1

1

2

3 ON THE LANDFALL OF NORTHEASTERN PACIFIC TROPICAL JOHN,

4 LANE AND PAUL (2006) OVER NORTHWESTERN

5

6 LUIS M. FARFÁN

7 Centro de Investigación Científica y de Educación Superior de Ensenada B.C. (CICESE)

8 La Paz, Baja Sur, Mexico

9

10 ROSARIO ROMERO-CENTENO and G.B. RAGA

11 Centro de Ciencias de la Atmósfera

12 Universidad Nacional Autónoma de México

13 Mexico City, Mexico

14

15

16 Submitted to Monthly Weather Review

17 American Meteorological Society

18 16 July 2009

19

20 Corresponding author address: 21 Luis M. Farfán 22 CICESE, Unidad La Paz 23 Miraflores 334, La Paz, BCS 23050, México. 24 E-mail: [email protected]

1 2

25 ABSTRACT 26 27 This study is focused on the life cycle of Tropical Cyclones John, Lane, and Paul in the

28 Northeastern Pacific Ocean. These systems developed during the relatively active season of 2006

29 and made landfall over the coast of Northwestern Mexico. Based on a regional network, heavy

30 rainfall was documented throughout the Peninsula and mainland Mexico. A

31 gridded analysis reveals storm motion related to large-scale circulations including an

32 over the and a middle-latitude system approaching the western .

33 Tropical John resulted in a maximum of total rainfall in the range of 350–510 mm,

34 concentrated along the storm track that moved over the peninsular mountains. These accumulations

35 set new records of daily rates with respect to the available dataset from the base period 1970-2005.

36 Later in the season, Tropical Cyclones Lane and Paul made landfall over the mainland and brought

37 low to moderate rainfall accumulations over relatively large communities in the state of .

38 Lane moved close to the coast and was a category-three hurricane at landfall with impact over the

39 southern . In contrast, Paul followed a recurving track that moved around Isla

40 Socorro and arrived to land as a tropical depression. An analysis of the official forecasts issued one

41 to three days prior to landfall reveals some track errors for John and Lane, more likely due to

42 limitations in the temporal coverage of upper-air soundings available from the operational network

43 established in central and western Mexico.

2 3

44 1. Introduction 45 The west coast of North America routinely experiences landfall of Tropical Cyclones

46 (hereafter referred to as TCs) from the Northeastern Pacific Ocean, a basin located east of 140°W.

47 During recent decades, several studies have been focused in TCs that resulted in landfall over the

48 coastal areas of Mexico. For example, Serra (1971) applied track information from the period

49 1921-1969 to document that TCs moving across the coastline had maximum frequency from

50 August through October. Jáuregui (2003) found that between 1951 and 2000, 65 hurricanes1 (88

51 tropical storms) made landfall in Mexico’s west coast and that 60% (49%) of them occurred in the

52 Northwest region of the country. Some of these TCs continued moving northward and, eventually,

53 they had influence on the weather conditions over the southwestern United States (Smith 1986;

54 Garza 1999).

55 In the present study, Northwestern Mexico is defined as the region located north of 20ºN

56 and west of 104ºW. As shown in Fig 1, this region includes the mainland states of , Sinaloa

57 and as well as the Baja California Peninsula. The topography includes mountain ranges that

58 are oriented in a southeast-northwest direction: the Baja California range and the Sierra Madre

59 Occidental. The latter is a wide range (200-300 km) located over the mainland with maximum

60 elevations over 2000 m, while the former is a narrow range (50-100 km) with maximum elevations

61 just above 1000 m. Elevations above 1500 m are located over the state of Baja California, north of

62 28°N. Northwestern Mexico has a coastal length of 6408 km (National Institute of Geography in

63 Mexico, INEGI, http://www.inegi.gob.mx/inegi/default.aspx), which represents 57% of the total in

64 the whole country including the Pacific Ocean, the Gulf of California, the Gulf of Mexico and the

1 In this paper the term hurricane refers to maximum sustain winds > 33 m s-1, tropical storm to winds from 17 to 33 m s-1 and tropical depression to winds < 17 m s-1.

3 4

65 coastlines. The state of (BCS), south of 28°N, has coasts along

66 both the Pacific and the Gulf of California equivalent to 42% of the total length in the northwest.

67 In 2005, 9% of the nation’s population lived in Northwest Mexico and the largest

68 communities were located close to the coast, over relatively flat terrain and below 300 m above sea

69 level (Fig. 1, Table 1). Sinaloa is the most densely populated state (45 inhabitants km-2) while BCS

70 is the least populated state (7 inhabitants km-2) in the region. These facts suggest that, upon TC

71 landfall, major differences should be expected between the impact from storm tracks that go over

72 the peninsula from those moving across the mainland.

73 Historical records for the Northeastern Pacific Basin, issued by the U.S. National Hurricane

74 Center (NHC), reveal the development of more than 500 named TCs during the period 1970–2006.

75 As also shown in Fig. 1, there were 63 landfall hits over Northwestern Mexico where the majority

76 of them occurred along the coasts of BCS and Sinaloa with, roughly, equal number of strikes per

77 state. Figure 2 displays the TC tracks of a couple of groups that, during such 37-year period, were

78 identified to make a first landfall over either the peninsula or the mainland. The figure inserts show

79 the temporal frequency of TC incidence, determined by dividing each month into three periods of

80 10 or 11 days.

81 The upper panel of Fig. 2 shows 27 TC tracks that first moved over the BCS. The incident

82 systems approached the peninsula from the south and most of them occurred from late August

83 through early October (see insert in Fig. 2a). Twelve TCs moved over land through the Gulf of

84 California, east from 110°W and across the eastern peninsula. Fifteen systems made landfall

85 through the Pacific coast and seven of them crossed the gulf to, eventually, reach the mainland.

86 During the same period, 25 TCs tracked only over the mainland (Fig. 2b). This group developed

87 later in the season with the highest frequencies occurring during the last third of September (22%)

4 5

88 and all of October (61%). Note that prior to landfall, and by Isla Socorro, the mainland tracks

89 tended to acquire an eastward component of motion and the preferred landfall region is bounded by

90 21°N and 28°N. Twenty-one systems moved across the state of Sinaloa, two over Nayarit and

91 another two over southern Sonora.

