• Adams, D. K., , and A. C. Comrie, 1997: The North American monsoon. Bull. Amer. Meteor. Soc, 78 , 21972213.

  • Anderson, B. T., , and J. O. Roads, 2002: Regional simulation of summertime precipitation over the southwestern United States. J. Climate, 15 , 33213342.

    • Search Google Scholar
    • Export Citation
  • Anderson, B. T., , J. O. Roads, , S-C. Chen, , and H-M. H. Juang, 2001: Model dynamics of summertime low-level jets over northwestern Mexico. J. Geophys. Res, 106 (D4) 34013413.

    • Search Google Scholar
    • Export Citation
  • Balling Jr., R. C., , and S. W. Brazel, 1987: Diurnal variation in Arizona monsoon precipitation frequencies. Mon. Wea. Rev, 115 , 342346.

    • Search Google Scholar
    • Export Citation
  • Barlow, M., , S. Nigam, , and E. H. Berbery, 1998: Evolution of the North American monsoon system. J. Climate, 11 , 22382257.

  • Berbery, E. H., 2001: Mesoscale moisture analysis of the North American monsoon. J. Climate, 14 , 121137.

  • Bryson, R. A., , and W. P. Lowery, 1955: Synoptic climatology of the Arizona summer precipitation singularity. Bull. Amer. Meteor. Soc, 36 , 329379.

    • Search Google Scholar
    • Export Citation
  • Carleton, A. M., 1986: Synoptic-dynamic character of “bursts” and “breaks” in the southwest U.S. summer precipitation singularity. J. Climatol, 6 , 605623.

    • Search Google Scholar
    • Export Citation
  • Cavazos, T., , A. C. Comrie, , and D. M. Liverman, 2002: Intraseasonal variability associated with wet monsoons in southeast Arizona. J. Climate, 15 , 24772490.

    • Search Google Scholar
    • Export Citation
  • Chen, F., , and J. Dudhia, 2001: Coupling an advanced land surface hydrology model with the Penn State–NCAR MM5 modeling system. Part I: Model implementation and sensitivity. Mon. Wea. Rev, 129 , 569585.

    • Search Google Scholar
    • Export Citation
  • Dai, A., , F. Giorgi, , and K. E. Trenberth, 1999: Observed and model-simulated diurnal cycles of precipitation over the contiguous United States. J. Geophys. Res, 104 (D6) 63776402.

    • Search Google Scholar
    • Export Citation
  • Douglas, M. W., , and S. Li, 1996: Diurnal variation of the lower-tropospheric flow over the Arizona low desert from SWAMP-1993 observations. Mon. Wea. Rev, 124 , 12111235.

    • Search Google Scholar
    • Export Citation
  • Douglas, M. W., , R. A. Maddox, , K. Howard, , and S. Reyes, 1993: The Mexican monsoon. J. Climate, 6 , 16651677.

  • Douglas, M. W., , A. Valdez-Manzanila, , and R. G. Cueta, 1998: Diurnal variation and horizontal extent of the low-level jet over the northern Gulf of California. Mon. Wea. Rev, 126 , 20172025.

    • Search Google Scholar
    • Export Citation
  • Dudhia, J., 1989: Numerical study of convection observed during the Winter Monsoon Experiment using a mesoscale two-dimensional model. J. Atmos. Sci, 46 , 30773107.

    • Search Google Scholar
    • Export Citation
  • Gao, X., , S. Sorooshian, , J. Li, , and J. Xu, 2003: SST data improve modeling of North American monsoon rainfall. Eos, Trans. Amer. Geophys. Union, 84 , P457,. P462.

    • Search Google Scholar
    • Export Citation
  • Garreaud, R. D., , and J. M. Wallace, 1997: The diurnal march of convective cloudiness over the Americas. Mon. Wea. Rev, 125 , 31573171.

    • Search Google Scholar
    • Export Citation
  • Gochis, D. J., , W. J. Shuttleworth, , and Z-L. Yang, 2002: Sensitivity of the modeled North American monsoon regional climate to convective parameterization. Mon. Wea. Rev, 130 , 12821298.

    • Search Google Scholar
    • Export Citation
  • Grell, G. A., 1993: Prognostic evaluation of assumptions used by cumulus parameterizations. Mon. Wea. Rev, 121 , 764787.

  • Guichard, F., , D. B. Parsons, , J. Dudhia, , and J. Bresch, 2003: Evaluating mesoscale model predictions of clouds and radiation with SGP ARM data over a seasonal timescale. Mon. Wea. Rev, 131 , 926944.

    • Search Google Scholar
    • Export Citation
  • Higgins, R. W., , and W. Shi, 2001: Intercomparison of the principal modes of interannual and intraseasonal variability of the North American monsoon system. J. Climate, 14 , 403417.

    • Search Google Scholar
    • Export Citation
  • 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. Climate, 10 , 26002622.

    • Search Google Scholar
    • Export Citation
  • Higgins, R. W., , Y. Chen, , and A. V. Douglas, 1999: Interannual variability of the North American warm season precipitation regions. J. Climate, 12 , 653680.

    • Search Google Scholar
    • Export Citation
  • Hong, S-Y., , and H-L. Pan, 1996: Nonlocal boundary layer vertical diffusion in a medium-range forecast model. Mon. Wea. Rev, 124 , 23222339.

    • Search Google Scholar
    • Export Citation
  • Hsu, K. L., , X. Gao, , S. Sorooshian, , and H. V. Gupta, 1997: Precipitation estimates from remotely sensed information using artificial neural networks. J. Appl. Meteor, 36 , 11761190.

    • Search Google Scholar
    • Export Citation
  • Johnson, R. H., , T. M. Rickenbach, , S. A. Rutledge, , P. E. Ciesielski, , and W. H. Schubert, 1999: Trimodal characteristics of tropical convection. J. Climate, 12 , 23972418.

    • Search Google Scholar
    • Export Citation
  • Jolliffe, I. T., , and D. B. Stephenson, Eds.,. 2003: Forecast Verification: A Practitioner's Guide in Atmospheric Science. Wiley and Sons, 240 pp.

    • Search Google Scholar
    • Export Citation
  • Mitchell, D. L., , D. Ivanova, , R. Rabin, , T. J. Brown, , and K. Redmond, 2002: Gulf of California sea surface temperatures and the North American monsoon: Mechanistic implications from observations. J. Climate, 15 , 22612281.

    • Search Google Scholar
    • Export Citation
  • Mo, K. C., , and H. M. H. Juang, 2003: Influence of sea surface temperature anomalies in the Gulf of California on North American monsoon rainfall. J. Geophys. Res.,108, 4112, doi:10.1029/ 2002JD002403.

    • Search Google Scholar
    • Export Citation
  • Negri, A. J., , R. F. Adler, , R. A. Maddox, , K. W. Howard, , and P. R. Keehn, 1993: A regional rainfall climatology over Mexico and the southwest United States derived from passive microwave and geosynchronous infrared data. J. Climate, 6 , 21442161.

    • Search Google Scholar
    • Export Citation
  • Negri, A. J., , R. F. Adler, , E. J. Nelkin, , and G. J. Huffman, 1994: Regional rainfall climatologies derived from Special Sensor Microwave Imager (SSM/I) data. Bull. Amer. Meteor. Soc, 75 , 11651182.

    • Search Google Scholar
    • Export Citation
  • Reiter, E. R., , and M. Tang, 1984: Plateau effects on diurnal circulation patterns. Mon. Wea. Rev, 112 , 638651.

  • Sorooshian, S., , K. L. Hsu, , X. Gao, , H. V. Gupta, , B. Iman, , and D. Braithwaite, 2000: Evaluation of PERSIANN system satellite-based estimates of tropical rainfall. Bull. Amer. Meteor. Soc, 81 , 20352046.

    • Search Google Scholar
    • Export Citation
  • Sorooshian, S., , X. Gao, , K. Hsu, , R. A. Maddox, , Y. Hong, , H. V. Gupta, , and B. Iman, 2002: Diurnal variability of tropical rainfall retrieved from combined GOES and TRMM satellite information. J. Climate, 15 , 9831001.

