P1.20 A Comparison of GPS-Measured Precipitable Water at Bartlett, NH with Radiosonde Measurements in the Northeast
Donald M. Dumont and Joseph Zabransky, Jr. *
Plymouth State College, Plymouth, NH 03264
1. INTRODUCTION
GPS measurements of integrated precipitable water (IPW) are moving rapidly from the experimental to the operational mode. NOAA’s Forecast Systems Laboratory (FSL) is currently sampling some 62 sites and eventually expects to increase that number to 1000 with an average spacing of 100km between sites, Gutman and Holub (2000). Efforts are also being made to incorporate GPS data into operational forecasts models for use on AWIPS terminals (Gutman, 2000). Such rapid progress is dependent on a variety of studies that have been done comparing IPW measurements from GPS sensors with comparable data determined from radiosonde measurements as well as direct inter-comparisons among GPS sensors. e.g. Derks et al. (1997), Elgered et al. (1997), Emardson and Johannsson (1998), Emardson et al. (1998) and Tregoning et al. (1998). These studies involve relatively coincident measurements of IPW, in time and space and have yielded good agreement between observing systems, e.g. correlation coefficients as high as 0.95 and rms differences as little as 0.1-0.2cm. Their relatively inexpensive cost has also added greatly to the rapid deployment of GPS systems.
Still, it is recognized that measurements from a GPS network, at times, are likely to show significant IPW differences from radiosonde data because GPS measurements are more localized than radiosonde measurements. Nevertheless, this study attempts to compare IPW data at a specific GPS site (Bartlett, NH) with IPW values determined from a network of radiosonde stations in the Northeast. The novel aspect of this analysis is that none of the radiosonde sites are physically coincident with the GPS location.
In May of 2000, FSL installed a GPS monitoring site at Bartlett, NH in connection with Project Ground Winds, a cooperative program between The University of New Hampshire (UNH) and Mount Washington Observatory (MWO) to determine the feasibility of using a satellite mounted lidar system to measure atmospheric winds. Plymouth State (PSC) quickly became interested in the data set at Bartlett because it is close to Plymouth and because PSC will be participating in the SuomiNet Program and will have its own GPS monitoring system.
2. DATA AND METHODOLOGY
Calculated IPW values from radiosonde data for
* Corresponding author address: Joseph Zabransky., Natural Science Dept., Plymouth State College, Plymouth, NH 03264; e-mail: .
this study came from six sites in the New England and surrounding region. The sites included Gray, ME (GYX), Albany, NY (ALB), Caribou, ME (CAR), Chatham, MA (CHH), Brookhaven, NY (OKX) and Maniwaki, QB (WMW). Both 00Z and 12Z data were examined for inclusion in the study. IPW measurements from the Bartlett GPS site were matched to the radiosonde data by combining three 30-minute averages around standard radiosonde observation times. 00Z GPS data included averages of 2345, 0015 and 0045Z values and 12Z data were based on 1145, 1215 and 1245Z values. Data comparisons covered the period from May 24 to September 15, 2000.
The data presented do not include instances when Albany, NY or Gray, ME were missing data. There were ten such occurrences at 12Z and two occurrences at 00Z during the period of the study. In addition, there were four occasions when data were excluded because synoptic conditions were not uniform throughout the New England region or when a valid GPS measurement was significantly at odds with the regionally-computed IPW value. Examples of the latter were error differences of a factor of two or more which probably were related to local (mesoscale) conditions which seriously influenced the GPS reading. One such case is analyzed and presented as part of this study.
The analysis involved first comparing the IPW values at Bartlett to calculated IPW values from each of the surrounding radiosonde stations. However, to make the comparison more spacially consistent, an objective Barnes analysis was performed on the radiosonde data using the surrounding six stations. This analysis weighted and extrapolated the data at each radiosonde site to a grid point coincident with the geographical location of Bartlett; these adjusted values were considered to be the most appropriate for the comparison. The grid point spacing on the Barnes grid was approximately 120km.
3. RESULTS
GPS IPW values compare well with the IPW values derived from the objective analysis. The scatter diagram in Figure 1 is a good example of this for the 12Z period. The correlation coefficient for this time period was 0.91 and the rms differences averaged 0.39cm. Similar results occurred for the 00Z time period where the correlation coefficient was 0.92 and the rms differences averaged at 0.37cm.
