Rodger I. Thompson

An Enlightenment About the Wonders and Opportunities of Infrared Studies with NICMOS

A compendium of NICMOS descriptions and operational modes, intended for the interested scientist.


NICMOS stands for the Near Infrared Camera and Multi-Object Spectrometer, a general purpose instrument for carrying out near infrared observations with the Hubble Space Telescope (HST). The primary NICMOS capability is accurate, high sensitivity imaging at high spatial resolution with three cameras at different magnifications. Grisms provide multi-object spectroscopy in the widest field camera. The NICMOS spectral range is 0.8 to 2.5 microns for both imaging and grism spectroscopy. Section 3.1 on page 15 gives the appropriate parameters for the filters and grisms.

NICMOS emphasizes the HST imaging characteristics. The three cameras operate simultaneously on adjacent, but not contiguous, fields with diffraction limited imaging at 1.0 and 1.75 microns (cameras 1 and 2) and wide field imaging with camera 3. Section 6.1.1 on page 18 gives the detailed imaging parameters. Observers should select the appropriate combination of cameras, observation times, and filters or grisms for their investigation. This document will help you in that choice as well as give you a description of the NICMOS capabilities.

All observers, both Guaranteed Time Observers (GTOs) and General Observers (GOs) obtain access to NICMOS through proposals to the Space Telescope Science Institute (STScI). Proposals must contain all of the information needed to plan and execute a NICMOS observation. Section 10.0 on page 33 describes the proposal process for both GTOs and GOs. Observers should check with the STScI for the current details on proposal submission and requirements.


This section describes the NICMOS cameras and their characteristics. Imaging observational modes are covered in Section 8.0 on page 26. NICMOS has three imaging fields, one for each camera. Each field has a different magnification and a dedicated 256 x 256 NICMOS 3 HgCdTe detector array. Operation of each camera is separate from the others which means that filters, integration times, readout times and readout modes can be different in each. However, all cameras share the pupil mirror (Field Offset Mirror or FOM) that provides small offset capability, therefore all cameras switch fields simultaneously if you move the mirror. If you use this mirror for background subtraction by median filtering, the observations in all of the cameras must be coordinated.

FIGURE 1 shows the arrangement of the three NICMOS camera fields.

Table 1 gives the basic imaging characteristics of the cameras.


Each camera has twenty filter positions on independent filter wheels. Section 12.2 on page 43 gives tables of the current filter selection.


2.1.1 Point Spread Function

The following figures give the theoretical point spread function for the HST. These figures assume a perfect 2.4 meter mirror with a 0.333 linear obscuration.







When we develop the predicted PSFs for the NICMOS optics we will replace these figures with those. We do not expect the final PSFs to differ in character from the ones show above.

2.1.2 Encircled Energy Functions

The following plots show the encircled energy fraction vs. angular radius under the simplification of a monochromatic wavelength at the center of the J, H, and K bands. These plots should be useful in calculating the amount of flux in any pixel.





There are several ways to characterize the NICMOS performance. The following sections give information on several observing modes and performance characteristics. Observers with special requirements should contact a member of the NICMOS team before launch or the NICMOS Instrument Scientist at the STScI after launch.

2.2.1 NICMOS Efficiency

Table 2 presents the efficiency factors used in the calculation of the sensitivity of the NICMOS cameras. In this table the efficiency of the mirrors is set to 0.95 although the goal for thermal suppression is 0.98. This provides a conservative estimation of the observable flux. Each filter has a different efficiency factor, therefore we leave the total efficiency as a factor times the filter efficiency.


In the calculation of the point source efficiency we will use an average filter efficiency of 0.77 for a total efficiency of 0.24.

2.2.2 Point Source Sensitivity Calculations

Several factors affect the final sensitivity of the cameras. Table 3 on page 10 displays the factors utilized in the sensitivity calculations. The NICMOS optics emissivity is one minus the product of the efficiencies of the NICMOS optical elements in Table 2 on page 9 except for the shield window, filter and detector which are cold. Again this is a conservative number relative to the emissivity goals of the instrument. The detector noise is for a two readout (beginning and end) read mode and is assumed to be independent of the integration time.


Equation 1 describes the zodiacal background.

where L is the wavelength in microns, T is the zodiacal dust temperature in Kelvins and B is the blackbody irradiance in photons/(second cm^2 micron steradian). Only the first or scattering term in the equation contributes to the background at NICMOS wavelengths for the assumed zodiacal dust temperature of 265 K. This is form is consistent with the background observed by the COBE satellite as determined from preliminary results supplied by Ed Cheng for 45 degrees out of the ecliptic. It is slightly different from the formulation used earlier for the SIRTF models. Figure 11 gives the background for camera 3 which has 0.2 arc second pixels. The flux for the other cameras is found by scaling by the ratio of pixel areas on the sky.

FIGURE 11. Camera 3 Background Flux

Since all of the NICMOS camera pixels are smaller than the full extent of the PSF the sensitivity depends on how the contributions in each pixel are summed. In most cases a straight, unweighted sum of between 4 and 15 pixels gives poorer signal-to-noise ratios than just taking the signal in the highest pixel. The following figures give the total object fluxes needed to give a signal-to-noise ratio of 10 in the central pixel for a point object centered on that pixel. Table 4 on page 11 gives the fraction of total object flux in that central pixel for the calculated wavelengths and cameras. These numbers come from the energy distribution charts in Section 12.5 on page 45. The next step will be the development of the proper weighting factors to produce the highest sensitivity.


2.2.3 Sensitivity Graphs

The following graphs give the sensitivity figures for the three NICMOS cameras. The bandpass at each wavelength is 10% of the wavelength and the achieved signal-to-noise ratio is 10. The graphs give the flux in Janskys needed to give a signal-to-noise of 10 in the time on the abscissa. Only the flux in the central pixel is used in this calculation. As an aid in interpreting the graphs the following table gives the magnitudes of 1, 10-3, and 10-6 Janskys in the I, J, H, and K photometric bands.

TABLE 5. Photometric Magnitudes of 1, 1E-3 and 1E-6 in the I,J,H, and K Photometric Bands




A Jansky is 10E-26 Watts / m^2 Hz. For those who are more comfortable with magnitudes than Janskys Table 6 gives the magnitude of 1 Jansky at J, H, and K.