92 Farfán (2004) examined general features of the large-scale environment associated with a

93 sample of six tracks that, during the period 1992-2002, made landfall over Baja California. By

94 using the NCEP-NCAR reanalysis, the author found that mean 500-mb height fields consist of an

95 anticyclone over the Gulf of Mexico and a wave trough traveling eastward across the western

96 United States. A similar analysis was performed here and the results indicate that during landfall in

97 the mainland, there is a southward shift of the wave trough and an eastward displacement of the

98 anticyclone’s center to be located just south of Florida. This configuration allows TC motion over

99 the along with the advection of a cooler and drier environment from the

100 Pacific Ocean, off the peninsula, into northern Mexico.

101 Further analysis of the 1970-2006 period suggests that stronger TCs tend to make landfall

102 over the mainland or over the southeastern edge of the peninsula. This is likely due, in part, to the

103 warmer SSTs over the Gulf of California with respect to the cooler waters of the California Current

104 in the Pacific (e.g., Mosiño and García 1974; Amador et al. 2006). Information on strong systems

105 at landfall is shown in the lower-left inserts of Figs. 2a and 2b. Some of mainland strikes are

106 associated with major hurricanes (this is, category-three or higher) in the Saffir-Simpson scale. In

107 fact, four out of 11 incident systems were major hurricanes while moving across the mainland

108 coastline. In contrast, 11 hurricanes moved through the peninsula with four of them along the

109 Pacific coast as category-one hurricanes and the rest of them went across the southern gulf. Only

110 Kiko (1989) and Paul (1982) were in the lower edge of the category-three hurricanes at landfall.

5 6

111 During the 2006 season in the Northeastern Pacific, from 27 May until 20 November, 18

112 named TCs developed and ten of them reached hurricane strength (Pasch et al 2009). These

113 numbers are slightly above the long-term averages for the period 1970-2006: 16.4 tropical storms

114 and 9.6 becoming hurricanes (Arguez 2007). In 2006, two systems made landfall in Northwestern

115 Mexico as hurricanes and one as a tropical depression. John and Lane arrived as category-two and

116 category-three hurricanes, respectively, and both became the third strongest TCs at landfall over

117 Baja California and over the mainland. Strong winds and heavy rainfall affected coastal

118 communities by causing extensive property damage and thirteen deaths (Pasch et al. 2009). No

119 landfall events occurred anywhere else in the basin, which makes Northwestern Mexico a unique

120 region to receive attention during the analysis of that season. The tracks of the landfalling TCs in

121 2006 are shown as thick lines in Fig. 2.

122 This study is focused on TC impact over the Baja California Peninsula and over the

123 mainland, and our specific objectives are:

124 · To document TC motion and intensity changes, as well as upper-level circulations, from the

125 NHC best-track database and a regional reanalysis.

126 · To determine the associated rainfall patterns from a network of stations operated by the

127 Mexican government. To compare these rainfall patterns and impact of selected case studies

128 with respect to the 1970-2005 climatology of TC landfall in the region.

129 · To evaluate the ability of the official forecasts, issued 24-72 hours prior to landfall, to

130 correctly predict tracks and discuss repercussions over populated areas.

131 This paper is divided into four sections. Section 2 describes the applied datasets. Individual

132 analysis from the case studies and official forecast performance are described in sections 3 and 4,

133 respectively. Finally, section 5 provides a summary and conclusion.

6 7

134 2. Datasets

135 During the last few decades, a significant source of TC observations is given by the satellite

136 surveillance and, in many cases, its imagery is the only available method for estimating motion and

137 intensity (Velden et al. 2003). Digital data from the Geostationary Operational Environmental

138 Satellite-11 (GOES-11), infrared and water vapor channels with spatial (temporal) resolution of 4

139 km and 8 km (30 min), respectively, are used to determine the cloud-cover structures and moisture

140 patterns. High-resolution (1-km, 15-min) imagery from the visible channel was applied to identify

141 significant features within the TCs of interest.

142 Rainfall estimates were obtained from the Tropical Rainfall Measuring Mission (TRMM)

143 product 3B42 version 6, which calibrates and combines precipitation estimates from different

144 instruments to adjust those from geostationary infrared observations. The gridded output has a

145 three-hourly resolution and 0.25x0.25-degree spatial resolution in a global belt extending from

146 50°S to 50°N. This product provides high temporal sampling for rainfall in the Tropics and it has

147 proven to be useful for a more detailed depiction of the TC structure (Hawkins et al. 2001).

148 To determine spatial and intensity patterns of significant precipitation episodes, data from

149 the ground-based network of rain gauges managed by the Comisión Nacional del Agua in Mexico

150 (CNA) are used. These records are available as 24-h totals ending at 1500 UTC. In the following

151 section, a set of 97 stations in BCS and 40 stations in Sinaloa are analyzed. The majority of stations

152 are located over relatively low-level terrain (below 300 m) and, therefore, the network is able to

153 provide only limited estimates at higher elevations. Some data from surrounding states (Sonora,

154 Durango and ) were incorporated to broaden our analysis. In order to set a climatological

155 perspective, daily records from the period 1970-2005 were examined and compared with the

156 accumulations associated with the landfall of TCs John, Lane, and Paul.

7 8

157 Data from the latest population census (INEGI 2005) are used to provide an estimate on the

158 distribution of major population centers within the area of interest. Table 1 provides the top-five

159 communities in the states of BCS and Sinaloa (geographical location shown in Fig. 1). The most

160 populated community in BCS is La Paz, the state capital, with approximately 37% of the state’s

161 population, while the rest of the communities are relatively small (<60,000 inhabitants). In contrast,

162 Sinaloa has a population of over two million inhabitants and Culiacán, the largest city, has 23% of

163 the total population in the state. Note that Sinaloa has a population five times larger than that in

164 BCS and that Los Mochis, Mazatlán and Culiacán are larger than La Paz.

165 The large-scale flow observed during the development of the selected case studies is

166 derived from the North American Regional Reanalysis (NARR, Mesinger et al. 2006) gridded data.

167 This is a dataset issued by NCEP that provides a three-hourly analysis system with a horizontal

168 resolution of 32 km and 29 isobaric levels over North America. In the present study, wind fields at

169 selected pressure levels are inspected to identify relevant circulations within the troposphere over

170 the Pacific Ocean, Gulf of Mexico, and continental areas of Mexico and the United States.

