AMPTE MEPA Comments

Link to Other Sources of Information for AMPTE

• JHU APL AMPTE Summary
• JHU/APL AMTPE/CCE Data Center
• NSSDC Page (links to all AMPTE spacecraft)

Basic Intrument Description

The instrument paper describing MEPA in more detail is:

           McEntire, et al., IEEE Transactions on Geoscience and Remote Sensing,
           Vol GE-23, No. 3, May 1985, pages 230-233  

The description below will make more sense if you read the instrument paper first.

We point out here that only the rate channel data for MEPA is available through our tool - no PHA data is included.

Magnetometer Data

Description of Particle Channels

MEPA consists of two heads -- the TOF head and the ION head. Each faces out perpendicular to the spacecraft spin axis and they both face the same direction.

The sector information is not preserved in our data interface. Rather we give the particle flow direction in some coordinate system. For MEPA data, this coordinate system is GSE. Thus each flux value has associated with it a theta (GSE lattitude) and a phi (GSE latitude) as well as an effective width in the theta and phi directions.

The TOF head can measure species and energy, while the ION head only measures the energy of the ions, which are mostly protons. In fact, the counts in the ION head are all assumed to be protons up to 1830 keV. So all channels below this are suumed to be all protons, and the 2 channels above this are assumed to be all alphas.

Note that in the original AMPTE datesets, there was a timing problem which required that 19.75 seconds (one Major Frame of telemetry) be added to time values extracted from the processing system. This correction has already been made in the particle data in this dataset.

Here is a table of the all the channels for MEPA. An explanation for the meaning of the "Spins Read Out" column is given later.

The last 6 rows in the table are engineering singles rates which are provided for convenience. The values for these channels are simply in counts/second, and they have no energy range, so the energy values are set to -1. See the instrument paper for details on the layout of the various detectors. Basically, MCP1, MCP2, and MCP3 provide start, mid and stop pulses for the TOF system, and the SSD detector is at the back of the TOF path. The "rate logic go" rate was a trigger which required signals from MCP1, MCP2 and MCP3 all within 150 ns. The "valid event" rate (which is the only value in the whole table which is not sectored, i.e., you get one readaout for every four spins) required a RLG signal, and also the TOF value between MCP1 and MCP3 had to be roughly twice that from MCP2 to MCP3.

Even though measurements are accumulated in 32 separate sectors, not every channel is read out every spin. A few channels are read out every spin, but most are read out either every other spin, or every fourth spin. The original MEPA data was divided into records of 4 spins each, and so the spin numbers in the table above tell you which of the 4 spins per record contain data for the corresponding channel.

For channels whose sectors were measured over multiple spins, the time associated with the data is the midpoint or average of all the measurement time intervals.

Also, the energy channels were alternated between the ION head and the TOF head. Three fourths of the time the energy channels came from the TOF head, and one fourth of the time the energy channels were taken from the ION head. In the original MEPA records, 3 records had energy channels from the TOF head, and 1 had energy channels from the ION head.

This means that three fourths of the time, there are no measurements from the ION head. Then when the ION head data starts up again, you will get 4 spins worth. Even then, the data will have significant spaces betwen points, because, for most channels, spins are skipped between the reading out of the accumulated counts in each of the 32 sectors.

The exact timing of each measurement may be slightly off due to the malfunction of a timing signature on the spacecraft. The sequence of the the data points is not affected. However, the relative difference between two measurements and the absolute placement of a measurement could both be off by at most 0.31 seconds. If you know about the internals of MEPA data records, this is half a minor frame. The start time of each record is assumed to be at the middle of the minor frame in whcih it occurs, because the timing info about the start point of the records within the minor frame was incorrect throughout the life of the mission. So subsequent measurements within records are all fine relative to each other, but measurements from adjacent records could be off by a little. Note that our mission independent data representation interface does not preserve the record number, so there is almost no way to reverse these timing uncertainties.

