HUYGENS Radar Altimeter
Digital Altitude Data Calibration
V3.1 
May 2005

Written by
    R. Trautner		RSSD		ESTEC

Re-formatted in ASCII by
    O. Witasse          RSSD            ESTEC

Approved by
    H. Svedhem		Radar Altimeter Science	ESTEC
    P. Couzin		Huygens System		Alcatel
    T. Blancquaert	Huygens System		ESTEC
    A.M. Schipper	Huygens Programme	Alcatel 
    J-P. Lebreton	Huygens Mission		ESTEC


Table of Contents

1	TABLE OF CONTENTS	
2	FIGURES	
3	DOCUMENT CHANGE RECORD	
4	INTRODUCTION	
5	RAW DIGITAL INTERFACE HRA DATA	
6	PWA ALTITUDE DATA	
7	TEMPERATURE AND ALTITUDE RELATED ERRORS	
7.1	Test data from FS balloon flight in Brazil	
7.2	Tests performed on FM and FS models (HRA ADP data)	
7.3	Altitude Error Correction Method	
7.4	Huygens Platform Temperature	
8	PROBE ATTITUDE AND RELATED HRA MEASUREMENT ERROR	
9	DIGITAL ALTITUDE DATA ERROR CORRECTION	
9.1	Assumptions	
9.2	Results	
10	CONCLUSION	
11	ACRONYMS	
12	REFERENCES	


Figures(see the PDF version)

Figure 1: Raw digital interface altitude data	
Figure 2: Digital interface altitude data after digital error correction	
Figure 3: FS B temperature related altitude error (PEASMA/Brazil)	
Figure 4: FS B temperature related relative altitude error at 33.3 km	
Figure 5: FS B temperature related altitude correction factor (parameter: altitude)	
Figure 6: FS B relative altitude error vs. altitude (parameter: temperature)	
Figure 7: Calculated (red) and measured (blue/pink) altitude correction factor	
Figure 8: Huygens platform temperatures and sensor locations	
Figure 9: Relative altitude measurement error vs. probe tilt angle	
Figure 10: Uncorrected and corrected digital HRA data	
Figure 11: Uncorrected and corrected digital HRA data - detail 1	
Figure 12: Uncorrected and corrected digital HRA data - detail 2
Figure 13: Uncorrected and corrected digital HRA data - detail 3	
Figure 14: Difference HRA A / HRA B before and after calibration	


Introduction

The Huygens Radar Altimeters (HRAs) have provided digital altitude information during the descent of the Huygens probe. 
While the digital data of both radars has been acquired via the probe telemetry, the additional data acquired by HASI-PWA 
is only available for channel B due to the loss of the telemetry data on channel A.

The digital data has been affected by a hardware bug, which caused the upper bits of the digital 15-bit altitude 
word to change their logical state in a seemingly random fashion. In addition, due to the 15-bit limitation of 
the digital altitude word any radar data above 32767 m is affected by a register overflow. In addition to these 
digital data errors, the radar has been found to be sensitive to temperature changes, and is affected by systematic 
errors that depend on the probe altitude. The raw digital altitude data has been processed in order to eliminate 
these known errors and to calibrate the measured altitude. After a first calibration using initial estimations of 
the radar temperatures a temperature sensitivity analysis and data assessment showed that more accurate 
estimates of the radar temperature were required. This data has been made available by ALCATEL and ESTEC. 
The present document provides an updated description of the processing steps and calibration data involved in the 
correction of systematic errors and in the data calibration, as well as information on the accuracy of the delivered dataset.

 
Raw digital interface HRA data

As described in the introduction, the raw digital data is affected by some problems corrupting individual bits of the data, 
and / or causing a register overflow. The same behaviour was observed on the flight spare units that were flown on a CNRS/CNES 
balloon flight in Brazil in December 2004. The following figure shows the unprocessed digital HRA data.

