2. IN SITU OBSERVING SYSTEM
2.1 The Criticality of In situ Observations
Finding A:
Both the long record of radiosonde observations, which extends back to at least 1958, and the high vertical resolution of radiosonde soundings provide unique information not available from remote sensors. To gain a better understanding of the magnitude and mechanisms of climate variability and change, it is necessary to have the high vertical resolution information from radiosondes on a global scale. In situ measurements are necessary globally to characterize the significant local climate variations. The density of the observations is currently insufficient to fully understand the nature of the atmospheric vertical structure over large ocean regions, as well as several largely unpopulated land masses (IPCC, 1996; NRC, 2000a). Not only are there spatial gaps in the current radiosonde network, but the network is also in decline (NRC, 1999, 2000a).
Finding B:
The potential for using satellite data to create useful CDRs is most convincingly demonstrated with independent, in situ observations. Utilization of both satellite and in situ datasets is providing a better understanding of the earth system than either dataset in isolation (NRC, 2000a). Satellite atmospheric datasets provide temperature information globally, but only for relatively coarse vertical layers, whereas radiosondes provide high vertical-resolution climate data, but only for limited regions. The European Centre for Medium Range Forecasting's (ECMWF) experience with ERA-15 suggests that there are not enough radiosondes in the tropics at present to calibrate the operational infrared satellite record, to perform bias corrections, etc. (Stendel et al., 2000; Trenberth et al., 2000).
Recommendation:
In situ measurements, including radiosonde observations, must be sustained. Their acquisition should be based on a strategy designed to provide a reasonably dense spatial and temporal distribution of observations. Radiosonde launches should be standardized and expanded in coverage, especially in the tropics. When instrument packages change, simultaneous launches of the old and new instruments, at regionally representative sites in the field, should be performed. These simultaneous launches should take place under a range of conditions, over at least one annual cycle, and at both day and night. The objective should be to determine instrument biases to a sufficient level to allow adjustment of the data for continued long-term climate monitoring.
2.2 Stability of Radiosonde Network
Finding:
Prior to 1988, the U.S. radiosonde network was the only major radiosonde network to have implemented no major instrument changes over several decades. But since that time, the long-term stability of U.S. radiosonde observations has been compromised by changes in the types of radiosondes employed (Angell, 1999; Christy et al., 2000). Current procurement practices are not designed to ensure a stable network and tend to encourage frequent instrument changes 3. These instrumentation changes typically introduce changes in the error characteristics of the measurements that are not fully understood and that reduce the temporal homogeneity of the overall dataset. Radiosonde data from other countries show significant effects of instrument changes, which are very difficult to appropriately adjust and which have a large impact on resulting temperature trend estimates (Gaffen et al., 2000).
3 |
A notable exception is the consistency through time of the instrumentation employed in the network of 63 stations used in the studies of Dr. J. Angell (e.g., Angell and Korshover, 1975; Angell, 2000). |
Recommendation:
NOAA should attempt to minimize the number and frequency of changes in instruments and observing methods in its radiosonde and other in situ systems. Although future technologies may offer improved operational observation capabilities, a major factor in evaluating instrument changes should be the continuity of the climate record. The continuity of existing long and/or particularly high-quality or geographically-critical data records should be promoted by NOAA. Other nations should be encouraged by NOAA (through the World Meteorological Organization) to follow those aspects of the U.S. observational procedures outlined in this report that help ensure high quality radiosonde-based CDRs.
2.3 Adaptive Strategies
Finding:
Construction of reliable CDRs requires consistent observing practices. Adaptive strategies (wherein the timing and location of radiosonde launches vary according to synoptic weather conditions) may introduce seasonally- and geographically-varying biases and render the data unsuitable for climate because the continuity of the essential climate monitoring network may be disrupted. These biases increase uncertainties in temperature trend estimates (Elliott and Ross, 2000).
Recommendation:
Consistent twice-daily radiosonde releases should be maintained at a reasonably dense network of stations, nominally several dozen stations in the United States. Research is needed to formulate more quantitative recommendations regarding the necessary spatial density of a climate monitoring network. Adaptive procedures should not compromise the integrity of such a basic network. The United States should encourage other nations to adopt similar policies.
