Hey, good morning, everyone. So I want to start off by a show of hands. Who here has actually heard of the Northern Mariana Islands? All right, almost everyone. Who actually knows where it's at? I was actually quite surprised.

Well, I didn't, and I didn't know until I actually moved there about three and a half years ago. So just to kind of give everyone an idea of where we're at, essentially, if CNMI is a US protectorate, it's located about 15 degrees north latitude in the Pacific.

And if you drew a line roughly from, south from Japan and east from the Philippines, right where they intersect would be where the Mariana's Trench is, and that's where you'd find the Mariana Islands. So the island chain consists of about 14 islands. Some count 15, depends on who you talk to.

But basically what we have here is 14 islands, and as you go down south, Saipan is the capital, and this is the largest island and the most developed. This is where I live. Most of the population lives here. There's about 50,000 people throughout the islands. Going south of that is Tinian and Rota,

and this is where the other populations of humans tend to occur. Anything north of that is roughly uninhabited. There's a few families and a couple people that live in some of these islands. But despite humans, it actually hosts a whole slew of animals and plant species that are endemic and endangered and not located anywhere else in the entire world.

So that's where I come in. I work for the local division of Fish and Wildlife. We're largely funded by United States Fish and Wildlife through their WISPR program, which is the Wildlife and Sports Fish Recreation Program, which gets funding from the Pittman-Robertson Act.

So believe it or not, as Americans, everyone thinks we have, everyone has guns and ammos. There's actually a conservation tax on every firearm and every bullet that's sold, and that goes to do two things, provide game management for hunting and other sports, sporting-type things, as well as conservation of endangered species.

In Saipan, we mostly don't have any game species that we manage, so all our funding actually goes to conserving endangered species. So we do this through a variety of methods, and one is actually provide technical guidance for land use activities and permits. So if you want to develop your land, you apply for a land permit.

You come in, it goes through multiple agencies. One makes sure there's not gonna be soil erosion problems, et cetera. It gets to our agency. We actually survey for endangered species, and we give a yes or no based on whether or not there's a presence of one on your property. We also do more interesting surveys

where we actually, at least once a year, go up to those northern islands. We camp out there. We live there for two or three weeks. We go by helicopter boat, do these massive biological surveys. It's pretty amazing. It's like National Geographic style. It's awesome, and we collect that data, and we revisit these areas once every roughly 10 years.

And we have a whole bunch of other projects going on like using radio telemetry where we tag some of these endangered species, follow them with transmitters, calculate home ranges, geotag bats, especially the Marianas fruit bat, have a data logger that collects information.

It logs like every, I forget, maybe like every 20 minutes so we can see spatial representation of habitat use and things like home ranges. But I'm actually not gonna talk about any of those cool projects, and we're actually gonna go back to this whole technical guidance permitting activity. But we will talk about these things in a little bit.

It'll all tie together. So when I first arrived in Saipan three and a half years ago this was essentially the permitting database. It's all paper copies. Very little, maybe in an Excel file or something along those lines. But it's actually a really important data set because like I said before, if you go out and you wanna clear your land,

you apply for a permit. Our texts in the past used to go out with a paper copy, actually survey for the bird species, write it down on multiple dates, and also draw these like handmade maps. And that would actually be the spatial footprint of your permitting activity.

So it had evolved at some point where we were collecting GPS points using like those little handheld garments. Sometimes you'll just have the point for the waypoint actually located here. Other times you would actually have the coordinates. But this is no good. This is all sitting in these essentially these file cabinets.

There's a whole bunch of important data that we could be using just sitting there. And one thing I wanna kinda reiterate is that there could be multiple surveys for one spatial footprint so that they come back multiple days and do surveys depending whether or not they find a rare species

and things along those lines. So I saw as a major priority was we need to establish some sort of centralized database, get all this data loaded up, and that's what we did. So we took 15 years worth of data. I think it was like 2,700 permits.

And Joe and I, this is my GIS technician, started entering it all in and actually digitizing the spatial footprint. So what we actually used was Google Drive. So we just set up a Google Sheet. Super easy tech, easy to do. Nice thing, it's cloud-based. In the Marianas we have tons of typhoons. There's a lot of typhoon damage.

We don't have to worry about the database getting destroyed in some sort of catastrophe. Another beautiful thing about this is actually we can have multiple users at it at the same time. So as I'm entering historic records, Joe can be on the same Google Sheet entering in records. As new permits are also coming in,

we can have another tech all editing at the same time. And it's version, so we can always look at the history of edits. If someone makes a mistake, completely erases the database, no big deal, we just reset back to an earlier version. And another nice thing about this is I'm able to go in through Python and actually pull this out into a pandas data frame and save it as a CSV file and then use it later for further analysis and manipulation.