    • Search Google Scholar
    • Export Citation
  • Stensrud, D. J., , R. L. Gall, , S. L. Mullen, , and K. W. Howard, 1995: Model climatology of the Mexican monsoon. J. Climate, 8 , 17751794.

    • Search Google Scholar
    • Export Citation
  • Wang, W., , and N. L. Seaman, 1997: A comparison study of convective parameterization schemes in a mesoscale model. Mon. Wea. Rev, 125 , 252278.

    • Search Google Scholar
    • Export Citation
  • Warner, T. T., , and H-M. Hsu, 2000: Nested-model simulation of moist convection: The impact of coarse-grid parameterized convection on fine-grid resolved convection. Mon. Wea. Rev, 128 , 22112231.

    • Search Google Scholar
    • Export Citation
  • Wilson, C. A., , and J. F. B. Mitchell, 1986: Diurnal variation and cloud in a general circulation model. Quart. J. Roy. Meteor. Soc, 112 , 347369.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Modeling domains. The numbers and letters identify locations that are discussed in text. The dashed line indicates the Continental Divide

  • View in gallery

    Monthly accumulated precipitation (mm) from different sources in Jun and Jul 2002. The gray dashed line indicates the leading edge of region of heavy monsoon rainfall

  • View in gallery

    (a) Modeled (left) Jun and (right) Jul mean upper-air sounding at (top) 1200 and (bottom) 0000 UTC from site 10. Wind vectors are shown with half bars equal to 2.5 m s−1. Dashed line indicates moist adiadatic trajectory for a parcel lifted from the surface. Cross-hatched area indicates amounts of CAPE for the surface parcel, and “F” indicates LFC. (b) A comparison between modeled and observed soundings at 0000 and 1200 UTC 7 Jul 2002. The wind vectors are shown with half barbs equal to 2.5 m s−1. The model location was at site 10 (refer to Fig. 10), while the observation was at Brownsville, TX (BRO; again refer to Fig. 1). Dashed line shows parcel trajectory lifted from boundary layer. Gray-shaded area indicates negative energy or CIN, “C” indicates LCL, and “F” indicates LFC

  • View in gallery

    (Continued)

  • View in gallery

    (a) Modeled 5-day-mean sounding over location A in Mexico at 0000 UTC Jun 2002. The wind vectors are shown with half barb equal to 2.5 m s−1. Location: 23.17°N, 106°W. Land use: wetland. Dashed line shows parcel trajectory lifted from boundary layer. Gray-shaded area indicates the observed layer of CIN, “C” indicates LCL, and “F” indicates LFC. (b) Same as (a) but for location in Mexico B at 0000 UTC Jun 2002. Location: 25°, 108°W. (c) Same as (a) but for location 3 (i.e., Tucson, AZ) in Jul. (d) Same as (c) but for observed sounding

  • View in gallery

    (Continued)

  • View in gallery

    (Continued)

  • View in gallery

    (Continued)

  • View in gallery

    Rainfall (mm day−1) evolution from different sources for 1 Jun 2002–30 Jul 2002, where A and B are sounding locations in Fig. 4. Data that were equal to 0 are plotted as white color. The panel of PERSIANN during 11 Jul–20 Jul was plotted with only 5 days' data.

  • View in gallery

    The 10-day-mean potential height at 500 mb. The contour interval is 20 m, and “X” indicates high center

  • View in gallery

    (a) Modeling diurnal variations of rainfall in Jun 2002. The time is MST. (b) Same as (a) except from PERSIANN data

  • View in gallery

    (Continued)

  • View in gallery

    Modeled mean rainfall diurnal variation over selected sites in Jul 2002. All the time in the figure is MST. The solid line represents PERSIANN data, the dashed line represents the modeling results, and the thin dashed line represents the gauge data

  • View in gallery

    (Continued)

  • View in gallery

    Modeled mean wind streamline in Jul at the σ = 0.9535 level. The bald arrows indicate main wind directions and their diurnal variations

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 12 12 2
PDF Downloads 1 1 0

Model Study of Evolution and Diurnal Variations of Rainfall in the North American Monsoon during June and July 2002

View More View Less
  • 1 Department of Civil and Environmental Engineering, University of California, Irvine, Irvine, California
  • | 2 Department of Atmospheric Science, The University of Arizona, Tucson, Arizona
  • | 3 Department of Civil and Environmental Engineering, University of California, Irvine, California
© Get Permissions
Full access

Abstract

Rainfall evolution and diurnal variation are important components in the North American monsoon system (NAMS). In this study these components are numerically studied using the fifth-generation Pennsylvania State University–National Center for Atmospheric Research (PSU–NCAR) Mesoscale Model (MM5) with high resolution (12-km grids) in contrast to most previous model studies that used relatively coarse spatial resolutions (>25 km grids). The model was initialized at the start of each month and allowed to run for 31 days.

The study shows that, in general, the model results broadly matched the patterns of satellite-retrieved rainfall data for monthly rainfall accumulation. The rainfall timing evolution in the monsoon core region predicted by the model generally matched the gauge observations. However, the differences among the three precipitation estimates (model, satellite, and gauge) are obvious, especially in July. The rainfall diurnal cycle pattern was reproduced in the monsoon core region of western Mexico, but there were differences in the diurnal intensity and timing between modeled and observed results. Furthermore, the model cannot capture the diurnal variation over Arizona.

Modeling results showed heavy monsoon rains shift northward along the western Mexico coast in association with the northward evolution of the subtropical highs. This is consistent with previous data analyses. The rainfall diurnal cycle was associated mainly with sea–land/mountain–valley circulations over western Mexico and adjacent oceans.

The simulations show that the model has deficiencies in predicting precipitation over the Gulf of Mexico. The model cannot reproduce the low-level inversion above the marine boundary layers and thus does not generate enough convective inhibition (CIN) to suppress the convection. The model also cannot produce realistic variations of day-to-day atmospheric conditions with only a single initialization at the start of the month.

Corresponding author address: Jialun Li, Department of Civil and Environmental Engineering, University of California, Irvine, E-4130 Engineering Gateway, Irvine, CA 92697-2175. Email: jialunl@uci.edu

Abstract

Rainfall evolution and diurnal variation are important components in the North American monsoon system (NAMS). In this study these components are numerically studied using the fifth-generation Pennsylvania State University–National Center for Atmospheric Research (PSU–NCAR) Mesoscale Model (MM5) with high resolution (12-km grids) in contrast to most previous model studies that used relatively coarse spatial resolutions (>25 km grids). The model was initialized at the start of each month and allowed to run for 31 days.

The study shows that, in general, the model results broadly matched the patterns of satellite-retrieved rainfall data for monthly rainfall accumulation. The rainfall timing evolution in the monsoon core region predicted by the model generally matched the gauge observations. However, the differences among the three precipitation estimates (model, satellite, and gauge) are obvious, especially in July. The rainfall diurnal cycle pattern was reproduced in the monsoon core region of western Mexico, but there were differences in the diurnal intensity and timing between modeled and observed results. Furthermore, the model cannot capture the diurnal variation over Arizona.

Modeling results showed heavy monsoon rains shift northward along the western Mexico coast in association with the northward evolution of the subtropical highs. This is consistent with previous data analyses. The rainfall diurnal cycle was associated mainly with sea–land/mountain–valley circulations over western Mexico and adjacent oceans.

The simulations show that the model has deficiencies in predicting precipitation over the Gulf of Mexico. The model cannot reproduce the low-level inversion above the marine boundary layers and thus does not generate enough convective inhibition (CIN) to suppress the convection. The model also cannot produce realistic variations of day-to-day atmospheric conditions with only a single initialization at the start of the month.