171 Previously, this reanalysis has been successful to study the development of tropical systems. For

172 example, Higgins and Shi (2005) were able to apply NARR fields to identify Northeastern Pacific

173 TCs and their relationship with surges of low-level moisture over the Gulf of California.

174

175 3. Results

176 Track and intensity evolution of the three TCs that made landfall over Northwestern

177 Mexico during the 2006 season are shown in Fig. 3. In early September, John moved over southern

178 BCS as a category-two hurricane, and it passed close to the major population centers in the state.

179 Later in the season, and Tropical Depression Paul moved over the southern Gulf of

8 9

180 California to cross the mainland south of Culiacán, Sinaloa. Note that, as soon as Lane made

181 landfall, the wind speed decreased dramatically from 57 m s-1 to reach the tropical-storm threshold

182 in a matter of 12 hours (insert in Fig. 3). This is likely due to frictional effects and reduction of air-

183 sea interactions imposed by the circulation’s motion over the steep topography of the Sierra Madre

184 Occidental (Fig. 1). In contrast, John experienced slow weakening to become a tropical storm in

185 more than 36 hours. This fact is related to the low elevations and reduced width of the Baja

186 California mountain range as well as to the marginal presence of the gulf at the right side of the

187 low-level circulation.

188 Figure 4 shows the spatial distribution of the accumulated rainfall derived from TRMM

189 during the life cycle of the three case studies. While moving over the ocean, large accumulations (>

190 250 mm) are observed around the John’s center of circulation (Fig. 4a). In contrast, during landfall,

191 moderate amounts (50-150 mm) are estimated along most of the coast of Northwestern Mexico,

192 including half of the peninsula. On the other hand, no rainfall over the peninsula and Sonora is

193 associated with the passage of Lane but an enhanced area is found over the southern gulf, west of

194 Nayarit (Fig. 4b). This pattern may be associated with relatively strong winds at the surface and the

195 storm core being close to the continent. Finally, the development of Paul (Fig. 4c) became a limited

196 source of rainfall for a portion of the southern gulf. Here, one reason to explain this pattern is based

197 in both a weakening trend and a relatively fast forward motion of the cyclonic circulation while

198 approaching central Sinaloa (Fig. 3).

199 At landfall, GOES-11 infrared imagery from Hurricanes John (Fig. 5a) and Lane (Fig. 5b)

200 show well-defined areas of high cloud tops around the circulation centers with minimum

201 temperatures below -60°C. The corresponding visible images (not shown) include spiral bands and

202 an , which are consistent with their hurricane strength. Landfall specifics are taken from the

9 10

203 post-storm reports and advisory archives available at NHC (http://www.nhc.noaa.gov/pastall.htm).

204 According to NHC estimates, Lane (57 m s-1) was stronger than John (49 m s-1) and it had larger

205 size with respect to the mean radii of tropical storm (164 km versus 88 km) and hurricane (36 km

206 versus 31 km) winds. In contrast, Tropical Depression Paul (Fig. 5c) arrived as a compact and

207 weakening (<13 m s-1) circulation with a reduced area of high cloud tops a couple of hours before

208 its final position.

209

210 3.1

211 This system was first detected, on 28 August, south of the Gulf of Tehuantepec (12.3°N,

212 94.6°W in Fig. 3). The NARR fields depicted in Fig. 6 indicate that, during the following few days,

213 the large-scale flow was controlled by a mid-level anticyclone centered over most of Mexico and

214 the southwestern United States (Fig. 6a-c). The anticyclone’s southern flank was providing a

215 steering current from the southeast, which is consistent with John’s displacement parallel to the

216 coastline of the Pacific Ocean (see dots in Figs. 6a-c). In addition, the fields indicate that the

217 vertical shear of horizontal winds between the lower (850 mb) and upper troposphere (200 mb) was

218 relatively low, in the range from 5 to 15 ms-1. This condition is known to be favorable for TC

219 intensification (e.g., Zehr 2003) and may play a role in the limited changes of intensity prior to

220 landfall, from 31 August through 2 September (insert in Fig. 3). As shown in Fig. 5a, landfall

221 occurred early on 2 September in an area just east of San José del Cabo.

222 Total rainfall from John’s passage over BCS are shown in Fig. 7, which is based on the

223 CNA rain-gauge network from 31 August to 4 September. This period includes a couple of days

224 prior to landfall and is consistent with persistence of high cloud tops in the GOES-11 infrared

225 imagery. Additionally, Table 2 presents the top-ten accumulations and the corresponding stations

10 11

226 are found to be within 30 km from the storm track. The pattern of significant in-situ rainfall over

227 the peninsular mountains is consistent with TRMM estimates (Fig. 4a) and both sources

228 characterize the lack of rainfall over the northwestern corner of the state, west of 113°W. However,

229 in-situ observations are able to identify substantial amounts of total rain that are underestimated by

230 the retrieval algorithm used by TRMM.

231 Considerable accumulations (> 440 mm) were recorded at stations 1 through 4 in Fig. 7,

232 which are located above the 300-m level. According to the census information (INEGI 2005), most

233 of these sites are relatively isolated settlements with less than 300 inhabitants each. The

234 development of flash-flooding episodes was likely around locations with steep topography such as

235 Agua de San Antonio (station 4, altitude 825 m). In contrast, the five largest communities listed in

236 Table 1 received total accumulations in the range from only 49 mm (Ciudad Constitución) up to

237 265 mm (Loreto). This fact suggests that limited to moderate rainfall-related damage was expected

238 to affect the majority (67%) of the state population that these communities comprise.

239 Landfall positions (triangles in Fig. 7) of earlier TCs listed in the lower-left insert in Fig. 2a

240 indicate a preference for arrival south of 24°N. This fact imposes a high risk to the population in

241 the southern peninsula and indicates that John made landfall over a coastal area that has been

242 affected several times, by category-three and category-two hurricanes, during the last few decades.

243 It is interesting to indicate that, during the following 24 hours, John moved over La Paz as a

244 category-one hurricane, and as a tropical storms 65-km east from Ciudad Constitución as well as 50

245 km west from Loreto.

246 Another outstanding aspect from Table 2 is that maximum rates from John, 270–449 mm d–

247 1, set new records at several of the CNA stations in the region. These records are compared with

248 daily observations collected during the period 1970–2005. Most of the previous records (< 425 mm

11 12

249 d–1) were associated with the approach or peninsular landfall of TCs in August or September such

250 as Liza (1976), Lester (1992), Isis (1998), Juliette (2001), Ignacio (2003), and Marty (2003). Some

251 of the above TCs also set records at the largest cities in BCS; however, during the passage of John

252 (2006), no new records were set at the major cities.

