Introduction To The Micro-Pulse LiDAR Network

All right, so welcome back everyone to this fifth and final part of our RSAT training series on the NASA Atmospheric Composition Ground Networks, which support air quality and climate applications.

Today I'm joined by Dr. Judd Welton, who's a principal investigator of the NPLNet network out of the NASA Goddard Space Flight Center. And in this part, our objectives will be to identify the basic characteristics of the MPLNet instruments, which are used by NASA for ground-based active remote sensing of aerosols, clouds, and the planetary boundary layer, or PBL. We'll also be recognizing how MPLNet sustains long-term global observations of the atmosphere and supports air quality and climate applications, and how it can complement satellite observations as well.

And finally, we'll be showing you how to access relevant MPLNet data for given locations and application purposes. Just a reminder of the other networks we've gone through in this training so far. We've looked at Aeronet and Pandora, which are passive remote sensing instruments looking at aerosols and trace gases, respectively. And then in our last session, we learned about TollNet, which is an active instrument, also using LiDAR, which looks at ozone in the atmosphere. And today, we'll be learning about, as I said, MPLNet, the micropulse LiDAR network, which is another active remote sensing instrument, but this time looking at aerosols and the planetary boundary layer and clouds.

As a reminder, if you have any questions during this session, please put them in the question and answer box in the WebEx. There's three dots in the lower right corner of the WebEx interface that will bring up the Q & A box. You can put your questions in there at any time during the training, and we'll answer all those questions together at the end of the training session today. We'll also be collecting the answers together in a question and answer document, which will be posted on the website about a week after the training.

And now, without further ado, I'll hand things over to Dr. Judd Welton to teach us about the Micropulse LiDAR Network, or MPLNet. Over to you, Judd. Thanks, Carl, for the introduction. I'll be discussing the Micropulse LiDAR Network today. Before I get started, I want to recognize my colleagues. I'm the principal investigator, Judd Welton, at NASA Goddard Space Flight Center. I have a staff shown on the slide here that are comprised of people working on the operational aspects of the network, as well as my science team.

And of course, we have all of our international network partners that contribute substantially to the project. And finally, ARINET, who we've worked closely with since the beginning. Here's the agenda for my presentation. I'll begin with an introduction and an overview on LIDAR basics. Then I'll talk about the MPLNet project and the data we provide. I'll do a tour of the NPLnet website and demonstrate how you access our data. I'll go over some case study examples, and then I'll conclude with a discussion on how you can access global LiDAR data beyond NPLnet.

So I'll begin with our introduction. NPLnet was funded in 1999 and began operations in 2000. The original goal was to develop a long-term global LiDAR network to profile aerosol and cloud vertical distribution and properties at Aeronet sites. With the goal of providing information for the bulleted items shown here, MPLNet is funded by the NASA Radiation Sciences Program and Earth Observing System, and we have significant contributions from our site partners. The pictures shown here are a few of our sites around the world.

The green beam that you see there is from our LIDAR. It is normally not visible. These are long exposure photographs. So I'll begin with an overview of backscatter LIDAR, which is a type of LIDAR we use in the MTLNet project. And note that LIDARs are like radars, but instead of using microwaves, we use laser light. Radars profile clouds, rain, severe weather. They use microwaves with a longer wavelength because that's ideal to look at scattering from the thick, larger particle layers like clouds and rain.

Backscatter LIDARs profile aerosol and cloud structure, and we do that using visible light with a shorter wavelength, which is more ideal to look at smaller particles like aerosols. The LIDARs work by transmitting short pulses of laser light into the atmosphere. A small portion of that light backscatters at 180 degrees from particles in the atmosphere. It's collected by a telescope and recorded using the detector and a data system. The NPLnet uses this type of elastic backscatter LIDAR. There are also advanced LIDARs that also profile aerosol in clouds using different techniques such as Raman and high spectral resolution.

Also other types of LIDARs that measure winds, ozone, et cetera. I'll focus on elastic backscatter LIDAR in this presentation. So the lidars obtain optical proxies of the aerosol and cloud concentration composition and shape. We start by getting the altitude or the range of the particle layer, which we determine from the time between the pulse transmission and the reception of the signal back in the lidar. The light's attenuated as it travels up and back through the atmosphere by scattering and absorption.

That in total is referred to as extinctions. So it's the backscatter and extinction parameters in our measured signals that are the optical properties of aerosol and clouds. That's what we want to get from the data. The laser lights often polarized in one orientation, and the scattering by non-spherical particles like dust or ice clouds depolarizes the light by an amount proportional to the non-sphericity of the particles. So the polarized LIDARs can measure what we call the depolarization ratio.

And that is proportional to the particle's degree of non-sphericity. So again, the idea is that the remote sensing data from our LiDAR provides optical proxies for the vertical profiles of aerosol and cloud concentration, composition, and shape.

On this slide, I'm showing an example of a LiDAR signal on the left. The x-axis is the signal strength, and on the y is the altitude. The blue line is a molecular-only signal, so that's assuming there were no aerosol and clouds present. The green line is a measured LIDAR signal that's been calibrated. You can see that the signals start out very high, and then they gradually decrease with range. That's because the signal is being continuously attenuated as it propagates through the atmosphere. You'll also notice on the right, this is the depolarization ratio.

And again, there's a molecular baseline. The molecular atmosphere does depolarize light a little bit. And in green, you can see the measured volume depolarization ratios, and they're measured in percent. So on these plots, anywhere you see the green signal or the green depolarization ratio greater than the molecular background, that is due to the presence of aerosols or clouds. The micropulse LiDAR network uses the micropulse LiDAR. That's shown here on this slide. It's the first instrument to the left.

It was the first commercially available autonomous eye-safe LiDAR, and it's eye-safe at a green wavelength. It's suitable for network operations because it can run continuously. It was originally developed at NASA Goddard Space Flight Center in the early 1990s. It was patented and then licensed for commercial use. The license has since expired. But over time, the instrument has continued to be sold by numerous companies. And at the moment it's available from Droplet Measurement Technologies. The instrument acquires atmospheric profiles from 250 meters to 30 kilometers at one minute temper resolution.

And this instrument became polarized around 2008. Unfortunately, the MPL itself is no longer available. We're now using the mini MPL shown to the right. This was an offshoot development of the original MPL. It's the same basic design and performance. It's smaller, easier to ship, but it does have a slightly lower signal to noise ratio. So the primary job of MPL-Net is to install and operate these sites and collect and archive the data. We then calibrate the instruments and transform the raw data to a level one signal product.

And this is a signal in the form that is usable. We retrieve cloud and aerosol and PBL properties, and we also provide level 1.5 and 2 quality-assured products. Finally, we provide the data products to the public, and all of our data is open and available, just like Aeronet. And then, of course, we assist with interpretation of the data and collaborative research. So this slide, I will discuss how we transform raw data to the atmospheric parameters and the signal that we then use to retrieve backscatter and extinction.

So as a reminder, it's these two parameters represented by beta and sigma, backscatter and extinction, that we're after. I'm showing the signal plot to the right again. And at top, this is the equation form of the signal shown in the plot. We call this NRB, or normalized relative backscatter. I've broken up the terms into molecular and particle, or aerosol and cloud. So you see that our signals we're measuring are directly proportional to the backscatter at a given altitude or range. Then they're attenuated by two transmission terms.

The first transmission term here is due to molecular scattering. The other transmission term in the end is due to aerosol and cloud attenuation. And this integral term in the exponent is actually the aerosol optical depth from the surface to that range. So there's a connection back to ARINET, for instance, that you've heard earlier in the training. There are a number of calibrations and instrument parameters that we have to process in order to produce the LiDAR signal you see on the right. Those include the laser energy, the solar background, the sun does emit some light at the laser wavelength, and we have to subtract that.