Corresponding author address: Jialun Li, Department of Civil and Environmental Engineering, University of California, Irvine, E-4130 Engineering Gateway, Irvine, CA 92697-2175. Email: jialunl@uci.edu

1. Introduction

The circulations of the North American monsoon system (NAMS) substantially affect the summer weather and climate of the southwest United States and Mexico. The NAMS is characterized by multiscale variations that include interannual, seasonal, and daily features in time, and continental, regional, and local patterns in space. The climatologic and synoptic features of NAMS have been studied systematically (see the reviews of Douglas et al. 1993; Adams and Comrie 1997; Higgins et al. 1997; Barlow et al. 1998). The mesoscale phenomena of NAMS, such as local circulations, low-level jet (LLJ), monsoon convection and its seasonal northward evolution, diurnal variations of wind, temperature, and rainfall, etc., are important features of the system that still need study and better understanding (Berbery 2001).

Previous studies have addressed many significant mesoscale features for the NAMS. Reiter and Tang (1984) analyzed multiyear sounding data and found strong diurnal variations of thermodynamic processes within the planetary boundary layer (PBL) of NAMS. Using field data from the Southwest Area Monsoon Project (SWAMP) in summer 1990, Douglas and Li (1996) and Douglas et al. (1998) documented a weak nighttime low-level jet (LLJ) from the northern Gulf of California to the southwestern Arizona desert that reaches maximum strength in the early morning and is considered important for low-level transport of water vapor northward in the NAMS (Mitchell et al. 2002) west of the Continental Divide.

The evolution and diurnal cycle of monsoon rainfall have been studied with rain gauge data (Douglas et al. 1993; Higgins et al. 1999) and satellite rainfall estimates (Negri et al. 1993, 1994). Douglas et al. (1993) and Higgins et al. (1999) found, as Cavazos et al. (2002) reviewed, that the mean evolution of the monsoon from Mexico to the United States is characterized by the regular northward progression of heavy precipitation from southern Mexico in early June. The NAMS spreads northward along the western slopes of the Sierra Madre Occidental (SMO) into portions of the southwestern United States by early July (see Fig. 2 in Higgins et al. 1999). The precipitation evolution is related mainly to the northward displacement of the high pressure centers over the eastern Pacific and western Atlantic Oceans (Carleton 1986). Additional factors, including a shift of the axis of the surface subtropical ridge, the development of a pronounced anticyclone at the jet stream level, and a thermally induced trough in the desert of the Southwest, also caused rainfall variations. All of these process variations are linked to seasonal adiabatic heating over the northern Mexico plateau and the southwest United States (Barlow et al. 1998). However, it needs to be verified whether a numerical model can reproduce this progression and associated phenomena.

Using 35 yr of hourly rainfall data in July and August at 45 locations, Balling and Brazel (1987) found maximum frequencies of nocturnal rains occur over low-elevation portions of central Arizona, with daytime storms in mountain areas. From satellite rainfall estimates based on the Special Sensor Microwave Imager (SSM/I) images, Negri et al. (1993, 1994) identified diurnal cycles of rainfall along the western coast of Mexico: convective storms occur offshore during the early-morning hours, with several local maxima around concave-shaped areas of the coastline. During the afternoon and evening period, deep convection reaches its highest development over land with marked maxima along the western slope of the SMO. Using rainfall estimates from combined Geostationary Operational Environmental Satellite (GOES) and Tropical Rainfall Measuring Mission (TRMM) data, Sorooshian et al. (2002) documented inverse diurnal rainfall patterns between the Isthmus and Gulf of Tehuantepec and between the southern Mexican coastal area and offshore in the eastern Pacific. Their results show morning offshore rainfall west of southern Mexico distorting the ITCZ into a broad north–south zone of active convection, then extending westward to about 120°W. It is only this far-eastern Pacific Ocean region of the ITCZ rainband that displays a distinct diurnal signal in the boreal summer season.

Although satellite remotely sensed rainfall data provides unprecedented, integrated patterns of global precipitation, many studies (e.g., Garreaud and Wallace 1997) have noted that deficiencies of satellite rainfall estimation—namely, indirect rainfall estimation and limited rainfall sampling in space and time—can affect the results and reduce their accuracy. Numerical studies of NAMS have the advantage of high time and space resolution. Wilson and Mitchell (1986) demonstrated that the simulation of climate in general circulation models (GCMs) will be degraded if the diurnal cycle is not resolved adequately. Dai et al. (1999) analyzed the diurnal patterns of precipitation simulated from the National Center for Atmospheric Research (NCAR) regional climate model (RegCM) using 60 km × 60 km horizontal grids and found substantial weaknesses in the three available cumulus convection schemes [Grell, Kuo, and Community Climate Model Version 3 (CCM3)]. Anderson et al. (2001) reproduced the LLJ from the northern Gulf of California to southwest Arizona using the National Centers for Environmental Prediction (NCEP) Regional Spectral Model (RSM) with 10 km by 20 km horizontal grids, and mainly focusing on the Gulf of California. Stensrud et al. (1995) reproduced the observed convective diurnal variations over the western slope of the SMO in fourth-generation Pennsylvania State University (PSU)–NCAR Mesoscale Model (MM4) simulations enhanced by the special observational data using 25 km by 25 km horizontal grids and initializing the model every 24 h. They found that the model overestimated convective frequencies over mountain areas, as well as morning rainfall, which was possibly related to the convective parameterization scheme they had used. Berbery (2001) analyzed 3 yr of forecast precipitation from the NCEP Eta Model (48-km horizontal grids) and found the diurnal rainfall variation over the SMO to be much weaker than the satellite estimates. He considered these differences reasonable because the satellite rainfall was estimated from the maximum instantaneous rainfall sampling in the afternoon (the highest rainfall period of a day) and the model forecast is integrated over time. The Eta-predicted rainfall was highly overestimated over and near southern Mexican coastal areas; therefore, it was difficult to diagnose the characteristics of rainfall in that region. In addition, Berbery found no rainfall diurnal cycle over central Arizona in the Eta-predicted rainfall (see his Fig. 4). Mo and Juang (2003), using NCEP RSM at 30-km horizontal grids, have studied the influences of SST in the Gulf of California on precipitation within the NAMS. In their 4-yr (1997–2000) monsoon simulations, they reproduced the diurnal variation of rainfall over the SMO, as well as a low-amplitude diurnal cycle over the Gulf of California.

Most previous model studies used relatively coarse spatial grid spacing (>25 km) that was not able to resolve well the mountainous topography in the NAMS region. This study employed the fifth-generation Penn State–NCAR Mesoscale Model (MM5) with higher resolution (i.e., grids at 12 km) in the primary study region. In previous research, wet monsoon years were usually selected as the case study (e.g., Anderson and Roads 2002; Gochis et al. 2002). The rainfall during NAMS 2002 was normal in general, but showed strong localized characteristics. (The reader can refer to the U.S. Climate at the Glance on the National Climatic Data Center Web site at http://lwf.ncdc.noaa.gov/oa/climate/research/cag3/cag3.html.) In June, the rainfall in Arizona and New Mexico was below normal. The rainfall in Texas and Mexico was near normal. In July, the rainfall in Arizona, New Mexico, and Mexico was almost normal, while rainfall in Texas was above normal for the entire summer, with an extreme rainstorm occurring in early July over southern Texas. The NAMS in 2002 was selected for study because it was more typical than the rainfall conditions in other studies.

The objectives of this study were to answer several questions: 1) To what extent can high-resolution MM5 simulations capture the evolution and diurnal patterns of NAMS rainfall for the summer of 2002? 2) Are the rainfall diurnal variations over central Arizona, eastern Mexico, and other areas basic features of NAMS resulting from different processes? 3) What are the major deficiencies of model predicting NAMS summer rainfall?

