dinsdag 15 december 2015

Work of a student: starting his own blog!

The DopTrack satellite tracking station is in full operation these last weeks. In the Space Minor at the university during my module "Satellite Operations", students are visiting the station and record satellite signals, especially Delfi-C3 and the ISS are very popular. The students then need to determine the Time of Closest Approach (TCA) and the corresponding frequency (FCA). This requires some FFT, but I am confident that they will manage.

Furthermore, our intern student has done work on the ground station. His job is to estimate the precision of the Doppler measurements of DopTrack. The last few weeks he has been working at the theory behind the Doppler measurement and some of its characteristics. So with the motto "More scientists and engineers should blog", I asked him to write a blog post about his finding. His results give a clear description about the Doppler measurement. Please follow the next link for his work:

It is a nice read and saves me writing a complete new blog post ;).
Unfortunately, Delfi-C3 stopped transmitting after 100 seconds during this students' recording, but as you can see the sky is full of other satellites broadcasting their information down to Earth.

maandag 16 november 2015

PhD almost over, so lets write a little about it

The ice sheets and glaciers on our Earth are currently melting, which causes the global sea level to rise. This has of course major impacts on the 44 percent of the world’s population living in coastal areas. This melting is happening already since 24,000 years ago, during that time enormous ice sheets, up to 3-4 km thick, covered lands like Scandinavia and Canada. Scientists have done measurements of this rising sea level using the coral reefs at Barbados that show a sea level rise of 120-130 meters since the start of the de-glaciation. Imagine this for a moment; England was not an island at that time.

Currently, we see a global sea-level rise of 3 mm/year, however locally the rate of sea-level change can vary significantly. For example, at the coastlines of the Netherlands we observe a 2.3 mm/yr sea-level rise, whereas the center of Sweden and Finland has a 7.8 mm/yr sea-level drop. The cause of these particular variations is that not only the sea-level changes, but also the solid ground underneath us moves up and down.
Anual Tide Gauge record of two stations: Hoek van Holland (blue) and Ratan (red, middle of Scandinavia). I fitted a linear trend line through the data to estimate the relative sea-level change of those two locations. Data can be found here.
The huge ice sheets of 3-4 km thickness, mention earlier, have pressed the Earth’s crust downward which caused mantle magma to flow outwards. Just like when you would lie on a waterbed. Some part of the water beneath you would flow away from you, such that other parts of the bed rise. When you get out of the bed, the water will flow to its equilibrium state. Similar effects will happen to the surface of the Earth after the big ice sheets have melted. So, the land is now rising again where the historical ice sheets were situated (Scandinavia), resulting in a sea-level drop in that area. This sea-level drop is caused, because the solid Earth is faster moving upwards than the sea level. Therefore, relatively in Scandinavia the land is coming up. One funny consequence, somebody once told me, is that Sweden and Finland have world’s best legal regulations for land division, because every year they get more land and need to divide it between them.

A schematic representation of post-glacial rebound of the crust.
(courtesy of the Canadian Geodetic Survey, Natural Resources Canada)
This motion is called post-glacial relaxation of the Earth’s crust and it can bias local observations of the actual ongoing sea-level change. So, to really understand the ongoing sea-level change we need to understand this relaxation. To do so we need information about the physical state of the magma underneath our feet. This region is called the mantle of the Earth. We need to know the temperature, density, and viscosity of the mantle. Temperature (how hot something is) and density (how heavy something is) are easy to recognise, but viscosity is sometimes difficult to understand. Viscosity is a way to tell how sticky a certain fluid will flow. For example, maple syrup has a higher viscosity then water. It takes maple syrup much longer to get out of a bottle than, for example, water. If we would fill our waterbed with maple syrup, the sloshing of the bed while you were moving on top of it, would go much slower. You can imagine that the viscosity of the mantle is therefore important to know when studying the relaxation motion.

However, the exploration of the mantle is extremely difficult. NASA scientists often say that humanity knows more about the farthest places in outer space, than the interior of the Earth. The deepest hole mankind has ever constructed is only 12 km deep (Kola superdeep borehole), whereas the mantle in continental areas (like Scandinavia) starts at 30 km deep until the core of the Earth (2890 km deep). We have only scratched the surface. So, instead of going there and study the mantle in-situ, we have to think of observation techniques that can remotely tell use something about the physical state of the mantle: measurements that use seismic waves, gravity field, and magnetic field, or old geological outcrops that give information about the composition of mantle rocks.

Unfortunately, ground measurements and satellite observations give us incomplete information and do not show us the complete picture. Seismic observations give us information about seismic wave speed, which are affected by composition, temperature, and density, whereas satellite gravimetry can penetrate to deeper layers of the Earth, but only shows us the density structure. Geological field studies give us hints about the history and composition of the Earth crust, but only give us information about what is found on the surface. Somehow, we need to combine all these different techniques in an ingenious way to successfully explore the deep Earth. So, how can we improve the exploration of the deep mantle, by combining gravity, seismic, and geological observations? With the end goal to better understand the relaxation of the Earth’s crust, such that we can improve our predictions of the current sea-level change?