Interestingly enough, that signal noise for us is the entire signal for Aeronet. And we, of course, have the range and the diagnostic temperatures. There are numerous calibrations that must be applied to the data. They include accounting for the performance of the detector, laser detector crosstalk, and we call this after pulse. That's an artifact of the MPL design. It's a transceiver, so when the laser fires, the detector sees a little bit of that signal, and we have to subtract it. Then we have receiver overlap, which is a calibration for the near-range signal loss, and I'll discuss that on the next slide.

And then finally, we calibrate the polarization to produce accurate depolarization ratios. There's one instrument parameter we do not remove from the signal, and that's called the LiDAR receiver efficiency parameter, or C, sometimes called the LiDAR constant. And that's shown in the equation here. So this is what would convert our signal units to SI units. And in the appendix of the slide deck, I provide more in-depth discussion of the LiDAR signals, calibration, and our processing techniques. So on this slide, I'm showing the same two signals in green, and in red, what the signal looks like with no calibrations applied.

So on the right, you can see if there's no after pulse calibration applied, we end up with slightly too high signals due to that noise term. It may not look like much on the slide on the scale shown here, but these can interfere with our retrieval algorithms. In addition, we do a slight range correction. You can see a little bit of difference in range at the base of the cloud here at about 12 kilometers. The overlap calibration is shown at the bottom, where you can see using the LiDAR, we're not able to accurately image signals closer to the surface.

So we account for that using a separate receiver channel that has a very wide field of view and a very short overlap. Polarization calibrations are done with both lab and measured data, as an example shown to the right.

I'll discuss interpreting LiDAR data on the next few slides. Our LiDAR data is often displayed as a 2D representation of the atmosphere shown on the right, where the x-axis is time and the y-axis is altitude. The data variable you're plotting at any time and altitude is represented by a color map as shown here. So I'm showing the LiDAR signals in the top plot from one of our NPLnet sites in Barcelona, and you can see aerosol near the surface in some high altitude clouds around 10 kilometers. And I want to focus on these mid-tropospheric layers around 5 to 7 kilometers.

If you look at them, they look like the signal is weaker or less than the signals near the boundary layer. The same with this layer here towards the end of the day, if you can make it out. And I want to remind you that, again, the signals are being attenuated as they propagate through the atmosphere. So you cannot make a conclusion about the concentration in these layers relative to those below from the LIDAR signal alone. We have to do retrievals of the aerosol properties to do that. On the bottom, I'm showing the volume depolarization ratio.

And again, I want to remind you that this variable indicates the sphericity of the particles, not the concentration. So the higher values of depolarization indicate more non-spherical particles. And you can see that in the ice clouds up around 10 kilometers relative to the aerosols near the boundary layer. But the volume depolarization ratio we measure from the signals is not the same thing as the aerosol or the cloud depolarization. So again, we have to do retrievals of the properties to get an accurate measure of what particle depolarization is.

And that's shown on the next slide here. And I'll discuss our aerosol retrievals later in the presentation. But if you look at retrievals of aerosol backscatter, you'll see that those weak signal layers aloft actually correspond to backscatter that is very comparable to what's near the surface, the same near the end of the day. And if you look at the aerosol depolarization ratio retrievals, you'll see that they're higher than the volume depolarization ratio. So next, I'll move on to discussing the project and our data.

On this slide, you can see our sites on the map. These are sites from 2000 to current. The green dots are operational sites. The red are closed, and the yellow are planned. We have 85 sites in total, uh, 28 are currently operational and we have about 10 sites in planning phase that changes year to year. We also have near complete co-location with Aeronet. Uh, that was the goal of the project. RMPLnet history has involved three version releases. And again, these are similar to Aeronet. Version one and two span the 2000 to 2021 timeframe.

We're currently in version three, which was released in November 21. Version releases include things like new instrumentation, data products, or data processing capabilities. All the version two and three data are currently available on our website, and there's a link at the bottom here.

The instruments that we deploy must be put inside of an environmentally controlled enclosure, and that's shown on the pictures here. For both the mini-MPL and the MPL, The enclosures have an optical quality window on the roof, and we sometimes install this tube on the top to block more of the sunlight and increase our signal to noise. There are controls on the outside for the environment and the temperature and humidity inside, and an Ethernet cable for data communication. We can also use Wi-Fi if needed.

We're a federated network like Aeronet, which includes both NASA and domestic and international partners. Data are acquired continuously day and night at a one minute temper resolution and 30 or 75 meter vertical resolution. The raw data are all transmitted back to our central server and automated processing of the data occurs and the products are available in near real time, which for us is hourly. Our data product suite is shown here. We have four products organized along themes and each of them contain variables and diagnostics.

Our NRB product contains the signals, We have a cloud product, aerosol, and PBL product, and I'll go over each of those in a second. All of our data files are in the standardized format. They're met CDF4 CF compliant, and we have full error propagation from raw data to the final variable. All of the error variables are in each of the product as well. The level one and level 1.5 products are available in near real time, and we have QA flags in all of the products. This is our product level table.

And again, it's aligned with Aeronet. So we have level one products. All of the data are available in near real time. There's no QA screen applied. Our level 1.5 products shown in the middle are the same, except we're applying a near real time quality of screen to the data. We're utilizing the NASA GEOS 5 model for all of our neurological inputs to calculate the molecular quantities. Finally, we have Level 2 data shown here, and these are available after post-calibration, and for the aerosol product, after Aeronet Level 2 data are available.

I mentioned we have numerous QA and confidence flags in the data product, and I'll talk more about those during the web tour. Next, I'll discuss each product in more detail. Our NRB product is shown here. The relevant variables are the LiDAR signal, the volume depolarization ratio, the error variables, and the QA flags. The references are shown here. To the right, I'm showing signal data from NASA Goddard Space Flight Center site on May 28th of this year, and the volume depolarization ratio below.

You might notice that there's a very intense layer of depolarizing aerosol in the mid-troposphere, and I'll come back to that later in the presentation. But in addition to the signal and volume depot data, we have diagnostics. So in green is the laser energy, in blue are the temperatures. The solar background is in yellow. And the purple line here is the surface temperature from the GEOS-5 model. Our cloud product is shown here. The relevant variables are cloud base and top heights, a cloud mask.

We have hourly and daily cloud fractions, both column and then organized by altitude level. estimates of thin cloud optical depth and extinction, and cloud phase.

The reference for the cloud product for the first four bullets up top is shown here. And the left-hand plot demonstrates that we were able to improve our detection capability for thin clouds from version 2 to the current version 3 system. The plot shows cloud occurrences versus altitude. And you can see around 10 kilometers, we are improving our cloud detection. Our cloud phase variables are shown on this page, and there's a reference at the top for that. The top plot, you see the LiDAR signal as the front is moving through.

The middle plot is the volume depolarization ratio, and the bottom is the cloud phase demonstrating there's ice clouds near the top transitioning to mixed phase in purple. And then down near the surface, these gray layers are water clouds. On the right is an example of some utilization of the data for science. There's supercooled liquid water fractions from NPLnet compared to Calypso. The LiDAR in Calypso is called Calliope. Now, I'll just mention we have many publications examining Cirrus clouds with NPLnet and Calypso data.

Probably more important for this training program, however, is to note that our NPLnet version 3 cloud data were used to validate the new thin Cirrus cloud screen in Aeronet version 3. Next, I'll talk about our aerosol product. The relevant variables are the aerosol top height or the top of the highest aerosol layer, backscatter and extinction profiles, that LIDAR efficiency parameter, new term, the LIDAR ratio, I'll talk about in a second, and then various other diagnostic variables. Essentially, our retrievals are the same for aerosol as we had back in version one, but we've added addition of a new lunar AOD data from Aeronet.