So Joe's also going in, he's drawing the historic polygons, sometimes from those maps, sometimes using the waypoint information. But the idea here is all we're gonna do is just have a single field with the permit ID that's gonna link it back to the database with the permitting surveys.

Make sense so far? All good, perfect. So that took about a year, but all the data is in there. And moving forward, we never wanna do that again. So we're gonna automate the data collection and essentially the management. And since we're usually out in the jungles, we want like a really rugged GPS enabled tablets.

We're using Juniper Mesa IIs. We have about five of these. And there's two applications that we run for data collection. For the spatial mapping, we actually use InterMaps Roam, which is based off of QGIS, it's a data collection app. But enabled to collect the tabular data,

since there could be multiple surveys, many relationships to one in terms of the spatial footprint, we designed these PiQT interfaces. And we'll look at that a little bit more. And the reason there is because we have a really complex survey system that is just not gonna work with most likely Qfield or Survey123

or anything along those lines. The nice thing is once the field tech goes out, they can come back. It'll all sync via Wi-Fi to our network. And then we have these Python scripts that run and essentially update the tabular and spatial databases back on the Google Drive.

And then kind of one thing that came out of this was all of a sudden the director really wanted maps produced for every one of these. We automated that process too. So we just loop through each permitting feature in the permit areas, zoom to it, query out all the other polygons,

do a small little calculation that adjusts the scale and then automatically print out the map. And originally this was done using ArcPy, but I've now almost navigated the using PiQGIS, which I think everyone should be happy about. So it's actually not that complicated, but this script just runs every night, automatically produces maps. No one has to worry about making maps.

So it's easy to say, but I think it's best displayed through this series of amateur home videos that I've stitched together and edited probably with some sort of free software as well, I'm sure. But anyways, so I'm gonna take it away.

So let's hope it works. So this is Marlin. Marlin receives land use permit. So she gets it and she starts entering the information into the Google sheet, such as the owner of the property,

the land use type activity and things along those lines. Kika is our other field technician. She goes out in the field. First thing she's gonna do is start actually mapping out the footprint. So she can do it manually, or she can walk around and take GPS points, but that's the Intramets Roam software. And I actually wanna pause it if I can.

At this point, what's actually not displayed is there is a little field form that pops up and just asks for the permit ID. So the unique identifier, it's not displayed here, but I think that's really important to tie it into what I'm later gonna talk about. So then Kika opens up this PyQT app that I built

and it's dynamically linked to what was entered in on that permitting spreadsheet. So these are all the active permits that she's choosing from. She selects a permit. It loads up the information. She can make sure that she's at the right site with the right landowner. There's another tab to collect site information. And this last tab is actually the species survey. So Kika's gonna go select her name, which is Francis.

Kika's her nickname. Start the survey. It shows the time. She's listening for birds. She's starting to hit the spin boxes. She's making counts for different species. So right now she's got two Micronesian starlings, two Bridal white eyes, and a bittern. After X number of minutes, she ends her survey.

That allows her to save the results. And this actually gets saved as a single line CSV file that later comes back, syncs up to our network, script runs, uploads it to the Google Drive. And now her survey is actually back onto that centralized database.

So at night, scripts run again, and it actually shuttles all this information using that permit ID into its own directory. So if the directory is there, it throws the data in there. If it's not, it actually creates it. And we're able to pass it through Python, through HTML, and then make these PDFs

that actually should display the survey results. That can be later used for reporting. So there's a survey, three surveys. Like I said before, we also automatically produce these maps every night as well. And the first time that ran, it ran for days because it produced like 3,000 maps or something.

But they're all there. They were also able to do some kind of neat and nifty things using Python as well, which is produce these monthly reports. And we could actually grab the spatial information from that as well if we link it back with the polygons and get maybe the amount of land that was permitted that month and throughout the year and things along those lines.

And finally, with all these scripts running that are producing these outputs, I actually have them email me and let me know every night whether they succeeded or failed. So there's a good example of a permitting download that failed, but the other scripts kind of ran and things along those lines.

So to re-date the whole process, all the tech does now is enter in the data onto the Google Drive, go out, map the boundary, do their survey, comes back. These usually actually sync up and automate at night. So the next day, they have all this output available for them. Makes everyone's lives way easier.