This is where my PhD research is hopefully answering a few questions. By studying the gravity field of earth, we know that continental plates are floating on the mantle just like an iceberg. Here, a small tip above the water line means a large submerged root to keep the whole iceberg floating. This information can be used to estimate the density of the mantle and validate it with the observed gravity field. Geological studies give us information about the type of rocks and their composition found in the area. Also, geological theories state that old continental areas are colder than young continental areas. These clues about the temperature can be validated with seismic studies. For instance, certain seismic waves are more sensitive to the temperature profile of the magma than compositional changes. By combining the initial estimated density of the mantle with seismic measurements of temperature and composition, we are able to construct a complete physical model of the deep mantle. Finally, the explored physical state of the deep mantle can be used to simulate the vertical motion of the Earth’s crust due to the melting of the huge ice sheets. By changing the physical parameters within their uncertainty, different results of the vertical motion can be obtained. These results are then compared to geological and geodetic observations the vertical motion. Hopefully after many iterations, we hope that these new computer models of the mantle can reduce the uncertainty in the physical state and better predict the motion of the solid Earth.

In the end, by combining the gravity observations and crustal structures from geological and seismic studies in sophisticated computer models, we can simulate the slow movement of the solid Earth and will improve our estimation of current sea-level change by removing the bias in local sea-level change measurements.

dinsdag 27 oktober 2015

How to make a spectrogram or waterfall plot

In a few blogpost ago, I showed you several figures of a satellites transmission. You know, those beautiful S-curve plots with all the colours.  Those are called spectrograms or we like to call them waterfall plots, because the radio transmissions streams up in time like a reversed waterfall. To see such a waterfall plot live in action, the University of Twente opened their antennas for you to listen and see (see link). Yet, how to make such a figure?

To start with you need a recorded radio transmission. When the website of the TUDelft tracking station is up and running you can download some data there. However, (Start NerdAlert!) you can also buy your own little USB stick that acts like a software defined radio (SDR) receiver like the Logilink DVB-T. For only 20 euros you have your own satellite tracking station. NOTE: Beware! You need the RTL2832U chip version, because a lot of open software is written for that device. It is almost plug and play, you need some software, but all is available on the internet. Download gnuradio for example, which is easy to learn, after only a few tutorials you are able to record your own data. This USB stick SDR development is a little bit of a revolution in radio transmission-world. It allows software to manipulate the radio-signal and let it do whatever you want. (OK StopNerdAlert!)

Logilink DVB-T USB stick with antenna works like a SDR and it is so small.
No back to the radio frequency spectrogram construction (OK, that also sounded nerdish), I have made a 6-step plan to convert the raw radio signal into a spectrogram.

1.     Check that the dataset is a complex signal;
2.     Define length of signal, sampling-rate, and center frequency;
3.     Define the time-step for the waterfall plot (trade-off between freq. & time resolution);
4.     Perform FFT for a segment of length equal to the times-step, iterate through complete signal;
5.     Represent the results in a waterfall plot;
6.     Choose color scale to better visualise signal.

Step 1: Check that the dataset is a complex signal
From the SDR-device or the USRP in the tracking station the data files are stored in the .fc32 format, which is a binary representation of a complex signal. You need the complex signal, because then the Fast Fourier Transform (FFT) will generate a full band representation of the signal. If the signal would be real (so not complex), only half of the spectrum would be visible. You would get two mirrored spectra. Here, a small MATLAB code to read the data into an matrix you can manipulate:

f = fopen ([filename '.32fc'], 'rb');
      t = fread (f, [2, Inf], 'float');
      v = t(1,:) - t(2,:)*1i;
      [r, c] = size (v);
      v = reshape (v, c, r);
      You can set the count in fread to Inf, such that the complete file is read or a part of the file and iterate through the data. In the end you will have a complex vector, v, that contains the complete information of the recorded signal.

Step 2: Define length of signal, sampling-rate, and center frequency
For our recordings mostly we use 900 seconds of recording time, this is the average time for a satellite pass, the tuning frequency for Delfi-C3 is 145.870 MHz, and we use a sampling rate of 250 kHz, because of hardware specifications of our USRP. We also downsample the data to 50 kHz, which is still usable and takes up less memory space: 250 kHZ file is 1.7 Gb and 50 kHz file is 360 Mb. 

Step 3: Define time-step for spectrogram
Here, you can play with time and frequency resolution. For now our aim is to have high accurate frequency determination every second, so we use Dt = 1 sec. You can play around with it, of course. Be careful with the memory of your computer, many times I have pushed my computer over the memory edge.

Step 4: Perform the FFT on the small defined part of the signal and iterate
The Fast Fourier Transform algorithm is a powerful piece of software. It can be very fast when handled properly. MATLAB has its inbuilt FFT algorithm, which works extremely fast. For Python there are many different options (SciPy version is unfortunately not as fast as MATLAB), and of course any well respected programming language should have the FFT algorithm implemented. 