And of course, then with polarized data, we now have aerosol depolarization ratios. So the first thing we do in the aerosol retrieval is to apply a cloud screen at one-minute temper resolution using our cloud product. There's no clouds in the example I'm showing here, but if there were, those minutes would be removed from the signal average. We then apply a running 20-minute signal average to the data, and these cloud screen statistics are kept and used in the QA flags. The next step is to apply our aerosol top height algorithm.

And the references are shown here. I don't have time to go into the detail. But our aerosol top height parameter is now shown overlaid on the bottom image. So anywhere above that, we determine there are no aerosol present. Our aerosol property retrievals are discussed on this slide. And for the retrieval of the aerosol properties, we utilize the ARINET optical depth as a constraint. So if we go back to the LIDAR equation I showed earlier, You'll note that we have two parameters we want, the particle backscatter and the particle extinction.

So we have one equation and two unknowns, and on top of that, it's a transcendental equation. However, solutions are provided by Fernald and Klett. They require the introduction of a new parameter we call the LIDAR ratio, which is below. LIDAR ratio is the extinction to backscatter ratio. And it's not just a trick used to solve the equation. The LIDAR ratio is actually inversely proportional to inherent properties of the aerosol, namely the single scatter albedo, which is the absorption of the aerosol, and the phase function at 180 degrees, which is related to the size and shape.

So the idea with the solution here is to set an a priori value of the LIDAR ratio. They typically just pick a number of 50. Usually this parameter varies from 20 to 80. Then they replace the extinction in the equation with the LIDAR ratio times backscatter, and you solve for backscatter. However, by setting a number for the LIDAR ratio a priori, you can introduce large biases in the retrieval. So for MTLNet, we solve this by constraining the retrieval, where we force the extinction to integrate to the ARINET optical depths.

So what we do is first set the LIDAR ratio to a value of, say, 100. We solve for the backscatter profile. Then we calculate a new value of the LIDAR ratio as shown here. It's the aerosol optical depth from Aeronet divided by the integrated backscatter profile. We repeat this and iterate it until the LIDAR ratio converges. This method does not require operary specification of the LIDAR ratio. It calculates a column average value in the solution. So an example of this retrieval for one of the ARINET observations on the day I showed in the last slide is in the corner here.

This is the extinction profile. All of the ARINET data are included in our MPLnet aerosol file. So you have a correlation of the ARINET and the MPLnet retrievals. They include the optical depth, the angstrom exponent, our calculated LIDAR ratio, of course, the extinction and backscatter, and even the sun photometer or the ARINET derived LIDAR ratio.

So on this day, I'm showing in green in the bottom, these are the aerosol optical depths from Aeronet. So every time there's an aerosol optical depth observation, we can do the retrievals I showed before. So now on the top, you have backscatter profiles at the Aeronet times. And on the bottom, these are the LIDAR ratios in orange retrieved from the process. You'll notice that there are gaps in the data where Aeronet has no observations. We can resolve that by using the LIDAR efficiency parameter.

We can calculate C by using the aeronet optical depth as shown below. So if we go to an altitude above the aerosol top where there's no aerosol and there's no clouds, the LIDAR equation simplifies to the form shown here, and we can solve for the system constant. So in purple, these are the retrievals of our system constant shown on the bottom plot. In order to fill in the gaps, we interpolate the C values to our one minute temple grid shown here. So now we've filled in the day. And these can be used to calculate the column aerosol optical depth at the times without the ARINET observations.

So now we've filled in the aerosol optical depth for the entire day. Now, wherever we have an aerosol optical depth, we can do the aerosol retrievals. and fill in the rest of the day with the full backscatter profiles and the LIDAR ratio. So our final aerosol product contains both constrained and interpolated data. The interpolated data will be lower quality. I mentioned there are a number of QA flags in the product, and one of them includes an indication of what type of optical depth was used. Was it day, night, or interpolated?

These are the references for our aerosol retrievals. And before I continue, I wanted to mention a caveat. The retrievals shown here Assume the lidar ratio is constant with altitude. This can bias the results if the lidar ratio actually changes. That's a limitation of backscatter lidar, and you can solve this by use of a Raman or HSRL lidar. So lastly, we have our PBL product. Relevant variables here are the mixed layer height, which is our proxy for the PBL height. And we also have a mixed layer aerosol optical depth, which is the optical depth only in the mixed layer.

A reference is shown here. On the right, these are comparisons of the PBL derived from Radisson launches with that from our mixed layer depth and MPLnet. On the next slide, I'm showing our LiDAR signal over a two-day period. I'm superimposing virtual potential temperature and demonstrating our mixed layer heights. The algorithm actually retrieves three potential heights and down selects to a final. That's shown in red on the plot. So you see that we're able to capture the diurnal evolution of the PBL layer while it's at night, as it grows in the morning, peaks in the afternoon, and then begins to collapse in the evening before growing again the next day.

Here's the 10-year climatology of the mixed layer height and the column aeronet optical depths and the mixed layer optical depths at Goddard. And if you look at the optical depths, you'll see aeronet's fairly constant day to night, but you'll see the mixed layer AOD decrease it in the evening and increase into the day as the PBL or the mixed layer grows and entrains more aerosol. You'll notice there's a gap of about 0.05 optical depth, and this indicates that on average about 0.05 of the aeronet optical depth is above the boundary layer.

So in summary, these are some images of the data that we provide from NPLnet. For a three-day period on the left at Goddard, there's signals on the top. On the bottom, you see aerosol backscatter, our cloud phase mask, and the mixed layer height in brown. On the right is the volume depolarization measured at our site in Barbados, with a Saharan dust layer over the boundary layer. And when you look at the aerosol depolarization ratio below, you can clearly identify the dust layer above the marine boundary layer with weaker depolarization.

Summary of products and development is in the appendix. So next I'll do a tour of the website and demonstrate how you access our data. Topics we'll cover will be project information, network slides, browsing and downloading data, and using our APIs. This is our MPLNet homepage. There's a description of the project at the top. There's a random image of the day on the bottom, which changes daily. First, I'll start with a discussion of our project material. If you look on the left side, you'll see a menu bar.

We go into project and click on version information. This page describes the various versions that we've had up to the one that's currently operating, version three. If you go back and we click on product information, these are the same product and data level tables that I showed in the presentation. Here we'll click on the aerosol product. All of these pages bring you to a description of the product and some references. At the bottom, there's a table which contains all of the information in our product files.

We have the variable, the definition, type, units, and if there's a CS standard name. We also have the global variables described in here as well. There are also buttons to click on QA flags and QA criteria. So if I click on QA flags and I'll scroll down, you'll see flag AOD. For the aerosol product, this is the QA flag I mentioned that tells you whether or not the data were from sun photometer or aeronet optical depth, lunar optical depth, or LIDAR derived. That would be the interpolation. So if you want more information on our products, they're available on this product information page.

Next, you'll see a formats link here. If you click this, this will present an NC dump style representation of our data file, if you're more familiar with that. It's also useful if you want to see what each variable's attributes look like and what they are. Next, I'll discuss looking for sites and our data. So if you go into the data section and we click browse V3 data, a map will appear with our sites. You can, of course, navigate around the map. You can also click on any of the icons and a quick look image will appear of the LiDAR signal with links to the site page.

All of the other information is displayed in tables below. They're organized by the active sites, oops, sorry, and then the inactive sites below. The first column contains the site name with links to the site page. We have lat-lons. If you click these, they bring you to Google Map to view the site. The ARINET site name is shown in this column, and if you click on that, it brings you to the ARINET Synergy Tool. For now, I'll click on the Goddard Space Flight Center site. There's some metadata information at the top and then a calendar view below of where and when we have data.

