Artificial airglow It is possible to enhance the natural airglow by transmitting a high power high frequency (HF) radio beam that modifies the ionosphere when the ionospheric conditions are favourable (i.e. high enough electron concentration and at high latitudes little or no auroral activity). The enhanced airglow is far too weak to be visible by the unaided eye but it is detectable from the ground with sensitive imagers and photometers. This section gives a very short introduction to the phenomenon of artificially enhanced airglow. When a powerful HF-beam is transmitted into the ionosphere a number of wave-wave and wave-particle interactions might occur. Here these processes are only mentioned as an intermediate step in the energy transfer from the radio wave to the excitation of the atoms and molecules which emit the enhanced airglow. Figure 2.4: Location and geometry of the HF induced artificial airglow observations made with ALIS. The points marked with S, M, K, T, A, N represent the position of the ALIS stations in Silkimuotka, Merasjärvi, Kiruna, Tjautjas, Abisko, and Nikkaluokta respectively. The yellow point marked with E represents the location of the EISCAT site in Ramfjordmoen outside Tromsø. When the O-mode radio wave propagates into the ionosphere it is reflected at the altitude at which its frequency equals the local plasma frequency. In the region just below the reflection altitude the amplitude of the E-field of the radio wave swells and rotates from transverse to parallel to the magnetic field. This creates strong wave-wave and wave-particle interactions known as Langmuir turbulence. It has been shown that this creates supra-thermal electrons, theoretically by e.g. Weinstock and Bezzerides (1974); Gurevich et al. (1985); Perkins and Kaw (1971); Weinstock (1975) and Wang et al. (1997), and experimentally by e.g. Carlson et al. (1982) and Fejer and Sulzer (1987). This is the dominant dissipation mechanism at mid latitudes, such as for Arecibo, where a vertically transmitted pump wave propagates at about 40 to the local magnetic field. At high latitudes, such as for northern Scandinavia where the HF-wave propagates nearly parallel to the magnetic field, the Langmuir turbulence is present but in addition at altitudes a few kilometres lower than the HF pump reflection height most of the pump energy may be dissipated by the excitation of upper hybrid turbulence. The excitation of upper hybrid turbulence is particularly strong for pump frequencies not near an harmonic of the ionospheric electron gyro frequency and sufficiently long pumping (several seconds). These interactions create striations which are density and temperature pertubations that are elongated along the earth's magnetic field. These structures cause anomalous absorption of the HF pump wave ( Robinson, 1989; Gurevich et al., 1996). When the Langmuir and/or upper hybrid turbulence dissipate energy, e.g. by increasing the plasma temperatures or accelerating electrons, a small fraction leads to excitation of atomic oxygen to the $O(^1D)$ and $O(^1S)$ states thus causing enhancements in the 6300 and 5577 Å airglow. Several mechanisms have been proposed to explain the enhanced airglow. There are two main theoretical models for HF-pump enhanced airglow: one model suggests that the airglow is caused by $O(^1D)$ excitation of the high energy tail of a Maxwellian electron distribution ( Mantas and Carlson, 1996; Gurevich and Milikh, 1997; Mantas, 1994), and the other suggests that the $O(^1D)$ state is excited by accelerated electrons ( Weinstock and Bezzerides, 1974; Gurevich et al., 1985; Perkins and Kaw, 1971; Weinstock, 1975). The only theory which quantitatively relates the enhanced airglow to observable ionospheric parameters $(T_e, n_e)$ is the thermal theory given by Mantas (1994). This model suggests that the excitation of the $O(^1D)$ state is caused by excitation of the high energy tail $(E>1.96$eV$)$ of a thermal/Maxwellian electron distribution, where $$O(^3P_2)+e_{th} \rightarrow O(^1D) + e_{th}$$ (2.20) Under the assumption that the electron distribution is Maxwellian, it is possible to calculate a ``thermal excitation rate'' $\alpha$ by integrating the product between the