INSTRUMENT TEST REPORT NUMBER 67 Calibration of Paroscientific Model 05 Pressure Sensor for use at Heard Island. Kent Gregory & Paul Huysing Physics Laboratory, OEB 15 July, 1994 Authorisation Jane Warne Senior Physicist Physics Laboratory 7 pages including 1 appendix page 1 of 7 ITR67
1. INTRODUCTION Prior to despatching Platform Transmitter Terminal (PTT) number 6741 to Heard Island, the Paroscientific pressure sensor model no. 05 serial no. 5550G (hereinafter pressure sensor ) had its calibration checked by the Physics Laboratory. The aim of the tests was to find if the pressure measurement uncertainty was within the Bureau of Meteorology's requirement of ±0.5 hpa for pressures from 900 hpa to 1050 hpa and temperatures -10 C to 10 C [1]. Heard Island is a sub-antarctic station positioned at longitude 53 South, latitude 73 East, and experiences mean air temperatures between -1 C to 3.5 C throughout the year []. The PTT will be installed about 5 metres above sea level and will transmit measurements once per minute to the Bureau of Meteorology via ARGOS.. EXPERIMENTAL METHOD The intended site for the PTT meant the pressure sensor would be operating in low temperatures, centred around 0 C. Therefore, the pressure sensor was placed in a refrigerator at 0 C for 17 hours prior to and during testing. The pressure sensor and the reference barometer were connected to pumps via hosing. The pressure was cycled once between 900 and 1050 hpa in 5 hpa steps and allowed at least 4 minutes to stabilise before each series of readings was taken. Pressure measurements were transmitted from the PTT to an ARGOS receiver every 60 seconds. The same method was used to calibrate the PTT at 3 C. A total of 47 readings at 0 C and 33 readings at 3 C were taken. The number of readings varied because some transmitted PTT messages were not received by the ARGOS receiver. A personal computer was used to collect the readings from the reference barometer (direct connection via IEEE-488) and the PTT (serial connection to an ARGOS receiver). The PTT transmitted a coded string of hexadecimal numbers which represented the various physical measurements. The pressure was coded as the first two numbers of this string. These numbers were converted into a pressure reading by the following formula; P = 900 + ((W1x4) + (W / 64)) x 0.15...(1) where P = pressure in hpa W1 = first transmitted number (integer from 0 to 55) W = second transmitted number (0,64,18, or 19). The formula limits the resolution of pressure measurements to 0.15 hpa. Furthermore, the formula is only applicable for pressures in the range 900 hpa to 1050 hpa and cannot generate pressure values above or below these limits. page of 7 ITR67
The reference pressure barometer used in these tests was a Druck (DPI140) Digital Pressure Indicator (s/n 79-89/) with a total uncertainty of ±0.08 hpa 1 directly traceable to the WMO RAV standard. 3. RESULTS The data from the tests appear in Appendix A. Figures 1 and are plots of the reference barometer readings versus the corrections of the pressure sensor at 0 C and 3 C respectively. The measurements made near 900 hpa at both temperatures were not used in the analysis (explained in Section 4.) Table 1 contains the mean and standard deviation of the corrections at 0 C and 3 C. U 95 is the total uncertainty of the pressure sensor measurements and was generated using the following formula; 95 σ corrections) + (U ref ) U 95 = ( ) t + (mean of where U 95 t 95 σ U ref = the total uncertainty of the pressure sensor measurements over the range 95-1050 hpa. = the t statistic to produce a 95% level of confidence = the standard deviation of the corrections (estimate of the standard deviation of the entire population) = uncertainty of the reference sensor (0.08 hpa). Table 1. Mean and standard deviation of the corrections, and U 95 of the pressure sensor measurements. All measurements in hpa Temperature Mean Standard deviation U 95 0 C -0.11 0.18 0.46 3 C -0.0 0.116 0.5 Linear least squares regression was used to fit a straight line to each data set. The lines of best fit appear with the corrections in Figures 1 and. Results of the regression and U 95 values applicable to the corrected measurements have been calculated and appear in Table. 1 Uncertainty was calculated using the root sum square method to 95% confidence. In this report, a correction is defined as the reference sensor measurement minus the test sensor measurement. page 3 of 7 ITR67
Figure 1. PTT pressure sensor response at 0 C 0.5 0.4 0.3 0. 0.1 0-0.1-0. -0.3-0.4-0.5 900 95 950 975 1000 105 1050 Reference pressure (hpa) Figure. PTT pressure sensor response at 3 C 0.5 0.4 0.3 0. 0.1 0-0.1-0. -0.3-0.4-0.5 900 95 950 975 1000 105 1050 Reference pressure (hpa) Table. Results of regression analysis. page 4 of 7 ITR67
( m measured pressure[hpa] ) Correction [hpa] = + c 1000 Temperature m c r U 95 0 C -5.41 5.300 0.901 ±0.17 3 C -.074.011 0.771 ±0.14 4. DISCUSSION The pressure conversion algorithm used by the PTT limits the range of pressure values to between 900 and 1050 hpa. At pressures above the theoretical limit of the algorithm, the PTT returned the value of 1050 hpa. It is expected, therefore, that for pressures below 900 hpa the PTT will return the value of 900 hpa, however this was not confirmed. At 0 C, the correction of the pressure sensor near 900 hpa was 0.06 hpa and deviates significantly from the expected correction of 0.43 hpa. It appears that the deviation in correction is a consequence of the sensor and not the algorithm, implying that the behaviour of the sensing element is non-linear below 95 hpa (see Figure 1.). This means the linear regression results supplied in Table are only applicable in the pressure range 95 to 1050 hpa. The most applicable calibration temperature for the PTT pressure sensor was 0 C, according to the information provided in []. Therefore, the pressure measurements made by the PTT at Heard Is. should be taken to have an uncertainty of ±0.46 hpa. Note that this uncertainty does not include additional uncertainty due to drift. Based on the data collected, the uncertainty of the PTT pressure sensor at 3 C is almost half that of the uncertainty at 0 C (compare U 95 values in Table 1). The linear regression statistics show that if a line of best fit was applied there would be a significant improvement in accuracy at both temperatures. It is interesting to note that the ratio between m 0 C and m 3 C is approximately the same as the ratio between c 0 C and c 3 C. A consequence of this is that the two lines of best fit intersect at a correction value of almost 0 hpa (-0.014 hpa). This correlation is indicative of a residual temperature dependency of the pressure readings. The nature of this relationship was not determined due to the short time available for tests prior to deployment. The PTT manufacturers claim the barometer is temperature compensated by the electronics [3]. However, from the data, it appears that the compensation algorithm could be improved. The algorithm is built into the electronic processing equipment in the PTT and therefore can only be changed by the manufacturer. 5. CONCLUSION page 5 of 7 ITR67