In the original MEPA data, there was a discrepancy (of one major frame) about the correlation of major frame number with the absolute time. This problem has been corrected in the MIDL version of MEPA data.

The MEPA Gain Shift Problem

The gain change caused the energies of all channels to increase over the mission. If the detector was rested, the effect became less for a while (in which case the energy boundaries would come back down for a while. Fortunately all the channels changed in unison, so that simply keeping track of the gain as a function of time is sufficient to know the actual energy boundaries measured by the on- board electronics.

The gain value as a function of time is contained in a file "gain.dat" The original copy of this file which came with the fortran code only went to some ime into 1987, but more gain data was extracted off of a hand plotted chart from Dick McEntire, and so the gain data was extended out to the beginning of 1989.

To represent MEPA data in a mission independent way, it was decided to create "Cannonical Channels" representing the energy ranges originally intended by the instrument designers. Then the flux from the measured channels (whose energies had shifted by some amount as determined by the gain) would be re-distributed into the cannonical channels, so that the data can always be reported in the channels with fixed energy boundaries.

So while looking at the data, all you will see are the Cannonical Channels. At times, some of the Cannonical Channels are absent, especially the lower energy ones. This can happen when the gain shift pushes the energy boundaries so high that the lowest Cannonical Channels do not overlap at all with the actual measured channel boundaries. For the case where the measured channel boundaries were higher than any Cannonical Channel boundaries, we did not create new Cannonical Channels, we sinply discarded the information above the highest Cannonical Channel energy.

Details about the Creation of the Cannonical Channels

The approach taken is the following:
Each measured channel belongs to a spectrum of some kind. Take the proton TOF species channels, for example: there are channels P0, P1, P2, which represent a 3 point spectrum. For each measured channel, the differential flux is computed, and then the spectral slope for each segement is computed.
differential flux = alpha * E^gamma

Given a functional form for the flux, the integral flux in an arbitrary energy range can be found by integrating the functional representation of the flux over the desired energy range. Of course the desired energy ranges are the energy ranges of the cannonical channel bouandaries. Once the integral flux for these channels is found, it can easily be converted to differential flux, which is what we store for MIDL data.

There are complications. For some spins, not every point in the spectrum is measured. To overcome this, the counts from surrounding measurements are used to get a time-interpolated value for the missing measurement. But the time interpolated channels are not used to add flux to the cannonical channels - they are only used to determine an accurate spectral slope. For a given cannonical channel, it is important to know what fraction of the cannonical channel is overlapped by measured channels. Only the cannonical channels with the most significant overlap are reported, and its never the case that more cannonical channels are reported than were measured. There can be fewer cannonical channels reported, however, because a limit is set requiring a certain minimum overlap requried for a cannonical channel in order for it to be reported as having a measurement. These kinds of considerations are important when you have sparse spectra being measured, because if you are measureing channels which are not adjacent, they each may straddle different cannonical channels, and its possible that none of the cannonical channels has even 50% coverage by measured channels. If all the measured channels are adjacent, then its very likely you will get more cannonical channels with sufficient coverage, because where the coverage from one measured channel ends, the coverage for the next adjacent channel begins.

The measured channels are functionally represented by power law segements, with each measured channel having one of two possible representations. Channels on the end of the spectra (the first and last channels are both considered "end" channels) are represented with a single power law function determined by the flux in the end channel plus the flux in the adjacent channel. This single power law is valid from the low energy of the measured channel to the high energy of the measured channel. For channels "inside" the spectrum (i.e., not on either end), a two segment power law function is used, because there is the slope from the previous and the next channels to consider. In both cases, after the power law function(s) is (are) determined, the differential flux given by the functional representation is normalized over the entire measured channel energy range so that it exactly matches the differential flux which was actually measured in this measured channel.

Then for each cannonical channel, the overlapping measured channels are found, and the integral flux contribution from each is added. Then the total integral flux can be found, and then the differential flux is just the integral flux divided by the energy.


back to Space Department home page
back to MIDL home page

Maintainer: