4. Capítulos de resultados
4.3 Fase 3: Validar los resultados del algoritmo de análisis técnico en el mercado de divisas
4.3.4 USDCAD:
3.5.1. 28-29 March 2012
The N IC T Rayleigh lidar w as operated on the night o f 2 8 -2 9 M arch 2012 from 2215 LST through 0442 LST and is first night o f m easurem ents using the high pow er Pow erlite 9030 N d:Y A G laser (PL9030). From 2116-2358 LST, m easurem ents w ere taken w ith the PL8020 steered into the inch telescope. The PL8020 w as beam steered into the
24-inch telescope by optim izing signal in the 6 0-65 km altitude range. The 41-24-inch telescope w as then steered into the laser beam and bore-sighted by m axim izing signals in the 60-65 km altitude range. The tw o telescopes took synchronized data sets from 2215 -2 35 8 LST w ith the PL8020 (M R 28R Y .005-M R 28R Y .011, M R 28R Z.001-M R 28R Z .007). The PL9030 w as then visually steered into the PL8020 in the sky. The steering procedure w as repeated by optim izing signals in the 60 -65 km altitude range in both telescopes w ith the PL9030. The telescopes took synchronized data sets from 0035-0442 LST w ith the photon counting error profiles are shown in Figure 3.14. The signal level per-laser-pulse is plotted as a function o f set (16,000 laser pulses) in Figure 3.15. The full lidar signal statistics are given in A ppendix A. The key system param eters are given in Table 3.6.
Table 3.6 . R ayleigh lidar acquisition param eters for 2 8 -2 9 M arch 2012.
Date Telescope
24-inch PL8020 0.3 2258-2358 8/1-16/16 64,000 0.891 1.0 03/28/12 Cassegrain
41-inch PL9030 1.0 0035-0111 8/1-16/16 64,000 3.64 2.0
The N IC T system yields a signal o f 0.891 photon counts per-laser-pulse in the 60 -65 km altitude range. The extended system yields a signal o f 3.64 photon counts per-laser-pulse in the 60 -65 km altitude range, a factor o f 6.46 tim es that o f the N IC T system. From 40
45 km, the N IC T system signal w as 29.9 photon counts per-laser-pulse and the N IC T
system signal w as 108 photon counts per-laser-pulse, 3.61 tim es larger than the N ICT because signals are only integrated over 64,000 laser pulses. It is clear in Figure 3.14 that the extended system signals are m ore statistically robust than the N IC T system signals telescope w ith PL9030 laser (green solid line) and 24-inch telescope w ith PL8020 laser (gray line w ith open squares).
(a) 28 - 29 M arch 2012 M arch 2012 (2215-0442 LST). Signals m easured through the (a) 24-inch telescope and (b) 41-inch telescope.
telescope by m axim izing signal in the 60 km to 65 km altitude range. The first tw o and last tw o sets represent the signal from the PL8020 and PL9030, respectively. A schem atic o f the m ethod o f data acquisition is shown in Figure 3.17. The specific sets m easured w ith the PL8020 and PL9030 are listed in A ppendix A. The data w as processed to yield individual m easurem ents w ith each laser by using bad files (A R8020RZ.B A D and A R9030RZ.BA D ) that are read by the data processing program s (see W ang [2003] for a discussion o f the processing method). The key param eters are shown in Table 3.7 and the full lidar signal statistics are given in A ppendix A. The total lidar signal and photon counting error are shown in Figure 3.18.
The total signal per-laser-pulse m easured on 4 April 2012 (Figure 3.19a) clearly shows the difference in the tw o lasers. Sky conditions are stable over the course o f observations, w ith background signals rising at set 15 in Figure 3.19a, corresponding to a sun elevation angle o f approxim ately -1 2 ° [USNO, 2012]. Betw een the 40 and 80 km altitude range, the ratio o f signals stabilizes at approxim ately 2.4. B elow ~40 km the signal is dom inated by pulse pileup and above ~90 km signals are dom inated by background skylight.
Time
□ P L 8 0 2 0 □ P L 9 0 3 0
Integrated photon count profile
--- ►
Integrated photon count profile
Figure 3.17. Schem atic illustration o f real tim e data set acquisition and processing m ethod on the night o f 4 April 2012.