For Goddard, this goes back to 2001. Anywhere you see a link, we have data. For instance, if you look into the future, say October, November, there are no links available. So you can click on any one of these and go to the data browse page, but you can also click on the month headers. And here you'll get a month view with quick looks. These are helpful if you're looking for a particular event or some sort of atmospheric structure of interest. You can click on any one of these and also go to the site, to the data browse page.

For now, I'll go back to the calendar view and we'll select May 28th of this year. This will bring up our browse page. So on the top will be the LiDAR signal, the volume depolarization ratio, And if you notice, I'm using the same day that I showed for the NRB product example before with this depolarizing layer overhead. The next plot window is an overlay of the aerosol backscatter, our cloud phase. You can see some water clouds here. And then in brown, the mixed layer height. And on the bottom are some diagnostic plots.

You'll notice each plot window has various plotting controls below. So you can control the plot and dynamically change various parameters. You can also change variables you'd like to plot. If you look at the top, we have controls to change the start and end date of the plot, as well as the altitude. So I'll demonstrate that by changing the start time. We'll go back one day to May 27th and forward one day to May 29th, and we'll change the altitude range from zero to 15 kilometers. Now, if you hit any plot button in any of the windows, it will replot all of the data.

So now we're seeing three days of data from 0 to 15 kilometers. In addition, you can add variables to plot. If you look down below each window, there's a green cross. If we hit that, now you can select any level and product that you would like to plot. Here, I'll select level 1 PBL. I'll choose the mixed layer top variable and hit plot, and it will overlay it on the image.

If you notice for each of these, there's a line plot button as well. So if I click that for the aerosol product, it brings you to a line plot display. And in this case, it will show you the backscatter. So there's various ways to look at our data dynamically on the website. For accessing our data files, you click download data here under the data section of the menu. And before I begin, I'll mention our data policy, which is also on the side menu. We ask that you adhere to this before use of any of the data.

It is openly available as you'll see in a second. We have two ways to access the data here. One is the data portal. If you click this in version three, there's an FTP style portal to access the data so you can manually download it. We also have a data ordering tool below and here I'll select the data we were just looking at. So I'll go down to our Goddard site. I'll pick May 27th as the start and May 28th as the end date, sorry, May 29th as the end date. I'll select the level one aerosol product

and then hit search. So here it will tell you that three data files were found and you can click the download button here. agree to our data policy in the pop-up window, and it will compile the files you requested into a zip and download to your hard drive.

You can also select to remove the QA variables flags or subset any of the variables to make the files smaller. Next, I'll move on to our APIs. Those are available under Web Services. We have APIs to download the data plot and look at metadata. So if you click on Download, it will describe how to download the files with our API. That's our third method of accessing our data. If you click on plot, this describes our plotting API. And as an example shown here, if you want to produce dynamically an image with our aerosol backscatter, the cloud phase, and the mixed layer height, here's the URL string example.

So if you click this, you'll notice it will just return that image, not the entire website. Lastly, we have a metadata API where you can download CSV, KML, or JSON formats that are in a Galleon standard. And I'll talk about Galleon as another project at the end of the presentation. As an example, if you'd like to get all of the active sites in a CSV file, there's an example link here. If you click this, you'll see you get the site name, site coordinates, information, status, and then a link to the site page in the CSV file.

If you wanted more information, you could use our standard Galleon JSON format. So if we want to look at Goddard site on October 1st, 23, here's an example. I'll click this and it returns a JSON with some header information at the top, then a station element with all of the station metadata, a contact element with all of the PI science and technical contacts, the date, and then a data list of all the variables available. In this case, this is our signal. So it has the units, it has hours of the day.

It tells you which hour of the day has data and the status, the data level, instrumentation, et cetera. More importantly, in the metadata file are links to the browse page and links to the data file itself. So you'll see the link here. If you click this, that will actually begin to download the data file for that day to your hard drive. So in this JSON format has everything you need to access state site information, as well as access the data. And that concludes the web tour. Next, I'll discuss some case studies using MPLNet data.

MPLNet has contributed to many research projects, journal publications, presentations, et cetera. I'll only show a few examples here. MPLNet has had a global impact on aerosol cloud and PBL research, as shown in this Web of Science citation map back in 2022. We've also contributed to unforeseen research projects, including the ice cube neutrino study in Antarctica. Here I'll focus on some aerosol related examples relevant to this training. I'll start with smoke, which is becoming increasingly important.

We've contributed to a lot of wildfire smoke research, and I'm showing a listing of various publications here, organized by Southeast Asia and North American and European smoke. studies of volcanic plumes, and then a series of papers actually examining both. And we have more of these in press right now. I want to highlight the paper by Osborne et al., where they were looking at the impact of Raikouki volcano eruption in 2019 on flood operations over the UK. The UK Met Office began seeing high-altitude plumes appear about one week before they forecast the arrival of the volcanic ash.

This ended up being smoked coming from fires burning in Alberta that began a week prior. So this paper discusses that, but what I wanted to highlight is that the large fires are becoming increasingly prevalent, and we're having smoke blanketing the northern hemisphere for most of the year. And right now, a lot of the larger fires are dumping smoke into the atmosphere at same altitudes that we typically see with volcanic eruptions. So we're getting an incredibly complex environment for most of the year.

I can illustrate that with this next slide, what I call the wall of smoke example from our MPL net measurements at Goddard back in June of last year. You see some heavy layers here present.

And through analysis, we can identify the lower altitude plumes, the ones that actually eventually reached the surface with smoke from Eastern Canada, namely Quebec. So this was the smoke that blanketed much of the Eastern U.S. that time of year and generated a lot of news stories. In mid troposphere between 5 to 10 kilometers, these plumes are from Western Canadian fires in Alberta. And this layer up high above 10 kilometers is actually smoke from older fires after circumnavigating the globe. So it's important to consider even one fire can produce a complicated environment based on its life cycle, how long it's burning.

It will produce different smoke properties and varied layer heights. If you add in multiple fires with different fuel types, the complication increases. And I can illustrate that by looking at the volume depolarization on this day, where you'll notice that the lower plumes coming from eastern Canada have lower depolarization, which indicates the particles were more spherical. This could be due to something with the smoke shape, or it could also be due to more sulfate being transported with the plume than those from western Canada.

So the point I'm making here is using remotely sensed data, especially passive observations, must be done carefully to avoid misinterpreting what you're looking at, especially with satellite data. And these conditions are becoming more prevalent for most of the year. Another smoke study is shown here. These were Quebec fires back in 2002. In this case, the authors were able to use the signals measured over Goddard and look at the plume heights as they arrived and do back trajectories to follow the smoke back over the source of the fire and determine the injection heights over the fire.

They then validated and compared them to what was being used in the model. Next, I'll discuss some satellite support. This slide focuses on the Calypso mission, which has a LIDAR on board called Calliope. We've been supporting their version three product development. This will be the final release for this mission, which unfortunately has ended already. Namely, we're providing LIDAR ratios from undersampled regions. This is part of a project joint with MERA, shown here on the left. If you're interested in MERA, you can visit the website.

The version 5 aerosol product from Calypso will be quite different over the ocean, where they're actually using LIDAR ratios measured from the instrument. And MPLNet and some other projects are also providing our information to help them generate a more accurate aerosol product over the ocean. We're also supporting an ESA mission called EarthCare that recently launched. ESA EarthCare mission includes ATLID, which is the next satellite LIDAR. It's in commissioning phase now. Data will be available soon.

ATLID's a polarized high spectral resolution LIDAR operating at 355. We're a member of the validation team, and some of our tasks are shown on the left. We're validating the aerosol cloud and boundary layer heights. We're also comparing their precipitation product with one of our own that is in development. And we're doing some joint analysis of Cirrus cloud forcing using MPLNet and EarthCare data. The pictures shown here are a demonstration of using the Calliope data as a proxy for EarthCare to look at cloud fraction and occurrence, aerosol heights, and the PBL height.

