NOTE: THIS IS AN EARLIER VERSION OF THE PROPOSAL THAT PROPOSED MORE OBSERVING SITES, AND HAD A LARGER BUDGET. THE BUDGET HAS BEEN REDUCED TO ROUGHLY HALF OF WHAT WAS PROPOSED HERE...

STRENGTHENING THE METEOROLOGICAL SOUNDING NETWORK

OVER THE EASTERN PACIFIC OCEAN AND THE INTERTROPICAL AMERICAS

TO SUPPORT PACS

PRINCIPAL INVESTIGATORS:

Michael W. Douglas

National Severe Storms Laboratory

Norman, OK 73069

Walter Fernandez

University of Costa Rica

San Jose, Costa Rica

Abstract

The present atmospheric sounding network over Central America and Northern South America does not reliably depict the structure of the Intertropical Convergence Zone (ITCZ) over the region. Over the eastern tropical Pacific no soundings are routinely made, except for recently renewed soundings in the Galapagos Islands. Thus, the variability of the ITCZ on all time scales, from daily to seasonal and interannual, can only be described over much of the PACS domain from a satellite perspective. The lack of detailed sounding information prevents many questions from being answered in a confident manner. Such questions include:

(1) Do major systematic errors exist in the routine NMC and ECMWF analyses over the eastern tropical Pacific? How closely does the actual mean structure of the ITCZ in the far eastern Pacific resemble that routinely analyzed by operational centers?

2. Does the current radiosonde network in central America provide an accurate depiction of the lower-tropospheric flow in this region?

(3) What large-scale meteorological conditions are associated with dry and wet spells over Central America during the May-October rainy season?

The primary objective of this proposal is to establish an enhanced atmospheric sounding network over the eastern tropical Pacific Ocean and bordering regions for a period of up to 18 months. At the end of this period we believe that enough data (both meteorological and operational experience) will have been collected to determine the desirability and feasibility of sustaining the network operation for the duration of PACS. The observations will include radiosonde and pilot balloon soundings, to be made in 4 countries. These data will be available to support many PACS investigators. During the first year of operation we will establish the feasibility of operating the stations for an indefinite period and determine which network configuration is most cost-effective for addressing PACS objectives. We intend to address the three questions posed above once data from one boreal warm season has been obtained by the network.

1. Introduction

The Intertropical Convergence Zone (ITCZ) over the eastern Pacific Ocean, as the region of convergence between the northern Hemisphere easterly trade wind flow and the southeast trades of the southern Hemisphere, is the locus of some of the highest precipitation values found anywhere on the globe. This relatively narrow zone of intense precipitation undergoes an annual cycle that is not well understood (PACS, 1994). Even the mean structure of this zone has not been well-described. There is also a major uncertainty in the amount of annual rainfall associated with the zone (PACS, 1994). However, even if the rainfall and its variability can be determined accurately, an understanding of the variability of preciptation requires a better knowledge of the atmosphere in and surrounding the ITCZ.

During the boreal summer, the ITCZ appears to lie over, or very close to, parts of Central America (Sadler, et al., 1987, Fig. 1). Variations in the position and intensity of cloudiness and rainfall have been used to explain some of the variability of the warm season rainfall over Central America and southern Mexico, though observations, especially meteorological soundings, have generally been lacking to support such assertions. In a similar manner, the ITCZ has been associated with the formation of tropical storms; periods of frequent tropical cyclogenesis being associated with an "active" ITCZ.

Any attempt at quantifying the ITCZ structure and its variability based on direct meteorological observations has been hindered by a lack of soundings (and even surface observations) to the south of the ITCZ. All regular radiosonde stations presently lie north of the ITCZ (Fig. 2); only the site in Panama may be near the axis during the July-October period. The tropical eastern Pacific Ocean is a region of major uncertainty in the meteorological fields above the surface, in large part due to the lack of inhabited islands from which to establish atmospheric , sounding stations. The deduction of the atmospheric circulation over the region comes from the blending of observations from surface ships, cloud tracked winds derived from satellite observations (mostly in the lower and upper troposphere) and infrequent reports from commercial aircraft. These observations, together with satellite soundings that provide only limited information on the tropical thickness field, are assimilated by global forecast models to produce what is considered to be the most reliable depictions of the tropospheric flow over the region. Unfortunately, it is well known that such depictions depend strongly upon the characteristics of the assimilating model in data voids (Lambert, 1988, Trenberth and Guillemot 1995), and all too frequently observations that deviate substantially from the model first guess are eliminated from the analysis cycle. Thus, widely separated, infrequent observations may not be assimilated correctly even when they are available. Since the observations over this region are from mobile platforms (ships, aircraft) or observations that otherwise vary irregularly in space and time (cloud-drift winds), there are no independent long-term observations that can be used to validate the accuracy of the mean fields derived from the assimilation-based analyses. Without such observations, the observed intraseasonal and interannual variations in the cloudiness (and derived precipitation fields) cannot be convincingly explained in terms of variations in the tropospheric flow and thermodynamic fields.