2. Numerical modeling

a. Study domain

Three nested domains were used in the simulations (Fig. 1). Domain 1 covers the whole United States, Mexico, southern Canada, Central America, the northern part of South America, and the surrounding oceans with a 108-km horizontal grid mesh. Domain 2 covers Mexico, western and central United States, and surrounding oceans, including the eastern tropical Pacific Ocean and the Gulf of Mexico, with a 36-km grid. Domain 3 is at 12-km resolution and covers the most active area of the NAMS region, eastern Mexico, Texas, and Oklahoma.

b. Model physics

MM5 provides multiple options and schemes to represent a variety of physical processes. The most important scheme to choose for summer rainfall simulations is the convective parameterization scheme (CPS). Many previous studies (such as Wang and Seaman 1997; Warner and Hsu 2000; Gochis et al. 2002; Guichard et al. 2003) suggested that, although performance may vary with rainfall types and model configurations, the Grell (1993) CPS produces reasonable rainfall patterns. In this study, the Grell CPS with the simple ice explicit moisture adjustment scheme (Dudhia 1989) was used. Additional model physics schemes selected for the study include the cloud radiation scheme (Dudhia 1989), the Medium-Range Forecast (MRF) boundary layer scheme (Hong and Pan 1996), and the NCEP–Oregon State University (OSU)–U.S. Air Force–National Weather Service Hydrologic Research Laboratory (NOAH) land surface model (Chen and Dudhia 2001). The vertical coordinate of the MM5 is a terrain-following coordinate system. In this study, 28 vertical sigma layers were employed from the surface to the top of atmosphere at 50 mb. From surface to top, the half layers are 0.9975, 0.9886, 0.9733, 0.9535, 0.9292, 0.8999, 0.8648, 0.8236, 0.7761, 0.7226, 0.6636, 0.6005, 0.5349, 0.4687, 0.4039, 0.3468, 0.2854, 0.2342, 0.1846, 0.1504, 0.1174, 0.0905, 0.0681, 0.0499, 0.0353, 0.0236, 0.0142, and 0.0051. There are 10 layers below 700 mb.

c. Model initialization and forcing

The NCEP reanalysis data for June and July 2002 was used for model initialization and boundary forcing. The reinitialization was conducted at 0000 UTC on the first day of June and July and the forcing along the boundary of domain 1 (D-1) was updated every 6 h.

In the simulation, the newly available, weekly 4.63-km Moderate Resolution Imaging Spectroradiometer (MODIS)/Terra SSTs were used to force the model water boundary instead of using the NCEP reanalysis skin temperature. The MODIS sensor was designed with higher sensitivity and better signal-to-noise ratio than the predecessor Advanced Very High Resolution Radiometers (AVHRRs) onboard National Oceanic and Atmospheric Administration (NOAA) satellites. The MODIS SST data is derived from two types of infrared brightness temperature data sensed from the Terra platforms: one is retrieved from the midinfrared 3.8–4.1-μm channels, and the other is from the thermal 11–12-μm channels, respectively. For more details, see the Web site http://modis-ocean.gsfc.nasa.gov. This replacement was based on three considerations: 1) the MODIS radiometer was designed to improve its predecessor sensor. Therefore, MODIS SSTs were expected to possess better quality. 2) MODIS SST data have high spatial resolution, which can enhance the SST forcing at the outer two computation grids. 3) preliminary studies (Gao et al. 2003) indicate that MM5 produced more realistic monsoon rainfall when forced by MODIS SSTs.

d. Rainfall observation data

To evaluate the model results, two independent rainfall datasets are used as references. One is the NCEP 0.25° grid rainfall data (available at http://www.cpc.ncep.noaa.gov/products/precip). The NCEP data is interpolated from daily rain gauge measurements (hereafter referred to as NCEP gauge data) and covers the continental United States and Mexico. Therefore, it has been used in many NAMS studies (Higgins and Shi 2001; Higgins et al. 1997, 1999). However, it should be noted that, because of the mountainous topography in Mexico and the southwest United States, the rain gauges in the region are sparse and heterogeneous (more gauges are located in accessible flat valleys than in the mountains), which could affect the accuracy of the rainfall data. Satellite-based rainfall estimates from the Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks (PERSIANN) system (Hsu et al. 1997; Sorooshian et al. 2000, 2002) were employed to compare rainfall over oceans and mountains and study rainfall diurnal cycles. The PERSIANN data, retrieved from combined geostationary infrared and TRMM microwave information, has a global coverage at hourly and 0.25° resolutions. As mentioned above, both of these two rainfall datasets possess some deficiencies; therefore, they were used in this study as independent references, rather than absolute ground-truth data to analyze the model performance in reproducing monsoon rainfall. In addition to these two rainfall sources, the station hourly gauge data at several selected locations were also employed to examine the model performance.

3. Results

a. Monthly rainfall

Figure 2 compares three independent monthly rainfall maps for June and July when the NAMS developed from onset to maturity. In June, the model showed the leading edge of heavy monsoon rain west of the Continental Divide (bold dashed lines) extending northward to approximately location B in Fig. 1. The model forecast nose of heavy rainfalls is between that of gauge data to the south and that of PERSIANN farther to the north. The PERSIANN estimates showed heavy rainfall (>180 mm month−1) extending north to about 30°N, along the SMO, and had the heaviest amounts. In the model and gauge maps, heavy rainfalls were constrained to the lower (southern) SMO. Apparently only light rainfall (50–120 mm) fell in the mid-SMO during June. The PERSIANN data also showed heavier rainfall over eastern New Mexico, Oklahoma, and most of Texas than did the model and gauge data. The model produced similar rainfall patterns over the eastern tropical Pacific Ocean and offshore sea surface outside the southern and western Mexican coasts, in comparison with the PERSIANN data, but overestimated rainfall over the Gulf of Mexico and underestimated over southern Mexico and the eastern Pacific relative to PERSIANN.

In July, all analyses show the monsoon rainfall had moved northward along the SMO to Arizona. In general, compared with gauge observations, the model agreed reasonably well with the monsoon rainfall core over the SMO and the southwestern United States. However, the model results have some deficiencies. First, in early July, abnormal storms occurred over southern Texas and resulted in severe flooding. This event was recorded in both gauge and PERSIANN rainfall maps, but the model did not reproduce this severe storm event at the right location. In the model, two severe storms were predicted about 2° west and also 2° east of the actual storm, respectively. Furthermore, in southern Mexico, the model underestimated the rainfall. Finally, the model underestimated rainfall, relative to PERSIANN, over the eastern Pacific Ocean and continued overestimating rainfall over the Gulf of Mexico. The latter seems a common problem that has been noted in other MM5 simulations (e.g., Gochis et al. 2002), and we have examined reasons for this recurring problem.

The PERSIANN tendency to overestimate over land and underestimate over water could be attributed to the following two reasons: 1) Garreaud and Wallace (1997) suggested that rain estimation with IR data tends to overestimate the areal extent and lifetime of the actual convective systems because of the persistence of cold, anvil cirrus clouds. Specifically, cloud tops over land at the SMO could be very high and cold even after heavy rains have ended. This could cause rainfall estimation systems based on the height of the cloud top (IR brightness temperature), like PERISANN, to overestimate rainfall. 2) On the other hand, IR satellite techniques might underestimate rainfall over water from low-top, relatively warm clouds. Previous studies have shown that as much as 28% of the rainfall over tropical regions is from relatively shallow and warm congestus clouds (Johnson et al. 1999).

The results for both months (especially July) indicate the model substantially overestimates convective rainfall over the Gulf of Mexico in relation to PERSIANN estimates. Figure 3a shows the model forecast monthly mean upper-air soundings for June and July at site 10 in the Gulf of Mexico (see Fig. 1). In Fig. 3a, convective available potential energy (CAPE), lifting condensation level (LCL), and level of free convection (LFC) are indicated. LCL and LFC in the figure are both located at about 940 mb, where the atmosphere was almost at the saturation condition. Above the LFC, the positive buoyant energy (i.e., CAPE) could reach over 1000 J kg−1 while there was no negative energy [i.e., convective inhibition (CIN)] below the LFC. This modeled atmosphere shows the potential for convective rain at any time of day because the modeled atmospheric state was unstable without CIN under LFC. Also, from the 3-h CAPE estimates (not shown), forecasted CAPE reached a maximum (>1100 J kg−1) from 0600 to 0900 UTC and reached a minimum (<800 J Kg−1) from 1800 to 2100 UTC. The buoyant energy was so high that the model could easily generate precipitation over the area through the Grell CPS. The Grell CPS would work to remove the convective energy once the model atmosphere is unstable in a column. (For more details, readers can refer to MM5 documents online at http://www.mmm.ucar.edu/mm5/mm5-home.html.)