The FFT algorithm has a parameter that tells how to bin the calculated frequencies. In MATLAB this parameter is called NFFT. Always have this parameter higher than the length of your sample that is inputted in the FFT algorithm to get high frequency resolution. To increase performance of the FFT algorithm, adjust NFFT such that it is an exponential of 2. Then start iterate through the signal:

for i = 1:endSignal         
            % Select certain part of the signal
            b = 1+ (i-1)*lengthPart;
            e = lengthPart + (i-1)*lengthPart;
            SigPart = v(b:e);
            % Perform FFT algorithm on selected part
            Y = fft(SigPart,NFFT);
            line = 2*abs(Y(1:NFFT))./lengthPart
            % Store the FFT result in Waterfall matrix
           Waterfall(i,:) = line;

In the end you will get a matrix filled with frequency information from the recorded signal.

Step 5/6: Represent the Waterfall plot / choose appropriate color scale
The data is now in a matrix that you can plot with MATLAB using imagesc, or plot it to a .bmp file. Choose the option that works for your purpose and software. Lets see one of my results.
Spectrogram of a Delfi-C3 pass for a 15 min recording at 145.870 MHz using a sampling-rate of 250 kHz.
By changing the colorbar, you can visualise the satellite signal much better. A beautiful S-curve due to the Doppler effect. Good luck!

dinsdag 28 juli 2015

Observing gravity on a moving ship

In the beginning of the 20th century the gravity field of the Earth was only measured on land. Of course, to obtain a complete image of the Earth's gravity field the oceanic area (covering 74% of the Earth) should also be inspected. However, in those days that was a difficult task. Gravity could be measured by several different instruments, but for them to work a stable platform was needed. This is difficult to arrange on a rocking ship. For example, the falling mass principle was used by Simon Stevin (1548-1620), who dropped balls of lead from the New Church in Delft, Netherlands. Later, a dutch sailor and scientist called Christiaan Huygens (1629-1695) found the relation between the period of a pendulum and gravity (I used this relation earlier in the my blog).

This is now called the Christiaan's Huygens law. With it the operator was able to measure gravity with a single pendulum, but needed a stable platform, such that horizontal movement did not interfere with the motion of the pendulum. There were already some attempts to solve this problem (e.g. the St├╝ckrath apparatus), but it was Vening Meinesz, who in 1923 designed, built and tested his instrument, showing that it was possible to observe the gravity field on the oceans of the Earth with great accuracy.

During my lectures about professor Vening Meinesz and his work, I often get the question how his instrument worked. So, I will try to explain its basic principles in this blogpost. If you want to know the exact working I have to redirect you to the professor's publication "Theory and Practise of Pendulum Observations at Sea", Vening Meinesz (1929), but beware its is heavy-reading.

Lets start with the basics of a pendulum in motion. And to keep it simple, I am talking about the spherical-chickens-in-vacuum-kind of pendulum. What I mean is that I neglect air resistance, elongation differences because of temperature changes, slowing down due to friction in the rotation point, finite stiffness of the beam, and other small effects that deviate the stable motion of a pendulum. Below you can see my spherical-chickens-in-vacuum pendulum in motion:
Motion of a free pendulum (don't get hypnotised)
I have released this particular pendulum from a starting angle of 20 degree. Due to the gravity pull on the mass (the bob), the pendulum starts to swing back and forth between 20 and -20 degrees.  The motion of the pendulum can be described by a simple linearised differential equation. The derivation of this relation (in several ways) can be found here.

The angle of the pendulum is depicted by theta, gravity is given by g, and the length of the pendulum is l. I have created the figure above with this relation, a propagator (see this blog post), and some MATLAB coding. So, back to the gravity! When you know the length of the pendulum and measure its period you are able to determine the gravity that the pendulum is experiencing. And NO! This is not the same everywhere on the globe, I already explained this here. A problem arises when the pendulum is not fixed to a stable platform, like a moving vessel on the ocean. I tried to illustrate this by enforcing a harmonic horizontal acceleration to the free moving pendulum (see red bar).
Motion of a forced pendulum. The red bar indicates the strength and direction of the horizontal acceleration. 
You can see that the motion of the pendulum is disturbed and a single unambiguous period is difficult to determine. This is where Vening Meinesz apparatus comes into play. He used two similar pendulums with equal length, swinging in the same plane (well actually three, but lets stick to the basics).
Two pendulum swinging in the same plane. Vening Meinesz measured the distance between the two pendulums.
He let the two pendulums swing in opposite manner and measured the difference of the two angles (or the distance between the two bobs does the same trick) and determined its period. The same thing can be done with pendulum that are experience a horizontal forcing, like the rocking of the boat. The beauty of this subtraction of the motion of both pendulums is that the horizontal forcing is cancelled. This can be seen in the following illustration:
Two pendulum swinging in the same plane and experiencing horizontal forcing
Both pendulums experience irregular motion due to the horizontal forcing (e.g. swinging of the vessel), however the distance between the two bobs is regular. This a little bit difficult to see, so I have plotted the horizontal motion and distance over time:
The horizontal motion of the two pendulums and their distance over time
The irregular motion of the individual bobs is clearly visible, but the horizontal distance (or the difference in their angles) is regular. The period of this distance is similar to the period of the pendulums if they stood in a stable environment. This remarkable piece of mathematics and physics enabled professor Vening Meinesz (and a submarine) to measure the gravity field of the Earth on the oceans. It kick-started the dutch and global marine research in the solid earth. Up until 1956, Vening Meinesz' apparatus was the only way to measure the gravity field on the oceans with great accuracy.