And our GIS tech doesn't have to make a map every time they go out. They're like, someone goes out to do a survey. So up to date, I launched this in January 2018 and there's 685 permits that have been entered using this system. And by taking the centroids of those polygons,

we can actually map the detection spatially as point counts. And ends up we've entered 6,300 new species detections recorded. So what that means is it's a species count per day, per sample effort. So per species account is one point per date.

So the benefits of this, obviously, it's more efficient data entry and collection. There's less errors in transferring or entering data. And the future steps is to help kind of automate this process for our other types of surveys. So we'll actually have these Python field survey things

with some sort of spatial linked up. So build similar apps for other surveys. But this is fine, this is great. But I don't think this is too impressive. People do this all the time, every day, right? The big picture is by taking all this data from all these different surveys and having that spatial component

with the species detections, we can actually pull it all together into this master species database. So we run a Python script that does this. So all the other surveys I didn't talk about, that's all the data coming in too. And ends up we have almost a quarter million records over 37 years. And this is data that's just, which was either sitting on someone's computer,

sitting on a paper copy just stored somewhere. And this is a really amazing robust data set that helps fulfill our mission needs. So lets us understand and provide better management of the species in the CNMI. And we can do this through looking at species detections over time. And build things like species distribution models using R.

And basically associate habitat associations, plan for basically better population management. So I think I'm running out of time, but just kind of a recap. Basically there's a whole bunch of free and open source solutions out there that actually can help streamline this whole process.

And in the end it benefits us and it benefits the species that we're trying to protect. And with that, thank you. And I'll open up the floor for questions. Thank you Bradley. Five minutes for questions. Anyone?

When was the last survey that you did of the whole island that you said you'd do on a 10 year basis? So we did a survey last year on Saipan. This year we'll be, in the upcoming months we'll be going to Anatan, which is a volcanic island just above us.

Probably spend two weeks there. Collecting vegetation information, invert bird surveys and things along those lines. So we're supposed to go every year to one of the islands if we can.

Sort of a simplistic question. What about when you're doing permit work, is there a cadastral survey? So like presumably people know what land they're on. Very, yeah that's a long story. Yeah there's another agency that manages that

and they've been a little bit slow on updating it. So there's been issues before where the lot information doesn't quite match up or there's miscommunication or something along those lines. So that can be a problem. But in general there are boundaries marked at every lot. So yeah, well actually we require

that the people have their property already, the boundaries already delineated. So they hire a surveyor outside of us because our services are free. They actually mark the boundaries and then our technician goes out. If the boundaries aren't marked, our technician's not gonna machete through the jungle trying to track down where this person's lot is.

So yeah, so there are boundaries. Nice, thanks. How did you go in terms of managing the change? Like was there a big challenge in training the technicians and the users or even selling it to the organization as well?

Yeah, so this was kind of in tears. So I had this vision. I wish I showed my outline on my whiteboard which is just chaos. So first was actually setting up the database in the Google Drive. Getting the field tablets took quite a while.

It's government time and it's government, or island time on the government. So there's a couple stages where I just had them uploading GPX points and actually manipulating it that way. So it being so simple to use Intramaps Roam,

especially when you're just drawing a boundary and filling in one field, that was quite easy for people to get used to. The database, not that big of an issue. All our technicians are actually younger so I feel like they're more exposed in these types of computer programs and things along those lines. And the field survey, everyone loved that little Python app

because it's the same as the field form. It's just way quicker and less writing and things along those lines. I should add like the beauty of having it back, upload back to the cloud, or at least onto the Google Drive, is that if there is an error, that's like where it permanently resides. They can correct like a data entry error there and it's actually gonna flow back down

into the reports and things along those lines. Anyone else? That was great, thanks. Is your code open source and where can I get the code from?

Ha ha ha. Ha ha ha. It's on Bitbucket but it's not, well, I'm more than happy to share it with you but it's not quite like an open program necessarily. So do you do any special surveys

when a typhoon comes around? So, not always. There's multiple surveys that go on. There's a breeding bird survey that the data's collected every quarter. So the last typhoon that really hit us bad was U2. I don't know if anyone heard about that

or without power and water for about a month. We actually re-surveyed immediately after U2. So some surveys we are actually trying to see if we can get information about pre-imposed typhoon effects. And one project that I'm gonna be working on within the next year is actually taking hopefully some multiple band,

multiple spectra satellite imagery and looking at vegetation resiliency following typhoon events at multiple resolutions and we're looking at a major typhoon event in 2015. Another island got hit in 2016. We just got hit in 2018. So I think we'll have some pretty good idea

about how the vegetation seas might rebound and things along those lines. But yeah, that's certainly something of interest to us. That's it? All right, great. Thank you very much. You get that yourself. Thank you. Thank you.