Continuous monitoring of fault-controlled CO2 degassing in the Los Humeros Volcanic Complex, Mexico

Anna Jentsch1,2,*, Walter Duesing2, Egbert Jolie1

1 Helmholtz Centre Potsdam GFZ German Research Centre For Geosciences, Potsdam, Germany

2 Institute of Geosciences, University of Potsdam, Potsdam, Germany

*ajentsch@gfz-potsdam.de

Characterization of significant temporal variations of soil CO2 flux in an active fault zone in relation to

▪ Environmental parameters (air temperature, barometric pressure, precipitation, relative humidity, wind speed, and wind direction)

▪ Natural/induced seismicity

▪ Geothermal exploitation (injection)

The latter two aspect play an important role to better understand the hydraulic connection and communication between subsurface and surface along structural discontinuities

Motivation

Monitoring networkStudy site

LHVC - Los Humeros Volcanic ComplexTMVB – Trans Mexican Volcanic Belt

Mexico

TMVB LHVC

▪ Los Humeros is a very large caldera complex (20 km diameter) hosting an active geothermal system producing 95.7 Mwe

▪ 28 production wells, 3 injection wells▪ Main structural trend is N-S, NW-SE, NE-SW and E-

W▪ Reservoir permeability is low to intermediate ->

mainly controlled by a dense fault and fracturenetwork

▪ The Los Humeros-Maztaloya fault is the longest fault with the highest vertical displacement (up to 80 m) crossing the active geothermal system

▪ In total seven CO2 flux chambers measuring once every hour ▪ Deployed on the Los Humeros fault zone▪ Areas of low and high degassing were chosen▪ Installation of a weather station recording air temperature, air pressure, rainfall,

humidity, wind speed, and wind direction▪ Sensors for soil temperature and soil humidity had technical problems and did

not record data▪ Soil temperatures were measured in-situ in 50 cm depth before the network was

set-up▪ Record time: 5 months, April - September 2018

S

E N

WView from NE to SW

Results

Box-plots of all stations showing the distribution of data including their minimum, maximum and median values.

The process of weather over the 5 month monitoring period. Red lines show data gaps due to technical problems.

Results

.

▪ Results show highly variable CO2 flux for each station

▪ Station 6 has the highest measured flux (mean 424 g m-2d-1), whereas station 5 measured lowest flux values (mean 3.5 g m-2d-1),

▪ We observe a general trend in almost all stations showing an increase in CO2 flux towards July and August, and slowly decreasing in September

▪ There is a data gap in all stations from the 8th of May until the 17th of May which we consider in the following analysis

N

Rose diagram showing wind speed and wind direction

▪ Mean horizontal wind directions between 210 and 12°▪ Maximum wind speed was reached in May with 21-24km/h▪ The main wind directions are reaching the monitoring network either

parallel (coming from N-NW) or perpendicular (SW)

Interpolated soil temperatures

▪ Maximum measured temperature in 50 cm depth was 97 °C, which corresponds to the location of station 6, mean FCO2 = 424 g m-2d-1

▪ Black rectangles with number show the location of CO2 flux stations▪ Decrease of soil temperatures towards the N and W (down the slope)

1

2

3

6

4

5

7

Methods applied for data analysis

1. Spearman correlation matrix to combine environmental parameters with CO2 flux

2. Change point analysis on CO2 flux datasets to find significant changes in their temporal distribution

3. Principal component analysis applied on a wavelet power spectrum to identify cycles and see abrupt changes

4. Spearman correlation coefficient between CO2 flux and Injection rates

Spearman Correlation matrix with station data and weather

Abbrevations: 1-7 – CO2 flux stations; Temp - Air temperature; Prs – Ambient pressure; Rh – Relative

humidity; Prec – Precipitation; Ws – wind speed; Wd – Wind direction

Correlation between CO2 flux and environmental parameters:

▪ Inverse moderate to strong correlation between air temperature, windspeed and CO2 flux.

▪ The highest coefficient is seen in station 3; R = -0.62 for temperature and R= -0.65 windspeed

▪ There is a weak positive correlation (R = 0.4) between some of the CO2 flux stations (3,4,5,6)

and atmospheric pressure and relative humidity

Correlation between the different environmental parameters:

▪ Air temperature and windspeed are very strong positively correlated (R= 0.96)

▪ Air temperature and windspeed to relative humidity are strongly inverse correlated (R= -0.65 and R= -0.6)

Change Point Analysis

The function changepoints (CP) was used to detect abrupt changes in the mean of the CO2 flux data.