electron distribution $f_e(E)$ and the integral $ø(^1D)$ excitation cross section $\sigma_{O(^1D)}(E)$ ( Mantas and Carlson, 1991): $$\alpha(T_e) = C\cdot\int_{1.96}^\infty f_e(E,T_e)\cdot\sigma_{O(^1D)}(E)E^{0.5}dE$$ (2.21) giving an electron impact excitation rate ( Mantas, 1994) of $$\alpha(T_e) = 0.502\cdot \sqrt T_e\frac{(9329+T_e)}{(51813+T_e)^3}\exp(-22756/T_e) ^3/$$ (2.22) where $T_e$ is the local electron temperature. This model has been capable of reproducing observed time variations of airglow and electron temperatures for some reported events ( Mantas and Carlson, 1996; Mantas, 1994). However, the model requires unrealistically high electron temperatures $(T_e \gtrsim 20000$   K$)$ to reproduce the observed 5577 Å to 6300 Å ratios ( Bernhardt et al., 1989; Haslett and Megill, 1974). Furthermore, recent (1998) E-region observations of 5577 Å emissions and emissions from $N_2(1P)$ band at 660 nm that require electron flux at energies above 9 eV ( Djuth et al., 1999) appear to be inconsistent with excitation from the tail of a thermal electron distribution. Paper IV and Paper V reports the time variation of the three-dimensional distribution of the enhanced 6300 Å airglow produced by the EISCAT-Heating facility ( Rietveld et al., 1993) near Tromsø, Norway on 16th February 1999. It is found that there are discrepancies between the predictions by the thermal model and the observations. From the EISCAT UHF radar measurements of electron temperature and electron concentration, the theoretically predicted airglow has a peak volume emission rate 15 to 20 km below the observed altitude of maximum emission. Further, the airglow enhancements are predicted to be between a factor of 2 and 3 above the natural background airglow intensity compared with the observed enhancement of 50 %. If the electron temperature measured by EISCAT is reduced so that the thermal model produces the observed enhancement of 50 %, the time variation of the airglow does not match -- the modeled airglow has a lag time from the start of the HF-pulse before the airglow intensifies. This lag time is absent in the observations. A further shortcoming of the thermal model is that it assumes that the electron distribution remains Maxwellian irrespective of the large energy-dependent cross section for excitation of the neutral atoms and molecules. For each inelastic collision one electron loses a significant amount of energy; for excitation of the $O(^1D)$ state the energy loss is 1.96 eV, and for excitation of $N_2$ vibrational states the energy loss is 0.2888 eV per vibrational level. This has been reported to produce a significant depletion of the electron distribution function in the energy range 2 - 3 eV: Stubbe (1981) calculated the ``modifying effects of a strong electromagnetic wave upon a weakly ionized plasma...'' taking both the wave-electron and electron-neutral interactions into account with parameters typical for the ionospheric D region; for auroral precipitation several authors have calculated significant bite-outs in the electron flux at altitudes up to 150 km (e.g. Rees, 1989; Konovalov et al., 1995). Bernhardt et al. (1989) calculated the modification effect on the electron distribution function of excitation of the vibrational states in $N_2$ to be so significant that it effectively increased the excitation threshold of the $O(^1D)$ state to 3.1 eV. In paper VI an attempt has been made to estimate the magnitude of the electron distribution function modification due to the electron neutral interaction. The results indicate that at altitudes below 220 - 230 km the electron distribution has a significant bite-out above 2 eV. This bite-out decreases the $O(^1D)$ excitation at altitudes lower than 220 km, and the total $O(^1D)$ excitation for typical HF enhanced electron temperature profiles would be decreased by at least 30 %. Further it is shown that transport effects cannot be neglected. Altogether this indicates that a complete model for HF enhanced airglow has to take electron-neutral interaction into account as well as transport effects. This level of description requires that the HF wave-plasma interaction models must produce quantitative disturbed electron distributions.