To obtain the extended source sensitivity in Janskys per square arc second take the sensitivity per pixel then divide by the pixel area in arc seconds. This accounts for the differences between the cameras since the read noise and dark currents are constants per pixel whereas the background and photon noises are proportional to the angular sky area for uniform extended sources. Table 7 gives the extended source sensitivity for the case where all of the pixels are added together to produce a single measurement for one arc second.





This section describes the NICMOS spectroscopic capabilities. Section 8.7 on page 31 describes spectroscopic operation. NICMOS utilizes grisms to achieve multiple object spectroscopy in the spectral range between 0.8 and 2.5 microns. The grisms reside in the filter wheel for camera 3, therefore the spatial resolution of the spectroscopy is similar to the spatial resolution of camera 3. The filter wheel contains three grisms to cover the wavelength range with a spectral resolution of 200. The two shorter wavelength grisms exploit the low natural background of HST while the longer wavelength grism is subject to the thermal emission from the HST and NICMOS optics.

3.1 General Grism Description

A grism is a combination of a prism and grating arranged to keep light at a chosen central wavelength undeviated as it passes through the grism. Grisms are normally used to create spectra in a camera by rotating the grism into the normal camera beam. The grism then creates a dispersed spectrum centered on the location of the object in the camera field of view. NICMOS utilizes this mode of operation without any slit or aperture at the input focus so that all objects in the field of view display their spectra for true multi-object spectroscopy. The resolution of a grism is proportional to the tangent of the wedge angle of the prism in much the same way as the resolution of gratings are proportional to the angle between the input and the normal to the grating. The NICMOS grisms have an interference filter coated on their entrance faces to limit the bandpass of the spectrum. This is necessary to prevent overlap of orders and reduce thermal emission from the telescope. Since the NICMOS grisms do not have an input slit or aperture, there is not a reduction of the background flux found in slit dispersing systems. This is not a significant problem in the shorter wavelengths, but the long wavelength grism has a high background flux.

The NICMOS grisms have a resolution of 200 per pixel. The resolution is a compromise between wedge angles, spectral and spatial coverage, sensitivity to alignment, and limiting flux. The thickness and clearances in the filter wheel along with the requirement of confocality with the images through the filters limits the thickness of the grisms and therefore their wedge angles. Table 8 indicates the NICMOS grism parameters. For a constant resolution the central wavelengths are space in equal logarithmic intervals which sets the central wavelengths. The bandpasses are set by the filters to reduce background and to limit the spatial extent of the spectra. All of the grisms have CaF as their optical material and a 2 mm thickness at the center. The tangent of 5.875 degrees is approximately 0.1, therefore there is about 1.6 mm of wedge in the grism going from 1.2 to 2.4 mm at the extreme edges. The clearances on the filter wheel are tight therefore higher spectral resolutions which require larger wedge angle would be difficult to accommodate.

TABLE 8. NICMOS Grism Parameters

3.2 Expected Performance

3.2.1 Sensitivity

Grisms A and B are in spectral ranges with very little background, however, grism C is in the range of strong thermal emission from the HST and NICMOS optics. Since the grisms are slitless each pixel receives the background radiation from the full bandwidth of the grism and filter combination. Table 9 shows the dramatic increase in the background flux for grism C. Use grisms A and B when possible. Grism C is for the longer wavelengths only. Figure 15 shows the expected performance of the grisms in the same type of flux versus time plot used for the camera performance.


FIGURE 15. Grism Sensitivity

The 0.9 micron line is from grism A, the 1.2 and 1.6 micron lines from grism B, and the 2.0 and 2.4 lines are from grism C.

3.2.2 Spectral Performance

As an example of what to expect with a R = 200 per pixel spectrum Figure 16 is a high resolution ground based spectrum of NGC 4151 reduced to a resolution of 200 per pixel.

FIGURE 16. An Example of an R 200 Spectrum

The region of this spectrum around Paschen -Alpha and the other marked areas is contaminated by telluric water vapor absorption which of course will not be present in the NICMOS spectra. Careful inspection of the width of the He 1.0830 and the Paschen -Beta line indicates that the broad line component of these lines is evident.



NICMOS contains a very simple coronagraphic capability in Camera 2. The cold pupil stop for the camera is a more detailed masking of the pupil which includes the spiders and the pads of the HST pupil. The field dividing mirror of Camera 2 contains a hole of 0.3 arc seconds radius which acts as the occulting spot for the coronagraph.


6.1 Optics

6.1.1 The Imaging System

7.0 Detectors

The NICMOS detectors are 256 x 256 pixel arrays of HgCdTe photodiodes read by a switched MOSFET multiplexer. Each pixel has an independent readout circuit in the multiplexer which is accessed by shift registers for the columns and rows of the detector. The HgCdTe photodiode array is attached mechanically and electrically to the silicon multiplexer by indium bump bonds, one bond for each pixel. For redundancy, the single silicon multiplexer chip is divided electrically into four independent quadrants of 128 x 128 readout circuits. Each quadrant has its own shift registers, output amplifiers, clock inputs and power inputs. Failures in one quadrant do not affect the operation of the other quadrants.

The range of sensitivity for the NICMOS detectors is 0.8 to 2.5 microns with an average readout noise of 35 electrons and a dark current on the order of 0.1 electrons per second. Section 7.3 on page 26 presents the detailed characteristics of each detector.

7.1 General Operation

7.2 Readout Modes

7.2.1 Two Sample Readout

The two sample readout is one of the simplest available modes and is probably the mode of choice for short integrations of relatively high signal to noise objects. The first action of the two sample readout is a single pass through the detector resetting each of the pixels. This is immediately followed by a second pass through the detector non-destructively reading and storing the pixel value. This marks the beginning of the integration. The final action is a second non-destructive reading of the detector at the end of the integration time, which marks the end of the integration. The returned image is the difference between the second and the first pass pixel values. The integration time is defined as the time between the first and second read of the first pixel in the array

The two sample readout has the advantage of requiring very little CPU time, which allows other integrations and actions to proceed simultaneously with the integration. It also has the minimum time for output amplifier operation which minimizes the amplifier glow contribution. This method does not discriminate against cosmic ray events and does not check for saturation levels in the integration.

