of R95p and its corresponding linear trend in MON is shown in Figure 6b. Linearly regressing R95p onto the PDO shows a significant correlation (r = 0.39, p < 0.02); after removal of the PDO signal, the trend is not statistically significant, but the residuals’ variability is large. Since MON does not show any significant trends during winter (Figure 7a), the significant trend in R95p during the summer

7 of 10

D14111


D14111

(R99p). ANM has significant positive trends in the number of days with precipitation greater than 10 mm (R10mm) and in the precipitation exceeding the 95th percentile (R95p), with increases of 4.3 mm decade−1 in R95p in the 1960– 1997 period, as indicated by the linear trend in Figure 7b. These two indices are also significant at annual timescales in ANM (Figure 5a). Cayan et al. [1999] document that heavy winter precipitation (above the 90th percentile, R90p) in the SW‐USA is favored during El Niño years. Consistent with Cayan et al. [1999], there are statistically significant (p < 0.01) positive correlations between raw (0.48) and detrended (0.44) winter time series of R10mm (∼ R90p ≥ 13 mm) in ANM and the antecedent summer ONI time series. [32] CAL shows a significant decrease in the maximum precipitation accumulated in 5 days (RX5day, Figure 7). It also shows a significant trend in R10mm, as observed on the map of Figure 7 and in the time series anomalies in Figure 7c, with an increase of 4 winter days in the precipitation greater than 10 mm during the study period. Cayan et al. [1999] and Cavazos and Rivas [2004] show that heavy precipitation (exceeding the 90th percentile, which is approximately above 10 mm in the southern part of CAL) is associated with moderate to strong El Niño events and to neutral ENSO conditions. Our results also indicate a significant positive correlation between R10mm and the antecedent summer ONI time series (r = 0.49, p < 0.01); after removal of the ENSO signal, the residuals still show large variations. According to Higgins et al. [2000] a large proportion of extreme precipitation events in CAL are linked to both intraseasonal (Madden‐Julian oscillation) and interannual (ENSO) fluctuations [Mo and Higgins, 1998].

5. Summary and Conclusions

Figure 7. (a) Statistically significant winter (Nov‐Mar) trends in each of the six precipitation regions. Nov‐Mar time series anomalies and linear trends (dashed lines) of (b) R95p in ANM and (c) R10mm in CAL. One (two) asterisks in Figure 7a indicate statistical significance at the 90% (95%) level. season is the main contributor to its corresponding annual trend in Figure 5a. [30] The other region that has a significant trend during the summer is CAL. Although summer is the dry season in CAL, it shows a significant increase in the SDII. 4.4. Winter Trends of Daily Precipitation Indices [31] The map in Figure 7 shows the significant linear trends of the winter (Nov‐Mar) precipitation indices. Two regions (DES and TEX) have significant positive trends in the total winter precipitation exceeding the 99th percentile

[33] A rotated PCA was applied to monthly precipitation in NW‐Mex and the SW‐USA resulting in six precipitation regions, which were used to estimate regional trends of 10 daily precipitation indices. The ten indices, that include several indices related with extreme precipitation, were calculated in the 184 selected stations in the study area at annual and seasonal (Jun‐Oct and Nov‐Mar) timescales during the 1960–1997 period. The annual and seasonal indices (in the form of anomalies relative to 1960–1997) were averaged over the meteorological stations corresponding to each of the six precipitation regions. Then, regional trends of the 10 precipitation indices were estimated. The slope was determined using Sen’s slope estimator and the statistical significance of the trend was computed using the non‐ parametric Mann‐Kendall test. [34] The annual precipitation indices showed a larger number of statistically significant trends than the seasonal indices, especially in Arizona‐New Mexico (ANM) and in NW‐Mex (MON and BCS), due to the combined contribution of summer and winter indices. Significant and positive annual trends common to ANM, MON, and BCS are from the indices of extreme precipitation exceeding the 95th (R95p) and 99th (R99p) percentiles. The most relevant feature of the summer (Jun‐Oct) precipitation indices is the occurrence of a significant positive trend in R95p in MON, mainly due to tropical cyclones, with the summer index being the main contributor to the annual trend. With the exception of the trend in MON, the most important contri-