maandag 15 juni 2015

DIY satellite-pass planner

The new lecture year is coming and the satellite communications and tracking practicum is about to start. In this lecture, students have to plan a satellite pass, record the downlink signal, and post-process this data, such that they can recover some parameters like: Time of Closest Approach (TCA), carrier frequency of the satellite, and error estimates compared to TLE (see previous blog posts: here, here, and here). For the first part, the students need to compute when the satellite is flying overhead. There are many websites and apps that are able to give you 5 day predictions, but I wanted to see if I could do this myself. As a little boy I always had the tendency to say: 'I can do that too!'. Up until the time my father said: 'Son, you can only say "I can do that too", when you at least have done it ones.' So, living by that wisdom I want to show you how to build your own satellite-pass predictor.

To validate my efforts, I am going to generate the same results as a pass-prediction website I usually go to (see link).  I have inserted the NORAD ID for the Delfi-C3 satellite and this is what the website gives me:

The results of N2YO at Sunday 14 June 2015 10:00 o'clock
So, I have to reconstruct this list: Start time Azimuth start, TCA, Azimuth TCA, Elevation TCA, End time, Azimuth End. The colours per row illustrate the visibility of the object, however I am not interested in this, because in the radio spectrum the satellite is always visible (well when his radio is turned on, this is not always the case with Delfi-C3). Let's start!

To predict when a satellite is passing overhead, the orbit of the satellite is needed. This orbit can be predicted for 5 in the future, because of its predictive behaviour (its not like predicting the weather). The uncertainty in this non-chaotic system is well below the one-minute prediction uncertainty of the timing requirements. The Americans knew this too, so they are tracking and recording all objects in space using radar telescopes. The great thing of this is that they are making all these recordings available for the public (well not the super secret spy satellites, but then go to this blog, if you want to know more about them). Nevertheless, Delfi-C3 is NOT a super secret spy satellite, so its track record is available. I used the latest at this moment:

1 32789U 08021G   15164.24672859  .00009977  00000-0  80803-3 0  9994
2 32789 097.6719 224.3843 0011775 057.7548 342.8056 14.99867522386494

This is called a Two Line Element, or TLE. This TLE is updated daily and gives a state vector of the satellite, or in other words, says where the satellite was and how fast it was flying at a certain moment of time. With this data, you are able to propagate and predict the satellite's orbit. You can obtain this TLE from the www.space-track.org website, after you have created an account. To extract the TLE use the following shell line:

curl -c cookies.txt -b cookies.txt -k https://www.space-track.org/ajaxauth/login -d 'identity=your_login&password=your_password&query=https://www.space-track.org/basicspacedata/query/class/tle_latest/ORDINAL/1/NORAD_CAT_ID/32789/orderby/TLE_LINE1%20ASC/format/tle' > TLE.txt

This will print the latest TLE of the Delfi-C3 in a text file named TLE.txt, use your own password and username of course. The space-track website gives good instructions for any OS you use. Also, they make propagation software available to do your own propagation. The propagator that they use is called SGP4 and is easy to use. Instead of rebuilding this, I have re-modified their propagator for my purposes (Sorry dad, but in my defence, I have build propagators in my Masters study, so I can really do that too ;) ). I have modified the SGP4 propagator software, such that it gives me a 5 day orbit prediction. Please, contact me when you are interested in the software.

--Just a side note-- 
The 5day predicted orbit is referenced to a reference system that is called TEME, which we (my office-mate and me) pronounce like TEME. It is a little bit weird reference frame. David A. Vallado the author of "Fundamentals of Astrodynamics and Applications" (He should know!), says the following about the TEME reference system:

The AFSPC analytical theory (SGP4) produces ephemerides in a "true equator, mean equinox" TEME. An official operational definition of TEME is very difficult to find in literature.

So, even the experts are a little bit fuzzy on the subject. Continuing reading his work, Vallado gives a thorough and detailed description of the TEME reference system. However, the TEME reference system stays a little bit vague. That's why I think our pronunciation is correct. If you disagree, please follow this link
--End side note--

The orbit is situated in a inertial reference frame, however our station is located on a rotating Earth. Therefore, we need to transform the orbit into a Earth fixed rotating reference frame. After, this quick transformation, you are able to compute the azimuth and elevation of the satellite. Also, you can compute the vectorial angle between the satellite and the ground station. The satellite is in view, if this angle is smaller than the "horizon angle" for that particular station. The "horizon angle" can be deduced by making a sketch of the situation:

The length of the satellite's position vector (r) when it is flying over the horizon is equal to the radius of the Earth (R) plus the height of satellite's orbit (h). To calculate the horizon angle (phi), you can make use of the cosine rule:

cos phi = R / (R + h)