Figure 3b shows an example of two model forecast soundings at site 10 and the simultaneous observed soundings at Brownsville, Texas (BRO, which is located just west of site 10), on 7 July 2002. The model-predicted soundings were similar to the monthly mean (Fig. 3-1), which was saturated at low layer and had positive buoyant energy (CAPE > 1100 J kg−1) above the LFC. The LFC was located at about 950-mb height with no CIN below the LFC in the lower layer. This atmospheric state generates rainfall easily within the model. However, the observed soundings indicate that there was substantial CIN and an inversion and dry layer above the marine boundary layer. In comparison with the modeling result, the observed soundings have these differences: 1) the LFC is higher, at about 700 mb; 2) CAPE was lower (around 600–730 J kg−1); and 3) there was CIN below the LFC. Under such conditions, convection has difficulty developing because of the hostile dry layer and CIN, and there was little observed rainfall in the real world. This is a serious deficiency in the model forecast thermodynamics that leads to a heavy bias in the rainfall simulations over the Gulf of Mexico. This deficiency could also influence the rainfall simulation in the nearby regions of southern and western Mexico and the Great Plains of the United States.

In southern Mexico (over site 9 in Fig. 1), the modeled atmosphere showed stable conditions with no CAPE to support convection (not shown). In addition, the mean precipitable water was less than 15 mm. These simulated environments caused the underestimation of rainfall in southern Mexico in July.

The model poorly simulated the day-to-day rainfall for this long-term integration. Table 1 gives the numbers of rain days at the specific sites (see Fig. 1) for July. The gauge rainfall data is from 0.25° × 0.25° interpolation data (NCEP gauge data). The PERSIANN data has the same resolution as NCEP gauge data. At site 1 (Flagstaff, Arizona), both PERSIANN and the model slightly overestimated the number of rain days in comparison with the NCEP gauge data. This was the same result for site 3 (Tucson, Arizona). At site 11 (Amarillo, Texas), the model performed well in comparison with the NCEP gauge data, but PERSIANN had relatively low rainfall days. In the monsoon core areas (sites 4 and 5), both PERSIANN and the model performed well on rainfall days. The model also had more rain days over nearby mountainous locations (sites 7 and 9) in comparison with NCEP gauge data, except at site 9. It seems that both PERSIANN and the model can get reasonable or higher rain days in comparison with NCEP gauge data in or near mountains, while over the low-elevation areas (sites 2, A, and B), results depend on the specific locations.

PERSIANN had relatively low rain days over water areas (sites 6, 8, and 10) in comparison with the NCEP gauge data, while the model overestimated the rain days over site 10. Again, it suggests that PERSIANN is underestimating rainfall over water areas.

To further evaluate the modeled daily rainfall performance, the day-to-day rain or no rain was examined at the selected stations over United States using the so-called signal detection theorem (Jolliffe and Stephenson 2003). The model performance is classified into four categories: 1) “hit”—the model-predicted rain day matches the gauge observation, 2) “false”—the model-predicted rain day has no rain in observation, 3) “miss”—the model-predicted no-rain day is rain day in observation, and 4) “correct rejection”—the model predicted no-rain day is no rain in observation. Table 2 shows the scores in the four selected stations in the United States in July 2002 comparing both the model and PERSIANN results with the gauge observation. In this table, the gauge data was from the point gauge stations, while PERSIANN data is the same as in Table 1. Therefore, the rain days in Table 2 may be different from those in Table 1 because the former were point observed while the later were interpolated in the area. The table indicates that the scores of both PERSIANN and the model “hit” gauge observations are relatively low in comparison with the total observed rain days (i.e., the sum scores of hit and missing). The numbers of “false” rain days were greater than or equal to the total observed rain days. If the frequencies of the “correct rejection” are also considered as correct prediction, the accuracy of both the PERSIANN and the model to predict the real rain and no-rain days is 50% to 60%. The results show that the model did not reproduce the real-world rainfall observations very well in day-to-day examination, although it did reasonably well at the monthly time scale. These deficiencies are probably related to both the long forecast period following the initialization on the first day of each month and to inaccurate forecasts of the atmospheric thermodynamic structure (refer back to Fig. 3).

b. Rainfall evolution

Figures 4a and 4b display 5-day-mean model forecast soundings at 0000 UTC over sites A and B (see Fig. 1), respectively, during June, that is, the period of monsoon onset. At site A, CAPE was always present in the 5-day means, although it varied from 600 to 1800 J kg−1 at different time periods. The LFC occurred around 700 mb with little variation. CIN (gray-shaded area in Fig. 4a) existed but was very small during some periods in June. This condition could cause the higher rainfall probabilities there. The model forecast 13 rain days in June at this location. This is reasonable because rainfall begins in early June at that location (Higgins et al. 1999). Figure 4b demonstrates the following information: There was no potential for convection at site B before 15 June. There was CAPE after 15 June but it was relatively small (<750 J kg−1) in comparison with that in Fig. 4b. This means the monsoon onset occurred at site B on or about 15 June. This is consistent with the climatological results. Also, at site B, LCL occurred at about 880 mb and LFC occurred at about 650 mb, with variation depending on the time period. Furthermore, there was always CIN (gray shaded in Fig. 4b) over the lower boundary layer. CIN at site B was larger than that at site A, and CAPE was less than that at site A. These thermodynamics differences led to less model rainfall over site B than that over site A (see Figs. 2 and 5).

Figure 4c is the modeled 5-day-mean sounding at the Tucson (site 3) region in July 2002. The results show that CAPE existed and there was potential for convection since the second time period. From the day-to-day rainfall analysis, model rainfall began on 5 July 2002. PERSIANN began recording rainfall on 9 July 2002, while gauge had rainfall recorded on 8 July 2002. Figure 4d is the same as Fig. 4c except for the observed soundings. There are two differences between Figs. 4c and 4d: 1) the model CAPE was relatively small (200–600 J kg−1), beginning on 5 July, and ending on 20 July (the model result shows no CAPE after 21 July) while observed CAPE was relatively large (>750 J kg−1), beginning on 11 July and continuing to the end of the month. 2) The wind directions were different. Above 500 mb, the modeled winds were dry westerly most of the time while the observed winds were southerly and/ or easterly after 11 July 2002. This modeled atmospheric condition resulted in no potential for convection after 21 July, while the gauge data recorded that there was rainfall on 22, 23, and 27 July, and PERSIANN had rainfall indicated for at least 8 days.

The evolution of NAMS precipitation is most visible through the time series of rainfall distribution. In Fig. 5 detailed patterns of rainfall evolution are presented for all 10-day intervals for June and July. The PERSIANN data are missing for 13, 14, 15, and 20 July. The figure shows that monsoon rainfall progresses northward from southern Mexico in early June, along the western slopes of the SMO, reaching into Arizona in early July. This characteristic is consistent with climatology based on long-term observations (Douglas et al. 1993; Higgins et al. 1999). Additional characteristics indicated by PERSIANN and the model include: 1) rainfall offshore southern Mexico expanded westward, and 2) rainfall over the eastern Pacific Ocean (west of 106°W) expanded northward. These variations are possibly related to the northward movement of the ITCZ in the summer and interactions between sea and land rainfall regions over the southern Mexico coast and ITCZ rainfall (Sorooshian et al. 2002).

Figure 5 indicates that the model underestimated rainfall over northwest Mexico, Arizona, and New Mexico after 21 July 2002. In Fig. 4c the model's predicted soundings for Tucson show the reason for this predicted, but not observed, break in the monsoon north of 28°N. The model predicted a dry and stable environment during this period due to an intrusion of deep westerly winds that did not occur.