8 of 10

D14111


bution to the annual trends comes from the winter season. Four of the six regions in the study area have significant positive trends in extreme winter precipitation (R10mm, R95p or R99p) during the study period. The variability of the annual indices that show statistically significant trends in extreme precipitation (R10mm, R95p and R99p) are partially linked to natural variations resulting from the combined effects of ENSO and PDO and, in most cases, the trends are explained by the PDO; however, after removal of the PDO and ENSO signals, the residuals still show large interannual variations. In this context, Gershunov and Cayan [2003] found significant regional skill at predicting heavy precipitation (above the 90th percentile) in the SW‐USA with lead times of several months during ENSO and non‐ ENSO conditions. However, from the climate change perspective, the Fourth Assessment Report (AR4) of the Intergovernmental Panel on Climate Change [IPCC, 2007] showed that most of the general circulation models (GCMs) of the Coupled Model Intercomparison Phase 3 (CIMP3) were unable to adequately reproduce extreme events, decadal variations and the ENSO evolution. Thus, it will be important to see if the new GCM experiments to be used in the CMIP5 to produce the Fifth Assessment Report of the IPCC [Meehl et al., 2009b] (http://cmip‐pcmdi.llnl.gov/cmip5/) will show relevant improvements in these major issues. [35] Our forthcoming work will focus on investigating temperature and precipitation extremes from a process‐ oriented perspective in order to understand the dynamical changes associated with the North American monsoon system under current and climate change conditions. With this research we contribute to the crosscutting theme activities of the Variability of American Monsoons (VAMOS‐ CLIVAR) on extreme events and anthropogenic climate change [e.g., Berbery and Marengo, 2009]. [36] On a final note, monitoring the climate system is of vital importance to diagnose and predict weather and climate extremes and their changes, which are known to have both positive and negative impacts on society and the environment. NW‐Mex and the SW‐USA need a larger number of meteorological stations with sufficient quality‐controlled daily data to be able to validate seasonal predictions and to reduce the uncertainty of regional climate change projections, especially those related with extreme events. [37] Acknowledgments. This study was supported by CONACYT‐ Ciencia Basica 2005 and UC‐Mexus CONACYT. We thank Sasha Gershunov for useful discussions and for his insightful comments and suggestions to earlier versions of the manuscript. We also want to thank two anonymous reviewers for their relevant suggestions. FClimDex was developed on behalf of the ETCCDMI by the Climate Research Branch of the Meteorological Service of Canada.

References Alexander, L. V., et al. (2006), Global observed changes in daily climate extremes of temperature and precipitation, J. Geophys. Res., 111, D05109, doi:10.1029/2005JD006290. Allen, M. R., and W. J. Ingram (2002), Constraints on future changes in climate and the hydrologic cycle, Nature, 419, 224–232, doi:10.1038/ nature01092. Berbery, H., and J. A. Marengo (2009), Editorial, CLIVAR Exchanges, 48, 15–16. Brito‐Castillo, L., A. Leyva‐Contreras, A. V. Douglas, and D. Lluch‐Belda (2002), Pacific Decadal Oscillation and the filled capacity of dams on the

