Moist Convection in the North American Southwest and the North American Monsoon Experiment (NAME)
Thunderstorms and moist convection are fascinating and Arizona and Northwestern Mexico are great places in the summer to study them. The summer of 2006 produced the 4th wettest in Tucson since 1949 making Tucson look more like northern Sinaloa than Arizona. In one 2 hour period in late July in east Tucson, 5 inches of rainfall in a 2 hour period set records for water flow in the Rillito river.
The
North American Monsoon Experiment
The North American monsoon typically produces about half of the annual rainfall in Tucson each year and a larger percentage further south in northwestern mainland Mexico. To first order, it results from heating of the continental interior that reverses the annual average direction of the pressure gradients and therefore the winds. The North American Monsoon Experiment is an ongoing NOAA sponsored effort to understand warm season precipitation in North America, and improving our ability to predict it on short term to interannual time scales.
During the summer of 2004, the North American Monsoon Experiment (NAME) Enhanced Observing Period (EOP, see Higgins et al., 2006 for details) was run in northwestern Mexico and the US southwest to develop better understanding of the mechanisms influencing warm season precipitation and ultimately to improve its representation and prediction in models. Current numerical weather prediction and climate models predict many feature of warm season rainfall rather poorly due to a strong dependence on small-scale dynamical processes, topography and rapid diurnal evolution. Convective parameterizations, which are crucial in some for predicting precipitation ranging from a few hours to decades often have difficulty accounting for such small scale, fast acting processes. Also, the monsoon affected region of the southwest U.S. and, particularly, northwest Mexico has historically been poorly observed which presents challenges for model initialization, validation and refinement.
Since precipitation condenses from atmospheric water vapor, understanding the patterns and movement of water vapor in the pre-storm environment is critical to improving precipitation forecasts. IR and visible satellite atmospheric measurements are limited to frequent cloud tops during the North American Monsoon (NAM) area and therefore cannot determine properties of the air below the cloud tops. Satellite microwave observations of PWV can only be made over large bodies of water because of surface emissivity variations over land.
To capture the diurnal water vapor variations in the critical mountain areas where the convection takes place, we implemented Global Positioning System (GPS) receivers and surface meteorological instrumentation, at 6 locations in the Sierra Madre Occidental Mountains (SMO) in Sonora and Chihuahua to measure precipitable water vapor (PWV) during the EOP. These data complement other datasets collected during the 2004 EOP, particularly the rain gauge observations collected at similar locations (e.g. Gochis et al., 2004). The data we acquired during the 2004 NAME EOP is summarized in our report to NSF.
I summarized our early findings in a talk at the NAME 8th Scientific Working Group
meeting in Tucson August 17, 2006 which can be viewed as a pdf. Our initial
findings are discussed in Kursinski
et al. (2008).
Quantifying the sensitivity of convective precipitation forecasts to errors
in the initial precipitable water conditions
One of our research goals for the NAME EOP was to assess the diurnal cycle of PW in the Weather Research and Forecasting (WRF) model run at very high resolution (1.8 km) to resolve moist convection without parameterizations. As we began evaluating the WRF model forecasts using our NAME data, we discovered immediately that the inaccuracies in the WRF moisture forecasts were tied more to the initial conditions defined by NOAA ETA analyses rather than inadequacies of WRF itself. My student at the time, Walter Kolczynski, as his masterÕs research performed an initial sensitivity study to determine the sensitivity of WRF convection and precipitation forecasts to the accuracy of the PW estimates used to initialize WRF. In 2008 we refined this sensitivity study which we documented in our paper in the April 2008 issue of CLIVAR Exchanges.