253

254 3.2 Hurricane Lane

255 As opposed to John’s track, Lane was associated with landfall over the mainland and lack

256 of measurable rainfall over most of the peninsula and the state of Sonora (Fig. 4b). This situation

257 involved a wave trough that propagated into the western United States and an anticyclonic

258 circulation over the eastern Gulf of Mexico (Figs. 6d-f). Such configuration is consistent with mid-

259 level, steering flow from the south and TC displacement across the Sierra Madre Occidental,

260 through Sinaloa. Animation of the corresponding water vapor imagery and inspection of the NARR

261 precipitable water fields (not shown) are used to detect the arrival of a large dry air mass from the

262 Pacific Ocean and lack of convection over most of northwestern Mexico.

263 Just before landfall, Lane intensified into a category-three hurricane (Pasch 2009) and

264 moved across the coastline, just north of Mazatlán. The relatively low values of 850-200 mb wind

265 shear, shown by light shading in the NARR fields (Figs. 6e-f), are consistent with the observed

266 intensification that occurred over the southern Gulf of California (Fig. 3). Also, during Lane’s

267 passage SSTs were in the range of 28-30°C (NOAA 2006), that is another favorable element to

268 support TC intensification.

269 Figure 8 and Table 3 show a rainfall maximum over 250 mm at Siqueiros (station 1) and

270 Mazatlán (station 3) in southern Sinaloa, while a secondary maximum (46-211 mm) occurred over

271 the central portion of the state (stations 2 and 4-10). Most of the rainfall was collected between 16

12 13

272 and17 September while the center of circulation was approaching the coastline. Note that there are,

273 approximately, equal number of stations along the eastern and western flanks of the storm track

274 that is consistent with the symmetric pattern of deep convection depicted by Fig. 5b. This area

275 includes Culiacán with its relatively large population of more than 600,000 inhabitants. According

276 to Pasch et al. (2009), Lane was responsible for damage to streets and homes due to flooding in El

277 Dorado, Culiacán and Mazatlán.

278 In order to set a historical perspective, Fig. 8 includes landfall positions of earlier strong

279 TCs in Sinaloa from the period 1970-2005 (Fig. 2b). In particular, Olivia (1975) was associated

280 with landfall south of Mazatlán as a category-four hurricane that resulted in 30,000 homeless as

281 well as 500 injured and 30 killed (Baum 1976). Another important event was the landfall of

282 (category-two) Hurricane Tico that, in October of 1983, was responsible for deaths, sunken ships

283 and extensive flooding in Mazatlán (Gunther and Cross 1984). Even though (1976)

284 entered the mainland near Los Mochis, heavy rainfall and flooding occurred over southern BCS

285 that caused in 435 deaths in La Paz (Gunther 1977). In summary, strong hurricanes moving across

286 the southern gulf are associated with significant damage over Sinaloa and, occasionally, over Baja

287 California.

288

289 3.3

290 Paul was active from 21 through 26 October and followed a storm track similar to the

291 recurving cases shown in Fig. 2b. After 24 October, this TC acquired an eastward component of

292 motion that resulted in landfall west of Culiacán, Sinaloa (Fig. 3). While approaching the mainland,

293 the incident system was classified as a weakening tropical depression and GOES-11 imagery

294 indicates that it dissipated within a few hours after landfall. This fact is consistent with the absence

13 14

295 of high cloud tops in the image displayed by Fig. 5c. Paul’s weakening trend initiated in an area

296 west of Isla Socorro where SSTs, during the storm passage, were in the range 26-28°C (NOAA

297 2006) .

298 Animation of infrared imagery was used to determine that active cloud clusters developed

299 ahead of the center of circulation. The large-scale fields from the NARR suggest that the cloud

300 clusters and a moist air mass were advected into the mainland by relatively strong, middle-level

301 flow from the equatorial Pacific (Fig. 6i). The spatial pattern of the low-to-upper level flow

302 indicates that a large area of strong wind shear (>20 m s-1) was present over the Baja California

303 peninsula and it was part of a cut-off, cyclonic circulation moving toward the United States-Mexico

304 border. This configuration favored the eastward component of motion followed by Paul. As during

305 Lane’s landfall, the northward component of motion occurs while a large-scale, anticyclonic

306 circulation is located over the Gulf of Mexico.

307 According to TRMM estimates, between 50 and 150 mm (Fig. 4c) occurred over the

308 southern gulf and right of the center of circulation. In spite of Paul’s relatively weak strength at

309 landfall, significant accumulations were collected at stations around the strike point (Fig. 9). For

310 example, more than 200 mm were reported at stations 1 through 4 that are located just north of

311 central Sinaloa. Inspection of daily records indicates that most of the rainfall occurred on October

312 24 (Table 4), which is one day prior to landfall and when intense cloud clusters were developing

313 ahead of the cyclone’s center of circulation. Heavy rainfall resulted in two deaths in Sinaloa while

314 large waves and high surf caused another couple of deaths in BCS (Pasch et al. 2009).

14 15

315 4. Official forecasts

316 Official forecasts of TC track and intensity are routinely issued by the NHC and they are

317 derived from the analysis of multiple models (Rhome 2007). Because of the close interaction

318 between the NHC and the Mexican Meteorological Service (Rosengaus 2001), the official forecasts

319 are an important guidance tool available to emergency managers in Mexico. This information is

320 extremely useful to define specific coastal areas under warning and watch conditions that are issued

321 to the public.

322 A limited sample of the NHC forecasts issued during the development of TCs John, Lane

323 and Paul was selected. This set is shown in Fig. 10 and includes best-track positions and official

324 forecasts that were available, in a real time basis, over the Internet (NHC 2006). Forecasts are

325 released three hours after the nominal initial time ( 0000, 0600, 1200, and 1800 UTC) and each of

326 these cycles contains projections valid at 12, 24, 36, 48, 72, 96, and 120 hours (Rappaport et al.

327 2009). For this study, and to simplify the analysis of tracks issued prior to landfall, we choose to

328 inspect: 1) the 0300 UTC cycle from 30 August to 1 September (Fig. 10a); 2) the 2100 UTC cycle

329 from 13 to 15 September (Fig. 10b), and 3) the 0300 UTC cycle from 23-25 October (Fig. 10c).

330 These times correspond to approximately 72, 48 and, 24 hours ahead of John’s, Lane’s and Paul’s

331 landfall.

332 Note that with respect to John’s best track (Fig 10a), there was a westward bias in the first

333 forecast and this predicted no landfall at all. This fact resulted in no coastal areas of under watch

334 and warning around the peninsula. Subsequent forecasts did predict landfall, within 50 km west

335 from the actual strike site. The first track moving into Baja California was released 47 hours (at

336 0300 UTC 31 August) before actual landfall and the strike site was predicted to be near Cabo San

337 Lucas, with maximum winds of 57 m s-1. The NHC estimate of intensity at landfall was 49 m s-1

15 16

338 (Pasch et al. 2009) and was located 53 km away from the predicted site. Forecast tracks indicated

339 that the storm center would continue moving away (this is, west) from the Pacific coast and,

340 therefore, limited impact was expected to occur over the relatively large population centers along

341 the Gulf of California, such as La Paz and Loreto. Recall that the actual track was parallel to the

342 peninsula, as given by the thick line in Fig. 10a.