Figure 1: Raw digital interface altitude data

In a first processing step, the bit flip and overflow errors were corrected in order to obtain a consistent digital dataset. 
This was done using techniques developed for the processing of the data obtained from the December 2004 balloon test flight. 
Figure 2 shows the corrected dataset (only the data points where the radar was locked).

Figure 2: Digital interface altitude data after digital error correction


PWA Altitude Data

The digital altitude data provided by the HASI-PWA instrument was generated by measuring the duration of rising and falling 
edges of the radar ramp signal via the corresponding blanking signal. Although the quartz crystal oscillator on board of 
PWA would in principle allow a more accurate measurement than the RC oscillator in the radars, the radars have been 
calibrated using data from their digital outputs. There is no altitude calibration data available for the PWA dataset. 
The balloon test flight data shows that the PWA altitude data is equally affected by the temperature and altitude 
related systematic errors of the radar. Therefore it is preferable to use only HRA digital data for further processing. 

Temperature and Altitude related Errors
Test data from FS balloon flight in Brazil 

One of the main results of the Brazil test flight was that there is a strong impact of the radar temperature on the altitude 
measurement accuracy. In Brazil, altitude errors as high as -13.5% have been observed for platform temperatures of 47 deg C. 
The temperature and altitude error data obtained for the FS models exhibit a linear relationship, which is similar but not equal 
for both FS radars. Figure 3 shows the raw (red), true (green) and corrected (blue) altitude data for radar B on the Brazil flight.

Figure 3: FS B temperature related altitude error (PEASMA/Brazil)

During the Brazil test flight, the balloon remained at an altitude of ~ 33km most of the time. Therefore the temperature 
calibration information from the Brazil test is valid for that altitude. The temperature sensor was accommodated on the 
aluminium sheet of the payload platform, which was a good thermal conductor. Although the sensor was not very close to 
the radars it is assumed that the aluminium sheet provided a sufficient thermal bridge to use the sensor data for the 
radar temperature. The test data shows a relative altitude error, which is decreasing with decreasing altitude, indicating 
that in addition there is also an altitude dependent error. The following plot shows the relative altitude error at ~ 33.3 km 
altitude vs. temperature as observed in Brazil (FS radar).

Figure 4: FS B temperature related relative altitude error at 33.3 km

It appears that for constant altitude the temperature error can be described by a linear relationship with good accuracy. 

Tests performed on FM and FS models (HRA ADP data)

An investigation of the FS and FM test data acquired during the flight unit qualification confirmed the observed 
temperature and altitude effects. During the qualification tests, the FM and FS models were connected to the return 
signal simulators (RSS), which were used to simulate a range of altitudes from 150 m up to 10 km. The simulators are 
based on up/down modulators and delay lines. Their altitude / delay accuracy is estimated to be in the order of 1% or 
better at simulated altitudes of 1000m or higher. 

The temperature of the radars was varied in a range of -40 deg C to +20 deg C, and the digital interface altitude reading 
was recorded for all available RSS altitude settings. Furthermore, the supply voltage was varied in order to allow assessing 
the impact of power supply changes on the accuracy.

Figure 5 shows the altitude correction factor for FS B for simulated altitudes of 3005, 6018 and 10020m as a function 
of temperature. The data obtained during the Brazil flight (true altitude 33300m) is added in the temperature range 10 
to 25 deg C. The Brazil data as well as the high temperature part of the ADP data show a linear relationship with temperature. 
This linear relationship is also expected for the FMs in the 10 deg C to 25 deg C temperature range which is relevant for 
Huygens data.

Figure 5: FS B temperature related altitude correction factor (parameter: altitude)
The data available in the ADP also allows assessing the relative altitude error vs. altitude at constant temperature. 
Figure 6 shows the FS B altitude correction factor for 2 temperatures (-11,3 and 18,85 deg C). One data point from Brazil 
test flight data is added for the +18.85 deg C temperature curve. 