7.2.2 Multiple Initial and Final Sample Readout

This method is a variation of the two sample readout which replaces the single initial and final readouts with multiple readouts. After the initial reset pass, there immediately follows n non-destructive reads of the detector where n is specified by the observer in the proposal for the observation. The average of the n values is then stored as the initial value for each pixel. At the end of the integration there are again n non-destructive readouts of the detector with the final value for each pixels being the average of the n reads. The returned image is the difference between the averaged final and initial values. The integration time is defined as the time between the first read of the first pixel in the initial n passes and the first read of the first pixel in the final n passes.

The advantage of this method is a reduction in the read noise associated with the initial and final reads. In theory the read noise should be reduced by 1/sqrt(n) where n is the number of reads. This can be an advantage in reducing the noise in an intermediate time integration of faint sources. For integrations where source photon noise or dark current noise exceeds the detector read noise the multiple initial and final read mode does not offer an advantage. The multiple initial and final reads put a higher burden on the CPU and increase the total charged observing time by the time required for the extra reads. The anticipated read time per detector is 0.5 seconds. This mode does not discriminate against cosmic ray events and does not check for saturation.

7.2.3 Ramp Integrated Mode

The advantage of the ramp mode is the ability to detect and recover from cosmic ray events and the ability to set saturation limits in the observation. This can be very useful in images where the expected flux levels are not well known and in serendipitous or survey observations. Early indications are that some signal-to-noise improvement is expected over the two sample mode of observation. The disadvantage is that the observation is very CPU intensive which limits the cycle time of the samples especially if more than one detector is operating. At the present time we expect a cycle time of approximately 5 seconds per detector to process the ramp data. The minimum time between passes is therefore 5, 10 and 15 seconds depending on whether there are 1, 2 or 3 detectors operating in this mode. BASIC CONCEPT

NICMOS detectors are capable of image read outs which do not destroy the image on the detector. These "non-destructive" readouts allow the observer to monitor the image integration at intermediate times during the observation. This provides a method of catching cosmic ray events and monitoring for saturation during an integration. The ramp mode is a method of image processing that implements these features. The ramp integration mode divides the total integration time T into N equal time intervals t = T/N. In a normal integration the signal increase during each time interval t should be the same within the normal noise sources present in the measurement. The arrival of a cosmic ray will alter the signal increase during the time interval and will be therefore detectable as an anomalous signal during that time interval. The value of the total signal at each readout is compared with a maximum value to detect saturation. The total signal is the value at readout, which is different that the signal increase during the interval which is the value at readout minus the value at the previous readout. The action taken upon the recognition of a cosmic ray or saturation will vary according to the specific actions specified in the proposal for the observation. BASIC ARITHMETIC

The basic arithmetic for the ramp mode is the mean and standard deviation arithmetic for the multiple measurement of a quantity. Although the signal is increasing along a ramp which should be given by

the differences between successive readouts should be constant within the signal-to-noise as

In terms of the measured signal from the A/D converters the delta-signal for the ith sample is simply

FIGURE 17 , Example of a set of samples with a cosmic ray event

FIGURE 18 , delta-SIGNAL output from above example

The best measurement of the delta-signal after n samples is the simple mean.

In order to determine when a particular signal increase is significantly deviant the standard deviation sigma of the delta-signal after n samples must be calculated as well. Cosmic Ray Detection

Signal increases that are more than a predetermined number of standard deviations from the mean are then considered cosmic ray events and not utilized in the image or in the calculation of the mean and standard deviation. The equation for the standard deviation assumes that the noise is independent of the signal. If we were calculating the standard deviation of the signal this would not be true if the photon noise dominated the signal. It is, however, true for the signal increase since the time intervals between signal samples are all equal. The photon noise is associated with the increase in signal between samples rather than the total signal. The expected noise is therefore the same for all signal differences. At first glance this may seem not to be true if you argue that the photon noise of the ith sample is the square root of the number of photons and the noise of the i-1 sample is the square root of the number of photons at that time. The important thing to remember is that the value of the i-1 measurement is an exactly accurate measurement, within read noise, of the integrated signal at that time and we are then measuring the increase in signal over the time between samples. It is equivalent to resetting the signal to 0 each time and looking for the integrated signal at the next sample time. Saturation Detection

A signal is considered saturated if its value, not it increase, exceeds a predetermined limit. SIMPLE PROBLEMS

The first problem is of course with the first samples where the mean and standard deviation are not yet calculated. Cosmic ray checking should not be initiated until at least 3 samples and more likely 4 samples have been collected. Cosmic ray events that happen during these first samples will not be rejected. We accept this. Saturation checking, however, is always valid and will catch those cosmic ray events that saturate a pixel.

A second problem for very low signals is read noise and dark current. There must be a minimum value of sigma which reflects the read noise and dark current. For any reasonable integration time the read noise will dominate over the dark current for the time between subsequent reads. ACTIONS UPON DETECTION

Actions upon detection of either saturation or cosmic ray events range from simple marking of the pixel as defective to recovery of the best estimate of the signal from the available and possibly subsequent data. Cosmic Ray Detection Actions

There are several possible actions upon detection of a cosmic ray event. The following lists the possible actions which will be proposal selected. The observer should be able to select the action mode in the observation proposal. No action

In some cases the observer will select the ramp mode based on signal-to-noise considerations alone and will not want any cosmic ray detection or correction actions. This mode simply computes the simple mean and does not compute the standard deviation. No algorithms for cosmic ray detection or correction are implemented. Detection and marking

In this mode the detection of a cosmic ray event produces an indication that the pixel has been hit by a cosmic ray with no further action by the program. The presence of a mark indicates that at least cosmic ray detection has occurred for the pixel. Once a cosmic ray detection has been made the pixel is marked and no further action occurs. This is not a toggle but a latch so that a further detection does not change the marker. My simple view is that the pixel would not be taken out of the normal image processing. This is under the assumption that checking for a mark and altering the processing mode would be slower than leaving the pixel in the loop for normal processing. Marking a pixel as having a cosmic ray event should be as simple as setting the value of the appropriate 65536 array from 0 to 1. As stated above subsequent events would again set the appropriate bit to 1, not toggle it between 1 and 0. Detection and retention of previous signal