343 Lane was first placed over mainland Mexico in the official forecast released at 0900 UTC

344 15 September. This time corresponds to, approximately, 34 hours prior to actual landfall (1915

345 UTC 16 September) and subsequent forecasts had some position adjustments prior to the strike.

346 Figure 10b depicts the track, from the forecast issued at 2100 UTC 13 September, which followed

347 northwestward motion over the Pacific Ocean. Later, the 14 September forecast suggested a

348 northward path and landfall over the southern peninsula affecting its major population centers.

349 Motion over the southern gulf and landfall in northern Sinaloa was predicted only by the last

350 forecast. Again, with respect to best-track position at landfall, there was a westward bias in the

351 selected set of predictions.

352 More than four days prior to landfall, Paul was first placed over mainland Mexico in the

353 official forecasts from 21 October. Actually, recurvature around Isla Socorro and landfall over

354 Sinaloa was predicted in all of the chosen forecasts (Fig. 10c). These forecasts had position

355 changes but kept the landfall site over central Sinaloa and south of the actual landfall position (Fig.

356 10c). As in the case of Paul, the forecast track and landfall position were also reasonable for

357 in 2002 (Franklin 2003). Kenna made landfall over the state of Nayarit (K02 in

358 Fig. 1) as a category-four hurricane, in late October, while a wave trough propagated over the entire

359 Baja California Peninsula. Therefore, late season paths appear to be easier to predict, given the

16 17

360 development of mid-latitude circulations that are typical of the transition to cool-season conditions

361 in Northwestern Mexico.

362 A comprehensive investigation of the elements that explain why some of the forecast tracks

363 had a westward bias is beyond the scope of this study. However, we suggest that the lack of upper-

364 air observations in Mexico could play a significant role in the correct initialization of numerical

365 models used by the NHC. For this purpose, we made a survey of the observations taken during the

366 forecast periods prior to the landfall of TC John, Lane and Paul. In the first case, only 20% of the

367 maximum number of standard soundings (0000 and 1200 UTC) were taken at the 11 operational

368 sites shown in Fig. 10a. No soundings were taken at five of the nine mainland sites in Mexico;

369 additionally, Isla Socorro, La Paz, and Guaymas (Sonora) had no releases at all during such period.

370 The sounding coverage increased to 35% and 41% prior to the landfall of Lane and Paul,

371 respectively, with more stations active along the west coast. The latter may be used, in part, to

372 explain better track predictions during the approach of Paul to the mainland. However, there are

373 still questions about quality control and assimilation of these observations that need to be carefully

374 examined and they are suggested as part of a subsequent study.

375

376 5. Summary and conclusion

377 The goal of this study was to document the evolution of TCs that made landfall in

378 Northwestern Mexico during the relatively active season of 2006. Based on NHC best-track

379 positions, three Northeastern Pacific systems were selected for further analysis: John, Lane, and

380 Paul. They brought strong winds and heavy rainfall to communities in the Baja California

381 peninsula and the mainland state of Sinaloa. Inspection of data from satellite products, gridded

382 fields, and from a rain-gauge network resulted in the following set of key findings:

17 18

383 · Hurricane John had a track consistent with TCs that developed during late August or the

384 month of September (Fig. 2a). John made landfall over the southern portion of the peninsula, as a

385 category-two hurricane, and was steered by the western boundary of an anticyclonic circulation

386 located over northern Mexico. This event resulted in a well-defined period of extreme rainfall (up

387 to 506 mm) along the mountains with maximum daily rates (198-449 mm d-1) that were

388 comparable to, and in some cases larger than, the climatological records from the period 1970-

389 2005.

390 · Hurricane Lane developed during the approach of a wave trough over the southwestern

391 United States and the presence of an anticyclone over the Gulf of Mexico. Lane moved across the

392 southern Gulf of California and made landfall near the largest population centers in Sinaloa. At

393 landfall, it was a category-three hurricane that became the third strongest TC to move over the

394 mainland. Its passage provided large amounts (up to 297 mm) and significant rates (200-260 mm d-

395 1) of rainfall along the western foothills of the Sierra Madre Occidental but no rainfall at all over

396 the entire peninsula.

397 · Of the three selected TCs, Paul was the only recurving system that moved around Isla

398 Socorro to approach the mainland with an eastward component of motion. Paul had a track

399 consistent with the group of storms that developed late in the season (Fig. 2b). Its convective

400 activity, ahead of the storm center, was associated with limited rainfall (< 225 mm; ≤ 200 mm d-1)

401 over Sinaloa. Paul’s recurvature was linked to the intrusion of a Pacific cut-off, cyclonic circulation

402 and we propose that relatively strong wind shear and cooler SSTs were responsible for the dramatic

403 decrease of intensity after moving close to Isla Socorro.

404 · A simple evaluation of the official forecasts, issued by the NHC prior to landfall, reveals

405 track errors in the estimate of the strike position. Predicted tracks for Hurricanes John and Lane

18 19

406 show the persistence of a westward bias with respect to the actual tracks. In order to explain this

407 tendency, we suggest that the reduced number of upper-air soundings from the Mexican network

408 may play a significant role in the correct initialization of operational models used during the

409 forecast process.

410 In conclusion, the analysis of case studies that made landfall during a single season

411 represents a contribution to the current knowledge of TC impact over Northwestern Mexico. A

412 similar methodology to the one used in section 3 is being applied to other events from the period

413 1970-2005. This extension will provide a more general set of results and will enhance our

414 understanding of: 1) the role played by the location, intensity, and extent of large-scale circulations

415 over the storm tracks and 2) the spatial patterns of rainfall and maximum rates associated with the

416 incident systems. Further TC-related disasters are expected to occur and they become more

417 important under rapid growth of the coastal population and infrastructures. Recent landfalls in 2007

418 (Henriette) and 2008 (Julio, Lowell, and Norbert) confirm the need to continue the examination of

419 TC development along with an objective evaluation of the official forecasts.

420

421

19 20

422 Acknowledgements 423 This work was carried out with aid of the Inter-American Institute for Global Change

424 Research (IAI) CRN II #2048, which is supported by the U.S. National Science Foundation (Grant

425 GEO-0452325), and from the National Council on Science and Technology in Mexico

426 (CONACYT, Grant 23448). Satellite imagery was provided by the Unidata Program Center (UPC)

427 at the University Corporation for Atmospheric Research and the Space Science and Engineering

428 Center (SSEC) of the University of Wisconsin-Madison. Juan D. Amador, Juan García, Alejandro

429 González, and Adolfo Portocarrero from the Comisión Nacional del Agua (CNA) supplied the

430 rainfall observations.