D14111

rivers on the Gulf of California continental watershed, Atmosfera, 15, 121–136. Castro, C., T. McKee, and R. Pielke (2001), The relationship of the North American monsoon to tropical and North Pacific sea surface temperatures as revealed by observational analyses, J. Clim., 14, 4450–4473, doi:10.1175/1520-0442(2001)0142.0.CO;2. Cavazos, T. (2008), Clima, in Bahia de los Angeles: Recursos Naturales y Comunidad, Linea Base 2007, edited by G. D. Danemann and E. Ezcurra, pp. 67–92, SEMARNAT, Mexico. Cavazos, T., and S. Hastenrath (1990), Convection and rainfall over Mexico and their modulation by the Southern Oscillation, Int. J. Climatol., 10, 377–386, doi:10.1002/joc.3370100405. Cavazos, T., and D. Rivas (2004), Variability of extreme precipitation events in Tijuana, Mexico, Clim. Res., 25, 229–243, doi:10.3354/ cr025229. Cavazos, T., C. Turrent, and D. P. Lettenmaier (2008), Extreme precipitation trends associated with tropical cyclones in the core of the North American monsoon, Geophys. Res. Lett., 35, L21703, doi:10.1029/ 2008GL035832. Cayan, D. R., K. T. Redmond, and L. G. Riddle (1999), ENSO and hydrologic extremes in the western United States, J. Clim., 12, 2881–2893, doi:10.1175/1520-0442(1999)0122.0.CO;2. Comrie, A. C., and E. C. Glenn (1998), Principal components‐based regionalization of precipitation regimes across the southwest United States and northern Mexico with an application to monsoon precipitation variability, Clim. Res., 10, 201–215, doi:10.3354/cr010201. 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. Clim., 11, 3095–3111, doi:10.1175/15200442(1998)0112.0.CO;2. Diffenbaugh, N. S., F. Giorgi, and J. S. Pal (2008), Climate change hotspots in the United States, Geophys. Res. Lett., 35, L16709, doi:10.1029/ 2008GL035075. Douglas, A. V., and P. J. Englehart (2007), A climatological perspective of transient synoptic features during NAME 2004, J. Clim., 20, 1947–1954, doi:10.1175/JCLI4095.1. Durre, I., M. J. Menne, and R. S. Vose (2008), Strategies for evaluating quality assurance procedures, J. Appl. Meteorol. Climatol., 47, 1785– 1791, doi:10.1175/2007JAMC1706.1. Englehart, P. J., and A. V. Douglas (2001), The role of eastern North Pacific tropical storms in the rainfall climatology of western Mexico, Int. J. Climatol., 21, 1357–1370, doi:10.1002/joc.637. Farfán, L. M., and I. Fogel (2007), Influence of tropical cyclones on humidity patterns over southern Baja California, Mexico, Mon. Weather Rev., 135, 1208–1224, doi:10.1175/MWR3356.1. Favre, A., and A. Gershunov (2008), North Pacific cyclonic and anticyclonic transients in a global warming context: Possible consequences for western North American daily precipitation and temperature extremes, Clim. Dyn., 32, 969–987, doi:10.1007/s00382-008-0417-3. Frich, P., L. V. Alexander, P. Della‐Marta, B. Gleason, M. Haylock, A. Klein Tank, and T. Peterson (2002), Observed coherent changes in climate extremes during the second half of the twentieth century, Clim. Res., 19, 193–212, doi:10.3354/cr019193. Gershunov, A., and T. Barnett (1998), Inter‐decadal modulation of ENSO teleconnections, Bull. Am. Meteorol. Soc., 79, 2715–2725, doi:10.1175/ 1520-0477(1998)0792.0.CO;2. Gershunov, A., and D. Cayan (2003), Heavy daily precipitation frequency over the contiguous United States: Sources of climate variability and seasonal predictability, J. Clim., 16, 2752–2765, doi:10.1175/15200442(2003)0162.0.CO;2. Goodrich, G. B. (2007), Influence of the Pacific decadal oscillation on winter precipitation and drought during years of neutral ENSO in the Western United States, Weather Forecasting, 22, 116–124, doi:10.1175/ WAF983.1. Groisman, P. Y., and R. W. Knight (2008), Prolonged dry episodes over the conterminous United States: New tendencies emerging during the last 40 years, J. Clim., 21, 1850–1862, doi:10.1175/2007JCLI2013.1. Groisman, P. Y., R. W. Knight, D. R. Easterling, T. R. Karl, G. C. Hegerl, and V. N. Razuvaev (2005), Trends in intense precipitation in the climate record, J. Clim., 18, 1326–1350, doi:10.1175/JCLI3339.1. Gutzler, D. S. (2004), An index of interannual variability in the core of the North American monsoon region, J. Clim., 17, 4473–4480, doi:10.1175/ 3226.1. Higgins, R. W., Y. Yao, and X. L. Wang (1997), Influence of the North American monsoon system on the U.S. summer precipitation regime, J. Clim., 10, 2600–2622, doi:10.1175/1520-0442(1997)0102.0.CO;2. Higgins, R. W., J.‐K. Schemm, W. Shi, and A. Leetmaa (2000), Extreme precipitation events in the Western United States related to tropical forc-

