We have calculated the global distribution of atmospheric neutrons and their products by a Monte Carlo simulation of nucleon transport, in the internuclear cascade followed by neutron transport below 19 Mev. First, we present the results generated by monoenergetic primary protons and alpha particles entering the top of the atmosphere. Second, the kernels derived from the monoenergetic cases are used to determine the spatial and energy distributions of neutrons and their products from the protons and alpha particles in the cosmic radiation; solar modulation effects are included. The calculation is compared, in the 1‐ to 10‐Mev region, with the results of our fast neutron experiment; the agreement is within the uncertainties of the primary spectrum and of the experimental results over most of the atmosphere. The calculation is then normalized to the experiment in the fast neutron region. The results of the normalized calculation include the steady state neutron spectrum, the neutron production rates, the radiocarbon production rates, the neutron leakage rates from the top of the atmosphere, and the production rates of other nuclides. The normalized calculation reproduces experimentally observed slow neutron densities and the observed neutron flux and spectrum above 1 kev, and it predicts features of the atmospheric neutron morphology not yet observed. The points of agreement and divergence with earlier calculations are discussed, including the radiocarbon production rates and the neutron leakage rates during solar cycle 20, which is near the mean of the last 10 solar cycles.

From an analysis of the time variations during 1968–1971 of the fast neutron flux in the upper atmosphere (mean energy of response to primaries, 1–2 GeV per nucleon) versus those of ground‐based neutron monitors we have identified two classes of transient intensity decrease on the basis of differences in their spectral responses, time histories, and flare associations. Type I events are found to be classic Forbush decreases, sharp declines accompanying a geomagnetic storm sudden commencement, following by 1–3 days a large optical flare with radio noise and energetic particle production, whereas type II events are more symmetric in their time histories and are therefore not associated with a particular flare. There are also differences in the spectral responses of the two types. During a type I decrease the flux change of the lower‐rigidity cosmic ray particles lags the flux change of the high‐rigidity particles both in the decline and in the recovery, tracing out a hysteresis loop. During a type II event, if there is any hysteresis at all, the lower‐rigidity primaries tend to ‘overrecover’ in comparison to the higher‐rigidity primaries. Intercomparison of neutron monitor data for median response rigidities from 10 to 30 GV reveals that the spectral response in type II events is softer on the average for low‐rigidity (< 10 GV) primaries and harder for high‐rigidity (> 10 GV) primaries than that in type I events. Comparison of intermixed sequences of type I and II events with recurrences of active regions reveals an identifiable but complex evolutionary relation between decrease occurrence and active region development. Long‐lived (of the order of days) low‐energy (<1 MeV) proton events occur during all but one of the type II events identified, supporting an association with solar active region transits. We interpret type II events as either a subsequent evolution of type I (Forbush decrease) events or a quasi‐stationary ‘corotating’ spatial structure loosely associated with an active region. Therefore both type I and type II decreases occur in intermixed recurrence series at intervals of 20–30 days.

Direct measurements of the thermal ionospheric ion distribution function made from two sounding rockets launched into the expansive phase of an auroral substorm are discussed. Ion flows perpendicular (convective) and parallel to the geomagnetic field lines are derived from these observations at altitudes from 200 to 840 km. Equivalent convective electric fields were observed to vary between a few tens of millivolts per meter to over 150 mV/m and are shown to be highly correlated with energetic electron precipitation. High‐velocity (up to 2 km/s) parallel ion flows were observed at various times and also found to be correlated with auroral electron precipitation. A correlation analysis of perpendicular (or equivalent electric fields E⊥) and parallel ion flow velocities with energetic electron intensities I at various energies is presented. The form of the E⊥ versus the logarithm of I relationship is shown to be approximately linear over most of the observed E⊥ range. Ionospheric ‘line‐tying’ models are presented and compared with the measured E⊥ versus I relationship. The model is found to be most consistent with observations when magnetospheric forces are assumed to be independent of the instantaneous values of E⊥ and I rather than viscouslike. No evidence was found to suggest that decoupling of magnetospheric and ionospheric convection through parallel electric fields produced the observed anticorrelation of E⊥ and I.

The time dependent worldwide distribution of atmospheric fast neutrons has been determined in balloon and aircraft measurements from 1964 to 1971. The 1‐ to 10‐Mev neutron spectrum was measured with a phoswich detector employing seven channels of pulse height analysis. Solar modulation effects were greatest near the high‐latitude transition maximum, where the flux varied by more than a factor of 2 from solar minimum to the deepest Forbush decrease. Near solar maximum and during Forbush decreases the relation between the neutron flux in the upper atmosphere and the counting rate of the Deep River neutron monitor deviated from a single‐valued function. The differential neutron spectrum between 1 and 10 Mev can be represented, within the resolution of the detector, by a power law N(E) = AE−n, where n = 1.17−0.20+0.12 near the transition maximum, n = 1.08−0.20+0.13 at 3‐ to 5‐g/cm² atmospheric depth, and n at sea level is larger than these values and dependent on terrain. The spectral index remains the same to ±0.1 over the solar cycle at fast neutron energies. The fast neutron data are self‐consistent to ±7% from 2 to 300 g/cm² over the range of cutoff rigidity and solar cycle variations. The characteristics of the fast neutrons as outlined here serve as a basis for checking a Monte Carlo calculation of the entire neutron distribution and its products.

During solar particle events from 1968 to 1971 we observed increases in the fast neutron flux at high latitude and at 55‐ to 75‐g/cm² atmospheric depth. The increases correlated with the variations in the solar proton fluxes; the neutron yield per incident proton, above threshold, increased by a factor of 100 with increasing hardness of the proton spectrum. Within a factor of 2 the neutron specific yield fell on a smooth curve versus the spectral parameter P0, where the values of P0 were based on the SPME (solar proton monitor experiment) data from Explorer 34 and 41. The neutron yield from solar particle events was calculated from a Monte Carlo simulation of neutron production and transport in the atmosphere. We compare the observed fast neutron flux with that calculated using the solar proton spectra reported at the times of the measurements; the causes for variation among the reported proton spectra and between the calculated and the observed fast neutron flux are discussed. The calculation reproduced the results of experiments by others with moderated slow neutron counters in and above the atmosphere. We calculate that the contribution of solar particle fluxes to the production rates of neutrons, to the production rates of radiocarbon, and to the leakage rates of neutrons from the top of the atmosphere are 2–3 orders of magnitude below the galactic cosmic ray contribution during solar cycle 20.