SolarEnergeticParticle

DatasetInfo

Structure containing metadata about a loaded dataset.

Fields

• energy_channels::Dict{String, Vector{Float64}}: Energy channel information
• units::Dict{String, String}: Units for each data column
• instrument::String: Instrument name
• spacecraft::String: Spacecraft name
• data_level::String: Data processing level (e.g., "L2")
• time_resolution::String: Temporal resolution of the data

cusum_onset_detection(I, μ, σ; h = nothing, k = nothing, n = 2, N = nothing)
cusum_onset_detection(I, times, μ, σ; kw...)

Detect onset time using Poisson-CUSUM (Cumulative Sum) algorithm, where $I$ is the (intensity) time series, and $μ$ and $σ$ are the mean and standard deviation of the background, k is the control parameter.

The CUSUM function $S$ for each timestamp for the standardized intensity $I_z = \frac{I-μ}{σ}$ was then calculated as follows:

\[S[i] = \max(0, I_z[i] - k + S[i-1])\]

where $S[0] = 0$. When $S$ exceeded the hastiness threshold $h$, a warning signal was given. After N consecutive warning signals, an event was found and the first timestamp of these consecutive warning signals is identified as the onset time.

Note that:

• $μ$ and $σ$ are used to calculate the control parameter $k = \frac{n}{\ln(μ + n σ) - \ln(μ)}$ when k is not specified.
• $n$ is a coefficient that was usually set to 2 and the control parameter $k$ was rounded to the nearest integer value when $k>1$.
• The hastiness threshold $h$ was set to 1 when $k <= 1$ and $h = 2$ otherwise (followed the practice introduced by Huttunen-Heikinmaa et al. 2005).
• N are suggested to correspond to 30 minutes of threshold-exceeding intensity.

find_onset(I, bg_timerange; method = :cusum, kwargs...)

Perform onset determination analysis on time series I using background statistics from bg_timerange.

Arguments

• method: Detection method (:cusum or :sigma, default: :cusum)
• kwargs...: Additional parameters passed to the detection method

Returns

Named tuple with onset_time, bg_timerange, bg_mean, and bg_std.

See also: cusum_onset_detection, sigma_onset_detection

find_peak(I)

Find the peak time and value of a time series I.

select_channel(data, indices)
select_channel(data, predicate)

Select specific channels (columns) from data while preserving metadata structure, including labels.

Examples

# Select specific energy channels
I_low = select_channel(data.PH, 1:3)        # First 3 channels
I_mid = select_channel(data.PH, [4, 6, 8])  # Specific channels
I_high = select_channel(data.PH, 10:15)     # High energy range

# Select using predicates
I_odd = select_channel(data.PH, isodd)      # Odd-numbered channels
I_subset = select_channel(data.PH, x -> x > 5)  # Channels above 5

# Single channel for onset analysis
I_channel = select_channel(data.PH, 4)      # Channel 4 only

sigma_onset_detection(I, μ, σ; n = 2, N = nothing)
sigma_onset_detection(I, times, μ, σ; n = 2, N = nothing)

Detect onset time using n-sigma threshold method when intensity I exceeds μ + n*σ for N consecutive points.

Returns

Named tuple with time and N.

vda(times, energies::AbstractArray{<:Energy}; particle = :proton, mass = nothing)
vda(fluxes, background_timerange; onset = (;), kw...)

Perform Velocity Dispersion Analysis (VDA) to determine solar particle release time $t₀$ and path length $s$ using onset times across different energies.

Principle

VDA is based on the fact that solar energetic particles of different energies travel at different speeds from the Sun to the observer. Higher energy particles arrive earlier than lower energy particles if they are released simultaneously from the source.

The relationship between onset time $t$ and inverse velocity $β⁻¹$ follows:

\[t = t₀ + \frac{s}{c} · β⁻¹\]

where: $t$ is the observed onset time, $β = v/c = \sqrt{1 - (\frac{1}{1 + E/mc^2})^2}$.

Linear regression of onset times versus β⁻¹ yields the release time (intercept) and path length (slope × c).

Arguments

• onsets: Onset times for different energy channels
• energies: Particle energies (MeV) corresponding to onset times
• min_points=2: Minimum number of onset points required for analysis

Returns

A named tuple with:

• release_time: Estimated particle release time at source
• path_length_au: Path length in AU

See also: find_onset