9 of 10

D14111


ing, J. Clim., 13, 793–820, doi:10.1175/1520-0442(2000)0132.0.CO;2. Higgins, R. W., V. B. S. Silva, W. Shi, and J. Larson (2007), Relationships between climate variability and fluctuations in daily precipitation over the United States, J. Clim., 20, 3561–3579, doi:10.1175/JCLI4196.1. Intergovernmental Panel on Climate Change (IPCC) (2007), Climate Change 2007. Synthesis Report. Contribution of the Working Groups I, II, and III to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, edited by R. K. Pachauri and A. Reisinger, IPCC, Geneva, Switzerland. Kendall, M. G. (1955), Rank Correlation Methods, 196 pp., Charles Griffin, London. Kharin, V. V., and F. W. Zwiers (2005), Estimating extremes in transient climate change simulations, J. Clim., 18, 1156–1173, doi:10.1175/ JCLI3320.1. Knapp, A. K., et al. (2008), Consequences of more extreme precipitation regimes for terrestrial ecosystems, BioScience, 58, 811–821, doi:10.1641/ B580908. Kunkel, K. E., et al. (2008), Chapter 2: Observed changes in weather and climate extremes, in Weather and Climate Extremes in a Changing Climate. Regions of Focus: North America, Hawaii, Caribbean, and U.S. Pacific Islands, edited by T. R. Karl et al., chap. 2, pp. 35–80, U.S. Clim. Change Sci. Program, Washington, D. C. Magaña, V. O., J. L. Vázquez, J. L. Pérez, and J. B. Pérez (2003), Impact of El Niño on precipitation in Mexico, Geofis. Int., 42(3), 313–330. Mann, H. B. (1945), Non‐parametric tests against trend, Econometrica, 13, 245–259, doi:10.2307/1907187. Mantua, N. J., S. R. Hare, Y. Zhang, J. M. Wallace, and R. C. Francis (1997), A Pacific interdecadal climate oscillation with impacts on salmon production, Bull. Am. Meteorol. Soc., 78, 1069–1079, doi:10.1175/15200477(1997)0782.0.CO;2. Mason, S. J., and L. Goddard (2001), Probabilistic precipitation anomalies associated with ENSO, Bull. Am. Meteorol. Soc., 82, 619–638, doi:10.1175/1520-0477(2001)0822.3.CO;2. McCabe, G. J., M. A. Palecki, and J. L. Betancourt (2004), Pacific and Atlantic Ocean influences on multidecadal drought frequency in the United States, Proc. Natl. Acad. Sci. U. S. A., 101(12), 4136–4141, doi:10.1073/pnas.0306738101. Meehl, G. A., J. A. Hu, and B. D. Santer (2009a), The mid‐1970s climate shift in the Pacific and the relative roles of forced versus inherent decadal variability, J. Clim., 22, 780–792, doi:10.1175/2008JCLI2552.1. Meehl, G. A., et al. (2009b), Decadal prediction: Can it be skillful?, Bull. Am. Meteorol. Soc., 90, 1467–1485, doi:10.1175/2009BAMS2778.1. Mesinger, F., et al. (2006), North American Regional Reanalysis, Bull. Am. Meteorol. Soc., 87, 343–360, doi:10.1175/BAMS-87-3-343. Milly, P. C. D., J. Betancourt, M. Falkenmark, R. M. Hirsch, Z. W. Kundzewicz, D. P. Lettenmaier, and R. J. Stouffer (2008), Stationarity is dead: Whither water management? Science, 319, 573–574, doi:10.1126/ science.1151915.