The next most sophisticated choice for cosmic ray correction is retention of the signal that has been detected previous to the event. In this mode the value of obtained previous to the event detection is retained as the value of the signal for that pixel. Further processing of the signal is discontinued and the number of the integration interval where processing stopped is put into a 65536 element array. Pixels where no event was detected will have a value of 0 in the array. Detection and continuation of processing

This mode simply rejects all signals which deviate from the simple mean by more than the preset number of standard deviations. In this case the total number of samples used in the calculations of the simple mean and standard deviation will be less than the nominal amount of N. Upon detection of a cosmic ray event that sample is rejected and not used in the calculations. The total signal is checked to see if it has reached the saturation level. If it has not normal processing of the pixel continues on the next sample for the pixel which is not a cosmic ray event. If the saturation level is reached, as it will for the majority of the cases, processing of the pixel is discontinued in the same manner as for the detection and retention of previous signal case. The associated 65536 element array will contain the number of samples used in the calculation of the mean. The values for the pixels without cosmic ray events will be N. Saturation Detection Actions

The saturation detection options will be fewer than the cosmic ray actions since there is not an easy way to continue meaningful observations after onset of saturation in a pixel. No action

As with the cosmic ray detection an observer may wish not to worry about or try to detect saturation. In this case no checking for saturation occurs. Detection of saturation and retention of previous signal

In this mode when saturation is detected the simple mean previous to the saturated sample is retained as the signal. A 65536 element array will be generated that contains the number of the sample where saturation was detected. The value for a pixel for which no saturation is detected will be 0. This avoids redundancy with the case where saturation occurs on the last or Nth sample. Termination of observation

In this mode the observation is terminated when the number of saturated pixels exceeds a predetermined number. The observation proceeds in the same manner as the detection of saturation and retention of previous signal until that point. When the number of saturated pixels exceeds the predetermined number the observation is terminated and no further processing occurs. The check for the number of saturated pixels occurs at the end of the processing of all pixels for that sample. This insures that all of the pixels for a given sample interval are processed. The 65536 element saturation array will contain the number of samples obtained at termination or less in the case of previously saturated pixels. WHAT ARE THE OUTPUTS

The primary outputs of the ramp mode integration is an image of the simple mean of the individual pixels. This is the best determination of the value of the delta-signal for each pixel. In addition to the primary image there will be up to two 65536 arrays containing the cosmic ray and saturation outputs. The presence and content of these arrays will depend upon the parameters selected in the observation proposal. The content of the arrays is described below separately for cosmic rays and saturation. The number of arrays will vary from none for both cosmic ray and saturation off to two for both on. A MUCH HARDER PROBLEM

There is a much harder problem with the ramp integration method than the simple problems in "SIMPLE PROBLEMS" on page 22. The ramp modes assumes that the output of the detector increases in a linear way with time due to the incident photon flux and dark current. The true nature of the signal level versus time contains an additional component which is characteristic of the individual detector. The additional component is a decaying signal component that appears to be extremely repeatable for each readout. The evidence for this repeatability is that the difference of two images with equal integration times shows no trace of the extra component even if the images are separated by days in time. This repeatability allows subtraction of a known function to eliminate the additional signal component.

An example of the extra component for a flight candidate detector is shown in Figure 1 below.


Note that the scale is in electrons rather than ADUs so that this should be scaled for the number of electrons per ADU to get the expected signal level in the electronics. Correction procedure

The correction procedure is to fit the additional signal with a function f(t), hopefully with few coefficients, and subtract this component at each sample time. A straight polynomial is probably a bad fit, as is a straight log function. The UofA will provide the fit for each detector as delivered to Ball Aerospace. The function will be different for each detector and may be influenced by the electronics. In this case a new function will be required for fitting. UofA will use the nature of the additional signal component as part of their flight selection procedure and will try to pick flight detectors with small and easily characterized components. At this time it appears that a single function will be adequate for all pixels. If this is not true we will inform Ball Aerospace of the proper procedure.

7.2.4 Bright Object Mode

The minimum cycle time for reading a detector is on the order of 0.5 seconds and is set by the sample time of the A/D converter in the detector analog electronics chain. Some objects, such as the disk of Jupiter, saturate the detector in a time far less than 0.5 seconds. Observations of these objects requires a separate mode of operation. The bright object mode resets and reads each pixel before moving onto the next pixel. In this way an image of the object is built up one pixel at a time. The mode begins with the first pixel with a reset and an integration of t < 0.5 seconds set by the observer in the proposal for the observation. After t seconds the pixel is non-destructively read and the detector is clocked to the next pixel where the cycle is repeated. Since the four quadrants of a detector are read simultaneously the net time for the integration is 128x128 t or 16384 t seconds. In the case of Jupiter the integration time t is only a millisecond resulting in a 16 second integration time. For objects with saturation levels just below 0.5 seconds the total time could be quite large if the full time were taken. Proposers will need to judge the real integration time and signal-to-noise ratio required for the observation time and adjust t accordingly. The integration time in this mode is the time between the reset of the first pixel and the read of the first pixel. The execution time, however, is 16,384 times the integration time.

The advantages of this mode of operation is the ability to observe objects significantly brighter than the normal saturation limit of the detector. The disadvantages are several. First, some observations will take a long time for t values on the order of 0.1 to 0.01 seconds. If the object changes (planetary rotation) or if the telescope is unstable it will affect different parts of the image differently. There will not be a single PSF for the observation. Second, the offset of the detector is not removed in this mode of operation. In general, the signal is very high and the offset does not matter. In some cases it will and this can be a detriment to the signal accuracy. There is also no cosmic ray correction or saturation detection in this mode of operation.

7.3 Detailed Detector Characteristics

7.3.1 Camera 1

7.3.2 Camera 2

7.3.3 Camera 3


NICMOS is basically a very simple instrument in which the fundamental operation is obtaining an image. The multi-object spectroscopic, polarimetric, and coronagraphic operations all entail the same fundamental operation of image taking. Only the coronagraphic mode, which requires positioning an object on the coronagraphic spot, has an extra component to it. There are, however, several observer selected options for image acquisition that are available. Perhaps the most significant option is the opportunity to specify a sequence of observations that are carried out for a single target acquisition with HST. Below is a description of each of these options.