CP’s occur where the mean of the y -values changes significantly.

CP’s which occur in all stations are marked by a black arrows.

CP’s are a result of an increase or decrease of CO2 flux delimiting a period of significant change.

Wavelet analysis based on PCA (Principal component analysis) for the identification of cycles

▪ PCA detects linear dependencies between variables and replaces groups of linear correlated variables with new, uncorrelated variables referred to as the principal components (PCs).

▪ The diagram on the left displays the loadings of each station within PC1 and PC2. All stations are contained in PC1, whereas in PC2 some station are negativley loaded and others positivley. We interpret PC1 to contain processes related to environmental changes and seismicity.

▪ PC1 shows us the maximum variation of the original datasets (50%), whereas PC2 merely includes 16% of variations.

Data gap

▪ In a next step we used PC1 to calculate a Continuous Wavelet Transformation (CWT) and displayed the result in a Wavelet Power spectrum.

▪ We observe high power at a 24h frequency band of the wavelet power spectrum. This is due to the strong influence of daily variations from atmospheric temperature and wind speed, as well as to a minor extent from barometric pressure and relative humidity (as shown by R-values).

▪ A less prominent cycle can be seen at a 12h frequency band related to minor semidiurnal changes influenced by environmental parameters.

PC1

PC2

Data gap

▪ We distinguished five changes based on the results of our CP analysis

▪ These abrupt changes are seen in the wavelet power spectrum and disrupt diurnal cycles.

▪ They occur at end of April (I), begin of June (II), from mid June to mid of July (III), begin to mid of August (IV), and begin to mid of September (V).

▪ We interpret these sharp changes with the increased frequency of seismic events as well as the stronger rain periods, which especially occur in the month of June, end of July and August.

▪ The magnitudes of seismic events recorded inside the geothermal production zone are generally low and never exceed MLv of 2.1. Their x,y,z coordinates show a strong relation to injection wells.

-1.5-0.50.51.52.5

22.4 29.4 6.5 13.5 20.5 27.5 3.6 10.6 17.6 24.6 1.7 8.7 15.7 22.7 29.7 5.8 12.8 19.8 26.8 2.9 9.9 16.9 23.9

Magnitude (MLv)

0

2

4

6

22.4 29.4 6.5 13.5 20.5 27.5 3.6 10.6 17.6 24.6 1.7 8.7 15.7 22.7 29.7 5.8 12.8 19.8 26.8 2.9 9.9 16.9 23.9

Frequency

II IIII IV V

Rain (mm)

Spearman correlation coefficient between CO2 flux and injection rates

▪ We used the spearman correlation coefficient to show the relationship between total injected fluid (by the 3 injection wells) and CO2 flux.

▪ A low pass - butterworth filter was used to remove diurnal variations from the dataset

▪ An inverse correlation between CO2 flux and injection rates is observed

▪ Station 1; R = -0.57

▪ Station 5; R = -0.68

▪ Station 6; R = -0.56

▪ Station 7; R = -0.41

▪ These moderate to strong correlation coefficentsshow an immediate response of changes in thereservoir (incraese of reservoir pressure due toinjection) related to a decrease of CO2 flux

▪ A sudden pressure increase in the reservoir canseal permeable fractures and faults shortly until thereservoir pressure stabelizes and CO2 increasesagain

▪ Furthermore the lower density of cold injectionfluids into the high temperature reservoir can causea short term decrease in reservoir pressure

▪ CO2 flux results from the seven stations showed a wide range of values -> Very heterogenous subsurface (active fault zone) with different CO2 transport mechanisms (diffusive, diffuse-advective, advective)

▪ Windspeed (high wind speeds may dilute soil gas, pushing air into the upper parts of the bare soil), air temperature (atmospheric tides) and precipitation (effect on soil permeability, on a short term scale) show the strongest inverse correlation to CO2 flux

▪ Wavelet analysis shows diurnal variations induced by environmental parameters and abrupt changes in the cycles which are the reason of non-stationary phenomena like long and heavy rain periods (6 days; with max. of 22 mm a day) and the higher frequency of seismic events

▪ This is likely due to the saturation of the soil, mechanical disturbance of faults/fractures or enhanced mineral precipitation which is decreasing the permeability in the subsurface

▪ Change Points are in accordance to the abrupt changes seen in the wavelet power spectrum of PC1 and verify our findings

▪ Finally we could show an inverse correlation between CO2 flux and injection rates which gives evidence for a significant hydraulic connection of permeable fluid pathways between the geothermal reservoir and the surface

Conclusion