Top-of-atmosphere (TOA) cosmic-ray (CR) fluxes from satellites and balloon-borne experiments are snapshots of the solar activity imprinted on the interstellar (IS) fluxes. Given a series of snapshots, the unknown IS flux shape and the level of modulation (for each snapshot) can be recovered. We wish (i) to provide the most accurate determination of the IS H and He fluxes from TOA data alone, (ii) to obtain the associated modulation levels (and uncertainties) while fully accounting for the correlations with the IS flux uncertainties, and (iii) to inspect whether the minimal force-field approximation is sufficient to explain all the data at hand. Using H and He TOA measurements, including the recent high-precision AMS, BESS-Polar, and PAMELA data, we performed a non-parametric fit of the IS fluxes $J^{\rm IS}_{\rm H,~He}$ and modulation level $\phi_i$ for each data-taking period. We relied on a Markov chain Monte Carlo (MCMC) engine to extract the probability density function and correlations (hence the credible intervals) of the sought parameters. Although H and He are the most abundant and best measured CR species, several datasets had to be excluded from the analysis because of inconsistencies with other measurements. From the subset of data passing our consistency cut, we provide ready-to-use best-fit and credible intervals for the H and He IS fluxes from MeV/n to PeV/n energy (with a relative precision in the range [2-10\%] at 1$\sigma$). Given the strong correlation between $J^{\rm IS}$ and $\phi_i$ parameters, the uncertainties on $J^{\rm IS}$ translate into $\Delta\phi\approx \pm 30$~MV (at 1$\sigma$) for all experiments. We also find that the presence of $^3$He in He data biases $\phi$ towards higher $\phi$ values by $\sim 30$~MV. The force-field approximation gives an excellent ($\chi^2/$dof$=1.02$) description of the recent high-precision TOA H and He fluxes.

The level of solar modulation at different times (related to the solar activity) is a central question of solar and galactic cosmic-ray physics. In the first paper of this series, we have established a correspondence between the uncertainties on ground-based detectors count rates and the parameter $\phi$ (modulation level in the force-field approximation) reconstructed from these count rates. In this second paper, we detail a procedure to obtain a reference $\phi$ time series from neutron monitor data. We show that we can have an unbiased and accurate $\phi$ reconstruction ($\Delta\phi/\phi\simeq 10\%$). We also discuss the potential of Bonner spheres spectrometers and muon detectors to provide $\phi$ time series. Two by-products of this calculation are updated $\phi$ values for the cosmic-ray database and a web interface to retrieve and plot $\phi$ from the 50's to today (\url{http://lpsc.in2p3.fr/crdb}).

Particles count rates at given Earth location and altitude result from the convolution of (i) the interstellar (IS) cosmic-ray fluxes outside the solar cavity, (ii) the time-dependent modulation of IS into Top-of-Atmosphere (TOA) fluxes, (iii) the rigidity cut-off (or geomagnetic transmission function) and grammage at the counter location, (iv) the atmosphere response to incoming TOA cosmic rays (shower development), and (v) the counter response to the various particles/energies in the shower. Count rates from neutron monitors or muon counters are therefore a proxy to solar activity. In this paper, we review all ingredients, discuss how their uncertainties impact count rate calculations, and how they translate into variation/uncertainties on the level of solar modulation $\varphi$ (in the simple Force-Field approximation). The main uncertainty for neutron monitors is related to the yield function. However, many other effects have a significant impact, at the 5-10\% level on $\varphi$ values. We find no clear ranking of the dominant effects, as some depend on the station position and/or the weather and/or the season. An abacus to translate any variation of count rates (for neutron and $\mu$ detectors) to a variation of the solar modulation $\varphi$ is provided.