The mean height of the orbit can be deduced from the TLE and the radius of the Earth is also known, so the determination of the horizon angle is possible. When you have a relation for the horizon angle, a visibility plot can be made:
Zoomed in section of the 5day visibility plot
The value 1 means that the satellite is in view of the ground station. These slots come in 3 or 4 bars, some broader then others. This is of course a result of the particular geometry of the satellite pass. After a simple search algorithm that identifies and characterises all passes, I was able to calculate the parameters that we needed, listed in the beginning. I used a 10 degree elevation cut-off criteria, because I know N2YO is also using this. The end results is given as a text file:

The results of pass prediction by Broodt on Sunday 14 June 2015 at 10:00 o'clock
By inspecting the first and last figure of this blog, I can now say (within reasonable uncertainty): I can do this too!

woensdag 6 mei 2015

Measuring mass changes with satellites

In my latest paper, GRACE gravity observations constrain Weichselian ice thickness in the Barents Sea, I use an incredible satellite data set. Starting from 2003, NASA's GRACE satellite mission is producing monthly "images" of the Earth's gravity field. These observations are so accurate that certain mass transport is visible, if you look at the differences between these "images". This new data set gives tons of information about many different natural processes. Oh and the figures are so cool!

For this blog, I downloaded GRACE data (click here for information, and here for the data) from the period between 2003 and 2013, CSR release. For every spot on the Earth, I calculated the linear gravity change in this period, so not the seasonal variations. The results are shown in the figure below:

GRACE gravity change during 2003 and 2013 using a 'white' noise filter of ±0.5 microGal/year.
The red areas show localised gravity increase (or mass increase) and the blue areas show localised gravity decrease (or mass decrease). I have capped the negative areas (blue) at 2 mircoGal/yr, but some areas, like West Antarctic, can go up to -8 mircoGal/year. However, smaller signals are made visible by performing this truncation. Of course, mass can neither be created nor destroyed, so integrating this signal should give zero. It now looks a bit of to the negative side. The large localised negative mass losses in Greenland, Alaska, and West Antarctic are of course linked to melting of glaciers and land ice. The meltwater is redistributed over the entire oceanic surface of the Earth. This generates a long wavelength signal with a very small amplitude, which is not visible in this map, because of my 'white' noise filter. However, when this would be taken into account the integral of this signal would be zero.

But enough physical and mathematical mambo jambo! Look at the figure (click on it to zoom in), what do you see? The melting of land ice is prominent in Greenland, Alaska, and West Antarctic, but can also be seen in the Iceland, Pantagonia (South of Chili), and localised areas on the east coast of Antarctic. Other, negative areas are related to non-ice hydrology, such as the Mississippi basin and the drying of the Caspian Sea. Also, the signals in the Himalaya can be treated as groundwater motion. Some positive hydrological signal can be seen in the Amazon basin, Zambezi river basin, and in Ghana. 

Larger scale areas with positive gravity change are related to Glacial Isostatic Adjustment (GIA), which I am working on in my PhD research. The biggest signal is in Laurentia (Canada). Here, GRACE has given evidence for a multi-dome ice sheet during the last ice age. Can you find the two maxima? Another well known GIA area is Fennoscandia (Scandinavia and Finland), also here the positive signal is mantle material flowing back to the area due to the uplift of the crust, responding to the isostatic equilibrium. Antarctic shows GIA related positive gravity change, but due to the ongoing ice transport it is difficult to tell what is present-day ice transport and GIA-related magma transport. 

Moreover, the figure shows the effects of three very large earthquakes in the period of 2003-2013. Can you spot them? I will help you. The easiest visible is the 2004 Sumatra earthquake (9.1), which is clearly visible as an elongated gravity dipole. The gravity dipole of the 2011 Honshu, Japan earthquake (9.0) is also visible. The last one is a bit tricky, because of its smaller magnitude, the 2010 Chili earthquake (8.8), where the positive part of the gravity dipole is barely visible above the noise.

It is incredible to see that by measuring the tiniest movements of orbiting satellites, scientists and engineers have created this new and incredible data set, which allows us to observe and learn more about our home planet.

zaterdag 18 april 2015

EGU 2015 End Week Update: A Young Scientist participating in the Discussion

The General Assembly of the European Geophysical Union in 2015 has come to an end. I am filled with new information, new ideas, motivational energy, great experiences, and encounters with new and interesting people. Waiting for my flight back home, I have time to reflect on the final few days of this incredible experience.

Last blogpost, I gave an update of the first two days. I was working that day on my rebuttal for a new paper. I submitted the paper to a journal, which is using a short response time of two weeks, and with Easter and preparation for the conference, I had little time to work on it. On Wednesday, my personal program only needed me to be at the evening poster sessions, so during the day, I locked myself up in my hotel room and worked on the rebuttal (despite the beautiful weather). It is now with the editor for decision! Nevertheless, I had an interesting time in the evening.

Day 3: Wednesday 15 April 2015
Arriving late at the EGU conference site, I went directly towards the poster session (quickly getting a drink) to visit the first poster on my list. Poster sessions are great, because they enable you to better interact with the scientist and ask lots of questions. Also, for a Young Scientist, you can really engage in the discussion, in contrast to the presentation sessions, where you need to have a lot of guts asking a question in a room full of experts.