The Paroscientific pressure sensor model no. 05 serial no. 5550G inside PTT 6741 conformed to the Bureau of Meteorology s accuracy requirements of ±0.5 hpa[1] for the limited conditions listed below; = the uncertainty is valid for pressures from 95 to 1050 hpa, and temperatures from 0 to 3 C, = the uncertainty result was correct at the time of testing but makes no allowance for drift of the pressure sensor. Pressure measurements made near the bounds of its operating range should be treated with care. Before the sensor is used in higher temperature or altitude areas, a complete calibration should be performed. The sensor was not tested for long term drift or for temperatures outside the range of 0 to 3 C and therefore no comment can be made on how long the pressure sensor will remain within acceptable accuracy limits. It is recommended that whenever possible, the pressure sensor be checked against a transfer standard barometer. 6. REFERENCES [1]. Guidance Specification (Functional) for a General Purpose Automatic Weather Station, Bureau of Meteorology Specification no. A659, June 1989. []. Selected Surface Climate Data for Antarctic Stations, July 1967, Bureau of Meteorology, page 7. [3]. Facsimile, 14 April 1993, Remote Sensor Systems. page 6 of 7 ITR67
Comparisons at 0 C All measurements in hpa PTT Ref Ref-PTT Time 900.30 900.176-0.14 10:06:58 900.60 900.577-0.03 10:07:58 901.35 901.364 0.014 10:09:58 901.80 901.739-0.061 10:10:58 90.5 90.136-0.114 10:11:58 90.55 90.504-0.046 10:1:58 903.00 90.879-0.11 10:13:58 95.50 95.777 0.77 10:40:58 95.80 96.095 0.95 10:41:58 96.10 96.399 0.99 10:4:58 96.40 96.688 0.88 10:43:58 96.70 97.004 0.304 10:44:58 950.10 950.08-0.018 10:49:58 950.5 950.336 0.086 10:50:58 950.55 950.575 0.05 10:51:58 974.10 974.063-0.037 10:59:58 974.10 974.31 0.131 11:00:58 974.5 974.39 0.14 11:01:58 974.40 974.557 0.157 11:0:58 974.70 974.710 0.010 11:03:58 974.70 974.863 0.163 11:04:58 974.85 975.01 0.16 11:05:58 1001.40 1001.9-0.110 11:09:58 1001.40 1001.30-0.100 11:10:58 1001.40 1001.31-0.090 11:11:58 1001.40 1001.3-0.080 11:1:58 1001.40 1001.33-0.070 11:13:58 105.5 105.04-0.10 11:1:58 105.5 104.90-0.350 11::58 105.10 104.77-0.330 11:3:58 104.95 104.6-0.330 11:4:58 104.80 104.48-0.30 11:5:58 104.65 104.34-0.310 11:6:58 104.50 104. -0.80 11:7:58 104.35 104.11-0.40 11:8:58 104.0 103.98-0.0 11:9:58 104.0 103.86-0.340 11:30:58 104.05 103.75-0.300 11:31:58 103.90 103.64-0.60 11:3:58 1050.75 1050.43-0.30 11:40:58 1050.45 1050.16-0.90 11:41:58 1050.30 1049.87-0.430 11:4:58 1050.00 1049.59-0.410 11:43:58 1049.70 1049.31-0.390 11:44:58 1049.40 1049.04-0.360 11:45:58 1049.10 1048.76-0.340 11:46:58 1048.80 1048.49-0.310 11:47:58 APPENDIX A Comparisons at 3 C All measurements in hpa PTT Ref Ref- PTT Time 901.0 901.63 0.063 10:44:4 901.35 901.444 0.094 10:45:4 901.50 901.608 0.108 10:46:4 901.80 901.954 0.154 10:48:4 97.75 97.81 0.071 10:5:4 97.75 97.945 0.195 10:53:4 97.90 98.080 0.180 10:54:4 98.0 98.343 0.143 10:56:4 949.95 949.888-0.06 11:05:4 949.95 949.979 0.09 11:06:4 949.95 950.088 0.138 11:07:4 950.5 950.198-0.05 11:08:4 950.5 950.304 0.054 11:09:4 975.15 975.144-0.006 11:14:4 975.30 975.0-0.098 11:15:4 975.30 975.67-0.033 11:16:4 975.30 975.333 0.033 11:17:4 975.45 975.401-0.049 11:18:4 1000.05 1000.01-0.040 11:3:4 1000.05 1000.01-0.040 11:4:4 1000.05 1000.0-0.030 11:5:4 1000.05 1000.03-0.00 11:6:4 1000.05 1000.03-0.00 11:7:4 105.40 105.5-0.150 11:36:4 105.5 105.17-0.080 11:37:4 105.5 105.09-0.160 11:38:4 105.10 105.0-0.080 11:39:4 105.10 104.93-0.170 11:40:4 1051.35 1051.13-0.0 11:47:4 1051.0 1050.96-0.40 11:48:4 1050.90 1050.78-0.10 11:49:4 1050.75 1050.61-0.140 11:50:4 1050.60 1050.44-0.160 11:51:4 page 7 of 7 ITR67