Other case studies are available in the appendix. So I'll conclude with a discussion of accessing global LiDAR data. and here I'll discuss a project called GALION. GALION is a network of lidar networks organized through the WMO Global Atmospheric Watch Program. Myself and Lucia Mona are the co-chairs. We have guidance and direction provided by a steering committee of our fellow network heads in the table shown at the bottom, as well as the GAW Aerosol Science Advisory Group. We have a report that was published in 2008.

Galleon LiDAR networks that compose Galleon at the moment are listed here with the red dots, and all of them have signed agreements with WMO's contributing networks. More information and links about each of the Galleon networks is available on our website. There's a link here. The idea for Galleon is to generate global coverage closer to the scale of Eronet that you get with the passive measurements. We recognize the only way to do this is by working together and combining our networks into one network of LiDAR networks.

Our website is shown here. We provide information on Galleon as well as a search and discovery tool I'll show in a second. We also have automated handshakes with the WMO OSCR database, which contains all of the WMO ground-based observations and traceability back to the observational requirements. So the goal is to provide one source for you to visit to get aggregated LiDAR network information and data. I should mention, we're still updating the metadata from some of the networks. So the website is functional, but some portions of it are not available to the public yet.

This is our search tool. If you visit the website and click on sites here, we display all of the networks. Um, and you get a, now a cool, more complete picture of the global LIDAR network data and the distribution of our sites. These are color coded by each individual network. There's a table at the bottom. which will contain links to the network for each site. We also composite sites together if they're co-located. We have links to the site page for each of them, status, and then metadata information on the location.

The final columns are the WMO site name and its YGOS ID if it has one, and then a listing of the other networks and programs that are co-located at the site. All of the search features, including some of them at the top here, where you can search for sites by variable or those that only have data in the PBL or the Stratosphere, for instance, all of those search features are also available via API. I'll discuss some tools in development. These are not available to the public yet, but they will soon be.

We'll have metadata downloads, and that is in that Galleon JSON format I mentioned previously, which does contain links to the browse images and directly to the data files. We also have a tool which can analyze trajectories. These will extract sites along the track, and those are available for high split files, KML or CSV, or volcanic ash advisories, either in the simple text or XML format. So here I'll show a high split example. And I'll go back to that day I showed and ask that, what was that aerosol layer over Goddard on May 28th?

I'm showing it here again for reference. It's highly depolarizing and it looks like it may be dust, but it's very late in the season to have Asian dust transport to the US. So I utilize our high split tool to determine this. So when we go live with the enhanced tool, there'll be a section to upload a trajectory file. In this case, I select high split. And in the bottom, you see the tool will extract all of the galleon stations along each air mass trajectory. They're color coded and grouped together by trajectory.

I'll use this information on the next slide to determine what this layer is over Goddard. So here, if we look at the map of the extracted galleon stations, you'll see that we have data starting on May 28th at Goddard. And then two days earlier, we have data available from the Reno MPLNet site. but if we constrain ourself only to MPLnet data, the next data point we have is eight days earlier at our site in Israel in the Middle East. Using Galleon, we have access to more LiDAR stations and more data, including the Sapporo station in AdNet, which is the Eastern Asian LiDAR network, and then two sites in EarlyNet, which is the European LiDAR network.

So here again is the data shown at Goddard, and now I access the Quick Look data through the Galleon search tool for these stations. And you'll see here at the Reno station two days earlier, they are also capturing this elevated plume. And if we go in temporal order here, we look at the quick look available from the ADNET network, and their quick looks a bit different. They have the LIDAR signal at the top, depolarization ratio in the middle, and then because they use two wavelength LIDARs, they're able to take a ratio of the backs and produce a parameter which is similar to the Angstrom exponent with ARINET.

It looks at the size. So when the air mass went over the Sapporo station, unfortunately, it was at the end of May 23rd into May 24th, and there was a rain event. But immediately preceding that, there is evidence of aerosol reaching deeper into the troposphere, It's depolarizing, and they indicate large particles. So in the Sapporo site, we're also seeing indication of this layer. If we look at our Mimpial Net site in the Middle East, or the early net stations in the Middle East and in the Mediterranean, there's no evidence of this layer.

So we can conclude that the source of this must be somewhere between those stations, perhaps over northern China or Mongolia. And I'll note that one of the trajectories, the lower altitude one, actually stalls out for a bit in this region. So now if we move to NASA Worldview tool, we're able to look at satellite data and incorporate MODIS fires in the VIIRS deep blue aerosol type. We'll notice that there is evidence of a storm system moving through on May 23rd. Even some indication of dust above the cloud layer here indicating that it is at the same or similar altitude we're seeing in the LIDAR data.

The aerosol types indicate dust, some fire, and then some mixtures as well in this area, and there are some fires burning in Siberia during this period. So combining this together, based on the multiple galleon networks and the NASA satellite data, we can conclude that this layer of dust originated over Mongolia somewhere around May 23rd and was likely mixed with smoke from fires in Siberia. So I just wanted to present an example of doing analysis or your research incorporating more LiDAR data than just what's available from NPLnet.

So thank you for your interest in the NASA NPLnet project. And Carl, I'll hand it over to you. So thank you very much, Judd, for that excellent overview of the NPLnet system. And now I'm going to just close us out with summarizing the overall training on our NASA Atmospheric Composition Ground Networks, which support air quality and climate applications. Over the course of the five sessions of this training, we've covered four different NASA networks, the Aeronet Network, looking at aerosols, Pandora and the Pandonia Global Network, which is another passive remote sensing instrument, this time looking at trace gases, The tropospheric ozone LIDAR network, or TOLNET, an active remote sensor looking at ozone in the troposphere.

And finally, the micropulse LIDAR network, or MPLNET, which we learned about today, which gives information and atmospheric profiles of aerosols and clouds, including aerosol properties and the planetary boundary layer height. So between all these networks, there's a wide variety of different types of information that these networks offer on different properties of the atmospheric composition and structure. The networks have various sizes, and many of these networks also work in close cooperation or federation with other international groups and efforts.

For example, today we learned about the Galleon Network, which MPLNet is part of. And together, these give us really a global scope in our ability to observe the atmosphere and collect this information. And this information from all these networks together with information from satellites and atmospheric models really allows us to study various different aspects of the atmosphere, including aspects relevant for air quality and relevant for the Earth's climate in a wide variety of different areas and in a wide variety of different ways.

So hopefully throughout this training, you've learned a little bit about what each of these networks offer and how the data from these networks can be used together with information from other sources to address questions that are relevant to you and your work. At the end of today's session, we'll be posting to the training web page a homework assignment. This assignment is accessible as a Google form. You can submit your answers in that Google form. The homework assignment will be due in two weeks on September 5th.

And to receive a certificate from completion of the course, you do have to complete that homework assignment before that deadline of September 5th. And in addition, you will have had to attend all five of the live webinars that we've offered as part of this training series. If you complete all those requirements, you should receive a certificate via email about two months after the completion of the course. Here is our contact information for myself and Dr. Welton, together with information about RSET and some of our sister programs, if you have further questions.

And here's a summary of some of the resources we provided, not only on Eronet, but on all the other networks we've presented over the course of this training. And now we'll move on to the question and answer session. All right, so we've had quite a few questions and we'll try our best to answer them live during this session. Again, any questions we don't get to live will be answered in this document, which will be posted to the training website in about a week. So to start us off with questions one and two, we had a couple of questions at the beginning about the depolarization ratio, how it's calculated and what it can be used for.

