Permanent GPS Receiver Network in Northwestern Mexico E. R. Kursinski R. Bennett, R. Maddox, W. Kolczynski,, C. Minjares (U. Arizona) I. Minjares (UNISON)
Outline 1. GPS for addressing the need for water vapor observations in the NAM Tier 1 area 2. GPS instrumentation in NAM area 3. Selected Highlights of GPS NAME 2004 4. Future GPS-NAM network applications and ideas
1. Achieving the NAME objectives: The need for water vapor observations in the NAM area
Our Focus: Water Vapor in and around the SMO Goal: 4-D observations of water vapor, condensed water, temperature, pressure and winds Want to trace water vapor and condensed water from source to location & time of precipitation Very difficult to do No observations are fully capable of this Not sure how good the models are Analyses include moisture increments when observations are assimilated which lose track of the moisture path through the atmosphere
GPS Observation Information A network of GPS & surface observations provides a subset of the desired 4-D information: Diurnally-resolved water column and surface water at array of locations ~2.5-D Diurnally resolved near T & winds (~2-D) Denser network (and more clever processing) can start to provide more information through synergy: ex. water vapor winds Also (relatively inexpensive) long term monitoring for interannual variability
GPS Observing Geometry Typically 8 to 12 satellites line of sights (LOS) in view Determine atmospheric delay along each LOS Map each LOS delay to zenith delay Average the zenith delays to produce one zenith delay
Atmospheric Transformations in GPS-Met
GPS PW Instrumentation GPS GPS GPS GPS Antenna Barometer Puerto Penasco Thermometer Hygrometer
GPS PWV Accuracy Comparisons of GPS and Tucson radiosonde PWV indicate ~2 mm (1-sigma) discrepancies
2. GPS instrumentation in NAM Area
Existing Geodetic Network in North American Southwest
GPS PWV processed by NOAA FSL and Suominet From UA Atmospheric Sciences GPS PWV web site
UA & UNISON GPS NAME Instruments Site IDs Location Lat (ºN) Long (ºE) Elev. (m) Duration SA21/SA46 Tucson, AZ 32.23-110.96 786.3 2002 - present SA24 Douglas, AZ 31.37-109.69 1263.7 2002 - present SA27 Hermosillo, Son 29.08-110.96 216.8 2003 - present SA31 Phoenix, AZ 33.45-111.95 384.1 2003 - present SA33 Puerto Peñasco, Son 31.30-113.53 10.7 2003 - present NAM1/YESX Yecora, Son 28.37-108.93 1544.0 2004 - present NAM2 Creel, Chi 27.74-107.63 2337.3 2004 NAM3 Tesopaco, Son 27.84-109.37 433.9 2004 NAM4 Mazatan, Son 29.00-110.14 549.4 2004 NAM5/USMX Moctezuma, Son 29.82-109.68 654.2 2004 - present SA48 Tohono Oodham 31.92-111.86 709.4 11/2005-present
NAME GPS PW & Surface Obs.. Array R. Kursinski, R. Bennett, W. Kolczynski, M. Leuthold, R. Maddox, C. Minjarez I. Minjarez, C. Rarellan UNISON S. Gutman, K. Hollub FSL NOAA T. Van Hove UCAR U. Arizona Funded by NSF & NOAA Jay Fein Combine existing GPS instrumentation plus 5 new sets in Northwestern Mexico to Understand relation between PW and precipitation in and around the SMO Assess and improve knowledge and model representations of moisture and precipitation Monitor intra- & inter-annual variations Improve forecasting initialization
NAME GPS PW & Surface Obs.. Array Puerto Penasco Douglas Moctezuma Hermosillo Mazatan Yecora Tesopaco Creel Placed 7 additional instrumentation sets at 5 locations in Mexico GPS receiver, barometer, thermometer, hygrometer Measured column water vapor & surface meteorological conditions In convectively active area in SMO In clear and cloudy conditions (5 to) 30 minute sampling Resolved diurnal water vapor, temperature and pressure cycle Internet for near real-time access & monitoring FSL & Suominet provided GPS processing at 30 minute resolution Through end of NAME Summer 04 campaign
North American Monsoon Experiment (NAME)
3. NAME GPS Results Highlights
Yecora & Moctezuma PWV 2004 Monsoon Onset (at least) 2 scales evident: 1. Large scale flow of moisture 2. Hour to day-scale (created by moist convection?) Rain at Moctezuma Monsoon Onset at Yecora Rain at Yecora Pre-monsoon event The monsoon onset at Yecora comes earlier, on July 4. Yecora PWV then roughly levels off whereas Moctezuma PWV continues to grow. Moctezuma PWV grows appreciably higher than Yecora s 3 days later on July 7. June 14 July 24