The first person I met was a very interesting and enthusiastic scientist from the University of Leicester. He and his group are slowly, but gradually mapping the complete crust around the British Isles and far surroundings (including parts of my study area Scandinavia). They use different geophysical observations to explore structures in the crust. What I like about the work is that they also look at the uncertainties in their observations and models, which allows a modeller (like me) to really test their work. Furthermore, I wanted to meet him, because I was asked to make a gravity model out of their data. So, it was good to finally meet the man behind the great work. Here, some preliminary results:
The seismic wave velocity and density estimates of a cross-section of the data set
You can clearly see that the density distribution follows physical principles. The deeper the material the higher the density is due to the pressure of overlying rocks. This is not always the case in other commonly-used density models, so I am eager to study the effects on the gravity signal. Thanks to the meet-up, I have more motivation to finish my modelling when I get back to my university. Continuing the poster session, I went to a scientific field which I no nothing about: Core-mantle boundary studies. I met a very smart and interesting guy that was willing to explain to me (newby in the field) how they measured deviations in that region. Incredible, using earthquakes to observe the deepest regions in the Earth. He was even willing to send me some papers that explained the principles and uncertainties. All in all, a great experience!

Just before I wanted to go to the city centre, I met up with my German colleagues from the University of Kiel and asked me to join a talk organised by the German Geophysical Society. During this talk, I learned about magnetic anomalies and how to observe and interpret them. The speaker gave a great historical overview of magnetic field measurements and showed me (and many others of course), that old data first needed to be corrected for daily variations. If this was not done, you could only interpret the velocity of the expedition vessel, around 10 knots. Nowadays, by using multiple magnetometers, they are able to remove this signal and study the magnetic signal in the underlying rocks. Then, he showed some applications, which blew my mind. Thanks to magnetic reversals and pole rotations, he was able to study crustal material that was already subducted into the mantle, and see what processes worked 100 million years ago. Wow!

Day 4: Thursday 16 April 2015
Today, I started the day with presentations about hydrological studies. In the first talk, a new hydrological model was presented, WGHM. It worked quite well, however it overestimated snow-melt in the spring season. I was wondering, could this be related to the observations done by the scientists on day 2, looking at satellite photos to map snow cover. He found a decline in snow cover during spring. I did not get the change to ask the question. After a while, I found that this session became to complicated for a non-expert like me, so I left and just wandered the exhibition area:

Exhibition area at the entrance of the conference building
A lot of nice companies and publishing agencies, however no gravity instruments. Still, the ESA and Google stands were quite interesting. After an hour, the Geodynamics session about the lithosphere and upper mantle started, which would keep me busy up till 19:00 in the evening.

Here, I list some of the interesting information I learned in the session. In the Atlas mountain range, the lithosphere plays a large role in the formation of this mountain belt. However, both strong and weak modelled viscosity structures give similar results. Furthermore, I learned that petrological processes are very important in the study of subsidence and uplift of the crust. Rocks like metabasite, and water insertion can lift or drop the crust by several hundreds of meters in geological time-scales. I should dive in that literature a bit more. Also, I learned about step-faults and some other tectonic behaviours.

Halfway through, the geodynamic medal lecture was given. An incredible talk about some problems of the lithosphere. Different types of earthquakes can be estimated by looking at the stress field. With a viscous sheet model, the complex stress field in Turkey can be modelled reasonably well. Observations of mantle dripping could explain localised earthquakes and tomographic velocity anomalies. His talk was followed by a presentation of the highest resolution tomographic model that clearly observed the North-American craton. A separation between the Greenland craton was noticeable.

The session was finished with some local studies of the crust and upper mantle, but one talk stood out quite well. The group from the University of Dublin had made a great inversion of different geophysical data sets to construct crustal models. They showed that crustal modelling only worked if all data sets were combined: seismic velocity observations, electromagnetic measurements, gravity observations, and petrological data. This was quite an intimidating talk that motivated me to do even a better job. During the poster session that evening, I presented my ideas on the my crustal modeling and its effect of using GIA observables. Introducing this extra data set could give us more understanding of the temperature structures in the upper mantle. I could talk the complete 1.5 hour and even the Core-Mantle guy dropped by for support. The day was ended by the Vening Meinesz Medal lecture, where the complete history of GPS observations was shown. Great work, which really deserved the Medal. Back in the hotel, I made some last alterations to my presentation and practised it a few times. Tomorrow, I would talk about Vening Meinesz and his submarine expeditions.

Day 5: Friday 17 April 2015
I was scheduled as second presenter in the session: History of Geophysics. After an inspiring talk about Wegener, it was my turn. With some nervousness in my body, I climbed the stage and relied on my experience and enthusiasm. I had a lot to show and came in a little bit of time-issues, but the convener let me go my way. Overall, I got good responses and hopefully people will visit our website (click here). My talk was followed by more historical interesting presentations about: science in the Ottoman Empire, history of seismic observation (two talks), and magnetic/meteorological observations during the Novara Expedition.