8.1 Definition of Independent Operation

Each of the three NICMOS cameras contains an individual focal plane assembly (FPA) capable of operation independently of the other FPAs. The following two paragraphs define the concept of independent operation.

The main concept of independent operation is the ability to schedule each NICMOS camera operation without concern about the operational state of the other cameras. This pertains to all operations relevant to a camera, including positioning of filter wheels, initiating and terminating integrations, performance of the several readout modes of the detectors and the storage of the resultant data in the NICMOS computer system. The observer must be aware, however, for observation planning that the non-interference of filter wheel movements in one camera with operations in the other cameras is a design goal and may be somewhat compromised by the possibility of electrical noise from the filter wheel motors. The observer must also be aware that the storage capacity of the NICMOS computer system and the downlink capabilities of the HST spacecraft can limit the number of camera operations that can be accomplished within an observing period.

An important goal of the detector electronics system is immunity to electrical noise from detector clocking on the detector output signals during the sample-hold and analog to digital conversion times. If this is not achievable through appropriate circuit design and filtering the sampling and conversion of the signal must be done in the absence of clock pulses on other detectors. This will be achieved by synchronizing the pixel clock pulses on the cameras to avoid digital noise during the measurement of the signal levels on the detector diodes by the analog electronics. All of the detector operations are therefore slaved to a master clock at a basic beat rate of the pixel readout time of the electronics. Integration start times may be advanced or delayed by up to the value of the pixel access time to achieve the synchronization. At this time this constraint does not appear to affect any of the anticipated observations.

8.2 Single Image

The single image is the most basic NICMOS operation, however, more advanced operations are generally just a repetition of the single image operation. A single image produces one 256x256 image in a single camera. There are four classes of parameters associated with the operation; standard, background, non-standard, and spacecraft. The standard parameters are ones that must be specified for each image. Background parameters are specified for images employing background subtraction techniques. Non-standard parameters are parameters that are usually left at their default values but may be altered for "expert" observations. Spacecraft parameters are special configurations of HST. The following lists these parameters by category.

Standard Selectable Parameters

1. Camera or cameras

2. Filter (including grisms and polarizers)

3. Integration time

4. Detector Readout mode

5. Coronagraphic mode (requires acquisition)

Background Parameters

6. Positions of the Field Offset Mirror

Non-Standard Selectable Parameters

7. Bias Voltages

8. Gain

9. Real Time Acquisition

Space Craft Parameters

10. Roll Angle

Once the parameters have been specified a single image returns one image of 256x256 16 bit words representing the image at the end of the integration.

8.3 Detector Readout Modes

As discussed in Section 7.2 on page 18 there are several detector readout modes that may be employed in the acquisition of a single image. The choice of readout mode will be made on the proposal for the observation. The following are the present readout modes with a brief description. New readout modes can be implemented once the instrument is installed but will have to be requested and programed by STScI.

8.4 Background Corrected Images

Since NICMOS is an infrared instrument background flux subtraction is an important aspect of observations at the longer wavelengths. NICMOS contains a Field Offset Mirror (FOM) to aid in the background subtraction. This mirror, located at the first warm pupil inside the instrument, moves the field of view in all cameras by a selectable angular distance up to 1/2 the field of view in the largest camera, camera 3. This is a maximum motion of 21 arc seconds in the x and y directions. The general method of background subtraction using median filtering is to center the object on the camera and then taking a series of integrations (~16) with the field offset in different directions by approximately 10% of the field width. The background is computed as the median of the pixel values from the images. In most fields the median will exclude the pixel values that may have a star or other image on them. Dense or complicated fields will require more extensive analysis. For reasonable background subtraction 16 is the minimum number of required images. The advantages are that the object is always in the field of view, increasing efficiency, and that the background is observed at the same time as the object, reducing the effect of temporal changes in the background.

In the observing proposal the observer must select the number of images for the median filtering and a single integration time for all of the images. The positions for each integration must be also selected and be inside the range of the FOM. The positions are given as offsets from the first position with the object centered.

If the FOM is inoperable or if the observer prefers, the same function can be accomplished through small angle maneuvers of the HST. The advantage of this mode is that there is no alteration of the view factors of the detectors, which results in a more accurate background determination. This mode can not be employed while NICMOS is carrying out parallel observations with another HST instrument.

It is important to note that the FOM moves the field of view on all cameras when it position is altered. If more than one camera is in use the FOM dwell time must be longer than or equal to the longest integration time of any of the cameras in use. As an example if all three cameras are in use the appropriate integration times for cameras 1, 2, and 3 may be 5, 3, and 1 second respectively. The dwell time of the FOM would then be 5 seconds per integration, with cameras 2 and respectively utilizing only 3 and 1 seconds of the available time. The total observation time would 5 x 16 = 300 seconds if a 16 position observation is selected.

Normally all images will be transmitted to the ground for subsequent processing. The observing time in each camera is equal to the number of settings times the integration time for each observation. The execution time for this observation is the integration time plus the FOM mirror movement time which is expected to be 2 seconds per position in the worst case. Any of the single integration readout modes may be chosen for this type of operation.

The advantages of this mode is the ability to define and subtract a background flux for the observation without having to waste time observing blank background fields. The disadvantage is the large amount of data associated with the production of a single image and the reduction of the total field through edge loss of the field areas that are not observed in all of the observations. This mode also complicates the operation of more than one camera since the FOM motion affects all cameras.

In some cases the object will be too large for this type of operation or there may not be sufficiently empty sky near the object in question. In these circumstances a series of single images at different HST pointings must be utilized. In that case the operation is similar to a series of single images.

8.5 Multiple Image Observation

Normally a single integration on a target results in a single 256x256 image at the termination of the exposure. The unique non-destructive nature of the NICMOS readout offers an alternative of providing images at intermediate stages of the integration for return to the ground. This is distinct from readout modes which utilize non-destructive readouts which are processed on board NICMOS to produce a single final image for return to the ground.