So, if you want to provide a general answer to those questions. Thanks, Carl. Can you hear me? Yes, I can. Okay. Yeah, I know I was covering quite a bit of detail in this presentation. The depolarization ratio is one of the more difficult concepts to communicate. So, I understand why there were several questions on that. The depolarization ratio is a measurement of the two polarized states that the LIDAR emits into the atmosphere. It's typically what we call a copolarized and a cross-polarized state that are emitted into the atmosphere.

Most LIDARs use linear polarization. So there's a parallel or a perpendicular laser state emitted. And then they take a ratio of the returns at those two orientation states, and that is the depolarization ratio. The micropulse light are designed a bit more complicated, as it's a transceiver. So we're using a mixture of linear and elliptical light. But the idea is the same. We take a ratio of the signal returns from those two emitted states, and that's the depolarization ratio.

It requires calibration to be accurate, but the ratio is really a measure of the particle sphericity. So if you have a lower depolarization ratio at a particular altitude, that means there's more spherical particles. You have a higher ratio that indicates the presence of non-spherical particles. Okay, great. And again, I think you noted this for several things, but there are more details that are provided in an appendix to your presentation. and that's been uploaded to the training webpage. So if people have questions, there are some more technical details provided in those slides.

So question three, how many MPL net stations are at or co-located with the Aeronet stations? Yeah, both of our networks change, you know, the network size is in flux year to year, so I don't have an exact number, but probably about 98% or more. Event CalNet is co-located with Aeronet. Really, the only places where we are not are at extreme polar latitudes, like our site at the South Pole. We had another one in Svalbard. And I think all of the few research cruise deployments we've done have also been co-located with Aeronet, just not with the CMOS, and using their handheld sun photometer and the manned network.

Okay, thank you. So questions four and five, you did cover them during the presentation. They were asked a little bit early during the presentation. So we would refer people back to the presentation. And again, we'll be posting the recording of this training in the next few days to the training webpage. So I think we'll go to question six next. Does the molecular attenuated backscatter signal vary day to day? Yes, it can vary day-to-day and based on location. The molecular scattering is dependent on the meteorology, so the pressure really.

So we use the NASA GS-5 model to get our meteorological information at all of the stations, and then from that we calculate the molecular scattering. Okay. Question seven was two different questions, actually. The first was basically what makes MPL-I safe, and then second, what is the sampling frequency, I think, was the question of the system. Yes. The question included the optical depth. The depth in the atmosphere and the eye safety of the instrument aren't really the same thing. But the MPL was really the first visible LiDAR operating a visible wavelength that was designed to be eye-safe.

And you achieve that by using very low energy pulses. Typically, we're about six microjoules per pulse And then we also expand that beam through the transceiver to the full diameter of the instrument. So it expands to almost 20 centimeters in diameter. So the energy density is very low anywhere in the beam. And the only way we get good signal-to-noise addresses the second part of the question, and yes, we do that by using a high repetition rate laser, in our case, 2500 hertz. Okay, thank you.

So question eight, how do you distinguish between the depolarization from non-sphericity versus multiple scattering? Yeah, this question really could be applied to but to other aspects of what we're measuring, not just depolarization. But as a result of being an eye-safe design, the MPL and the mini-MPL use very narrow fields of view on our receiver, and we do that to reduce the solar background noise. So typically, we're about 100 microradian field of view, which is much narrower than a typical LiDAR system.

So since we have a narrow field of view, this really removes a lot of the multiple scattering problems that you encounter with other designs. If we're looking at a very thick layer, it's possible we could begin to see some multiple scattering, but we also tend to attenuate very quickly in those layers, really probably before we begin to detect any multiple scattering impact. Okay. Um, so question nine is how does the polarization induced in the, the instrument, um, and do you measure both parallel and perpendicular polarization?

I think you discussed this a little bit when discussing the depolarization ratio, but if you have anything to add here. Yeah, I think this is a similar question. I was answering them rather quickly, but, uh, I don't know if I have much more to add. Um, the, uh, Yeah, really getting too much into the detail here was beyond the scope of this presentation. So I can point you to our website where we have more information on this as well as links to other papers. The polarization design for the MPL was actually developed by colleagues of ours in the DOE ARM program.

And there's a paper they have out on that if you want to look at the original design. OK. Yeah, that's good. Okay, so question 10, how do you maintain all the time beam alignments for both TX and RX? Yeah, I guess maybe I misinterpreted this a bit, but well, I can address what I wrote and then I'll add a bit more. In terms of timing, the NPL data system controls the timing between the emission of the laser pulse and then reception of the signal. We use a photon counting detector for that. We do a small range correction during the calibration process.

But if they were talking about the alignment of the transmitted and received signal, then the NPL is a transceiver design. So we're biaxial and we don't have this type of problem. So there's no real boresight required with an NPL that's handled during the construction and initial alignment and calibration of the instrument itself. Okay. So question 11 deals with maybe plans to expand MPLNet to Ethiopia, and maybe we can expand this question. There were several questions related to basically how do I join the MPLNet, how do I purchase the instruments, and so forth.

So maybe you could just address those here. Sure. Yeah, this is probably one of my number one questions I get a lot. So specific to Ethiopia, unfortunately, no, at this time there aren't plans to deploy something. Um, in general, we, we really can't grow as large as Aeronet because we're using a LIDAR and the LIDAR cost and complexity is just much greater than the sun photometer. An inherent limit based on budget and, you know, uh, our staff size to handle all of that. Um, so we're a little bit more constrained in what we can deploy.

We do have a list that we maintain of planned deployments that is dependent on funding, but also availability. Much like Aeronet, we're using a commercial instrument. So if you purchase one yourself and are interested in joining the network, that does make it easier for us to discuss incorporating you into MPLNet. So if you're interested in buying one, I'd recommend that you contact the company. At the moment, our list of planned stations still includes sites we had planned during COVID. So once we clear that list, which should happen at the end of this year, we'll probably be more open to looking at other stations.

Okay, great. So question 10, can we measure stratospheric aerosol with MPLNet? Yes, the answer is yes. If you want to retrieve properties of aerosol with stratosphere, we would have to do much longer time averaging. We don't actually have any stratospheric aerosol products, but we can do some custom analysis where needed. However, in terms of just protecting the stratospheric aerosol layers, that is quite easy, especially at night with the MPL. Our signal data is acquired up to 30 kilometers. I think most recently we looked at the wrong eruption in Indonesia, and we detected that over Singapore site at about a 22 to 23 kilometer altitude, I believe.

Okay. Question 13. Are MPL-Net vertical profiles all zenith measurements, and does it have a different footprint or spatial representation, I guess, with respect to Aeronet, since Aeronet does direct sun scans typically? Yes. The answer to all of that is yes. MPLNet data are essentially nadir. We're looking straight up. We have a small angle tilt to avoid any specular reflection from ice clouds. But in terms of this question, we're basically pointed straight up. The aeronet instrument is looking at the sun and then, of course, doing sky scans for radiance.

But we do have different view angles. The MPL can be scanned with some additional equipment. We do not do that for MPLNet operations. OK.

Question 14, is the aerosol extinction calculated using the classical CLET inversion technique, and what was the LIDAR ratio used? I think you also, you know, covered this in the presentation to some extent, but if you want to add any details here. Yeah, I think this question came a little before that slide. So yes, we use this Fernald-Clitt inversion technique, but we use the error net optical depth as a constraint. So we don't specify or set the LIDAR ratio. We actually calculate the column average value.

Okay. So the next question was about how to download the MPLNet data. Again, you covered this in the presentation a little bit later after the question was asked, but the data are available through the MPLNet website. Okay. Yes. The one thing I may add is there may be some data we have not finished reprocessing in version 3 yet. That data would be available in our version 2 archive, but if you're interested or need some data in version 3 that are not available, you can contact me. Okay. Question 16 was another question related to basically how to join the network.

