A New Upper Air Database at NCAR

A new consolidated Upper Air Database (UADB), containing global observations beginning in the 1920s and continuing to the current date, has been created at NCAR. It has several product types and is a free and open access data resource.

Upper air soundings have been curated in the Research Data Archive (RDA) for more than four decades. Collectively, there are about 50 different sources totaling more than 85 million soundings. As separate entities these are inconvenient to use due to diverse data formats, different measurement units, occasionally limited area coverage, varying amounts of metadata and documentation, inconsistent quality assurance and control, and spatial and temporal overlaps that introduce duplicate or near-duplicate soundings between the data sources.

The UADB project has consolidated the RDA data sources into products that are accessible from a single web location, well documented, and provide the longest possible, duplicate-free, station time series. Historical sampling methods have been accounted for by creating two product lines, one with soundings that have wind (UADB-Wind) and another with temperature and humidity (UADB-TRH) measurements. This approach maximizes the amount of data consolidated, and when profiles have simultaneous temperature, humidity, and wind measurements, they are replicated in both products.

The first consolidation step was to “composite” stations with identical WMO station numbers across all data sources. The composited UADB-TRH product begins in 1943 with 5 stations, grows to a maximum in the 1980s with 1150 stations and then decreases to 850 stations in modern times. Nearly 1100 stations report for 20 years or more, and over 400 stations have time series of 50 years or longer. The UADB-Wind product starts in 1922 with 3 stations, increases to 1600 stations through the 1970s and then decreases to 1000 stations in the current years. About 1600 stations report for 20 years or more, and over 500 stations have soundings for at least 50 years.

The second consolidation step was to “combine” stations with different WMO station numbers, but that are located at identical latitudes and longitudes, or nearly so (within 20 km). This accounts for WMO station number changes and small position relocations. The result is that even longer station time series are created. Extensions occur at over 240 and 250 locations in the UADB-TRH and UADB-Wind products, respectively.

This presentation will discuss the source datasets, including the sounding provenance tracking, the metadata quality control and unit homogenization, the compositing and combining processes, and output data characteristics. The user data access for both the composited and combined UADB-TRH and UADB-Wind products will also be described.