Post Seismic Sea level measurements, Haiti

5-28 March 2010

Installation and initial data

Roger Bilham

University of Colorado, Boulder CO 80302

Summary Five pressure-sensing sea level monitoring devices were installed in Haiti in two visits 5-9 th and 24-28th March 2010. Fifteen-minute sampled graphic and numerical data with a resolution of 1 mm are publically available at http://www.datagarrison.com, name "geo" password "hobo" within 2 hours of real time. The data provide a time history of post-seismic vertical adjustments following the 12 January Haiti earthquake, and provide an absolute datum from which to measure the elevation of coral microatolls. Preliminary results show the epicentral area t be rising still. The data also provide a precise template for examining the elevation of uplifted sea-level notches raised in this, and in former earthquakes.

Fig.1 Tide Gauges, epicenter (star) and the approximate location of the subsurface fault , Haiti

'Table 1 Location of tide gauges , Haiti

Site	Location	North	West	Transmitter	data logger	Stevens #	sample sec
GONA	Petite Gonave	18.70420	72.80496	Iridium	energy pro	193888	44s direct
CIRA	Ca Ira	18.52499	72.65077	Iridium	microstation	196728	180s piezo
BLOC	Beloc	18.47886	72.66727	none	microstation	196730	180s piezo
JACK	Gros Jacques	18.43027	72.74943	Iridium	energy pro	193887	44 s piezo
PARA	Jeanty les Baines	18.43168	72.76044	Iridium	microstation	196728	60 s direct

Scientific Motivation

The 12 January earthquake produced no surface rupture, consistent with Insar imagery and teleseismic inversion that suggest that rupture terminated 2-4 km below the surface. Three interpretations are possible: that slip in this 2 km shallow subsurface region occurred prior to the earthquake as aseismic creep, that slip is currently occuring in this region as afterslip, and that slip will be released in a future earthquake. The tide gauge measurements are designed to address the second two of these possibilities. The magnitude of this future earthquake would be Mw=6.5 if it consists of a 4x35 km patch of the Enriquillo fault that slips ≈2 m, or the apparent deficit could be released as part of a larger earthquake.

Tide Gauge Operation

The sensor in each tide gauge is a Stevens Pressure transducer moulded to a 10-13 m length of electrical/pressure-equalization cable vented to the atmosphere. The capillary vent permits sea floor water column pressure to be monitored independent of atmospheric pressure. A dessicant trap ensures that the capillary tube is not blocked by condensation. The resolution is 1 mm and the range 3 m. Power to the sensor is applied from a 18 volt lithium battery via a current loop for 30 ms every 1-3 minutes. The resulting samples are digitally averaged and recorded every 15 minutes by either a 12 bit Onset Energy Pro data logger or by an 12 bit Onset Microstation equipped with an A-D current converter. The data logger is interrogated every two hours by a Solar Stream transmitter that uploads the accumulated data to the Iridium satellite system which forwards the data to a publically accessible web site where the data are available in calibrated engineering units (cm of sea level rise) and local time (GMT-5 hours).

Data Access

https://datagarrison.com/ user--geo password--hobo Click on site name.

Fig 2 Schematic electronics for logging and transmission, and interior view of transmitter box connected to 10 m of transducer cable.

The system is designed to operate for a year unattended at temperatures up to 50°C.

Description of Tide Gauge Installations

See Table 1 for coordinates and general specifications. Each tide gauge is designed to operate for a year unattended with maximum 2 hour latency telemetry, with the exception of BLOC whose data must be retrieved manually. GONA and PARA measure sea level directly with minimal hydraulic filtering, and the central three are installed in beach sand close to the shore, and measure piezometric pressures that closely follow sea level with a filtering time constant of approximately two hours.

GONA, Petite Gonave, SE of the island of Gonave. 18.70420, 72.80496

The instrument is located 200 m north of the island village of Petite Gonave at the SE end of Gonave Island. The island lies outside the zone of deformation affected by the 12 January earthquake. The sensor is supported by a galvanized steel ell-girder bolted to a vertical face of in-situ bedrock in a small cleft in the coastal limestone. The gauge is sheltered from direct wave impacts, and faces the mainland from the small island on which Petite Gonave has been constructed. It has no hydraulic filter but samples sea level every 44 seconds, storing a digital average every 15 minutes which is transmitted every 120 minutes via the Iridium satellite. The limestone above sea level proved sufficiently hard to blunt tungsten-carbide drills and it was necessary to secure the cable without the benefit of clamping bolts through karst holes and cracks to protect it from the disturbance.

Fig. 3. The island and village of Petite Gonave lies at the SW tip of Gonave Island. At the latitude of the tide gage station Petite Gonave Island is approximately 100 km east west. The rocks of the island are a silicious limestone exhibiting at sea level a prominent, deep solution notch with no evidence for recent changes in sea level.