Each intermediate image will contain a header and image, identical to a single image observation and will be treated as a single image observation. If it is part of a sequenced observation it will be treated as a series of single images in the sequence. The proposer will be responsible for the post processing of the multiple images. GTO time will be charged for one image equal in length to the total observing time. Any of the single image mode readout modes may be chosen for this type of observation.

8.6 Sequenced Observations

A sequenced observation is a pre-planned sequence of observations involving one or more cameras which returns the images as a set after the conclusion of the sequence. This mode adheres to the general rule of returning a specified amount of data at a specified time.

The following table gives a sense of a sequence of observation utilizing three NICMOS cameras. In general there is no requirement for the observations to start or stop in one camera at the same time as another camera unless the FOM is in operation as in the background subtraction mode described in Section 8.4 on page 28. Although the sample sequenced observations begin and end at the same time for each camera, there is no requirement for this to occur.


At the end of the time indicated by the total observation time all of the data will be ready for either storage in the CPU memory for transmission to the SDF at the designated time or for immediate transmission to the SDF if scheduled in the mission specification. The data will adhere to the following format. The images will appear in the order of their termination. In the case of simultaneous termination the images will appear ordered by their camera number, camera 1 first, etc. Each image will contain a standard header as if it was a single integration. If appropriate the data may be preceded by a sequence header containing information on the number and properties of the images in the sequence.

The individual exposure times for each observation is the time associated with the readout mode utilized as described in Section 8.3 on page 28. The total observing time is the time between the execution of the first observation to the termination of the last observation. The execution time will include the setup time for the first observations.

At this time there are no restrictions on moving filter wheels in one camera while observing in another camera. If there are optical or electrical restrictions on this we will incorporate them into the sequenced observation protocol.

The proposal for sequenced observation will contain the information required for each single integration and the order of the images for each camera. The images will be obtained in order by each camera.

8.6.1 Limitations on Sequenced Operations

In planning for sequenced operations the observer must be aware of the following limitations.

1. The FOM (field offset mirror) may not be moved during the integration period of any camera. The FOM moves the image fields in all cameras since it is located in the first or common optical stage of the optical system. The proposer is responsible for making sure that all camera integrations are finished before requesting a motion of the FOM.

2. The sequenced observation should not contain more images than can be stored in the NICMOS data memory. At the present time the NICMOS memory is capable of storing 80 individual images.

8.7 Multi-Object Spectroscopy

Multi-object spectroscopy observations are identical to imaging observations and can be carried out in the same manner as any of the imaging operations discussed above. In multi-object spectroscopy one of the grisms in the filter wheel for camera 3 will be selected in the same manner as selecting any filter. The observations then proceed via one of the readout and operation modes discussed above. Multi-object spectroscopy observations may appear as part of a sequenced observation.

The direction of dispersion is perpendicular to the radial direction in camera 3 where the radial direction is defined by a vector originating at the center of the field of view for camera 3 and pointing toward the center of the OTA axis. Some observations may profit from a particular orientation of the dispersion direction on the sky. It will be the observers responsibility to state a preferred direction of dispersion in the proposal submission. This will generally require a particular roll direction of the space craft to achieve this. A change of spacecraft attitude in either pointing or roll will require either a new single image or a new sequence of observations at the new attitude.

Although multi-object spectroscopy observations can stand alone with no supporting observations, it is anticipated that most observations will be paired with an image in camera 3 at the same pointing with an appropriate filter. This provides the location of each object in the field and aids in the identification of each spectrum. Because of this natural pairing it is anticipated that most spectroscopy observations will be in at least a two image sequenced observation.

Observing time for a spectroscopic observation is identical to that of an imaging observation in the same mode.

8.8 Polarimetry

Polarimetric observations are also identical to imaging observations and can be carried out in any of the imaging modes discussed above. Cameras 1 and 2 contain 3 polarizers each in their filter wheels. Polarization observations occur through simply selecting the filter location containing the desired polarizers and proceeding as in a normal image. The three polarizers in a filter wheel are identical with a 120o polarization axis separation. Observations in all three polarizers will then provide the linear polarization for any direction of polarization. A priori knowledge of a source may define a preferred orientation on the sky. As with the spectroscopic observations, it will be the observers responsibility to state a preferred direction of dispersion in the proposal submission. This will generally require a particular roll direction of the space craft to achieve this. A change of spacecraft attitude in either pointing or roll will require either a new single image or a new sequence of observations at the new attitude.

Although polarimetric observations can stand alone with no supporting observations, it is anticipated that most observations will be paired with an image in the same camera at the same pointing with an appropriate filter. This provides a continuum image of each object in the field and aids in the interpretation of the polarized image. Because of this natural pairing and the requirement for all three polarizers for a full set of Stokes parameters, it is anticipated that most polarimetric observations will be in at least a four image sequenced observation.

Observing time for a polarimetric observation is identical to that of an imaging observation in the same mode.

8.9 Coronagraphic Imaging

Coronagraphic imaging is identical to normal imaging except for the required acquisition sequence at the beginning of the observation. Full coronagraphic capability is implemented in camera 2 which includes a full diffraction limiting cold pupil mask and an occulting spot in the field splitting mirror for camera 2. The acquisition task is to center the bright central object on the occulting spot of the camera. This results in a signal minimum for the object and the accompanying reduction of scattered light into the field. Since most acquisition methods work best on producing a signal maximum at a location on the detector, the coronagraphic acquisition will be a two step process. First the central object will be centered on a predetermined position on the camera 2 detector, then the HST will move the object to the occulting spot with a small angle maneuver (SAM).

Since we want the best possible image quality in coronagraphic observations, the FOM will be initially centered in the highest image quality position for camera 2 and the object acquired with coordinates which will nominally center the object on the a predetermined spot close to the occulting disk. The true location of the central object will then be determined by centroiding or other algorithms to determine the actual position of the central object. From this information the required SAM to return the object to the central spot will be calculated and the movement performed. Since the central object will now be occulted there will not be a further attempt to refine the position of the source. The accuracy of the SAM is better than 1/4 of a pixel, therefore, the centroiding on the occulting spot should be as accurate as the original centroiding calculation of the object position.

After acquisition a coronagraphic image proceeds in the same manner as normal image. Coronagraphic images are not included in sequenced observations due to the acquisition required. Once the acquisition has been performed, however, simultaneous, asynchronous observations in all cameras will be supported. All of the image readout modes are applicable and the integration time is identical to that of the selected image mode. The total observation time will include the acquisition time.