20 21

431 REFERENCES 432 Arguez, A., 2007: Supplement to State of the Climate in 2006. Bull. Amer. Meteor. Soc., 88, s1– 433 s135.

434 Amador, J.A. E.J. Alfaro, E.J., O.G. Lizano and V.O. Magaña, 2006: Atmospheric Forcing of the 435 Eastern Tropical Pacific. Progress in Oceanography, 69, 101-142.

436 Baum, R.A., 1976: Eastern North Pacific Tropical Cyclones of 1975. Mon. Wea. Rev., 104, 475– 437 488.

438 Farfán, L.M., 2004: Regional Observations during the Landfall of Juliette (2001) 439 in Baja California, Mexico. Mon. Wea. Rev., 132, 1575–1589.

440 Franklin, J.L., L.A. Avila, J.L. Beven, M.B. Lawrence, R.J. Pasch, and S.R. Stewart, 2003: Eastern 441 North Season of 2002. Mon. Wea. Rev., 131, 2379–2393.

442 Garza, A. L., 1999: 1985–1998 North Pacific tropical cyclones impacting the southwestern United 443 States and northern Mexico: An updated climatology, NOAA Tech. Memo. NWS WR-258, 444 83 pp.

445 Gunther, E.B., 1977: Eastern Pacific Tropical Cyclones of 1976. Mon. Wea. Rev., 105, 508–522.

446 Gunther, E., and R. Cross, 1984: Eastern North Pacific Tropical Cyclones of 1983. Mon. Wea. 447 Rev., 112, 1419–1440. 448 Hawkins, J.D., T.F. Lee, J. Turk, C. Sampson, J. Kent, and K. Richardson, 2001: Real–Time 449 Internet Distribution of Satellite Products for Tropical Cyclone Reconnaissance. Bull. Amer. 450 Meteor. Soc., 82, 567–578. 451 Higgins, R.W., and W. Shi, 2005: Relationships between Gulf of California Moisture Surges and 452 Tropical Cyclones in the Eastern Pacific Basin. J. Climate, 18, 4601–4620. 453 INEGI, cited 2005: General Census of Population and Housing (in Spanish). 454 http://www.inegi.gob.mx/est/contenidos/espanol/sistemas/conteo2005/localidad/iter/default. 455 asp 456 Jáuregui, E., 2003: Climatology of Landfalling Hurricanes and Tropical Storms in Mexico, 457 Atmósfera, 16, 193–204.

458 Mesinger, F., G. DiMego, E. Kalnay, K. Mitchell, P.C. Shafran, W. Ebisuzaki, D. Jović, J. 459 Woollen, E. Rogers, E.H. Berbery, M.B. Ek, Y. Fan, R. Grumbine, W. Higgins, H. Li, Y.

21 22

460 Lin, G. Manikin, D. Parrish, and W. Shi, 2006: North American Regional Reanalysis. Bull. 461 Amer. Meteor. Soc., 87, 343–360.

462 Mosiño, P. A., and E. García, 1974: The climate of Mexico. Climates of North America, R. A. 463 Bryson and F. K. Hare, Eds., Vol. 2, World Survey of Climatology, Elsevier, 345–404.

464 NHC, cited 2006: NHC Archive of Hurricane Seasons. [Available online at 465 http://www.nhc.noaa.gov/pastall.shtml.] 466 NOAA, cited 2006: NOAA Optimum Interpolation (OI) (SST) V2. 467 [Available online at http://www.cdc.noaa.gov/data/gridded/data.noaa.oisst.v2.html.]

468 Pasch, R.J., E.S. Blake, L.A. Avila, J.L. Beven, D.P. Brown, J.L. Franklin, R.D. Knabb, M.M. 469 Mainelli, J.R. Rhome, and S.R. Stewart, 2009: Eastern North Pacific Hurricane Season of 470 2006. Mon. Wea. Rev., 137, 3–20.

471 Rappaport, E.N., J.L. Franklin, L.A. Avila, S.R. Baig, J.L. Beven, E.S. Blake, C.A. Burr, J.G. Jiing, 472 C.A. Juckins, R.D. Knabb, C.W. Landsea, M. Mainelli, M. Mayfield, C.J. McAdie, R.J. 473 Pasch, C. Sisko, S.R. Stewart, and A.N. Tribble, 2009: Advances and Challenges at the 474 National Hurricane Center. Wea. Forecasting, 24, 395–419.

475 Rhome, J.R., cited 2007: Technical Summary of the National Hurricane Center Track and Intensity 476 Models. Available online at: http://www.nhc.noaa.gov/modelsummary.shtml

477 Rosengaus, M., 2001. An Analysis of the 2000 Tropical Cyclone Season for Mexico. Hydraulic 478 Engineering in Mexico (in Spanish), 16, 5-23.

479 Serra C., S., 1971: Hurricanes and Tropical Storms of the West Coast of Mexico. Mon. Wea. Rev., 480 99, 302–308.

481 Smith W., 1986: The Effects of Estern North Pacific Topical Cclones on the Southwestern United 482 States. NOAA Tech. Memo. NWS WR-197, 229 pp.

483 Velden, C., J. Simpson, W. Timothy Liu, J. Hawkins, K. Brueske and R. Anthes: The Burgeoning 484 Role of Weather Satellites. Hurricane! Coping with Disaster. R. Simpson, Ed., American 485 Geophysical Union, 217-247.

486 Zehr, R.M., 2003: Environmental Vertical Wind Shear with Hurricane Bertha (1996). Wea. 487 Forecasting, 18, 345–356.

22 23

488

489 FIGURE CAPTIONS

490

491 Figure 1. Geographical area of northwestern Mexico and topographic contour intervals (shaded, m)

492 given in the vertical bar. Specific locations mentioned throughout the text are the states of Baja

493 California (BC), Baja California Sur (BCS), Nayarit (NAY), Sinaloa (SIN), and Sonora (SON).

494 Circled letters are the five largest communities in BCS and SIN, denoted in Table 1. Black dots

495 represent landfall positions from tropical cyclones during the period 1970-2005 and plus signs are

496 positions from Tropical Cyclones John (J06), Lane (L06) and Paul (P06) in 2006 and Kenna (K02)

497 in 2002. The dashed box outlines the area shown in Figs. 8 and 9.

498

499 Figure 2. Tracks of tropical cyclones that made landfall over (a) the southern Baja California

500 Peninsula and (b) the mainland of northwestern Mexico. Tracks are from the 1970-2006 period and

501 thick lines are tropical cyclones from the 2006 season. The filled dot represents Isla Socorro.