2. Objectives

Our work will involve two principle tasks. The first is to establish a prototype special meteorological sounding network in a section of the PACS domain that focuses on the Intertropical Convergence Zone in the far eastern tropical Pacific and surrounding land areas. The second objective is to use the special sounding observations to answer questions about the accuracy of the current observing network and operational analyses and aspects of the intraseasonal variability over the region. More specifically, we propose to use the data to address the following questions:

1. Do major systematic errors exist in the routine NMC and ECMWF analyses over this usually data-sparse region? How closely does the observed mean structure of the ITCZ resemble that routinely analyzed by operational centers?

2. Does the current radiosonde network in central America provide an accurate depiction of the lower-tropospheric flow in this region?

3. What large-scale meteorological conditions are associated with dry and wet spells over Central America during the May-October rainy season?

Other questions that PACS investigators can address with the special observations we propose to collect include:

4. What relationships exist between monthly mean meteorological conditions (winds, vorticity, divergence) measured by the enhanced sounding network and satellite-based estimates of precipitation and cloudiness over the same region?

5. What is the annual cycle of the depth of the boundary layer and the northward moisture flux in the eastern tropical Pacific? How do these observed fluxes compare with currently analyzed fluxes?

6. Are cross-equatorial wind surges are associated with the initiation of tropical storms in the northeastern Pacific?

7. What fluctuations in ITCZ circulation strength are associated with extended periods of tropical cyclogenesis in the eastern Pacific?

Finally, if the network proves to be feasible to operate indefinitely, a host of questions related to interannual variability of the flow over the eastern tropical Pacific and its relation to rainfall over South, Central and North America can be answered better.

3. Establishing the PACS special atmospheric sounding network

3.1 Establishing the sounding network

The proposed work will establish meteorological observing sites on the following islands:

Cocos Island 5.5_N, 87.0_W

Malpelo Island 3.9_N, 81.5_W

Observations are also proposed for the following site in South America:

Salinas, Ecuador 2.3_S, 81.2_W

The proposed observations at each site would be the following:

Cocos Island: Daily rawinsonde, 2 daily pilot balloon soundings, surface weather and rainfall measurements.

Malpelo Island: Twice-daily pilot balloon observations. Surface observations.

Salinas: Twice-daily pilot balloon observations. Surface observations.

The Inter-American Institute for Global Change Research (IAI) would facilitate some of the international exchanges required to establish the proposed enhancements to the regional meteorological network.

The radiosonde and pilot balloon stations on Cocos and Malpelo islands

The single most important component of the proposed observing system will be the deployment of atmospheric sounding systems to islands in the far eastern Pacific. These islands are currently sparsely inhabited, but informal correspondence with both Costa Rican and Colombian officials indicates a willingness to cooperate in the establishment of meteorological stations on the islands. We are proposing to establish pilot balloon sounding capability on both islands, with Cocos Island receiving an omega-based rawinsonde capability as well.

Both Cocos and Malpelo are very rugged islands, with steep slopes that make access difficult. Cocos is better suited for establishing a rawinsonde station, with anchorage and shore access in several protected coves. The topography (600m maximum elevation) prevents using conventional radiotheodolite systems (antenna cannot track unobstructed to low elevation angles), hence our plan to use an omega-based windfinding system, whose portability may allow it to be situated more easily on a high point for better reception.

Because both islands are expected to affect the windflow below 500 m altitude, we plan to carry out a series of measurements to measure the extent of this effect. By launching pilot balloons from the resupply ship sveral km away from the island (in a direction perpendicular to the low-level wind) and measuring the balloon motion with a theodolite on the island, we can calculate wind profiles relatively undisturbed by the island.