The following is the current draft proposal instruction sheet for NICMOS which will have to be filled out for each observational program. Please note the limitations sequenced observations listed in Section 8.6.1 on page 30.

                         NICMOS Proposal Instructions (Draft 7/5/94)
CONFIG  |   Mode    | Optional Parameters        | Comments 
NIC1    |   ACCUM   | NREAD = 1 - 25             | Exp Time > NREAD * 0.2 sec 
NIC2    |           | BKG_OFFSET = FOM, SAM      |  
NIC3    |           | BKG_SCALE = 0 - ?          | 
        |           | FOMXPOS = -? to +? arcsec  | 
        |           | FOMYPOS = -? to +? arcsec  | 
        |MULTIACCUM | NSAMP = 1 - 10             | Exp Time > NSAMP * 0.2 sec 
        |           | SAMP_TIMES = ST(n) n=3-10  | 
        |           | BKG_OFFSET = FOM, SAM      |  
        |           | BKG_SCALE = 0 - ?          | 
        |           | FOMXPOS = -? to +? arcsec  | 
        |           | FOMYPOS = -? to +? arcsec  | 
        |   RAMP    | NSAMP = 3 - N              | Exp Time >   NSAMP *  
        |           | CR_ELIM = NO, YES          |           (5,10 or 15 sec) 
        |           | SATURATION = NO, YES       | 
        |           | BKG_OFFSET = FOM, SAM      |  
        |           | BKG_SCALE = 0 - ?          | 
        |           | FOMXPOS = -? to +? arcsec  | 
        |           | FOMYPOS = -? to +? arcsec  | 
        | BRIGHTOBJ |                            | Exposure Time < 0.1 sec 
        |           |                            | Obs Time < 1638.4 sec  
NIC2    |   ACQ     |                            | ACCUM readout (NREAD = 1) 
        |           |                            | Reposition target via SAM  
        |           |                            |    from FSW 
|NIC|   |  |ALIGN|  | | FOCUS = -? to +? steps|  | Move corrector mechanism 
        |           | | XTILT = -? to +? steps|  | No science data taken 
        |           | | YTILT = -? to +? steps|  | 
Items enclosed by  |    |  are used for engineering purposes only  
                           and are not available to GOs 
                       NICMOS Internal Calibration Targets 
       Config |  Mode  | Internal Targets 
        NIC1  | ACCUM  | DARK, FLAT1, FLAT2, FLAT3 
        NIC2  |        |  
        NIC3  |        |  

NICMOS Science Modes


ACCUM - The detector array reset (i.e. setting the detector pixels to their

bias levels) is followed by 1 - 25 (specified by the NREAD parameter) reads

of the initial pixel values which are averaged onboard to define the initial

signal level. After the exposure time has elapsed, the final pixels values

are read NREAD times and averaged onboard. The data downlinked is the

difference between the initial and final signal levels for each pixel.

MULTIACCUM - The detector reset is followed by a single read of the initial

pixel values. At intervals specified by the SAMP_TIMES parameter, the

detector pixels are read out nondestructively NSAMP number of times. The

last read of the detector array ends the exposure. All of the data from the

initial and subsequent detector readouts is downlinked to the ground.

RAMP - As in the case of the MULTIACCUM mode, the initial detector readout

is followed by NSAMP number of nondestructive readouts. For the RAMP mode,

the intermediate readouts are at equal intervals during the exposure and

are not individually downlinked to the ground. Instead, the intermediate

signals are used onboard to calculate the slope of the signal vs time for

each pixel. The onboard calculation of the slope will also detect pixel

saturation and cosmic rays which will be flagged in the downlinked data

along with the slope up to that point.

BRIGHTOBJ - This readout mode must be used for very bright objects which

would saturate the detector pixels in less time than it takes to read through

the entire detector array. In this mode, each pixel is reset and then read

twice in succession with the interval between the reads being the exposure

time and the difference between the pixel values being the pixel intensity.

This process is performed on each pixel in succession, however each quadrant

of the detector can be operated in this mode simultaneously. Thus the

observation time for such an exposure is 16384 (128x128) times the exposure


NICMOS Support Modes


ACQ - Acquisition mode exposures are used to locate the target in the NICMOS

FOV for subsequent science exposures which are identified via the special

requirements: EARLY ACQ FOR <exp IDs>, INTeractive ACQ FOR <exp IDs>, and

ONBOARD ACQ FOR <exp IDs>. For the latter two cases, the HST is repositioned

to place the target in the aperture specified for the subsequent science

exposures. The target position and associated HST maneuver will be determined

via ground or onboard software respectively.

ALIGN - An engineering mode which is used to move the NICMOS internal focus

and tilt mechanisms via stored commanding for the purpose of on-orbit alignment

of the aberration correction optics. ALIGN mode exposures do not acquire

science data and must be interleaved with science mode exposures to obtain

data with the alignment mechanisms at a variety of positions for the ground

alignment process. The values of the optional parameters specify the positions

of the three mechanisms in motor steps relative to a nominal zero position.

The position corresponding to the stored commanding nominal zero is set via

real-time command which allows the specification of a preplanned scan of a

mechanism's position without knowledge of where in the mechanism range the

scan will be executed.

NICMOS Optional Parameters


NREAD - Specifies the number of detector readouts to be used in determining

the initial and final pixels values onboard for ACCUM mode exposures.

NSAMP - Specifies the number of intermediate nondestructive readouts for

MULTIACCUM and RAMP mode exposures.

SAMP_TIMES - Specifies the intervals at which the NSAMP nondestructive readouts

are performed in MULTIACCUM mode exposures.

CR_ELIM - Specifies if the onboard algorithm for cosmic ray detection is

to be used for RAMP mode exposures.

SATURATION - Specifies if the onboard algorithm for pixel saturation detection

is to be used for RAMP mode exposures.

BKG_SCALE - Specifies a scale factor (0 - ?) by which the predefined pattern

of FOV offsets for median background filtering is scaled. The number of

offsets will be a fixed value (>16) and there will be a different pattern for

each camera. A separate exposure as specified by the exposure logsheet line

is taken at each of the FOV offset positions. If BKG_SCALE is zero, the FOV

is not moved and a single exposure is taken.

