TheEregion at 69°N, 19°E: Trends, significances, and detectability

Abstract

[1] Earlier, it has been demonstrated that ionospheric parameters such as critical frequencies and layer heights exhibit long-term trends. These trends have been reported to be both positive and negative depending on parameter and geographic location, forming a picture that is difficult to explain in terms of ionospheric physics combined with the widely accepted climatic cooling of the underlying middle atmosphere. Here we present updated results from Tromsø (69°N, 19°E) confirming the previously identified negative trend in E region critical frequency (and therefore electron density) and altitude (commensurable with a shrinking of the underlying middle atmosphere). We note that the E region data sets are not necessarily best described by Gaussian distributions. We quantify uncertainties and required time series lengths for trend detectability on the basis of the true probability density functions, giving high credibility to climate change in the lower thermosphere.

Temperature trends at 90 km over Svalbard, Norway (78°N 16°E), seen in one decade of meteor radar observations

“…We may be allowed, therefore, considerable sureness in our trend determination. In fact, Hall et al [2011] found that the true probability density function of the analogous stochastic component for E region critical frequency implied shorter n * than if a Gaussian was assumed.…”

“… Weatherhead et al [2002] have subsequently built upon work by Weatherhead et al [1998] and Tiao et al [1990] and define a minimum time series length n * for which a trend can be considered detectable: where σ N is the month‐to‐month variability in the data as we determined earlier, ω 0 is the expected trend, φ is the autocorrelation in the month‐to‐month data (i.e., the autocorrelation at lag 1) Tiao et al [1990]. and Weatherhead et al [2002] use j = 3.3 for 90% probability that a trend ω 0 is detectable after n * ( ω 0 and n * having the same time units) assuming a Gaussian distribution (and also a data set with noise characterized as an autoregressive – AR(1) ‐ process) Hall et al [2011],. working with ionospheric E region height and critical frequency time series critically investigated the very distributions, finding the stochastic component of the critical frequency data set (for example) to be better characterized as fractional Gaussian noise.…”

Solar activity impact on the Earth’s upper atmosphere

“…After Qian et al (2008). (Hall et al 2011). Recent trends in the mid-and low latitude foE and foF1 are a slight increase, and in foF2 in day-time a very small decrease, probably due to changes in minor constituent chemistry and various temperature-dependent reactions and loss rates.…”

Complexity signatures in the geomagnetic <i>H</i> component recorded by the Tromsø magnetometer (70° N, 19° E) over the last quarter of a century

“…On the other hand, there has been relatively little focus on observables related to the terrestrial ionosphere and measured locally in order to examine the mapping of solar forcing to the Earth's surface. Hall et al (2011) examined complexity in the ionospheric E region in the auroral zone: the altitude, strength and persistence of the E region are of particular interest for radio communications and for studying the possible overall shrinking of the middle atmosphere due to climatic cooling (Roble and Dickinson, 1989;Rishbeth and Clilverd, 1999). In this particular study, the geomagnetic field characteristics represented by a local time series measured, on average, beneath the auroral oval at 70 • N, 19 • E (geographic) will be examined.…”

“…In order to eliminate these deterministic features from the data set to be studied, a boxcar filter is applied to remove fluctuations less than 24 h; the smoothed time series is then subtracted from the original to arrive at a set of residuals representing the stochastic component. A corresponding method was employed by Hall (2014) and discussed and tested by Hall et al (2011). As will be shown, an advantage of spectral analysis is that individual periodicities remaining in the (supposedly) stochastic residual show up as narrow spikes and, in practice, have an insignificant influence on the determination of the slope of the spectrum.…”