Figure 6: FS B relative altitude error vs. altitude (parameter: temperature)

Just like the temperature related error, the altitude related relative altitude error shows a tendency which can be described 
by a linear relationship with reasonable accuracy.

Altitude Error Correction Method

The observed effects show that digital interface altitude data from Huygens needs to be corrected for both temperature and 
altitude related errors. The available data indicate that linear relationships of error vs. altitude / temperature can be 
used. The observed linear relations of correction factors and temperature / altitude are applied using the following equations:


Acorr_T= Ameas * k_Ta	for a given altitude a,


And

Acorr_A = Ameas * k_At	for a given temperature T


These need to be combined into a single expression, which takes into account the corrections for both altitude and temperature:

Acorr = Ameas * ( k0 + k_T*T + k_A*Ameas )

The factors K_A and K_B are given by

K_A = AcorrA / d A
K_T = AcorrT / d t 

From the FS Brazil test data we obtain


AcorrT_A = 1,012240 + T * 0,00194		(FS A, at 33300m)
AcorrT_B = 1,007106 + T * 0,00267		(FS B, at 33300m)


Averaging these relations yields


AcorrT = 1,009673 + T * 0,002305		(FS A + B, at 33300m true altitude) 



From ADP test data on the FS models (+12V / -8V supply voltages) we obtain


AcorrA_A = 1,0233 + A *  8,5260E-07	(FS A, 18,90 deg C)
AcorrA_B = 1,0235 + A *  1,1406E-06   	(FS B, 18,85 deg C)



Again, averaging these equations yields


AcorrA = 1,0234 + A * 9.966E-07	(FS A + B, at 18,88 deg C)


The test data from Brazil provides the overall correction factor Acorr for the true altitude of 33300m (measured altitude 31618m) 
and a temperature of 18,88 deg C. 

This allows the calculation of k0 and the completion of the overall correction formula: 


Acorr = Ameas * (0,97788 + 0,002305 * T + 9,966E-07 * Ameas)


While this correction method is based on the FS data (which is the most complete calibration dataset), it is also applicable for 
the FM models due to the similarity of the individual models. A comparison of the correction formula results and the FM test 
data in the ADP shows that the deviation in the [+20 deg C .. -10 §C] temperature interval are in the order of 1%. Figure 7 shows 
the measured and calculated altitude correction factor for radars FM A and FM B at room temperature. The calculated factor is 
in good agreement with the average of the measured factors for A and B.

Figure 7: Calculated (red) and measured (blue/pink) altitude correction factor
 
Huygens Platform Temperature

The Huygens payload platform remained in a comfortable temperature range during the whole mission. However, the platform 
temperature changed significantly during the descent. The Figure 8 shows the temperature of sensors D5017T - D5020T 
which were accommodated on the Huygens platform (HRA side). The sensor positions are indicated in the sketch below. 
Data from sensors closer to the radars (D5005T, D5007T) is not available due to the loss of telemetry chain A. Therefore 
the temperatures of the radars have been estimated using thermal models. The simulation results indicate temperatures 
which are much lower than the measured platform temperatures. The results are shown in Fig 8. 
For the calibration of HRA altitude data the temperature data from these simulations is used.

Figure 8: Huygens platform temperatures and sensor locations


Probe Attitude and related HRA Measurement Error

It is understood that the attitude of the Huygens probe has an impact on the accuracy of the HRA altitude data. 
A deviation from the nadir position leads to an altitude reading, which is higher than the true altitude. However, 
due to the beam width of the HRA antennas (~8 degrees) only deviations exceeding several degrees would lead to 
significant measurement errors. The following figure shows the relative altitude measurement error as a function 
of the probe tilt angle for a perfectly focused beam. Even under these worst-case conditions, the altitude error 
does not exceed 1% for tilt angles of 8 degrees or below.