Fig. 4 Location of Petite Gonave Island and first three days of data from GONA

The transmitter is located 3 m above sea level on the exposed (knife-edged) eroded coral limestone. It uses a single box for data logger, transmitter and antenna, as illustrated in Figure 2, inclined at 45° to the south, and uses the 12 Volt internal supply of an Onset Energy Pro datalogger, and a bank of 4 AA lithium cells to generate the 18 V that powers the current loop. These batteries will last about 15 months, but should be replaced after a year (March 2011). The 6V solar panel powers only the iridium transmitter. Two channels of data are transmitted - the internal power of the data logger in volts and sea level converted at the tide gauge to sea level in cm of water density 1.03. The downloaded binary data are in GMT and the calibration constants can be examined when opened in Onset's Hoboware software. The transmitted voltage must exceed 12 V for correct operation. In principle the entire ystem could be arranged to run indefinitely from solar panels, however, its clock drift may reach 1 minute per month, and it is considered desirable that it be reset at yearly intervals. The instrument was started at 12:30 local 6 Mar 2010 with an accidental header of GMT-6 hours, and restarted with a GMT-5 hour header on 9 March.

CIRA near the village of Ca Ira, 18.52499°N, 72.65077W

The instrument is installed in the north corner of Ca Ira Orphanage west of Leogane. The microatolls 100 m offshore to the NW here were raised 20-30 cm in the earthquake. The sensor is bolted to a stainless steel bar that is clamped to a 60 cm long steel helical screw which reached a depth of 97 cm below the beach level in muddy sands without encountering resistance. A copper mesh was wrapped around the pressure port and this in turn was inserted into a porous foam inlet filter enclosed within a 4 cm diameter PVC jacket bolted to the base of the stainless bar. The height of the top of the stainless bar was referenced to a galvanized bolt hammered into the lintel of the doorway through the NW wall to the orphanage, and also to two offshore corals that had been sampled by the USGS (Rich Briggs).

The transmitter is located inside the protective fence of the orphanage and is mounted on awoodend plank at a height of about 2m. The current loop does not use the 6V power of the Onset microstation but instead uses the power from a pair of 9V lithium fire alarm batteries. These will power the system for two years but should be replaced annually along with the microstation batteries.

The first few days of data show that mean sea level is 83.6 cm above the zero pressure datum, i.e. 72.5 cm below the nearby bench mark, and 8.0 cm below the blue screw marker on the southern of the two microatolls, and 20.4 cm below the highest remaining point on the sampled northern microatoll. Given a tidal range of more than 36 cm and assuming that the highest coral was immersed at low tide suggests that the microatoll here has been raised by 18-20 cm.

Fig 4 First week of data from CIRA peizometric gauge, at Ca Ira, Haiti

BLOC , near the village of Beloc south of Leogane.

The tide gauge is installed near the village of Beloc 5 km to the south of Leogane, to the east of a colony of microatolls that have been raised more than 50 cm above sea level by the 12 january earthquake. The sensor was bolted to a stainless steel bar A copper mesh was wrapped around the pressure port and this in turn was inserted into a porous foam inlet filter enclosed within a 4 cm diameter PVC jacket bolted to the base of the stainless bar. The height of the top of the stainless bar was referenced to two 100 m offshore corals that had been sampled by the USGS (Rich Briggs) using a level and alluminium leveling staff.

The data logger is enclosed in a 4" PVC pipe sealed with a rubber gasket, and fastened using hose clamps vertically to a 1 m long angle iron driven into the berm above the pre-earthquake tidal level near a small village hut and thorn bushes. The atmospheric pressure equalisation tube faces downward, and the pipe is potentially vulnerable to vandalism. No transmitter was available for this unit. The batteries should be replaced in March 2011.

Fig 5. Corals 53 (18.4789N 72.66859 W) and 54 (18.47913,72.66882) are numbered 8 and 17

Leveling at Beloc: The pressure port was referenced to the offshore corals using a precise level with a 10 m back sight and a 120 m foresight. The asymmetry means that the data are probably not accurate to better than 5 mm.

Height of top of steel bar 1.00 m, pressure port zero is 57.9 cm below top of bar.

Height of B17 coral 1.05 m ie 5 cm below steel bar reference, and 52.9 cm about transducer zero

Height of B8 coral 0.449 m ie 55.1 cm above steel bar reference, 113.0 cm above transducer zero

JACK , South of Gros Jacques Island 18.43027N, 72.74943 W

The sensor was bolted to a galvanized steel angle girder bolted to the base of the east-facing end of a poorly constructed and partly damaged sea wall. The transducer port has no protective filter and behaved erratically in the first week of its installation. The blockage cured itself on 13 March, after which time it was possible to examine differnces in sea level relative to GONA. The smoothed difference curve indicates subsidence of 0.6 mm/day.