I think the only thing different here was Questions about some other instruments, but MPL-Net only supports the MPL-Net instruments. So let's go to question 17. How is the range-corrected backscatter signal normalized, and with what value is the normalization done for the range-corrected backscatter signal? Yeah, this was discussed in the slides a bit. I have more on this in the appendix. What we call our NRB signal, the normalization part of the name refers to the fact that we normalize the measured signal to the range, to the range squared to remove the range square fall-off and get some scattering.

And then we also normalize it to the emitted laser energy. And so that produces our signal, the attenuated backscatter signal, which you typically might be used to from Calypso. That involves removal of this LIDAR efficiency parameter or the LIDAR constant. We do not provide attenuated backscatter from NPLnet because it's more difficult to do that from the ground. We do have that variable in our aerosol product, so you could use that to then calculate attenuated backscatter from our signal product if you wanted.

Okay. Question 18, do we need to perform any background corrections? Well, I interpret this as you, the user, and the answers, no. At ZKLNet, we provide all of the calibration data processing. Yep. All right. So, question 19, I think there were a few questions about the interactions between NPLNet and Calypso. You discussed this to some extent in your presentation, but maybe you could just discuss a little bit more based on the answer here, the synergies between NPLNet and Calypso. Yeah, there have been some studies and publications.

The word you used is a synergy is probably a better term. There have been several. Most of them have been cloud-related. However, I'll mention that it is really difficult to do a direct comparison between Calypso or satellite LiDAR. For instance, now EarthCare is in orbit. It's very hard to do that direct comparison. between satellite LIDAR and a fixed ground station LIDAR because the swath of each instrument is so narrow. And, you know, the satellite's moving quickly. So it's really better to think of a synergy where you're combining the data from both.

And the satellite LIDAR provides a lot of spatial coverage, not much temporal information, especially from low Earth orbit. But the ground stations don't have much spatial information, but we have a lot of temporal information, especially in MKLNet because we're operating 24-7. Okay. So question 20, will the C values eventually be used for completing aeronet level two data in order to avoid problems of not being able to measure when clouds are present? And could this C factor be used to complete the data when there are technical problems in the instrument?

Yes. So our level two NPL net aerosol processing is the same as I described in the presentation. The only difference is that we would use level 2 ARINET optical depth as our constraint. For level 1 and 1.5 processing, we're using level 1.5 ARINET AOD. That's really the only difference. We do use the C values to track the health of the instrument along with the instrument diagnostics, and then, of course, to produce what we call our interpolated aerosol retrievals. But just like in level 1.5 data, those would be They would be level two data, but they would be marked as less consonant in our consonant flag.

Okay. So question 21, what is the agreement between the PBLs from NPLnet and from Radiosons? Again, it was discussed briefly on the PBL products. Our original PBL product or really mixed layer height, I should say, not PBL height, but a mixed layer height variable was discussed in the paper reference that I provided on that slide. And as part of that study, there was a comparison of our retrievals to the PBL height from Radiosonde data. There have been some other work like that, but a really large scale comparison between RNPLNet Mix layer heights and radius on PBL heights have not been done, but I'd love to see that.

Okay. Question 22, how is cloud screening done for MPLNet signals? Yes. So our data are all on a one-minute temporal resolution. For the aerosol product, we do a 20-minute signal average. So any minute of data that contains a cloud-based signal in our cloud product that is below the aerosol top height is screened from that 20-minute average. If a cloud is detected above the aerosol top, then the signal is left in the average, but that is flagged in our product. Ideally, aeronet data should be cloud-free, but we do typically see the presence of clouds, especially thin cirrus, above the aerosol layer.

even though ARINET has data. And this could be due to the different view angles, as we discussed earlier, or the potential of just missing that cloud in the ARINET processing. So we do flag that in our data. And that was part of what was used to help in the development of the version 3 ARINET data using our cloud product and the detection of these mist clouds in the aerosol product. Okay. Okay. Question 23, what are the differences between the mini MPL and the CL51 and 61 solometers? Yes, I was waiting for that question.

The differences are shrinking over time, but typically the MPL is still classed in the LiDAR category, not really a solometer. That's mostly because we're using a green wavelength in our case versus a near-infrared wavelength that's in most solometers. As a result, we just have a better signal-to-noise ratio than you typically see with solometers, and also a longer range. We do profile up to 30 kilometers. Most solometers are capped at much lower altitudes. The MPL is also polarized, and until the recent release of the CL61, that was not available in solometers.

So we are looking at intercomparisons now and more planned between the mini-MPL and the CL61. Okay. Question 24 is another question about basically, you know, how to expand or grow the network. I think you've already addressed that previously. So let's go to question 25. Can we discriminate between the aerosol types? And I guess the question is basically, which is more accurate between the Calypso and I guess the MPL LIDAR? Sure. So I'll say it is definitely more difficult to discriminate aerosol types with just NPLNet.

We really have only one wavelength, which limits what we can do. Calypso has two wavelengths, 532 and 1064 in the near IR. So they're able to get some information about particle size. We both have polarization measurements. So if you add the extra wavelength, Calypso can do a little bit more because they have information on particle size. But really, if you want to talk about aerosol types or aerosol properties, which are really what you use when you're looking at the type, it's better to use multi-wavelength Raman or HSRL instruments to do that.

For MPLNet, we can infer dust or smoke versus other aerosol types, but that's pretty much the limit. OK. Question 26 is another question about basically joining the network. What's the cost of a mini MPL? And that's provided. We recommend you go to the company's website to look that up or contact them. So question 27, basically should the Aeronet AOD data be available for the same sites as the MPLNet? Another question related to co-location. And then what maybe is the maximum distance between the sites?

Yes. So we do require co-location with ARINET, as I said before, except in situations where it's just not appropriate, like our extreme sites where it snows most of the year, it's too cold for the female. And in terms of the co-location, we limit the distance between MTLNET and ARINET to less than one kilometer, ideally much closer, especially if you're in a region with a lot of local aerosol sources. And again, I would mention there have been some early deployments in the network where we did not use aeronet optical depth.

And those, I think, were some early research cruises that predated man, the man portion of aeronet. But for now, we require co-location with aeronet. Okay. So questions 28 and 29 were both quite expansive questions. I don't know if you want to try to take some time to answer these now, or we can come back afterwards and maybe point to some papers or references for these. Sure. I can give a bit of a, well, I'll do my best with question 28. Okay. I guess I'll say that the, I guess they're kind of asking what are typical depolarization ratios for, say, ice, water clouds, and aerosols.

So water clouds will have very low depolarization. Ice clouds will have very high, typically 30% or 40% in that range. Aerosol depolarization ratios can vary pretty wildly. Spherical particles such as sulfate and marine aerosols like sea salt, which tend to be more spherical, unless they begin drying, have very low depolarization also. Dust has much higher depolarization. So if it's a thick dust layer, that's typically in the 30% range, somewhere there. If you're looking at the volume depolarization for dust, it can be lower.

As I tried to explain in the presentation, the volume depolarization is not the same as the particle depolarization. The volume depolarization ratio is a ratio of our signals. The aerosol depolarization is a retrieved product, so it's more accurate. Smoke depolarization can vary pretty wildly. It depends on the size and the age of the smoke particles, but they typically may be in the 15% to 25% range in depolarization. These are all kind of vague numbers. That's why I didn't want to give anything specific.

In terms of backscatter and extinction parameters for NPLNet, those are really not what we would use to explain to someone whether this was an ice or a water cloud. That would be more just depolarization. But if you want to begin looking at aerosol typing, which is, I think, what this meant, again, that can be very difficult with NPLNet because we're using a more simple LIDAR. If you have multi-wavelength Raman or HSRL you have more accurate backscatter and extinction profiles, and at different wavelengths, you can begin to do retrievals of the properties, like the index of refraction, look at the absorption, things like that, that help you differentiate aerosol species.