The complete dataset of Vening Meinesz of gravity observations during his 1934-1935 expeditions onboard the submarine K-XVIII
On Friday the GIA session was planned. So in the late morning, I visited the poster session where I had great discussions with my peers. I even met one of my Giants, who'd later had a great talk about the effect of GIA in oil exploration. Despite, the small amount of abstracts this year, the GIA sessions (poster and presentation) showed incredible research and it was great to see all the new ideas in the field. I even learned about a new dataset I should check out.

After the GIA session, I tried to follow some other talks about mantle plumes, but I found out that my brain was saturated. Grabbing a quick bite, I went to my hotel room and got an early night. This was my experience during the EGU 2015 General Assembly and I think I will need the rest of the week to reflect and organise all the new information. Hopefully, see you next year at EGU 2016.

woensdag 15 april 2015

EGU 2015 Mid Week Update: A Young Scientist among Giants

With the European Geophysical Union General Assembly halfway there, it is time for me to sit and write down all my experiences as a young scientist. From sitting behind your desk, interacting with max 3 living beings (one of them is my office plant), to literally meeting thousands of scientists from around the world can be a little bit exhausting. Therefore, I have my mid week moment-of-zen, which I use to write this blog.

Pre-EGU2015 adventures:
Due to a tight-scheduled revision of one of my papers, I had little time to design and print my poster, forcing me to drop by the print office on my way to the airplane. After a provocatively slow printer, I hurried to the airplane and got my flight to Vienna.

After extra long minutes my poster was ready to go!
Arriving at Vienna, I checked in my hotel, after a short train ride from the airport. During the waiting on the platform and the train ride, you could spot them already: other scientists, still a little weary from the flight and carrying to much luggage, but above all the characteristic poster holder. I forgot mine, so they all looked back at me with empathy, the first crease awls on my poster were visible. 

After a good night sleep I set the alarm early, I needed to get my entrance badge. At 08:00 o'clock registration would open, but at 07:30 lines of scientists started to form. Luckily, I was there at 07:29, so in front and got my badge after waiting half an hour.

Let the learning begin!
Day 1: Monday 13 April 2015
With my notebook as equipment I charged inside, took a quick look and went to my first session. The first presentation was about a new theory in plate generation. It neatly explained why Earth has plate tectonics and Venus does not! This speaker linked his theory to the tiniest elements in Geosciences, crystal structures. Wow, this would become a very nice week.

During the day, I learned about how crystal structures could have an effect on long-term behaviour of crust and mantle and advancements in seismic imaging. Not bad for a gravity scientist. The most interesting talk was about melt distributions in a convecting mantle. The speaker spoke about his theory that explained why geophysical data sometimes show a sharp LAB boundary and sometimes they do not. This intrigued me, because I run across this problem in my own studies. Using a theory called redox melting of CO2 in the mantle, the asthenosphere can have extra melting (Oxygen somehow lowers melting point) regimes above areas with a lot of diamonds (C). This results in a sharp signal in geophysical observations, like seismic tomography. This concept only works when there is convection in the mantle. Could it be that where no clear LAB signal is present, we have stagnant convection areas? Back home, I need to read up on this. 

During transfer between talks and posters, I met a lot of old friends and colleagues and in the evening, I ended up with some of them in a cool restaurant designed like a library. I slept like a baby.

Day 2: Tuesday 14 April 2015
After a warm, but good night sleep (broken airco), I was ready to fill my brains with knowledge. I decided to see what the cryosphere people were doing. One of the speakers talked about his life work, mapping snow coverage on the Northern Hemisphere, using satellite photography and many other techniques. One of the interesting (and maybe a little bit troubling) things was that he saw a decline in snow cover in the spring months during the 50 years of satellite data coverage. Could this be a climate change signal or do the seasons change a little bit with respect to our calendar?

During the same session, one of my (young) Giants was presenting his results on ice melting of the Northern ice sheets (Canada, Greenland, Iceland, Svalbard, Novaya Zemlya and other Russian islands). He combined GRACE gravity measurements and ICESat/Cryosat altimetry measurements to observe this ice mass changes with extreme spatial accuracy. You could see ice mass changes of individual ice sheets. Very cool! All three satellite observations fitted perfectly on top of each other. Hooray for the satellites and remote sensing.

After an interesting session about seismic anisotropy, I had a quick lunch with some colleagues of mine. Quick, because I did not want to miss the medal lecture of one of my other Giants. I have been reading papers of this medalist from the start of my PhD. They quickly helped me to get accustomed in geophysical modelling (my background was satellite orbit determination). I briefly met him on a field trip in the Italian Apennines and discussed topics like elastic strength of the lithosphere and mantle rheology. His talk was incredible and an inspiration to continue my work with more motivation and energy.

After the medal lecture, I went together with a friend, which I met on my first conference as starting-PhD and now coincidently sat next to me, to listen to a speaker that boldly stated that the Indian collision with Asia could not have been initiated before 20 Ma. All the geologist in the room silently (or less silently) struggled with this concept, but his geophysical modelling was very good and they could not break his theory.

