dirty_radome.png

This is the old receiver radome of the 94 GHz RPG radar from the Leipzig Institute for Meteorology (LIM), short LIMRAD94. It transmits radiation at this frequency (94 GHz) through the transmitter antenna vertically upwards into the atmosphere. When the emitted microwave radiation reaches clouds and precipitation particles, is scattered. The part of the radar signal which is returned back exactly to the radar is measured by the receiver antenna. Both transmitter and receiver are covered with a so-called radome, that is a thin hydrophobic membrane, somehow similar in its material properties to those of camping mats: The sheets are very light and soft but still firm. On the upper side, they are covered with a water-repelling layer.

before_after.png

These sheets have to withstand all kinds of weather: Storm, rain, hail, heat and frost, just to name a few extremes.  Also, dirt and small debris that is sucked in through the blower which blow strongly over the antenna to keep water droplets from accumulating on the radomes leave their traces. As a consequence, the radomes age gradually and loose their water repellent property. The manufacturer thus suggests to replace the radomes after 6 months. We exchanged both the transmitter and the receiver radome this week. In the picture above, you can see a comparison between the receiver radome, which has already been changed, and the transmitter radome, which is not changed yet. You can see that they lool considerably different: The old radome has some scratches and is a lot dirtier than the shiny, white and new radome on the receiver side.

Folie1-4.png

Changing the sheets was not that hard: First of all, of course the instrument had to be switched off. Then we started to remove all the screws fixing the old radome to the radar. It is fixed by a metal ring, which has to come off. Removing the old sheet, we could catch a glimpse at the inner radome which is covered by the outer radome, the one we changed. Then we put the new radome in place by fixing screws through the holes of the metal ring into the soft radome material. 

You have to be very careful not to puncture the new radome sheet.

Folie5-8.png

After both radomes were changed LIMRAD94 was switched on again. A short moment of tension before the first data coming in showed that everything was alright. :-)

 [TV, WS]

Last Friday and Saturday, the wind here in Punta Arenas was extreme, even for the local standard. It might not have been the perfect time to arrive here via airplane, but on the other hand, the landing was a real experience, something to remember later on. A striking thing about the weather situation was the heavy, stormy wind in connection with bright sunshine. Looking out of the window, you wouldn’t imagine almost being blown off the street when going out for dinner. The wind also wasn’t very cold, so being outside was actually quite enjoyable.

The Centro Meteorologico Maritimo reported a maximum wind speed of 58.2 knots, or 108 km/h. On social media, people shared the experiences they had with the unusually strong wind: Windows were broken and power outages were reported in parts of the city. Someone uploaded a video from an airplane landing on Saturday, in which one can see a little bit of the swaying of the plane, followed by a rather rough landing.

Lots of flights were canceled on Saturday and consequently, there was a big chaos at the airport. This complicated retrieving my suitcase, which was lost the day before when I flew to Punta Arenas. Luckily, the baggage (which contained some important cables and switches needed for the site) was there on Saturday.

The wind raised quite a lot of dust, which complicated car traffic in addition to the heavy wind gusts. So on top of having to steer against the wind, the view was also obstructed by a thick dust plume from time to time. A truck besides the road was actually overturned. Riding the taxi on the way from/ to the airport was quite exciting:

 

Image-12_smaller.png

 

We can also see this dust in our lidar data. The dust stays in the lower atmosphere, below 2 km altitude, as can be seen in the POLLY backscatter (1064 nm).

soil_dust.png

When checking our remote sensing data, we noticed that we cracked some of the standard color scales in our quicklook plots: The Shaun wind lidar quick look plots show only dark red arrows during the first half of February 16, starting in the second height bin. Below, surface friction slows down the wind at least a little bit so that it’s “only” around 20 m/s, or 70 km/h. Shaun measures the horizontal and vertical wind. In the graph below, the left panel shows the horizotal wind. Vertical wind was exceptionally high as well, which can be seen in the radar data: The Doppler velocity, which can be translated to the combined vertical wind + fall velocity of cloud particles, is shown in the left panel below. Also here, the color scale is not wide enough to accommodate the vertical wind speeds of sometimes almost 4 m/s:

Both graphs show the time on the horizontal axis and the height in the vertical coordinate. The color scale depicts the wind speed. Red means very high wind velocities.

 

horizontal_vertical_wind3.png

 

[TV]

 

During the first days of this week, we had quite special meteorological conditions at Punta Arenas. Persistent wind from the north brought unusually warm air to the Magallanes region. Together with the absence of clouds and hence strong heating by the sun, surface temperatures climbed to extraordinarily high values. On Monday (4 Feb), the maximum temperature at Punta Arenas airport reached almost 25°C and on Tuesday the maximum was 28°C. Long-term temperature records were set for several other stations throughout the region, as for example Porvenir and Puerto Natales with both above 30°C (https://twitter.com/meteochile_dmc/status/1092808761167831040). On Tuesday evening, the heat period came to a sudden end, when a cold front arrived. This is nicely visible in the surface observation (at the UMAG roof platform, shown below) as temperature drops by 15°C (red curve) and pressure increases (green curve) after 17:30 UTC (14:30 local). A nice feature is also the jump in temperature and drop in relative humidity (blue curve) directly before the front, which is probably caused by Föhn effects as the wind increased and backed toward westerly directions.

surface_met.png

But the airmass from the north did not only bring high temperatures, but aerosol as well. Our lidar PollyXT (see also prior blog posts here and here) detected several lofted layers between 1 and 5km height on Monday and Tuesday. Several plumes are visible in the backscatter signal. One with rather weak backscatter and low depolarization ratio on Tuesday and another one with higher backscatter and some features with higher depolarization ratio.

measurements_polly.png

The aerosol optical depth, a measure for total aerosol load in the atmosphere, reached values up to 0.1 (at a wavelength of 500nm). The spectral dependence between 440 and 870nm (called Ångström exponent) was around 1.5, indicating predominantly small particles with a diameter well below 1μm. Based on their optical parameters these particles are most likely smoke caused by wildfires, on Tuesday maybe mixed with some soil dust.

We were not yet able to pin-down the source of this aerosol plume unambiguously. Two source regions are possible: either the forest fires in the Region of Araucanía in Chile or the fires on Tasmania, Australia. Simulations with the HySPLIT (https://www.ready.noaa.gov/HYSPLIT.php) trajectory model (a model that traces the pathway of air parcels), show that the airmass, which was at Punta Arenas in 3km height on Tuesday 12 UTC (9 local) crossed both region during the prior 10 days. We are thus looking forward to have to solve yet another research puzzle.

sat_images_fire.png

20190205_12_03000_trajectories_map_cut.png

[mr]