The pilot balloon network

Currently, there is moderately good radiosonde coverage over the region of the western Caribbean Sea and Central America (Fig. 2). Radiosonde stations at San Andres Island, Kingston, Grand Cayman, and Belize permit a reasonable depiction of conditions on the Caribbean side of Central America. This contrasts with the nearly complete lack of stations in the Pacific Ocean, a fact that makes the central American radiosonde stations critical to estimating conditions over the eastern Pacific, at least near Central America. Unfortunately, most of the current radiosonde stations in Central and northern South America are located close to population centers at higher elevations and surrounded by mountains and sample the lower-tropospheric flow over the eastern Pacific littoral very poorly. The most adversely affected stations are Bogota in Colombia, San Jose in Costa Rica, Guatemala City in Guatemala (currently not operating) and Tegucigalpa in Honduras. Only Panama City is relatively unaffected by nearby topography. Thus, only one radiosonde station on the Pacific coast between Acapulco, Mexico and Lima, Peru is unaffected by local topographic effects. All are affected by the strong diurnal sea- and land-breeze circulations near the surface.

The proposed pilot balloon site near Salinas, Ecuador, to be operated for an entire year, is located on a peninsula that provides a reasonable representation of conditions over the ocean - much more so than a previous radiosonde station at Guayaquil (no longer operating). The observations at both Salinas and in the Galapagos will allow some estimation of the zonal asymmetry in the windfield.

This part of Ecuador also undergoes a strong seasonal cycle with a short but pronounced rainy season; the tropospheric variations associated with this seasonal variation have been poorly described.

To attempt to obtain a better estimate of the windfield over Central America and the nearshore eastern Pacific for a special observation period of 4 months during the summer (June-September) of 1997 we are proposing to establish pilot balloon sites in locations that will minimize local topographic effects on the windfield (Fig. 3). These sites are relatively flat and near the shore, relatively distant from high mountains, and should better describe the low-level southwesterly flow that enters Central America during the northern summer. This enhanced southwesterly flow, associated with a northward movement of the ITCZ, is associated with heavy rain spells on the Pacific side of Central America. Streamlines of monthly mean surface wind for September and October (Fig. 4) show southwesterly flow into the Pacific coast of Panama and Costa Rica; one proposed temporary site will be on the Nicoya peninsula to better measure this flow and its diurnal variation. The relatively strong winds that flow through the low-terrain in Nicaragua will be monitored by the Managua pilot balloon site, and a coastal site near Tapachula, Mexico will better describe the low-level flow in this region than Guatemala City, which lies at 1.5 km above sea level.

During the operation of the special summer pilot balloon network we plan to make soundings 4 times daily at the Central American soundng sites for a period of 30 days to determine better the diurnal cycle. This would allow a more accurate assessment of whether once- or twice-daily soundings really represent the mean conditions at each of the sites.

There are several reasons for using pilot balloons to measure the lower to mid-tropospheric windfield instead of employing alternative observing systems. Cost is the primary consideration, in that little capital outlay is initially required to obtain the necessary reconditioned theodolites. In addition, the entire system is rugged, is independent of electrical power, and can be transported relatively easily. Little operator training is required, compared with some observing systems; this makes it possible to establish pilot balloon stations in almost any location that can be reached by road (helium/hydrogen gas transport is the main limiting factor). Finally, since the hardware and labor costs are relatively low, a substantial network can be established and maintained for a given budget, and meteorological uncertainties due to poor spatial sampling, usually a major source of uncertainty in a network in the tropics, can be reduced substantially (Douglas, 1991).

The main disadvantage of using pilot balloons, that of lost data due to clouds, will be reduced in this project by locating the stations in areas where low cloudiness is a minimum at the key synoptic hours, and indeed in a climatological sense. Most sites are relatively dry (Managua, Nicoya Peninsula, Salinas, are all relatively dry (Wernstedt 1972, WMO 1979) and others may have little cloud cover at certain hours.

The processing of the pilot balloon data can either be done at each site or centrally; the option chosen will depend on the training of the personnel and personal computers available at each site. In any case, reliable telecommunications is desirable, both to communicate data and to report problems as they arise.

3.2 Monitoring data quality

We propose to monitor the data from the special network at the University of Costa Rica to determine if adjustments are needed to the network, or if there are problems with observations at particular sites. Data, which would be received by various means, (radio, fax, satellite relay or via INTERNET) would be displayed and subjected to quality control procedures (buddy checks, comparison with previous observations and climatological values, comparison with satellite imagery (water vapor and looped IR and visible imagery to assess cloud drift wind accuracy) and with the observer reports that will accompanying each observation. The data dissemination will not be delayed by this step; the monitoring will proceed in parallel with dissemination and will be used to determine possible problems that may require immediate attention.