Table 3.7. R ayleigh lidar data acquisition param eters for 4 April 2012.
The ratio o f the PL9030 lidar signal to the PL8020 lidar signal at 40 km based on a 2 km average is 2.31. The ratio increases to 2.37 at 45 km then rem ains at 2.38 betw een 50 km and 65 km. The constant ratio above 50 km is confirm ation that steering the laser beam into the telescope FO V rather than steering the telescope into the laser beam yields better alignment. The signal (integrated over 144,000 laser pulses) at 40 km based on a 2 km average is 1.82*105 photon counts and gives a count rate o f 2.53*106 counts/s m easured w ith the PL8020. U sing a conservative estim ate o f 8.7 ns for the detector dead time, the observed count rate gives a value o f x = 2 .2 0 * 10'2 or ~ 0 .02, and gives a 0 .02%
probability o f pulse pileup in the PL8020 and 41-inch telescope system. This results in the observed count rate being 98% o f the true or expected count rate for both a paralyzable and non-paralyzable detector system. The signal at 40 km m easured w ith the PL9030 and 41-inch telescope system is 4 .2 U 1 0 5 photon counts and gives a count rate o f 5.85*106 counts/s. U sing a detector dead tim e o f 8.7 ns, the observed count rate results in x = 5.09*10'2 or ~0.05. From Figure 3.4, at x = 0.05 there is a 0.12% probability o f pulse pileup in the PL9030. F or a paralyzable and non-paralyzable detector, the observed count rate is 95% o f the true or expected count rate.
N onlinearities in the raw photon count signal corrupt the tem perature because the lidar retrieval m ethod interprets the reduced signal as a decrease in atm ospheric density. The lidar retrieval m ethod depends on the assum ption o f hydrostatic equilibrium and uses the ideal gas law. Thus, low er densities appear as higher tem peratures. The tem perature profiles generated from the lidar signals integrated over 144,000 laser pulses is plotted in Figure 3.20a and the absolute value o f the tem perature difference is plotted in Figure 3.20b.
4 April 2012 (b) 4 2012
variability, the tem perature retrieved from the PL8020 data represents the average over a longer period o f time. A t 80 km, the 2 km averaged lidar signal m easured by the PL9030 and 41-inch telescope w as 5.18*102 photon counts, 3.00*102 photon counts m ore than the lidar signal o f 2.18*102 photon counts m easured by the PL8020 and 41-inch telescope. I attribute the difference o f 5.7 K at the low est altitude to signal contam ination from pulse pileup. A t 40 km, the 2 km averaged lidar signal m easured by the PL9030 and 41-inch telescope w as 4.21*105 photon counts, 2.39*105 photon counts greater than the lidar signal o f 1.82*105 photon counts m easured by the PL8020 and 41-inch telescope.
The w arm er tem perature due to the nonlinear lidar signals at low altitudes underscores the im portance o f correcting for pulse pileup. I f left unresolved, this w ould lead to a bias in reported tem peratures. The m agnitude o f the tem perature difference is significant, especially w hen w arm ing and cooling trends in the m iddle atm osphere are on the scale o f 1-2 K per decade. For example, lidar observations from the O bservatoire de H aute Provence (O HP) in France (44°N, 6°E) over a 20 year period yielded a cooling o f 0.4 K per year in the m esosphere and 0.1 K per year in the stratosphere [K eckhut et al., 1995].
3.6. Summary
In this Chapter, I have presented field tests from telescope trials in February, M arch, and April 2012, and laser trials in M arch and April 2012. These field trials w ere carried out w hile supporting the continuous routine operation o f the N IC T R ayleigh lidar system at Chatanika. The field test have allow ed m e characterize the effects o f nonlinearities in the photon counting receiver and determ ine an experim ental range o f detector dead tim es in the system. Based on tw o nights o f observations, the detector dead tim e estim ates ranged from 7.0 to 8.7 ns, larger than the 5 ns pulse duration o f the PM T and the 6.7 ns counting interval o f the M CS unit.