Mazatan-Hermosillo Comparison mid-july through mid-september 2004 Mazatan functioning DOY 200 (July 18) Mazatan is ~90 km east of Hermosillo at edge of SMO Elevation: Hermosillo 217 m, Mazatan 549 m Puerto Penasco Douglas Moctezuma Hermosillo Mazatan Yecora Tesopaco Creel
Hermosillo & Mazatan PW & q Overall PW Correlation: 0.55 Bias: 4.27 mm Cor: 0.88 Bias: 2.63 Cor: 0.34 Bias: 5.21 Monsoon Post-Monsoon Cor: 0.46 Bias: -2.58 mm Cor: 0.74 Bias: 0.05 mm Cor: 0.54 Bias: -1.23 mm August 17
Hermosillo-Mazatan Moisture Comparison PWV Hermosillo > PWV Mazatan BUT q Hermosillo < q Mazatan for mid-july to mid-august => Smaller average water vapor scale height over Mazatan Lower LCL over Mazatan (higher surface q,, lower surface T) Easier to form moist convection More daily precipitation at Mazatan Evaporation from wet surface maintains high near surface mixing ratios at Mazatan POSITIVE FEEDBACK during monsoon
Hermosillo-Mazatan Monsoon & Moisture Correlations Monsoon is characterized by relatively low correlations of PW and q mid-july to mid-august Higher correlation following mid-august several day dry spell Low correlation in PW and q presumably associated with frequent convective events over Mazatan (and Hermosillo) Higher correlation and smaller biases after mid-august indicates larger scale dynamical control of water vapor structure Local convective scale in SMO foothills less important after mid-august ( post-monsoon( post-monsoon ) 2 to 6 days of drying and circulation change apparently kills monsoon and diurnal cycle of moist convection near Mazatan
Monsoonal Diurnal Cycle of q (mid-july to mid-august) Larger diurnal cycle in q at Hermosillo suggests Diurnal evolution of mixed layer Deeper (?) over Hermosillo than Mazatan? entraining drier air aloft (than Mazatan) ) yielding minimum q in afternoon q at Mazatan is higher and generally varies less diurnally Shallower mixed layer? Air entrained from above BL is more moist? Larger moisture flux from surface is able to maintain moist surface air?
Hermosillo/Mazatan PW, T, T q & Hermosillo dt/dt & Precip Residual convective peak from earlier in the day? Westward propagating disturbances Downdrafts
Hermosillo/Mazatan PW, T, T q & Mazatan dt/dt & Precip Residual convective peak from earlier in the day? Westward propagating disturbances Downdrafts
Temperature and Rain: Hermosillo vs. Mazatan Mazatan Hermosillo
Convective Downdrafts and Precipitation C. Minjares thesis discussed signatures of downdrafts in Tucson Explained ~half of outliers in radiosonde-gps PWV comparisons Convective downdrafts are evident as Sharp decrease in surface temperature Often with near-coincident precipitation (if precip. meas.. available) with coincident well-defined peaks in PW» Possibly asymmetric: faster decrease than rise may or may not be a sharp change in surface q or pressure Sharp change in wind velocity would be useful diagnostic Propagating convective systems are evident in comparing Mazatan and Hermosillo downdrafts Consistent with satellite imagery Water isotope implication: lots of mixing=> messy & difficult to interpret
Convective Downdraft Differences: Hermosillo vs. Mazatan Downdraft temperature decrease is generally more sharply defined and obvious at Hermosillo Multiple events often evident at Hermosillo Overhead convection or more distant cold gust fronts? Downdraft change in surface q Similar q downdraft at Hermosillo & Mazatan Mazatan: : Typically seems to decrease (at least briefly) Hermosillo: : typically increases from relatively low values in late afternoon Guess: Rain is falling into lower RH air above Hermosillo More evaporation => colder, stronger downdrafts at Hermosillo