This monsoon rainfall evolution was associated with the northward expansion of the subtropical high (Carleton 1986). Figure 6 presents the mean geopotential heights for 10 days for the same periods shown in Fig. 5. In early June, the high pressure center was located in Mexico (near site B) with a maximum height of about 5860 m. The high pressure center shifted to southeastern Arizona (close to site 3) during 21–30 June 2002 at about 5900 m. At this time, the whole southwestern United States and northwestern Mexico were dominated by the subtropical high (see the 5880 line in Fig. 6). The evolution of the synoptic pattern favors the northward movement of subtropical moisture into the southwestern United States and supports monsoon rainfall in this region, as was noted by Bryson and Lowery (1955).

c. Rainfall diurnal cycle

1) Diurnal variation in June

The model rainfall diurnal cycles averaged every 3 h through the days in June are illustrated in Fig. 7a. Figure 7b shows the corresponding rainfall diurnal cycle obtained from the PERSIANN rainfall estimates. In the figures, time is mountain standard time (MST; 7 h behind UTC). Although PERSIANN rainfall always seemed higher than the model rainfall, the rainfall distribution patterns show many similarities. The rainfall over all land areas had minimum coverage and intensity in the morning from 0500 to 1100 MST, and maximum coverage during late afternoon from 1400 to 2000 MST. Over the northern part of the Gulf of Mexico in domain 3, PERSIANN estimates had high rainfall rates from 0200 to 1400 MST while the model forecasted high rainfall rates from 0500 to 1700 MST. However, both PERSIANN and the model estimates had no clear diurnal variation over the southern part of the Gulf of Mexico in the study domain. The rainbands along the eastern and western Mexican coasts and the offshore rainfall over the neighboring sea surface areas possessed strong, and inverse, diurnal signals in both the model and the satellite data.

2) Diurnal variation at selected sites in July

In July, the rainfall diurnal variations over Mexico, Texas, and their surrounding oceanic areas had no substantial changes from the patterns of June. Therefore, only the diurnal variations at specific sites are discussed.

The rainfall diurnal cycle curves at selected sites (see Fig. 1) are plotted in Fig. 8. In the figure, all the times are MST including the time at site 11 and the hourly rain is equal to total hourly rain divided by the number of rainfall days (see Table 1). In general, the PERSIANN rainfall intensities were higher than model results over land (sites 1, 2, 3, 4, 5, 7, and 11), but lower than model prediction results over the water (sites 6 and 8). The diurnal cycle of the gauge data is more distinct than either PERSIANN estimates or the model results. Both the model and PERSIANN estimates can reproduce the tendencies of the rainfall diurnal variation, but there were substantial differences in intensity and time.

The model predictions over land (sites 4, 5, and 7) show 1–3 h earlier for beginning of rain and 1–3 h later to reach peak, in comparison to PERSIANN estimates. There are a few possible reasons for the time differences. At the beginning of convection, the height of the cloud top is low, and thus the PERSIANN system, in which rainfall intensity highly depends on cloud-top brightness temperature, diagnosed that there was less rainfall. On the other hand, in Grell CPS, CAPE is gradually consumed, causing convection to be slow to reach its peak, delaying the rainfall peak time. This may partly explain why model rain amounts are smaller than PERSIANN estimates.

However, over water (sites 6 and 8), rainfall was overestimated in the model in comparison to PERSIANN estimates, although the tendencies of model rainfall variation were nearly the same as PERSIANN estimates. In site 8, the rainfall occurs from early morning to noon. The PERSIANN estimates occur about 1–2 h earlier than the model results.

Sites 1, 2, and 3 in Fig. 8 were selected in Arizona (see Fig. 1) because hourly gauge data were available. The three sites represent three distinct regions of Arizona. Site 1 (Flagstaff, Arizona) represents high-elevation mountains; site 2 (Phoenix, Arizona) represents low-elevation desert; and site 3 (Tucson, Arizona) represents southeastern mountain and desert regions of Arizona that are active in the monsoon. There are distinct rainfall characteristics in these three regions. For example, Balling and Brazel (1987) reported that most storms in the vicinity of site 2 occur in late evening and nighttime (1900–2300 MST), while rainfall occurs mostly in the afternoon at the other two sites. At site 1, the model-predicted rainfall begins after 1000 MST and lasts to 1900 MST, with light rainfall intensity. PERSIANN rainfall begins after 1300 MST and lasts to 2300 MST, also with light intensity. The gauge rainfall, however, was very concentrated from 1400 to 1700 MST, with strong intensity. Therefore, neither the model nor PERSIANN captured the observed diurnal cycle well at this site. At site 2, the model only generated very light rainfall while PERSIANN and gauge measured a strong diurnal signal. The PERSIANN diurnal cycle peaks early but has a similar start time to that observed. The model did not perform well for this site. Over site 3, the model reproduces patterns of the two rainfall peaks from afternoon to evening. However, the amounts of model rainfall were less than those of PERSIANN and the gauge measurement. Furthermore, the rainfall beginning time was also different among the three sources. Neither PERSIANN nor the model captured the rainfall occurring in the early morning that the gauge measured.

The model did not forecast the amount of rainfall and timing very well. The main reason could be that the model ran for an entire month without reinitialization. Berbery (2001) suggests that because Arizona is in the periphery of the monsoon region it is more challenging for any numerical model to correctly predict summer precipitation. However, the data for site 11 show that neither PERSIANN nor the modeling results agree well with the observation.

d. Model forecasts of diurnal circulations

Figure 9 presents mean modeling results showing low-level (σ = 0.9535, about 500 m above ground level) wind streamlines every 3 h in July 2002. Although the figure only shows every 3 h, the nature of the streamlines were analyzed hourly. Even so, many local circulation patterns are clearly shown in the figure:

  1. During 1200–1800 MST, the southerly wind over the northern Gulf of California flows directly to southwestern Arizona. From 2100 to 0300 MST, the southerly winds become relatively weak. Meanwhile, there was westerly flow across the border of California and Arizona to western Arizona. From 0600 to 0900 MST, there was no obvious southerly wind directly from the northern Gulf of California, but a distinct LLJ from the west of the SMO along the coast flows into southern Arizona. These results are consistent with the data analyses (Douglas and Li 1996; Douglas et al. 1998) and previous model results (Anderson et al. 2001). Such diurnal southerly winds bring rich atmospheric moisture into the southwestern United States.
  2. Along the western coast of Mexico, sea breezes from the Gulf of California, and eastern Pacific continue from noontime (1200 MST) to midnight (0000 MST), bringing rich moisture to the western slope of the SOM (Stensrud et al. 1995) and results in heavy rainfall on the mountain slopes. However, no land breeze from the coast to the gulf occurs. Thus there is neither wind convergence nor rainfall over the Gulf of California during the nighttime. From night to early morning (0400–0900 MST), the dominant winds were parallel to the SMO from the central Gulf of California to southern Arizona, reaching a maximum at about 0600–0700 MST.
  3. Over the southern Mexican Pacific coast, sea breezes from the ocean blowing inland to southern Mexico are present during 1200–1800 MST, and an inverse land breeze existed during the late night and early morning (0000–0700 MST). This complete sea-breeze/land-breeze circulation enhanced the diurnal rainfall pattern. The wind forecasts for June indicated no such clear sea-breeze/land-breeze circulation (not shown). In July, the model underestimated afternoon convective rainfalls over southern Mexico (see Fig. 2). The model's rainfalls offshore from south Mexico were triggered by the land breeze during the night.

4. Conclusions

Using a high spatial resolution (12 km) mesoscale model, rainfall variations for June and July 2002 over southwestern North America were simulated. The results indicate the following:

  1. The model can generally reproduce rainfall pattern matches in the NAMS core region on monthly scales but cannot reproduce the day-to-day amounts in detail. This situation was severe especially over Texas where the model incorrectly forecast the location of the unusual heavy storm.The model rainfall amounts were larger than the gauge observations but less than the satellite-based PERSIANN estimates over the SMO and Oklahoma. Over water, the model-simulated rainfall agreed reasonably well with PERSIANN over the eastern tropical Pacific but overestimated the rainfall in the Gulf of Mexico in both June and July. The overforecasts were shown to be due to poorly forecast thermodynamic structures over the Gulf of Mexico.
  2. Even using high resolution, just one-time initialization to predict for the following 30 days cannot produce realistic evolution of day-to-day conditions within the model.
  3. The model simulated the rainfall northward evolution reasonably well in time and location in comparison with the gauge data and PERSIANN estimates from monsoon onset to maturity. The model also indicated the northward shift and expansions of the subtropical high from western Mexico into southeast Arizona.
  4. Over Arizona (sites 1, 2, and 3), the model cannot capture either the diurnal pattern, rainfall intensity, or timing very well. The model can reproduce rainfall diurnal patterns but not the accurate intensity and timing in comparison with PERSIANN and the gauge data over the other selected areas.