A record of the day to day variation of IPW values for both GPS-measured and computed cases is shown in Figure 2. Because some data have been filtered out as earlier mentioned, the IPW trends show breaks; the agreement between the two data sets is unmistakable.
.
Figure 1. Comparison of computed IPW (via Barnes analysis) and GPS-measured IPW at Bartlett, NH for 12Z.
Figure 2. GPS and computed (via Barnes analysis) IPW values for Bartlett, NH at 00Z (May-September, 2000)
The preceding results indicate that GPS water vapor measurements can clearly be representative of regionally uniform atmospheric conditions, however they can also show definite exceptions in cases where synoptic conditions vary on a regional scale or when mesoscale phenomena locally influence the immediate region sampled by a GPS sensor. Such a case occurred at Bartlett on August 4, 2000 at 12Z. The IPW recorded at Bartlett at this time was 0.78cm while at the surrounding radiosonde stations IPW values ranged from 1.60 to 4.75cm. Excluding the high IPW value at Chatham, MA (4.75cm), caused by heavy precipitation associated with the cold front position, this range in IPW varies from 100 to 228 percent higher than the Bartlett value. A check was made to ensure GPS data validity at 12Z, Gutman (2000).
The post cold frontal conditions throughout much of New England and southeastern Canada are similar on the synoptic scale, so it is surprising that such a discrepancy should occur in the data. Some significant drying was occurring at Bartlett which is situated south-southeast of Mt. Washington. A look at the skew-T analyses from the surrounding raob sites shows a wedge of dry air between 700-800hpa. The skew-T from Albany, NY (Figure 3) is a good example of this.
Figure 3. Skew-T diagram at 12Z for Albany, NY on August 4, 2000
It is quite possible that some of this dry air may have descended locally to lower levels and been seen by the Bartlett GPS sensor. Figure 4, which shows he IPW trace for August 4, 2000, shows a rapid drying with a minimum spike followed by a recovery to regional IPW levels.
Figure 4. Variation in GPS IPW at Bartlett, NH from 3-5 August 2000. Minimum value occurs at 12Z on 4 August 2000.
4. CONCLUSIONS
With the exception of occasional cases where the GPS sensor is clearly looking at some relatively local phenomena above the site, GPS IPW values correspond very well to the regional measurements taken with radiosondes. More significant is that GPS measurements compare quite well with regional radiosonde observations that have been adjusted by objective analysis to the GPS measurement location. In this study for the Northeast, computed correlation coefficients were about five percent lower than corresponding values obtained in studies where the GPS antenna and radiosonde launch site were co-located. Likewise, rms differences between computed and GPS-measured IPW are similar in magnitude to numbers obtained from co-located sites.
It must also be recognized that GPS sensors see much more detail in the atmospheric moisture structure than do radiosondes and they will measure smaller-scale changes. Thus GPS measurements will contrast dramatically, at times, from regional radiosonde measurements. It is in this area that future GPS networks have the greatest potential to provide new insights to the moisture characteristics of the atmosphere.
5. REFERENCES
Derks, H. et al., 1997. GPS Water Vapour Meteorology Status Report, Koninklijk Nederlands Meteorologisch Instituut, pp.40.
Elgered, G., et al., 1997. Measuring Regional Atmospheric Water Vapor Using the Swedish Permanent GPS Network, Geophysical Research Letters, 24, 2663-2666.
Emardson, R. and M. Johannsson, 1998. Spatial Interpolation of Atmospheric Water Vapor Content Between Sites in a Ground-Based GPS Network, Geophysical Research Letters, 25, 3347-3350.
Emardson, R. et al., 1998. Three Months of Continuous Monitoring of Atmospheric Water Vapor With a Network of Global Positioning System Receivers, Journal of Geophysical Research, 103, pp. 1807-1820.
Gutman, S. and K. Holub, 2000. Gound-Based GPS Meteorology at FSL: A New Path Toward Better Water Vapor Measurements, FSL Forum, pp. 12-19.
Gutman, S., 2000. Personal Communication
Tregoning, P. et al., 1998. Accuracy of Absolute Precipitable Water Vapor Estimates From GPS Observations, Journal of Geophysical Research, 103, pp. 28701-28710.