502 Temporal frequency (percentage) of landfall strikes is shown in upper-right inserts. The lower-left

503 lists provide the five strongest hurricanes at landfall: name, year and maximum winds in m s-1

504 (Saffir-Simpson scale category).

505

506 Figure 3. Tracks of Tropical Cyclones John (circles, August 28-September 4), Lane (plus signs,

507 September 13-17) and Paul (crosses, October 21-26) in 2006. Initial and final positions are

508 represented by upper-case and lower-case letters respectively. Positions are marked every 6 hours,

509 numbers are dates of fixes at 0000 UTC and locations of maximum intensity (MAX) are indicated.

510 The filled dot represents Isla Socorro. Inserts show sustained wind speed (m s-1), maximum

23 24

511 intensity and time of landfall (vertical line). Dashed lines are the tropical storm (17 m s-1) and

512 hurricane (33 m s-1) intensity thresholds.

513

514 Figure 4. Rainfall accumulations from TRMM during the period (a) 28 August - 4 September, (b)

515 13 -17 September and (c) 21 - 26 October, 2006. The spatial resolution of the imagery is 0.25

516 degree and the vertical bar indicates the scale of total accumulations (mm). Tropical cyclone

517 positions are given by circles (John), plus signs (Lane), and crosses (Paul) at 6-hourly intervals.

518

519 Figure 5. Cloud-top temperature from GOES-11 infrared satellite imagery at landfall. The images

520 are from (a) Hurricane John at 0200 UTC 2 September, (b) Hurricane Lane at 1900 UTC 16

521 September, and (c) Tropical Depression Paul at 0400 UTC 26 October 2006. The spatial resolution

522 of the imagery is 4 km and the vertical bar indicates the calibrated scale of temperature (°C).

523

524 Figure 6. North American Regional Reanalysis fields at 0000 UTC from (a-c) 31 August through 2

525 September, (d-f) 15-17 September and (g-i) 23-25 October, 2006. Solid lines represent geopotential

526 heights (dm) and wind barbs (m s-1) at the 500-mb level. Shading is the 850-200 mb horizontal

527 wind shear with light, intermediate and dark tones representing values above 10, 20, and 30 m s-1,

528 respectively. Filled dots are the corresponding tropical cyclone positions at 0000 UTC as well as 12

529 hours before and after.

530

531 Figure 7. Total rainfall and top-ten reports (circled numbers) in Baja California Sur.

532 Accumulations (mm) are for the period 31 August-4 September 2006 and are shown with symbols

533 in the lower left insert. Best-track positions (large dots) and intensities (Hurricane=HR, Tropical

24 25

534 Storm=TS and Tropical Depression=TD) are indicated. Terrain elevations (shaded) are at 300-m

535 intervals. Asterisk is landfall site for John and triangles (name initial and year) are the strongest

536 hurricanes from Fig. 2a.

537

538 Figure 8. Total rainfall and top-ten reports (circled numbers) in Sinaloa. Accumulations (mm) are

539 from the period 15-17 September 2006. Best-track positions (large dots) and intensities

540 (Hurricane=HR, Tropical Storm=TS and Tropical Depression=TD) are indicated. Terrain

541 elevations (shaded) are at 600-m intervals. Asterisk is landfall site for Lane and triangles are three

542 of the strongest hurricanes from Fig. 2b (name initial and year).

543

544 Figure 9. Total rainfall and top-ten reports (circled numbers) in Sinaloa. Accumulations (mm) are

545 from the period 24-26 October 2006. Best-track positions (large dots) and intensities

546 (Hurricane=HR, Tropical Storm=TS and Tropical Depression=TD) are indicated. Terrain

547 elevations (shaded) are at 300-m intervals. Asterisk is landfall site for Paul.

548

549 Figure 10. Best-track and official forecast positions for Tropical Cyclones (a) John, (b) Lane, and

550 (c) Paul. Initial dates of forecasts are given in the lower-left inserts (symbol, day month and time).

551 Forecasts are valid at 12-, 24-, 36-, 48-, 72-, 96-, and 120-hours after initialization. The thick line

552 represents the best track and the filled dots are locations of upper-air stations in Mexico. Numbers

553 indicate released soundings at 0000 (left from slash) and 1200 UTC (right from slash) during the 3

554 days covered by the initialization of each case study.

555

25 26

556 TABLE CAPTIONS

557

558 TABLE 1. Population from largest cities in the states of Baja California Sur and Sinaloa.

559

560 TABLE 2. Rainfall accumulations (mm) from stations in Baja California Sur.

561

562 TABLE 3. Rainfall accumulations (mm) at 10 sites in Sinaloa, September 14-16 2006.

563

564 TABLE 4. Rainfall accumulations (mm) at 10 sites in Sinaloa, October 24-26 2006.

26 27 United States

Sierra Madre Occidental

K02 +

565 566 Figure 1. Geographical area of northwestern Mexico and topographic contour intervals (shaded, m) 567 given in the vertical bar. Specific locations mentioned throughout the text are the states of Baja 568 California (BC), Baja California Sur (BCS), Nayarit (NAY), Sinaloa (SIN), and Sonora (SON). 569 Circled letters are the five largest communities in BCS and SIN, denoted in Table 1. Black dots 570 represent landfall positions from tropical cyclones during the period 1970-2005 and plus signs are 571 positions from Tropical Cyclones John (J06), Lane (L06) and Paul (P06) in 2006 and Kenna (K02) 572 in 2002. The dashed box outlines the area shown in Figs. 8 and 9.

27 28

574

Figure 2. Tracks of tropical cyclones that made landfall over (a) the southern Baja California Peninsula and (b) the mainland of northwestern Mexico. Tracks are from the 1970-2006 period and thick lines are tropical cyclones from the 2006 season. The filled dot represents Isla Socorro. Temporal frequency (percentage) of landfall strikes is shown in upper-right inserts. The lower-left lists provide the five strongest hurricanes at landfall: name, year and maximum winds in m s-1 (Saffir-Simpson scale category). 28 29

Figure 3. Tracks of Tropical Cyclones John (circles, August 28-September 4), Lane (plus signs, September 13-17) and Paul (crosses, October 21-26) in 2006. Initial and final positions are represented by uppercase and lowercase letters respectively. Positions are marked every 6 hours, numbers are dates of fixes at 0000 UTC and locations of maximum intensity (MAX) are indicated. The filled dot represents Isla Socorro. Inserts show sustained wind speed (m s- 1), maximum intensity and time of landfall (vertical line). Dashed lines are the tropical storm (17 m s-1) and hurricane (33 m s-1) intensity thresholds.