Walter and Carlos Minjarez assessed the accuracy of the ETA PW analyses via comparisons with our GPS PWV measurements. WalterÕs results show that when the conditions are ripe for moist convection, errors of 5% in the initial PW field produce large and fundamental differences in the style of the moist convection in the WRF results. In comparison, we found that the 1-sigma errors in the 2003 ETA PW analyses were 7 -8% at Hermosillo and Puerto Penasco, Sonora, Mexico. The clear implication is that summertime severe weather prediction in the NAME region, particularly south of the border, is limited at present more by our knowledge of the water vapor distribution rather than our ability to model the convection. This situation could be improved upon dramatically by placing a small network of GPS receivers and associated surface meteorological stations in Northwestern Mexico (see below).
Deep convection and the water cycle in Brazil
I
am working with Rick Bennett (UA Geosciences), Seth Gutman (NOAA) and David
Adams who recently took a faculty position at the Universidade do Estado do
Amazonas in Manaus, Brazil on measuring water vapor in Brazil using GPS
receivers and surface meteorological instrumentation to better understand deep
convection and dynamics there. Dave presented initial GPS PWV results in August
2008 at the IRS conference in Brazil that are summarized in an extended
abstract. The Brazilians have indicated they are eager to work with us to develop
an expanded observing network in the Amazon basin.
Toward a permanent GPS and meteorological network in northwestern Mexico
A
goal of Rick Bennett (UA Geosciences) and mine is to establish a permanent GPS
receiver network in Northwestern Mexico to provide a multi-functional network
for atmospheric and solid earth research as well as surveying and to provide a
backbone for an internet network in Northwestern Mexico. Such a network could be assembled
relatively inexpensively either from existing older GPS receivers or new
internet ready receivers and low cost surface meteorological packages, some of
which are already in place in Mexico (Dave Gochis pers.comm.). The internet could be paid for by the
community where the internet would be placed as was accomplished in Mazatan
during the 2004 NAME EOP. A quick
summary of applications include
1) Atmospheric
research and operational forecasting
a) Weather forecasting which will use the
network to determine the upstream moisture before it flows into the semiarid
southwestern North America.
b) Hydrology which will use observations of
water in the gas phase to complement the rain gauge and radar network in the
NAM area.
c) Climatic monitoring which will use a long
term, precise and all-weather hydro-meteorological record of the important and
relatively remote and certainly poorly sampled NAM area.
d) Measuring the behavior and evolution of
summertime moist convection to provide critical constraints needed to
refine/develop convective parameterizations that will work in the NAM region
and other similar regions of the globe.
2) Solid earth
applications for a
high-rate GPS network in Mexico
a) High-precision
tectonics characterized by measuring plate boundary deformation in and around
the Gulf of California (a focus site for the NSF MARGINs program) and possible
diffuse deformation within the Mexican Basin and Range province.
b) Seismology
using surface waves (e.g., Larson et al., SCIENCE, 2003) and records of
near-field displacements captured by high-rate GPS receivers.
c) a
network complementing the US-based PBO facility by extending CGPS coverage into
northern Mexico, and other relatively smaller-scale CGPS networks in southern
Mexico.
3)
Surveying and mapping applications such as exist in
the Southwestern US in the greater Tucson and Phoenix metropolitan areas.
4)
an Internet accessible
phone/satellite network as required for data access which would provide a
series of internet hubs at relatively remote locations across Northwestern
Mexico, fulfilling the internet accessibility goal defined by the Mexican
government.
Future Data Assimilation ideas
I
am interested in data assimilation which combines model with observations to
produce a statistically optimal estimate of the state of the atmosphere (and
other geophysical systems). I hope
to create a quantitative, data assimilation characterization of moist
convection in the NAM area using our GPS and surface observations combined with
radar, rain gauge, lightning and satellite data and the high resolution WRF
model explicitly resolving moist convection to better understand warm season
precipitation and convection and ultimately their parameterization in climate
models. The intent is to work with
Jeff Anderson using the WRF Ensemble Filter data Assimilation tools in the Data
Assimilation Research Tool (DART) produced by the NCAR Data Assimilation
Initiative on this initiative.
References
Douglas
et al., 1993
Larson
et al., SCIENCE, 2003