A bench mark was installed on the western end of the top step of three steps leading to the north entrance of the house immediately to the west of the transmitter. The top of the steel bar is 67.0 cm below the step.

The transmitter is fastened to the top of a steel pole cemented into the ground within a wire fence belonging to a private citizen on the mainland south of the Island of Gros Jacques. The house is 100 m to the west of that belonging to Richard Boyer.

Although the site is part of an extended delta that with swamplike conditions 100 m to the south, the subsurface is close to a layer of corals similar to those that have established the island of Grand Jacques offsore. Thus the headland is probably quite stable. However, it is backed by distributaries that presumably fill with fresh water during the monsoon, hence it is likely that mean sea level here is contaminated to some degree by heavy rain and by the transient development of a freshwater lens.

Fig 6 Top:Map views of the Peninsula and island. Below: South looking view of JACK from the sea . The transducer is fastened to the low sea wall to the left.

The transducer was installed in an excavated water-filled hole 56 cm below sea level. The support system was then buried by coarse beach sand and large rocks. The first cycle of sea level recorded a water level range of about 40 cm as the tide receded and returned, but for the next week the pressure port was blocked.

Fig 7 Differences in level between GONA and JACK, and between GONA and CIRA.

PARA, at Jeanty les Bains, Grande Guave, 18.43168, 72.76044

The instrument is located at the Paradis Beach Club at Jeanty les Baines (near Grande Guave). The transmitter is fastened to the roof of the restaurant and the sensor is bolted to the north-facing dock below it. It is possible to stay in the undamaged hotel of the Beach Club for $30/night. The proprieter is Confident Jean Herold, Phone 506 9977

The sensor was bolted to a stainless-steel bar fastened to a plastic-coated, concrete-filled barrel. A copper mesh was wrapped around the pressure port and this in turn was inserted into a porous foam inlet filter enclosed within a 4 cm diameter PVC jacket bolted to the base of the stainless bar. The PVC jacket lies approximately 10 cm below the level of cobbles and gravel that here form the sea floor. A coral wave-cut platform is exposed on the beach to the west, and lies at no great depth beneath the shingle deposit adjoining the dock.

The cable is bolted up to the roof and protected where it crosses the dock walkway by a galvanized angle iron strip iron hammered into the concrete with non-remavable bolts.

Fig 8 Pressure sensor bolted to concrete foundation support column at PARA. The cable is bolted to the wall at 6 points on its way to the data logger and transmitter on on the roof

Fig 9 Location of the PARA tide gauge at the Parasis Beach Club, Jeanty les Bains.

Calibration

The precision of the gauges (least count) is 1.2 mm (1.165 mm sea water). The range is 3m.

Figure 10 Laboratory intercalibration of two tide gauges to determine zero datum.

Table 2 Calibration test 28 Feb-2 March.

	actual cm	cm 193888	cm193887
zero	2.9	-0.5	-0.5
cal2	28.2	25.12	24.29
cal 1	36.7	33.58	32.38
zero above base	2.95	2.95	2.55
calibration fresh water	-	0.9953	0.9518
calibration 1.03 gm/cc	-	1.025	0.9804

The gauges were fastened together in the lab on a steel rod and left in air for 12 hours recording an apparent water depth of -0.5 cm. They were then plunged into water at 20°C 36.7 cm deep for a day, which was reduced to 28.2 cm for half a day before removing them and exposing them to air pressure once more. The experiment reveals a datum offset of 2.95-2.55 cm caused by the elevated position of the port (zero datum) above the base of the transducer (measured as 2.95 cm), and a calibration error of 4% in 193887. The calibration error and an adjustment for sea-level density of 1.03 gm/cm³ is removed in the data displayed on the DataGarrison web page. The last two rows of Table 2 and the last two columns of Table 3 indicate the calibration applied to the pressure sensors.

Table3 Calibrations off three Stevens gauges sampled by ONSET A-D 4-20 mA current loop converters. The last two columns are the calibrations needed to correct the 4-20 mA range to 0-10 feet scaling for fresh water. They should be increased by 1.03 for sea water.

	a mA	a± mA	b mA/cm	b ± mA/cm	zero	cm/mA
600 196728	3.8943	0.041	.05075	0.0007	-76.735	19.7044
729 196780	3.9683	0.017	.05023	0.00031	-79.002	19.9084
196730	3.9872	0.063	.04972	0-.0011	-80.193	20.1126

Long term stability is anticipated to be 1mm/year but it is planned to test this experimentally by annual in-situ exposure to atmospheric pressure and recalibration of the gages.

Fig. 11 Calibration of three tansducers simultaneously in 1 m long vertical PVC tube. The transducers are fastened to a vertical bar which is lowered to various depths in the tube, which is maintained full of water. The plot of electrical output vs water depth yields both the calibration of the sensor and the null datum (the intersect on the graph). This permits absolute measurements of water depth referred to the input orifice of the transducer.