Furtherm ore, the field tests have allow ed me to characterize the perform ance o f the extended lidar system and determ ine the upper altitude accessible for tem perature retrievals. Table 3.8 sum m arizes the lidar signal integrated over 144,000 laser pulses
(64,000 laser pulses on 2 8 -2 9 M arch 2012) and the relative photon signal error from these field tests. Table 3.9 is the average 2 h lidar signals and relative photon signal error. From Table 3.9, the lidar signal m easured w ith the PL9030 and 41-inch telescope shows an im provem ent o f 6.5, 6.7, 6 .6, and 7.4 tim es the lidar signal m easured w ith the PL8020 and 24-inch telescope over the 60 -65, 70-75 , 80-85, and 9 0-95 km altitude range, respectively. The im provem ent o f 5.9 over the 4 0-4 5 km altitude range is contam inated by pulse pileup and therefore does not reflect the accurate im provement.
The m easured im provem ents are less than the expected im provem ent w ith the extended lidar system o f 8.4 com pared to the N IC T lidar system and reflect the system atic differences in system optics and electronics.
Table 3.8. Perform ance o f the N IC T and extended lidar systems.
Date Laser Telescope
NS (Photon Count)1/ANS / NS (%)
40-45 km 60-65 km 70-75 km 80-85 km 90-95 km 2/18
19/12 pps22 0 2 24-inch 53954/4.3 x10-1 1505/2.6x10° 285/5.9x100 49/1.5x10' 7/4.3 x101 3/28
24/12 pps220 2 41-inch 41-inch 152782/2.6x10-1 6042/1.3 x100 1154/3.0x100 173/8.4x100 26/3.1x101 4/24
25/12 pps220 2 41-inch 41-inch 168617/2.4 x10-1 6635/1.2x100 1262/2.8 x100 192/7.7x100 26/2.8 x101 4/03
1: Average of the signals presented in Table 3.6 scaled to signals integrated over 2 h.
These field tests establish that the upper altitude o f 80 km previously used for tem perature retrievals [e.g., Thurairajah et al., 2009] can be extended by 10 km.
However, the lidar signal at the low est altitude o f tem perature retrieval (40 km) is contam inated by pulse pileup and requires further developm ent o f a m ethod to system atically correct for this. These tests have dem onstrated that a system em ploying a dual telescope receiving system [Alpers et al., 2004] to detect signals from the low er atm osphere at reduced rates is not practical. B oth telescopes cannot be aligned w ith sufficient accuracy to ensure bore-sighting o f both telescopes w ith the laser. V arious m ethods have been developed to correct for count loss due to pulse pileup. These include having tw o channels w ithin the receiver to separate the incom ing light from low er altitudes (e.g., below 50 km) and from the higher altitudes (e.g., above 50 km ) [von Zahn et al., 2000], and com bining analog-to-digital (AD) and photon counting signal detection in the PM T [Liu et al., 2009].
The m axim um count rate for the receiver system is found using the experim entally determ ined detector dead tim e o f 8.7 ns and given an acceptable linear operating threshold. For exam ple, signals m easured by a system w ith a detector dead tim e o f 8.7 ns and w ith a 99.9% probability o f no pulse pileup correspond to the m axim um count rate o f 5.17*106 counts/s. To have 99.99% confidence that the signal is free o f pulse pileup, the m axim um count rate is 1.95*106 counts/s. To estim ate the true count rate from the 40 -45 km altitude range in the extended system, I scale the signal in Table 3.9 o f 312207 photon counts by the ratio o f the extended system lidar signal to the original lidar system from 90 -95 km (i.e. 7.4/5.9) to yield 3.92*105 photon counts. The scaled lidar signal o f 3.92*105 photon counts yields a count rate o f 5.44*106 counts/s, 1.05 tim es larger than the m axim um count rate o f 5.17*106 counts/s determ ined by the 99.9% threshold.
Therefore, if a beam splitter w ere placed before the PM T that directed 10% o f the incident light to another PM T, the count rate w ould fall below the 99.9% threshold. A 10% reduction to the incom ing signal w ould reduce the 80 km signal to 3.28*102 photon counts and w ould still allow for the upper altitude to be extended by 10 km.