The mechanisms that resulted in the rainfall diurnal cycle vary according to locations. Along the coastline, such as the SMO, eastern Mexico, and southern Mexico, the sea breeze and mountain–valley circulations are responsible for the rainfall diurnal cycle. The coastal land- and sea-breeze circulations, the solar heating over sloping terrain, and diurnal changes in frictional drag of the PBL induce diurnal variations in low-level convergence that control the timing of convective precipitation during the summer.

While there are some deficiencies in this high-resolution simulation, the results indicate that for long time periods of week to month, the predictions capture the large-scale processes and precipitation associated with the NAMS. Prediction of high-time-resolution details probably requires much more frequent initializations and many other improvements.

Acknowledgments

Primary support for this research was provided under the NASA/EOS Interdisciplinary Research Program (NAG5-11044), the NOAA GAPP Program (NA16GP1605), and the NSF-STC Program (Agreement EAR-9876800). The first author also would like to thank NCAR/Mesouser, Drs. Jaime E. Combariza, Jimmy M. Ferng, and other staff at the Computer Center and Information Technology, University of Arizona, for their help and support. Thanks also to John Glueck of the Tucson NWS Forecast Office and Steven Vasiloff of NOAA/NSSL for providing the hourly gauge data. The comments and suggestions provided by two anonymous reviewers were most useful and helped the authors to improve this manuscript. A final thanks to Ms. Diane Hohnbaum for editorial comments on the manuscript.

REFERENCES

  • Adams, D. K., , and A. C. Comrie, 1997: The North American monsoon. Bull. Amer. Meteor. Soc, 78 , 21972213.

  • Anderson, B. T., , and J. O. Roads, 2002: Regional simulation of summertime precipitation over the southwestern United States. J. Climate, 15 , 33213342.

    • Search Google Scholar
    • Export Citation
  • Anderson, B. T., , J. O. Roads, , S-C. Chen, , and H-M. H. Juang, 2001: Model dynamics of summertime low-level jets over northwestern Mexico. J. Geophys. Res, 106 (D4) 34013413.

    • Search Google Scholar
    • Export Citation
  • Balling Jr., R. C., , and S. W. Brazel, 1987: Diurnal variation in Arizona monsoon precipitation frequencies. Mon. Wea. Rev, 115 , 342346.

    • Search Google Scholar
    • Export Citation
  • Barlow, M., , S. Nigam, , and E. H. Berbery, 1998: Evolution of the North American monsoon system. J. Climate, 11 , 22382257.

  • Berbery, E. H., 2001: Mesoscale moisture analysis of the North American monsoon. J. Climate, 14 , 121137.

  • Bryson, R. A., , and W. P. Lowery, 1955: Synoptic climatology of the Arizona summer precipitation singularity. Bull. Amer. Meteor. Soc, 36 , 329379.

    • Search Google Scholar
    • Export Citation
  • Carleton, A. M., 1986: Synoptic-dynamic character of “bursts” and “breaks” in the southwest U.S. summer precipitation singularity. J. Climatol, 6 , 605623.

    • Search Google Scholar
    • Export Citation
  • Cavazos, T., , A. C. Comrie, , and D. M. Liverman, 2002: Intraseasonal variability associated with wet monsoons in southeast Arizona. J. Climate, 15 , 24772490.

    • Search Google Scholar
    • Export Citation
  • Chen, F., , and J. Dudhia, 2001: Coupling an advanced land surface hydrology model with the Penn State–NCAR MM5 modeling system. Part I: Model implementation and sensitivity. Mon. Wea. Rev, 129 , 569585.

    • Search Google Scholar
    • Export Citation
  • Dai, A., , F. Giorgi, , and K. E. Trenberth, 1999: Observed and model-simulated diurnal cycles of precipitation over the contiguous United States. J. Geophys. Res, 104 (D6) 63776402.

    • Search Google Scholar
    • Export Citation
  • Douglas, M. W., , and S. Li, 1996: Diurnal variation of the lower-tropospheric flow over the Arizona low desert from SWAMP-1993 observations. Mon. Wea. Rev, 124 , 12111235.

    • Search Google Scholar
    • Export Citation
  • Douglas, M. W., , R. A. Maddox, , K. Howard, , and S. Reyes, 1993: The Mexican monsoon. J. Climate, 6 , 16651677.

  • Douglas, M. W., , A. Valdez-Manzanila, , and R. G. Cueta, 1998: Diurnal variation and horizontal extent of the low-level jet over the northern Gulf of California. Mon. Wea. Rev, 126 , 20172025.

    • Search Google Scholar
    • Export Citation
  • Dudhia, J., 1989: Numerical study of convection observed during the Winter Monsoon Experiment using a mesoscale two-dimensional model. J. Atmos. Sci, 46 , 30773107.

    • Search Google Scholar
    • Export Citation
  • Gao, X., , S. Sorooshian, , J. Li, , and J. Xu, 2003: SST data improve modeling of North American monsoon rainfall. Eos, Trans. Amer. Geophys. Union, 84 , P457,. P462.

    • Search Google Scholar
    • Export Citation
  • Garreaud, R. D., , and J. M. Wallace, 1997: The diurnal march of convective cloudiness over the Americas. Mon. Wea. Rev, 125 , 31573171.

    • Search Google Scholar
    • Export Citation
  • Gochis, D. J., , W. J. Shuttleworth, , and Z-L. Yang, 2002: Sensitivity of the modeled North American monsoon regional climate to convective parameterization. Mon. Wea. Rev, 130 , 12821298.

    • Search Google Scholar
    • Export Citation
  • Grell, G. A., 1993: Prognostic evaluation of assumptions used by cumulus parameterizations. Mon. Wea. Rev, 121 , 764787.

  • Guichard, F., , D. B. Parsons, , J. Dudhia, , and J. Bresch, 2003: Evaluating mesoscale model predictions of clouds and radiation with SGP ARM data over a seasonal timescale. Mon. Wea. Rev, 131 , 926944.

    • Search Google Scholar
    • Export Citation
  • Higgins, R. W., , and W. Shi, 2001: Intercomparison of the principal modes of interannual and intraseasonal variability of the North American monsoon system. J. Climate, 14 , 403417.

    • Search Google Scholar
    • Export Citation
  • 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. Climate, 10 , 26002622.

    • Search Google Scholar
    • Export Citation
  • Higgins, R. W., , Y. Chen, , and A. V. Douglas, 1999: Interannual variability of the North American warm season precipitation regions. J. Climate, 12 , 653680.

    • Search Google Scholar
    • Export Citation
  • Hong, S-Y., , and H-L. Pan, 1996: Nonlocal boundary layer vertical diffusion in a medium-range forecast model. Mon. Wea. Rev, 124 , 23222339.

    • Search Google Scholar
    • Export Citation
  • Hsu, K. L., , X. Gao, , S. Sorooshian, , and H. V. Gupta, 1997: Precipitation estimates from remotely sensed information using artificial neural networks. J. Appl. Meteor, 36 , 11761190.

    • Search Google Scholar
    • Export Citation
  • Johnson, R. H., , T. M. Rickenbach, , S. A. Rutledge, , P. E. Ciesielski, , and W. H. Schubert, 1999: Trimodal characteristics of tropical convection. J. Climate, 12 , 23972418.