29 30

576

Figure 4. Rainfall accumulations from TRMM during the period (a) 28 August - 4 September, (b) 13 -17 September and (c) 21 - 26 October, 2006. The spatial resolution of the imagery is 0.25 degree and the vertical bar indicates the scale of total accumulations (mm). Tropical cyclone positions are given by circles (John), plus signs 30 (Lane), and crosses (Paul) at 6-hourly intervals. 31

578

Figure 5. Cloud-top temperature from GOES-11 infrared satellite imagery at landfall. The images are from (a) Hurricane John at 0200 UTC 2 September, (b) Hurricane Lane at 1900 UTC 16 September, and (c) Tropical Depression Paul at 0400 UTC 26 October 2006. The spatial resolution of the imagery is 4 km and the vertical bar 31 indicates the calibrated scale of temperature (°C). 32

579

Figure 6. North American Regional Reanalysis fields at 0000 UTC from (a-c) 31 August through 2 September, (d-f) 15-17 September and (g-i) 23-25 October, 2006. Solid lines represent geopotential heights (dm) and wind barbs (m s-1) at the 500-mb level. Shading is the 850-200 mb horizontal wind shear with light, intermediate and dark tones representing values above 10, 20, and 30 m s-1, respectively. Filled dots are the corresponding tropical cyclone positions at 0000 UTC as well as 12 hours before and after.

32 33

Figure 7. Total rainfall and top-ten reports (circled numbers) in Baja California Sur. Accumulations (mm) are for the period 31 August-4 September 2006 and are shown with symbols in the lower left insert. Best-track positions (large dots) and intensities (Hurricane=HR, Tropical Storm=TS and Tropical Depression=TD) are indicated. Terrain elevations (shaded) are at 300-m intervals. Asterisk is landfall site for John and triangles (name initial and year) are the strongest hurricanes from Fig. 2a.

33 34

581

Figure 8. Total rainfall and top-ten reports (circled numbers) in Sinaloa. Accumulations (mm) are from the period 15-17 September 2006. Best-track positions (large dots) and intensities (Hurricane=HR, Tropical Storm=TS and Tropical Depression=TD) are indicated. Terrain elevations (shaded) are at 600-m intervals. Asterisk is landfall site for Lane and triangles are three of the strongest hurricanes from Fig. 2b (name initial and year).

34 35

583

Figure 9. Total rainfall and top-ten reports (circled numbers) in Sinaloa. Accumulations (mm) are from the period 24-26 October 2006. Best-track positions (large dots) and intensities (Hurricane=HR, Tropical Storm=TS and Tropical Depression=TD) are indicated. Terrain elevations (shaded) are at 300-m intervals. Asterisk is landfall site for Paul.

35 36

585

Figure 10. Best-track and official forecast positions for Tropical Cyclones (a) John, (b) Lane, and (c) Paul. Initial dates of forecasts are given in the lower-left inserts (symbol, day month and time). Forecasts are valid at 12-, 24-, 36-, 48-, 72-, 96-, and 120-hours after initialization. The thick line represents the best track and the filled dots are locations of upper-air stations in Mexico. Numbers indicate released soundings at 0000 (left from slash) and 1200 UTC (right from slash) during the 3 days covered by the initialization of each case study. 36 37

586

37 38

587

588 TABLE 1. Population from largest cities in the states of Baja California Sur and Sinaloa.

589

590 Baja California Sur (512,170) Sinaloa (2,608,442)______

591

592 A La Paz (189,176) Culiacán (605,304)

593 B (56,811) Mazatlán (352,471)

594 C San José del Cabo (48,518) Los Mochis (231,977)

595 D Ciudad Constitución (37,221) (66,793)

596 E Loreto (10,283) Guamúchil (61,862)

597

598 Number of inhabitants is indicated in parenthesis.

599 Source of data: INEGI 2005

38 39

600 TABLE 2. Rainfall accumulations (mm) from stations in Baja California Sur.

601 602 Station name Total MAXDAY MAXREC Tropical cyclone

603

604 1 San Bartolo 506 449* 425 Liza (1976, Sep)

605 2 San Javier 480 400* 250 Marty (2003, Sep)

606 3 El Huatamote 456 436* 229 Ignacio (2003, Aug)

607 4 Agua de San Antonio 447 295 360 Isis (1998, Sep)

608 5 Santa Gerturdis 360 198 300 Isis (1998, Sep)

609 6 Los Planes 360 300* 240 Juliette (2001, Sep)

610 7 San José de Magdalena 353 313* 170 N/A

611 8 San Lázaro 340 270* 264 Marty (2003, Sep)

612 9 San Antonio Norte 290 290* 124 Marty (2003, Sep)

613 10 Guadalupe 287 222 227 Lester (1992, Aug)

614

615 Total = rainfall (mm) accumulations from 31 August through 4 September 2006

616 MAXDAY = maximum daily accumulation (mm d-1) from Tropical Cyclone John

617 MAXREC = record of maximum daily rainfall (mm) from 1970 through 2005

618 Tropical cyclone = name (year and month) associated with previous record

619 N/A = No tropical cyclone associated with this maximum

620 * = Daily report greater than previous record.

39 40

621 TABLE 3. Rainfall accumulations (mm) at 10 sites in Sinaloa, September 14-16 2006.

622

623 Station name Total 14 Sep 15 Sep 16 Sep

624

625 1 Siqueiros 297 27 49 223

626 2 San Lorenzo 268 5 3 260

627 3 Mazatlán 257 1 9 247

628 4 El Dorado 211 6 205 N/A

629 5 La Cruz 208 2 7 199

630 6 López Portillo 140 24 4 112

631 7 Sanalona 139 21 1 117

632 8 Culiacán 96 2 0 93

633 9 Badiraguato 66 0 0 66

634 10 López Mateos 46 0 0 46 635 636

637 N/A = Not available

638

40 41

639 Table 4. Rainfall accumulations (mm) at 10 sites in Sinaloa, from October 24-26 2006.

640

641 Station name Total Oct 24 Oct 25 Oct 26

642

643 1 Pericos 224 0 200 24

644 2 San Juan 222 0 148 74

645 3 Eustaquio B. 215 0 163 52

646 4 Guamúchil 215 0 163 52

647 5 Diaz Ordaz 147 0 101 46

648 6 Badiraguato 136 0 104 32

649 7 Culiacán 131 0 113 18

650 8 López Mateos 120 0 82 38

651 9 Obs. Culiacán 103 N/A 89 14

652 10 Andrew Weiss 95 0 80 15

653

654

655 N/A = Not available

656

41