4. Diagnostic analyses

By December 1997, at the end of the first wet season over Central America, it is our intention to perform the following diagnostic studies with the special sounding network observations:

1. Calculate the monthly mean soundings at all stations in the PACS network and compare them with similar mean profiles from the regular WWW stations in Central America and northwestern South America. Such a comparison will tell us if there is a sampling problem in the region, that is, if the current network adequately describes the flow over the region, or whether there are important details of the tropospheric flow that are missed. We suspect that the easterly gaps winds through the low terrain of Nicaragua will be underestimated, and that a return upslope flow off the Pacific in the lee of the high mountains of Costa Rica and Guatemala will only be detected with the special sounding network. Such flow features, though local and regional in scale, are extremely important for explaining the mean rainfall climatology of the region.

2. If the special island observations do not generally arrive in time to be assimilated into the NMC operational global analyses (contrary to our objectives), we would compute the difference between the actual observations and the NMC analyses at each site. The mean difference, and its standard deviation will provide an indication of the uncertainty that exists in the routine NMC analyses over this region of the ITCZ. This region is quite close, in relative terms, to a region of substantial data (Central America) and uncertainties farther offshore are likely to be higher. None the less, we hypothesize that uncertainties in the mean fields will be substantial.

If the data reliably arrives in time for assimilation into the NMC analyses, it is still possible to infer the difference between the PACS observations and the NMC analyses without the data, by examining mean NMC analyses for previous years. A comparison between the observations and analyses over a 3 month period should be sufficient to indicate any major differences, should they exist. This assumes that interannual variability will be relatively small compared with the differences between the NMC analyses and the observations. We expect that easily detectable differences to be apparent only at Cocos and Malpelo Islands, since these stations are located well-away from current WWW sounding sites.

3. To determine what large-scale meteorological conditions are associated with dry and wet spells over Central America during the May-October rainy season, we plan to composite the PACS network (and WWW) soundings with respect to raingauge and satellite-inferred wet and dry periods determined for specific sub-regions in Central America. Initially, the Pacific side of Costa Rica and Nicaragua would be one region to focus attention on, since it undergoes a pronounced dry season, and rainfall variability in the wet season is of importance to agriculture in the region. There is also a reasonable raingauge network over the region to compare with the satellite-imagery. The composite analyses can be used to determine large-scale flow regimes that are favorable for heavy rainfall over the region of interest.

References

Lambert, S.J., 1988: A comparison of operational global analyses from the European Center for Medium Range Weather Forecasts (ECMWF) and the National Meteorological center (NMC). Tellus, 40A, 272-284.

PACS, 1994: Pan American Climate Studies: A Scientific Prospectus. NOAA Office of Global Programs, Silver Spring, MD, 27pp.

Douglas, M.W., 1991: Cost-Effective upper wind observing networks for developing countries. The SWAMP example. In preprint volume for Lower tropospheric profiling. Needs and technologies. Boulder, Colorado, September 10-13, 1991.

Sadler, J.C., M.A. Lander, A.M. Hori, and L.K. Oda, 1987: Tropical Marine Climatic Atlas: Vol 2. Pacific Ocean, Dept. of Meteorology, Atlas UHMET 87-02, University of Hawaii, 27pp.

Trenberth, K.E., and C.J.Guillemot, 1995: Evaluation of the global atmospheric moisture budget as seen from analyses. J. of Climate, 8, 2255-2272.

Wernstedt, F.L., 1972: World Climatic Data. Climatic Data Press.

WMO, 1979: Climatic Atlas of North and Central America, Vol I: Maps of mean air temperature and Precipitation. World Meteorological Organization.

Figures

Figure 1. Mean surface streamlines for July (from Sadler, et al., 1987).

Figure 2. Current radiosonde stations near Central America. The station in the Galapagos is intermittent. Many others make soundings once a day.

Figure 3. Proposed PACS sounding stations. R/P indicates radiosonde and pilot balloon site, T/P indicates tethersonde and pilot balloon site, and P indicates pilot balloon site only. Circle indicates Clipperton, another potential island available for radiosonde station establishment, not explicitly part of this proposal.

Figure 4. September and October surface streamlines (From Sadler et al., 1987). Note southwesterly flow entering Pacific coast of Central America from Nicaragua southward.