2.4 In Situ Stratospheric Observations
Finding:
The stratosphere is a sensitive region of the atmosphere, in which natural and human-induced climate changes appear to be more readily detected than in the troposphere (Chanin and Ramaswamy, 1999; Ramaswamy et al., 2000). Stratospheric temperature climate observations have been derived from a disparate set of observational platforms, many of which are in decline. Rocketsonde observations have essentially ceased, and a limited network of lidar sites has taken their place (Keckhut et al., 1999). Radiosondes, which are capable of providing data in the middle stratosphere, generally do not reach this region because of balloon failure. Currently, about 60 percent of radiosondings in the U.S. network reach the 10 hPa level and only 3 percent reach 5 hPa (These statistics were provided by G. Mandt in his presentation to the RTO sub-panel on June 8, 2000). While these statistics represent an improvement in the past few years, they are not reflective of performance at tropical locations, where lower stratospheric temperatures cause balloons to burst at lower altitudes. In the global radiosonde network, observations often do not achieve the 30 hPa level (Gaffen et al., 2000).
Recommendation:
NOAA should place a higher priority on ensuring that radiosonde observations (both in the U.S. network and at foreign stations) routinely reach at least the 5 hPa level through, for example, the use of more resilient balloon materials. The accuracy of the data at all levels, including the stratosphere, should be well characterized, including any adjustments to the data to correct for radiation and lag errors. These improvements should be carried out for at least a designated subset of stations, if not for the entire network.
2.5 Radiosonde Metadata
Finding:
The production and interpretation of CDRs from radiosondes depend on station history metadata that are as complete and accurate as possible. Currently available station histories are several years out of date (Gaffen, 1996). Particularly in the United States, important changes in instrumentation, data processing algorithms, and observing practices have occurred in the 1990s (Angell, 1999; Gaffen, 1993) and are planned for the coming years (NWS, 1999).
Recommendation:
NESDIS/NCDC should work with the World Meteorological Organization to update and enhance existing radiosonde metadata information. This effort should not be limited to a selected network (e.g., the Global Climate Observing System Baseline Upper-Air Network or the core network of the Comprehensive Aerological Research Dataset) but should address the entire global network. Continuing data archaeology and documentation of resulting datasets should be supported.
2.6 Radiosonde Data Products for Monitoring Global Temperatures
Finding:
Neither NOAA nor other U.S. scientists are currently producing global and regional pressure-level or layer-mean temperature time series based on radiosonde data in a timely manner for monitoring climate variations and trends and for comparison with satellite-derived products. The most complete database suitable for such an effort is the Comprehensive Aerological Research Dataset of daily soundings, maintained by NESDIS/NCDC, yet no temperature monitoring data products are being disseminated.
Recommendation:
NOAA should take responsibility for developing and updating global, radiosonde-based CDRs and for making them available to both the research community and the public. These data should include the raw soundings as well as adjusted CDR products. The development of prompt web availability of these data, as well as general web access to the metadata, should be explored.
2.7 Next Generation In Situ Upper-Air Data
Finding:
The National Weather Service is in the process of upgrading the U.S. radiosonde network by replacing outdated computers at the ground stations and implementing radiosonde tracking using the Global Positioning System (GPS) (NWS, 1999). However, the basic temperature and humidity sensors will not be upgraded and are unsatisfactory for many applications. Because of the importance of detecting changes in the hydrologic cycle, and because of the key role of water vapor as a greenhouse gas, detecting changes in atmospheric moisture, particularly in the free troposphere and stratosphere, is important for understanding climate change. Radiosonde humidity observations are poor in the upper troposphere and useless in the stratosphere (Elliott and Gaffen, 1991), and no network makes routine in situ observations of humidity to form a CDR at these higher altitudes. The aforementioned implementation of GPS tracking may also introduce heterogeneities into the long-term upper-air wind record (WMO, 2000), which could affect estimates of, for example, moisture convergence.
Recommendation:
NOAA and other agencies should explore options for a significantly improved atmospheric sounding system, which would afford higher quality humidity observations in the global troposphere and stratosphere. Because changes in instrumentation and data processing methodology in the past one to two decades have significantly diminished the utility of the radiosonde record for climate purposes (see