Figure 9: Relative altitude measurement error vs. probe tilt angle

Therefore the altitude and temperature related errors will be the dominant effects for HRA altitude data. Attitude 
deviations can be taken into account as soon as attitude data becomes available.
 
Digital Altitude Data Error Correction
Assumptions

The correction of systematic HRA measurement errors is based on the following assumptions:

- The mechanism of the digital altitude data corruption is understood
- The FM and FS models radars are using the same design, and are built using identical components, and therefore show 
similar systematic errors
- The average attitude deviation from nadir does not exceed few degrees

These assumptions are confirmed by the following observations: The digital data corruption is the same observed in the 
test data obtained during the PEASMA stratospheric balloon flight. Correction algorithms have been developed and tested 
for the Brazil flight, and are applied to the Huygens data. The identity of the FS and FM design has been confirmed by 
the manufacturer. The similarity of FM and FS radars has been confirmed by the analysis of the data available in the 
ADP documents. All flight and flight spare radars show the same behaviour, which is a negative relative altitude error, 
which increases both with altitude and temperature. 

Results

After correcting the digital bit errors and the temperature and altitude related altitude error we obtain a dataset 
which indicates the probe with improved accuracy. The following figures show the corrected altitude data as a whole and in detail.

Figure 10: Uncorrected and corrected digital HRA data

Figure 11: Uncorrected and corrected digital HRA data - detail 1

Figure 12: Uncorrected and corrected digital HRA data - detail 2

Figure 13: Uncorrected and corrected digital HRA data - detail 3

Figure 14: Difference HRA A / HRA B before and after calibration

In general, the calculated altitude is increased with respect to the uncalibrated data during the first part of the descent. 
After ~ T0+5500 seconds the calculated altitude is lower. The difference of radar A and B altitudes is decreased significantly 
especially during the last part of the descent, providing evidence that the calibration actually improves the dataset.

Conclusion

The data obtained by the Huygens Radar Altimeters has been processed and improved by correcting the digital errors and by 
applying a temperature and altitude error compensation based on data from the FM ADP and FS model tests on the PEASMA 
balloon flight. The similarity of FS and FM models in terms of error characteristics has been confirmed and the deviation 
of the error model and ADP measurement data is in the order of 1% or better for all models. The estimated accuracy of the 
temperature data is 5 degC, corresponding to an altitude error of ~1.15 %. In the absence of attitude data, an additional 
altitude measurement error of 0.5% is assumed. Based on these assumptions, an analysis of the overall error of the calibrated 
altitude data indicated a relative error of 2.7%. This error corresponds to the accuracy of the derived altitude over terrain; 
for the reconstruction of the Huygens descent trajectory, the possible effect of varying terrain elevation needs to be taken 
into account. 

The calibrated data shows an altitude, which is increased by 5 % with respect to the original unprocessed data at T0+3000 sec. 
The deviation then decreases with decreasing altitude and reaches a minimum around T0+5500 sec. During the later part of the 
Huygens descent, the calibrated altitude is up to 4% lower than the  uncalibrated one. The agreement of radar A and radar B 
data has significantly improved in all parts of the descent. 

Further improvements of the dataset may be possible if the following data is made available:

- Data on the probe attitude during the descent
- Additional measurements of the altitude related error using available hardware (RSS and FS models)
- More accurate estimations of the radar temperatures during the descent

The corrected altitude dataset is made available in the following ASCII files:
HK_CDMS_RADAR_A_CORRECTED.TAB
HK_CDMS_RADAR_B_CORRECTED.TAB


Acronyms




ADP: Acceptance Data Package 
ALT: Altitude
ASCII
American standard code for information interchange
FM: Flight Model
FS: Flight Spare Model
HASI: Huygens atmospheric structure instrument
HRA:Huygens radar altimeter
PWA:Permittivity, wave and altimetry experiment
PEASMA: Name of the stratospheric balloon campaign performed in 12 2004
RA:Radar
RSS:Radar Return Signal Simulators