BKG_OFFSET - If BKG_SCALE is non-zero, this optional parameter specifies if

the pattern of FOV offsets is to be performed via the NICMOS FOM or HST SAMs.

FOMXPOS - Specifies the offset of the NICMOS FOV using the NICMOS FOM in

FOMYPOS arcsec on the sky for the exposure line on which it is specified.

These parameters allow the proposer to specify the pattern of FOM FOV offsets

for a background filtering observation. They cannot be specified in

conjunction with the BKG_OFFSET and BKG_SCALE parameters.

FOCUS - Specifies the relative position of the NICMOS focus mechanism in

motor steps.

XTILT - Specifies the relative position of the NICMOS X or Y tilt alignment

YTILT mechanism in motor steps.


11.1 Eric Becklin

11.2 John Black

11.3 Harland Epps

11.4 Ed Erickson

11.5 Fred Gillett

11.6 Don Hall

11.7 John Hill

11.8 Frank Low

11.9 Don McCarthy

11.10 John McGraw

11.11 Marcia Rieke

11.12 Nick Scoville

11.13 Brad Smith

11.14 Hy Spinrad

11.15 Rich Terrille

11.16 Rodger Thompson


Recently Tonry and colaborators (Tonry and Schhneider 1988, Tonry, Ajhar, and Luppino 1989) began utilizing the fluctuations in galaxy surface-brightness as a distance indicator. The concept uses the fluctuations to measure the average brightness of the stars in the galaxy, then compares the measurement to the average brightness predicted from a luminosity function for the galaxy. The ratio then directly gives the distance. For ground based observations only visible wavelengths are appropriate. Near infrared images are dominated by spatial and temporal variations in the OH and thermal backgrounds. Mid and far infrared observations have thermal dust emission, rather than stars, as their main component. Active optic techniques will not alleviate these problems Near Infrared Advantages

Once rid of the telluric backgrounds, however, the near infrared has several advantages in the surface-brightness determination of galaxy distances. Two advantages are observational. First, the intrinsic stellar spectrum of elliptical galaxies (the type used for this study) peaks in the near infrared. Second brightness fluctuations due to variegated extinction is far less in the near infrared than in the visible regions. The near infrared also has a distinct theoretical advantage. The final result rests on the accuracy of our knowledge of the galaxy's luminosity function. The near infrared luminosity function is far less sensitive to evolutionary effects than the visible luminosity function. The combination of these factors make NICMOS by far the most powerful practitioner of this technique and give it a unique opportunity to carryout these studies. HST Advantages

HST, especially with internally corrected instruments, has a very stable and known Point Spread Function (PSF), which makes the analysis much easier since the power spectrum of the PSF is the signature of true signal. The stability also greatly enhances the photometric accuracy of the method which translates directly into increased accuracy of the determination. The small size of the HST pixels also increases the magnitude of the fluctuation to flux ratio by reducing the number of stars per pixel for any given galaxy. Basic Concepts

In an area of galaxy defined by a camera pixel the average flux is Nfavg where N is the number of stars in the pixel and favg is the luminosity averaged flux of the stars, usually equal to the flux on the ordinary giant branch of the galaxy. The fluctuation in the flux from pixel to pixel in a local region is due to the statistical fluctuations in the number of stars in a pixel and should be equal to favg. The variance is then Nf2avg and the ratio of the variance to flux is just favg. If the absolute average flux is known from galaxy models then the distance is determined from the ratio. Techniques

The following flow chart gives some of the techniques and science inputs to the analysis of the data.

This is text below the split columns to see it we can keep the text flow going.

Proposed Objects for Observation Proposed Objects for Observation Sensitivity, Signal-to-Noise, and Required Time Calculations References

Tonry, J. and Schneider, D. 1988, A.J., 96, 807.

Tonry, J., Ajhar, E., and Luppino, G., 1990, A.J., 100, 1416.

11.17 Ray Weymann

11.18 Erick Young

11.19 Joint Programs



For those with access to mosaic there is a NICMOS information area maintained by the Space Telescope Science Institute at

12.2 NICMOS Team Members

TABLE 11. NICMOS Team Members

12.3 List of Filters for Each Camera




12.4 Sensitivity

12.5 Encircled Energy

12.6 Point Spread Function

The charts in this section show the fraction of energy falling in each pixel for monochromatic point spread functions (PSFs) at 5 wavelengths (1.0, 1.4, 1.75,2.15, and 2.5 mm). The charts marked center put the center of the PSF on the center of the central pixel, those marked offset put the center of the PSF at the intersection of the corners of four pixels. These cases indicate the extremes of the possible alignments. Refer to these charts for sensitivity and encircled energy calculations. For those who wish a digital copy of these charts, please contact Rodger Thompson.

Camera 1 Centered 1.0 Microns

Camera 1 Centered 1.4 Microns

Camera 1 Centered 1.75 Microns

Camera 1 Centered 2.15 Microns

Camera 1 Centered 2.5 Microns

Camera 1 Offset 1.0 Microns

Camera 1 Offset 1.4 Microns

Camera 1 Offset 1.75 Microns

Camera 1 Offset 2.15 Microns

Camera 1 Offset 2.5 Microns

Camera 2 Centered 1.0 Microns

Camera 2 Centered 1.4 Microns

Camera 2 Centered 1.75 Microns

Camera 2 Centered 2.15 Microns

Camera 2 Centered 2.5 Microns

Camera 2 Offset 1.0 Microns

Camera 2 Offset 1.4 Microns

Camera 2 Offset 1.75 Microns

Camera 2 Offset 2.15 Microns

Camera 2 Offset 2.5 Microns

Camera 3 Centered 1.0 Microns

Camera 3 Centered 1.4 Microns

Camera 3 Centered 1.75 Microns

Camera 3 Centered 2.15 Microns

Camera 3 Centered 2.5 Microns

Camera 3 Offset 1.0 Microns

Camera 3 Offset 1.4 Microns

Camera 3 Offset 1.75 Microns

Camera 3 Offset 2.15 Microns

Camera 3 Offset 2.5 Microns