    • Search Google Scholar
    • Export Citation
  • Jolliffe, I. T., , and D. B. Stephenson, Eds.,. 2003: Forecast Verification: A Practitioner's Guide in Atmospheric Science. Wiley and Sons, 240 pp.

    • Search Google Scholar
    • Export Citation
  • Mitchell, D. L., , D. Ivanova, , R. Rabin, , T. J. Brown, , and K. Redmond, 2002: Gulf of California sea surface temperatures and the North American monsoon: Mechanistic implications from observations. J. Climate, 15 , 22612281.

    • Search Google Scholar
    • Export Citation
  • Mo, K. C., , and H. M. H. Juang, 2003: Influence of sea surface temperature anomalies in the Gulf of California on North American monsoon rainfall. J. Geophys. Res.,108, 4112, doi:10.1029/ 2002JD002403.

    • Search Google Scholar
    • Export Citation
  • Negri, A. J., , R. F. Adler, , R. A. Maddox, , K. W. Howard, , and P. R. Keehn, 1993: A regional rainfall climatology over Mexico and the southwest United States derived from passive microwave and geosynchronous infrared data. J. Climate, 6 , 21442161.

    • Search Google Scholar
    • Export Citation
  • Negri, A. J., , R. F. Adler, , E. J. Nelkin, , and G. J. Huffman, 1994: Regional rainfall climatologies derived from Special Sensor Microwave Imager (SSM/I) data. Bull. Amer. Meteor. Soc, 75 , 11651182.

    • Search Google Scholar
    • Export Citation
  • Reiter, E. R., , and M. Tang, 1984: Plateau effects on diurnal circulation patterns. Mon. Wea. Rev, 112 , 638651.

  • Sorooshian, S., , K. L. Hsu, , X. Gao, , H. V. Gupta, , B. Iman, , and D. Braithwaite, 2000: Evaluation of PERSIANN system satellite-based estimates of tropical rainfall. Bull. Amer. Meteor. Soc, 81 , 20352046.

    • Search Google Scholar
    • Export Citation
  • Sorooshian, S., , X. Gao, , K. Hsu, , R. A. Maddox, , Y. Hong, , H. V. Gupta, , and B. Iman, 2002: Diurnal variability of tropical rainfall retrieved from combined GOES and TRMM satellite information. J. Climate, 15 , 9831001.

    • Search Google Scholar
    • Export Citation
  • Stensrud, D. J., , R. L. Gall, , S. L. Mullen, , and K. W. Howard, 1995: Model climatology of the Mexican monsoon. J. Climate, 8 , 17751794.

    • Search Google Scholar
    • Export Citation
  • Wang, W., , and N. L. Seaman, 1997: A comparison study of convective parameterization schemes in a mesoscale model. Mon. Wea. Rev, 125 , 252278.

    • Search Google Scholar
    • Export Citation
  • Warner, T. T., , and H-M. Hsu, 2000: Nested-model simulation of moist convection: The impact of coarse-grid parameterized convection on fine-grid resolved convection. Mon. Wea. Rev, 128 , 22112231.

    • Search Google Scholar
    • Export Citation
  • Wilson, C. A., , and J. F. B. Mitchell, 1986: Diurnal variation and cloud in a general circulation model. Quart. J. Roy. Meteor. Soc, 112 , 347369.

    • Search Google Scholar
    • Export Citation

Fig. 1.
Fig. 1.

Modeling domains. The numbers and letters identify locations that are discussed in text. The dashed line indicates the Continental Divide

Citation: Monthly Weather Review 132, 12; 10.1175/MWR2832.1

Fig. 2.
Fig. 2.

Monthly accumulated precipitation (mm) from different sources in Jun and Jul 2002. The gray dashed line indicates the leading edge of region of heavy monsoon rainfall

Citation: Monthly Weather Review 132, 12; 10.1175/MWR2832.1

Fig. 3.
Fig. 3.

(a) Modeled (left) Jun and (right) Jul mean upper-air sounding at (top) 1200 and (bottom) 0000 UTC from site 10. Wind vectors are shown with half bars equal to 2.5 m s−1. Dashed line indicates moist adiadatic trajectory for a parcel lifted from the surface. Cross-hatched area indicates amounts of CAPE for the surface parcel, and “F” indicates LFC. (b) A comparison between modeled and observed soundings at 0000 and 1200 UTC 7 Jul 2002. The wind vectors are shown with half barbs equal to 2.5 m s−1. The model location was at site 10 (refer to Fig. 10), while the observation was at Brownsville, TX (BRO; again refer to Fig. 1). Dashed line shows parcel trajectory lifted from boundary layer. Gray-shaded area indicates negative energy or CIN, “C” indicates LCL, and “F” indicates LFC

Citation: Monthly Weather Review 132, 12; 10.1175/MWR2832.1

Fig. 3.
Fig. 3.

(Continued)

Citation: Monthly Weather Review 132, 12; 10.1175/MWR2832.1

Fig. 4.
 Fig. 4.

(a) Modeled 5-day-mean sounding over location A in Mexico at 0000 UTC Jun 2002. The wind vectors are shown with half barb equal to 2.5 m s−1. Location: 23.17°N, 106°W. Land use: wetland. Dashed line shows parcel trajectory lifted from boundary layer. Gray-shaded area indicates the observed layer of CIN, “C” indicates LCL, and “F” indicates LFC. (b) Same as (a) but for location in Mexico B at 0000 UTC Jun 2002. Location: 25°, 108°W. (c) Same as (a) but for location 3 (i.e., Tucson, AZ) in Jul. (d) Same as (c) but for observed sounding

Citation: Monthly Weather Review 132, 12; 10.1175/MWR2832.1

Fig. 4.
Fig. 4.

(Continued)

Citation: Monthly Weather Review 132, 12; 10.1175/MWR2832.1

Fig. 4.
Fig. 4.

(Continued)

Citation: Monthly Weather Review 132, 12; 10.1175/MWR2832.1

Fig. 4.
Fig. 4.

(Continued)

Citation: Monthly Weather Review 132, 12; 10.1175/MWR2832.1

Fig. 5.
Fig. 5.

Rainfall (mm day−1) evolution from different sources for 1 Jun 2002–30 Jul 2002, where A and B are sounding locations in Fig. 4. Data that were equal to 0 are plotted as white color. The panel of PERSIANN during 11 Jul–20 Jul was plotted with only 5 days' data.

Citation: Monthly Weather Review 132, 12; 10.1175/MWR2832.1

Fig. 6.
Fig. 6.

The 10-day-mean potential height at 500 mb. The contour interval is 20 m, and “X” indicates high center

Citation: Monthly Weather Review 132, 12; 10.1175/MWR2832.1

Fig. 7.
Fig. 7.

(a) Modeling diurnal variations of rainfall in Jun 2002. The time is MST. (b) Same as (a) except from PERSIANN data

Citation: Monthly Weather Review 132, 12; 10.1175/MWR2832.1

Fig. 7.
Fig. 7.

(Continued)

Citation: Monthly Weather Review 132, 12; 10.1175/MWR2832.1

Fig. 8.
Fig. 8.

Modeled mean rainfall diurnal variation over selected sites in Jul 2002. All the time in the figure is MST. The solid line represents PERSIANN data, the dashed line represents the modeling results, and the thin dashed line represents the gauge data

Citation: Monthly Weather Review 132, 12; 10.1175/MWR2832.1

Fig. 8.
Fig. 8.

(Continued)

Citation: Monthly Weather Review 132, 12; 10.1175/MWR2832.1

Fig. 9.
Fig. 9.

Modeled mean wind streamline in Jul at the σ = 0.9535 level. The bald arrows indicate main wind directions and their diurnal variations

Citation: Monthly Weather Review 132, 12; 10.1175/MWR2832.1

Table 1.

Number of days with rainfall in Jul 2002. Rain day mean rainfall > 1.0 mm day−1 . Gauge miss date: 15. PERSIANN missed dates: 13, 14, 15, 20, and 31. The gauge data at sites 6 and 8 in the figure are interpolated from near land

Table 1.
Table 2.

Rain day statistics for Jul 2002

Table 2.
Save