D14111

Mo, K. C., and R. W. Higgins (1998), Tropical convection and precipitation regimes in the western United States, J. Clim., 11, 2404–2423, doi:10.1175/1520-0442(1998)0112.0.CO;2. Mock, C. J. (1996), Climatic controls and spatial variations of precipitation in the western United States, J. Clim., 9, 1111–1125, doi:10.1175/15200442(1996)0092.0.CO;2. Pavia, E. G., and F. Graef (2002), The recent rainfall climatology of the Mediterranean Californias, J. Clim., 15, 2697–2701, doi:10.1175/15200442(2002)0152.0.CO;2. Pavia, E. G., F. Graef, and J. Reyes (2006), PDO–ENSO effects in the climate of Mexico, J. Clim., 19, 6433–6438, doi:10.1175/JCLI4045.1. Peterson, T. C., X. Zhang, M. Brunet‐India, and J. L. Vázquez‐Aguirre (2008), Changes in North American extremes derived from daily weather data, J. Geophys. Res., 113, D07113, doi:10.1029/2007JD009453. Ropelewski, C., and M. Halpert (1986), North American precipitation and temperature patterns associated with the El Niño/Southern oscillation (ENSO), Mon. Weather Rev., 114, 2352–2362, doi:10.1175/15200493(1986)1142.0.CO;2. Seager, R. (2007), The turn of the century North American drought: Global context, dynamics, and past analogs, J. Clim., 20, 5527–5552, doi:10.1175/2007JCLI1529.1. Seager, R., et al. (2007), Model projections on an imminent transition to a more arid climate in the southwestern North America, Science, 316(5828), 1181–1184, doi:10.1126/science.1139601. Sen, P. K. (1968), Estimates of the regression coefficient based on Kendall’s Tau, J. Am. Stat. Assoc., 63, 1379–1389, doi:10.2307/2285891. Smith, T. M., R. W. Reynolds, T. C. Peterson, and J. Lawrimore (2008), Improvements to NOAA’s historical merged land‐ocean surface temperature analysis (1880–2006), J. Clim., 21, 2283–2296, doi:10.1175/ 2007JCLI2100.1. Stahle, D. W., E. R. Cook, J. Villanueva Diaz, F. K. Fye, D. J. Burnette, R. D. Griffin, R. Acuña Soto, R. Seager, and R. R. Heim Jr. (2009), Early 21st‐century drought in Mexico, Eos Trans. AGU, 90(11), 89–100, doi:10.1029/2009EO110001. Trenberth, K. E., J. Fasullo, and L. Smith (2005), Trends and variability in column‐integrated atmospheric water vapor, Clim. Dyn., 24, 741–758, doi:10.1007/s00382-005-0017-4. Vera, C., et al. (2006), Toward a unified view of the American monsoon systems, J. Clim., 19, 4977–5000, doi:10.1175/JCLI3896.1. Vose, R. S., R. L. Schmoyer, P. M. Steurer, T. C. Peterson, R. Heim, T. R. Karl, and J. Eischeid (1992), The Global Historical Climatology Network: Long‐term monthly temperature, precipitation, sea level pressure, and station pressure data, ORNL/CDIAC‐53 NDP‐041, 317 pp., Carbon Dioxide Inf. Anal. Cent., Oak Ridge Natl. Lab., Oak Ridge, Tenn. White, D., M. Richman, and B. Yarnal (1991), Climate regionalization and rotation of principal components, Int. J. Climatol., 11, 1–25. T. Cavazos and S. Arriaga‐Ramírez, Departamento de Oceanografía Física, CICESE, c/o PO Box 434844, San Diego, CA 92143, USA. ([email protected])

10 of 10