I finished the day with presentations and posters about how to use gravity in geosciences, more common ground for me. Interesting results, especially the fact that satellite-based gravity gradients are not sensitive to deep mantle effects, which could help in constraining crustal masses much better. Again a good day!

I now have to run to the poster session of today, because I want to know more about the core-mantle boundary and its effects on the Earth system...or I could sit in the sun in front of the entrance...

Scientists in the Sun

woensdag 1 april 2015

Gravity Expeditions at Sea

This week one of the major milestones of the Vening Meinesz project is a fact. The website is online, where you can follow the adventure of professor Vening Meinesz onboard the submarine K-XVIII! He sailed from Den Helder, Netherlands to Surabaya, Indonesia via a large detour, visiting ports on several continents. Along the way he measured the gravity field of the Earth with extreme precision. He did this with a pendulum apparatus onboard a moving vessel. A great achievement done by a great scientist. Please explore the website and tell me what you think of it.

Enter here!

Oh, for none Dutch speakers, just skip the introduction of the main app, the expedition of Vening Meinesz is written in English. Yet, if you can read Dutch, please check my interview in the TU Delta (University paper) about the project, here (page 18).

The captain and crew of the K-XVIII with the professor among them.

zaterdag 7 maart 2015

Observing the shape of the Earth

A few weeks ago, I visited a dear friend of mine at the University of Kiel. He is a professor in geosciences over there. He invited me to meet his team and learn about what they are doing. It was a great week with lot of learning moments. I advise every PhD student and scientist to once in a while visit other groups in your field to see their point of views. Also, it is nice to get out of your office and see the world. 

I want to write about one particular learning moment that week, which was a field tutorial with the LaCoste-Romberg gravimeter. This particular device is able to measure relative gravity variations up to 10 microGal. That is a relative variation of 9 digits after the 9.81 m/s^2 (in the Netherlands). This is the domain where you can feel the pull of the Moon, when it is orbiting above you. Wow!!!

The Lacoste-Romberg gravimeter in the field. 
The LaCoste-Romberg was invented already in 1932 by Lucien LaCoste and his teacher Arnold Romberg, who'd came up with a complete mechanical way to measure the gravity field. Inside this temperature-controlled box a bar with a proof mass is suspended with wires and springs. I have sketched this during my coffee break, with the forces acting on it. It was a long time ago, I had to solve a static problem and my drawing skills are not perfect.

The mass is held up by an inclined spring and two horizontal wires, such that its rotation point is moved towards the location where the horizontal wires are attached. This clever mechanism increases the sensitivity of the device by the square root of two. The bar with the mass is balanced by pulling or relaxing of the inclined spring, which can be done by rotating a millimeter screw on top of the device (the grey round nob in the first figure). If there is a difference in gravity between measurement stations, this can be observed by the difference in relative rotations of the grey nob.

According to the people at the university in Kiel, the LaCoste-Romberg is one of the most precise instruments to take outside and measure the gravity field of the Earth. So, me being a gravity scientist (artist or nerd, whatever), of course I wanted to try this out. One of the undergraduate students took me outside in the 'field'. There is an abandoned railway track close to the department of geosciences in the campus of the university of Kiel. The railway track is quite flat (only 3 meter difference between station 1 and station 12) and runs north south. This is ideal to measure the curvature of the Earth, or even more precise the parabolic shape of the Earth. Some readers of the blog could remember this post, where I showed a way to measure this, but my pendulum setup was not accurate enough. If I wanted to see the shape I needed to visit the north and south poles. However, now I had a much more accurate device, so was it possible to see the shape of the Earth?

The measurement campaign in Kiel, which runs almost perfectly north south. We walked towards the south so we would see a decline in the measured gravity signal.
In total, we observed the gravity field on 12 locations along a 1-km north-south track. I measured the location of the station with my GPS-device, which was a few meters accurate. The gravity measurements of the LaCoste-Romberg were corrected for tidal accelerations, rotation of the Earth, and the drop of 3 meter in the height of the stations. The measurements could be improved by also correct for drift in the device (forgot to do a benchmark, but we assumed little to no drift within the hour of the campaign) and letting the measurements be done by one person, but this was a tutorial, so some mistakes were made. However, we can still see a shift in the observed gravity field.

The results of the campaign in Kiel: (a) the absolute gravity of the measurements (red) and a the WGS-84 gravity model (black), (b) the relative gravity measurements (red) and a linear fit with uncertainties (blue) compared to the WGS-84 gravity model (black), and (c) the location of the stations in north-south direction.
We have equalised the absolute gravity of station 1 to the WGS-84 gravity model at that location, because the LaCoste-Romberg is a relative gravimeter. Both in the measurement as in the model, a declining trend is visible towards the south. The Earth is not perfectly round and thus you will move farther away from the center of gravity if you walk towards the equator (south on the northern hemisphere). The law of newton states that gravity will diminish if the distance becomes larger. Both the linear trend of the measurements and the gravity model show this behaviour. So finally, I have measured the shape of the Earth and had already proven that the Earth is not flat (see here), but now also that it is not perfectly round.