Okay, great. And then question 29, I guess, similar, I don't know if you want to address this a little bit now, but I guess the question is basically how to distinguish between aerosols either in the boundary layer or outside of it. It's pretty broad, but I guess if you want to provide some information. Sure. So our aerosol product provides an aerosol extinction profile, and our PBL product has the mixed layer height. So we can take our retrieved aerosol extinction profile and integrate that from the surface to the mixed layer height.

And that then is the aerosol optical depth in the mixed layer. So that's different than the aeronet optical depth that you're getting, which is the column. So that's all of the aerosol optical depth in the entire atmospheric column. So the mixed layer optical depth will always be less than the aeronet A or D. And I believe they're referring to the slide I showed on our new products we're developing, our level three products where we're looking at climatologies. And for the Goddard Space Flight Center, sorry, for our site at Goddard, over a 10-year period, there was a difference of about 0.05 aerosol optical depth at midday, the height of the mixed layer, compared to that of Aeronet.

So that indicates that on average at that particular site, there's about 0.05 optical depth in the free troposphere or also stratosphere, not at the surface. So if you have wider profiles of aerosol extinction and an idea of the PDL height or the mixed layer height, you can determine what the amount of optical depth that's in the boundary layer. I hope I answered that question. Yeah, I think that's a good answer there. Okay, so question 30. Is the AOD at the MPL wavelength obtained from simple interpolation of aeronet AODs?

Yeah, I'm not sure what simple, if this is considered simple, but we use kind of a standard two-order polynomial fit to the aeronet optical depths across their wavelengths. That's how we then get the optical depth at 532. Okay, yeah. Okay, question 31. Has MPLNet been used for large-scale aerosol transport, I guess specifically for volcanic eruptions? Yes. Yes, it has. There's a number of publications and studies people have done. You can see the book on our publication page on the website. Some I listed, I believe, in the case studies section.

Interestingly, it's sort of morphing into combined studies of volcanic eruptions and large fire and smoke events. But there are even a few more in process now. I think people are still looking at the Ryokoki eruption at this point with our data. But I will mention that the studies are being done not just with NPLNet, but data from the other LiDAR networks that are in Galleon, because together we get a denser coverage globally. And then I guess related to that, question 32 again asks about expanding the network, but with specific reference to South America.

So if you want to talk a little bit about the networks there. Sure. So, you know, with some budget constraints with NPLNet, my focus for expanding the network has been to fill in the gaps between my colleagues and the other LiDAR networks, most of whom are regional, such as EarlyNet in Europe, AdNet or ADNet in Eastern Asia, and LollyNet, who are in South America, or really Latin America, they're really working regionally to get a denser LIDAR coverage. So most of where I expand MPLNet are into regions where there is no LIDAR coverage.

So when it comes to LollyNet, I've been collaborating with them. We do have one site that's joint with LollyNet. We are talking about some more MPLNet deployments into the region. But if you're interested in that region in particular, I would recommend that you contact my LollyNet colleagues, and there's a link on our galleon website to go to the LollyNet page. Okay, great. So question 33 is specifically about the effectiveness in detecting Saharan sandstorms and their characteristics. Yeah, I mean, I can say the entailment instruments are very effective.

capable of detecting the presence of dust compared to other aerosol types. And that's not just in regions near the Sahara in this case, but even far downwind and remote. Not sure how much more to say about that. Again, if you really, in terms of detecting, yes. In terms of retrieving dust, accurate aerosol properties, especially in the case of mixed aerosol types. It's really better to go to more advanced instruments like Raman or HSRL instruments for that. Okay. So, question 34, what's the smallest size aerosol that can be reliably detected?

Yeah, it wasn't– not really thought of it that way, but– I would say that we're capable of detecting what I would call normal aerosol size ranges, but not, say, cloud condensation nuclei, which should have been CCN. We'll change that. Okay. So question 35, is it possible to reactivate an inactive station? Yes, and that happens somewhat frequently. Sometimes a site will go down for a longer period of time due to an instrument issue, and then it will become active again. And sometimes the more frequent scenario is that we did a field campaign somewhere that was short term for maybe a month or two, and then a few years later, we install a permanent station.

Okay. Question 36, are you aware of environmental agencies, and I guess this is referring to U.S. environmental agencies, using these instruments for exceptional event demonstrations related to the U.S. Clean Air Act? Or I guess if you know of examples in other countries as well, I guess. Yes, within the U.S., I don't know if, well, I should say for NTLNet, we do not have instruments that are at official EPA sites. I know of cases where I've been contacted and we pointed them to our data, but I'm not sure to what degree the data were used, probably not in an official capacity as we're a research organization.

Outside the US, we do have colleagues, especially in Taiwan, that are working more directly with the local EPA. In terms of how they were utilized for exceptional event studies, I could put you in contact with them if you get a hold of me. You really are doing the data processing and provision. The use of the data for situations like that tends to be a bit removed. Okay. But it does sound like it has been used outside the U.S. It sounds like Taiwan would be the people to talk to if people were interested in that.

Yes. I should mention there is another project called UCN, which is an effort joined with the EPA to measure the boundary layer height at ham stations. And many of those groups are doing that by deploying solometers. So there is a growing number of solometers at EPA stations. So you'll probably start to see certainly the use of that data and their measurements of mixed layer heights more and more as time goes on. Okay, great. Question 37, what is the advantage of NPL over a solometer? I think you addressed this to some extent in a previous question, but if you have anything else to add here, for example, is the unified solometer network going to be included in Galleon?

Yeah, so, well, I'll take these questions one on each here. So, again, the MPL still has a better signal-to-noise and a longer range profiling capability than solometers that are available right now. The CL61, since it is now polarized, is definitely closing that gap However, they're still at a near IR wavelength, and the data that I've seen so far is still definitely more noisy than the MPL, so you'll have a shorter range that they can profile during daytime. However, they're much easier instruments to deploy and work with, so there are tradeoffs between the two.

That's why you can't really say one's better than the other. they're built to do slightly different things. As I said, we have intercomparison projects underway now, and I probably have a more concrete answer after those are complete. In terms of UCN joining Galleon, I have had some discussions with them about that. And I'd say they're kind of at an early stage at this point. Okay. Well, I should mention other Solometer networks, I'm also in discussion with, and these would be your operational networks run by the weather services.

So in Europe, there's eProfile, and they will be joining or added to our Galleon website shortly. And then the ASOS network within the U.S., I would like to include their solometer data when it's available or metadata into Galleon as well. Okay, great. And then finally, question 38, what is the minimum laser pulse frequency to get good LiDAR retrievals? Well, for the MPL, that's fixed at 2,500 hertz, and it does not vary.

In general, what you're really after is not the pulse frequency, but a good signal-to-noise. So it's a trade-off between your laser pulse energy and the laser pulse frequency. So you're really after and emitted power to get the right signal to noise. And that's really dependent on your LiDAR design.

All right. Great. Thank you. So that was all the questions that we collected. And we'll, again, be posting this document after we finalize it to the training webpage in the next week or so. But, you know, thank you very much again, Dr. Welton, for that overview. And thank you to the MPLNet team. And more broadly, thank you to all the ground networks, Aeronet, TallNet, MPLNet, and Pandora and the Pandonia Network who supported this training. And finally, of course, you know, thank you to all our attendees for your attendance and your very interesting and insightful questions throughout the training.

Just as a reminder, the homework should be posted to the training webpage. You have about two weeks to complete that homework assignment, and if you do that and also have attended the five parts of this training, you should receive a training certificate in about two months or so afterwards. So, again, thank you very much, everyone, and we look forward to seeing you in future ARSA trainings. Thank you, Carl. Thanks, everyone.

The Story