Convective Downdrafts with no Precipitation (Measurement) Two obvious explanations are Virga Precipitation was too localized and was missed by gauge Both should be observed by radar (if there is a radar) Hermosillo example: DOY 203 02:30 UTC sharp increase in PW and decrease in surface temperature but nearby gauge did not measure precip No satellite data available in NAME data set Radar data show 15-20 dbz echoes move across Hermosillo GPS can observe what the water in the column is doing after the intense precipitation Under the anvil which hides behavior at IR and shorter wavelengths
Timing of Precipitation Downdraft identification complements sampling of convective precipitation and virga events by rain gauges Cold outflow from downdrafts covers larger area than rainfall event itself individual downdraft sensor can pick up more precipitation events than a rain gauge May be particularly useful for determining timing of precipitation vs diurnal cycle, a critical variable for climate and climate model realism
Residual convective plumes: Keeping track of the water There are a number of PWV peaks without the sudden cold temperature signature of a convective downdraft Example: 204 19:30 UTC PW Event over Hermosillo w/o evidence of downdraft Strong convective event over Yecora earlier in the day at 204 05:30 UTC Not simple westward advection because no Mazatan PWV signature Based on GOES IR, the PWV moisture may be coming from the south Surface q signature is very different from PWV signature suggesting this is a free troposphere structure, not a BL moisture structure Much of the relatively sharp PWV structure may be mid-level moisture left over from residual or fossil convective structure Such moisture may play a role in subsequent convection Mid and upper troposphere moisture important for climate
Sensitivity of NAM Precipitation to Initial PW Conditions We planned to evaluate WRF using our NAME data. However, the first comparison of WRF with GPS measurements revealed a problem with the quality of the initial conditions that overwhelmed any sensitivity to the quality of WRF So we performed a two step study 1. We estimated the sensitivity of WRF moist convection, specifically dpw/dt,, to changes in the initial PW field 2. We evaluated the quality of the ETA PW analyses in Northwestern Mexico via comparisons with the GPS PW
Sensitivity of NAM Precipitation to Initial PW Conditions 5 panels show dpw/dt at 3:30 PM MST on July 29, 2004. The percentage of the ETA analyzed PWV field used to initialize the WRF model run for the five panels is 90%, 92.5%, 95%, 100% and 105%. A uniform scaling is applied to the moisture in all WRF grid cells.
Sensitivity of NAM precipitation to initial conditions 90% 92.5% 95% 100% 105%
Sensitivity of NAM Precipitation to Initial PW Conditions Substantial differences in the 5 panels indicate magnitude and timing of moist convection and precipitation is very sensitive to the moisture initially available. Under wetter initial conditions, Moist convection initiated & evolved earlier The resulting convective storms intensified and propagated further to the west by 3:30 PM local time. ~10 m/s propagation speeds may be tied to mid-level wind velocity needs further examination.
Sensitivity of NAM Precipitation to Initial PW Conditions Step 2: Assessed quality of ETA PW analyses vs GPS PW ETA 1-sigma PW errors in Sonora are ~7-8% Location/Station Year GPS PWV ETA PWV Bias(mm) Standard deviation of ETA PW (mm) Tucson/SA21 2002 2.36 2.9 2003 2.57 2.6 Douglas/SA24 2002 2.10 3.9 2003 1.35 3.0 2004 1.88 3.3 Phoenix/SA31 2003-0.61 2.6 2004 1.20 3.1 Hermosillo/SA27 2003 1.61 4.0 2004 2.31 3.8 P.Penasco/SA33 2003-0.03 3.7 WRF exhibited large changes in precipitation with 5% PW changes Errors in 2003 and 2004 ETA PW analyses are too large for accurate NAM precipitation forecasts in Northwestern Mexico
TBD on Data Set The data from this special NAME GPS data has not been put in the NAME archive Need to complete QC and refine error estimates Put data in standardized data format Reprocess to fill in day-long gaps Refine the estimate Moctezuma surface pressure during first 40 days Reprocess to 5 minute intervals to better capture convective events Analyze the results!
