This goal of this project is to investigate the long-term behavior of the mid-latitude ionosphere near Washington, D.C., using historical ionosonde measurements obtained from 1936 to 1957 in order to extend existing data by more than two decades prior to the 1957 International Geophysical Year (IGY), when ionospheric observation became widespread. The archival data is currently stored on 35 mm films, and the first goal of this effort is to digitize the records using a film scanning device to be developed for this project. The data will be converted into ionospheric state parameters (electron density profiles) using an automated procedure, the Expert System Ionosonde Reduction (ESIR), previously developed by the investigators. The new data set will be analyzed in conjunction with an existing database of hand-scaled ionogram values, obtained from an ionosonde located near Washington, D.C. (1957-1968) and Wallops Island (1967-2009), in order to assess long-term trends in the ionospheric parameters. A specific scientific focus of the trend analysis is the quantification of secular cooling of the upper atmosphere, an outstanding question in aeronomy. The new, extended dataset will be made available to the wider community.
As the debate on climate change continues and researchers strive to identify new sources of data, the process of quantifying and quality controlling these measurements becomes essential for their use by the broader communities. This pilot study sponsored by the National Science Foundation (NSF) reports on one type of observation that was made in the early 1950s. These observations were made by researchers using radio waves to probe the ionospheric region of the upper atmosphere. The observations were recorded on 35 mm film, partially analyzed, and archived at world data centers. These ionogram films represent the longest record of ionospheric and upper atmospheric measurements in the history of mankind. Climate change in the lower atmosphere in the form of temperature increases manifests itself as a cooling trend in the ionosphere. Hence, these ionogram archives hold information that when combined with today’s ionosonde observations could establish if, indeed, an ionospheric cooling trend exists. This one-year pilot study had two objectives. The first was to develop automated software that could convert ionogram images obtained by digitization of the 35 mm archived images into coordinate-registered ionograms, ready for analysis. The second objective was to analyze these recovered ionograms to produce the electron density profile (EDP) that represents the ionospheric profile above the ionosonde location. It was also necessary to develop an algorithm that provided uncertainty parameters and levels of confidence in the EDP. Both objectives have been met, verifying the feasibility of recovering ionospheric information from film archives that are over 60 years old. The task of coordinate registration of the scanned 35 mm film involves several steps, allowing ionogram information to be extracted in an automated, consistent fashion, as follows: A brief manual review of ionograms eliminates unusable frames and verifies the date/time stamp in usable frames. Software equalizes the film exposure and identifies frequency/virtual height marker lines in the frames; depending on the scan resolution, pixels can be located to within 0.1 MHz at mid-HF and to within 3 km. Ionogram pixels are stored as relative amplitudes with frequency and virtual height coordinates, along with metadata identifying the sounder location, date, and time. Additional quality control checks may be applied to certain frames, such as those with blanketing sporadic E. The above steps provide a complete description of the ionosonde operation and epoch. Figure 1 shows an example of a processed ionogram. The Fort Belvoir ionosonde would generate five ionograms per hour (at 0, 1, 15, 30, and 45 minutes.) Of the 120 ionograms expected per day, typically less than 5% were lost due to equipment problems, with the remainder successfully coordinate-registered. The second part of the project involved demonstrating that these ionograms could also be successfully converted into scientific information about the ionosphere. This information usually took the standard form of parameters for the frequency and height of the two main ionospheric layers, namely the E layer and the F layer, followed by the electron density profile that encompasses the E layer and extends up to the F-layer peaks. SEC had previously developed automated ionogram analysis software, the Expert System for Ionogram Reduction (ESIR). In this project, we developed interface software using ESIR to analyze the coordinate-registered ionograms. ESIR provides quality control in the form of error bars and confidence flags which are required by users who will utilize these types of products in real-time operations, based on data assimilation. These software and analysis products have been successfully adapted to the film ionograms. Figure 2 gives one example of an optimal EDP along with a spread, representing the probable analysis uncertainty. If this spread falls within specific thresholds, the overall analysis could be flagged as "very good", "good", "poor", or "unsatisfactory". The two steps in the pilot study were applied to two weeks of ionogram films from Ft. Belvoir, Washington, D. C. The first week was December 1951, a period of moderate solar activity. The electron density profiles for this period are shown in Figure 3 and those for the week of December 1954 (quiet sun) in Figure 4. These two weeks of data confirm the ability of the SEC automated software to fulfill the objectives of the pilot study. A comparison of these two weeks in 1951 and 1954 shows the qualitative differences in the solar cycle. They also provide detailed weather information that has never before been extracted from these ionogram observations. Previously, these films had only been analyzed at hourly intervals, and the original analysis was unable to generate EDPs due to the technological limitations of the time. Hence, this two-week analysis can be hailed as a success in data mining.