4. Future GPS NAM Network Solid Earth Applications UNISON interest Summary of Utility Instrumentation at each site Possible locations Atmospheric Capabilities/Functions
Solid Earth applications for GPS networks in Mexico High-precision tectonics characterized by a) measuring plate boundary deformation in and around the Gulf of California (a focus site for the NSF MARGINs program) b) possible diffuse deformation within the Mexican Basin and Range province. Seismology using surface waves (e.g., Larson et al., SCIENCE, 2003) and records of near-field displacements captured by high-rate GPS receivers. A network complementing the US-based Plate Boundary Observatory (PBO) facility by extending GPS coverage into northern Mexico and Other relatively smaller-scale GPS networks in southern Mexico.
Existing Geodetic Network in North American Southwest
Spreading of Floor of Gulf of California
UNISON Interest in Mexican GPS Network UNISON is interested in future work with GPS stations for developing the following subjects: 1. Studying moist convective events in Mazatan and SMO. 2. Developing an early warning system for extreme storms and flooding. 3. Providing a infrastructure complementing the rain gauge stations. 4. Measuring the tectonic motion of Baja, and possible active extensional tectonics in Sierra Madre and neighboring areas. 5. Measuring subsidence in regions with overpumped aquifers. UNISON has offered to contact Mexican federal agencies, such as Comisión Nacional del Agua (CNA) and Comisión Estatal del Agua (CEA), to gain support for logistics, internet connection and more.
Instrumentation & Infrastructure for each site GPS Receiver Barometer Thermometer Hygrometer Anemometer Rain gauge Solar power meter Camera Computer plus Power Science location Safe location Internet
Preliminary Hardware Cost per Site $0-6K GPS receiver + antenna (less if old receiver) $3K Met package (P,( T, T D ) $1K Anemometer $0.5K Rain gauge $2-5K Mounting hardware, solar power (if necessary) $1K Computer $1K Travel $8-15K TOTAL Plus salaries
How closely spaced should the sites be? First order answer is closer is better so likely comes down to $ Sample a range of elevations to constrain processes and parameterizations Sufficiently dense spacing to sample all of the large convective events Dense enough to constrain/determine water transport NAME spacing was NOT dense enough to determine where water was coming from Depends on horizontal scale of water blobs» During monsoon, decorrelation length may get down to the scale of the convection, ~few km Can estimate water vapor motion from 3 relatively close receivers:» test this at Tucson Can transport be achieved using line of site and higher time resolution processing? Perhaps
Potential Locations Place instruments at a subset of present rain gauge locations along Mexican roads through the SMO
GPS/Surface Obs.. Information Summary Unique data set measuring complete diurnal cycle Independent of clouds Provides key observations in and around SMO Understand topographic influence and interaction with warm season convection and precipitation Measures water vapor evolution and history Complements precipitation observations Crucial information for understanding and modeling transport and source/cause of precipitation Determines Intra- and Inter-annual Climate variability Provides basic objective climatology Drives more correct and resolved climate analyses
GPS/Surface Obs.. Information Summary (cont d) Constrains precursor conditions for moist convection and evolution of moisture field associated with convection Measures properties of convective downdrafts Proxy for precipitation Refine diurnal timing of convection and precipitation complementing rain gauge and radar observations Strong and unique set of observational constraints for evaluating and improving models» downdrafts can be critical in evolution of convective events
GPS/Surface Obs.. Information Summary (cont d) Monsoon onset and end Increase in PWV indicates monsoon onset Increase in moisture spatial correlations indicates end of Monsoon Key constraints to evaluate model performance and improve model parameterizations Relation between surface conditions, water vapor, moist convection and precipitation Diurnal cycle of column and surface water vapor, surface temperature, pressure (winds in the future) Determine important scales of variability Downdrafts NWP: Better model initialization, better forecast models Establish feasibility & utility of observational network in Mexico
Summary of Network Functions Provide critical data for constraining processes, evaluating models and improving parameterizations Establish interannual NAM climatology Provide key data for NWP initialization Improve severe weather and flooding forecasts Determine subsidence due to aquafer overuse Measure tectonic motion Measure seismic waves Provide reference stations for surveying Provide internet in more remote areas