1 00:00:09,840 --> 00:00:11,273 - Hi, welcome, everybody. 2 00:00:12,720 --> 00:00:14,350 Welcome to our first talk 3 00:00:14,350 --> 00:00:17,143 in the technology and innovation track. 4 00:00:18,750 --> 00:00:21,260 I'd like to introduce Adam Zylka 5 00:00:21,260 --> 00:00:22,450 a research specialist 6 00:00:22,450 --> 00:00:24,060 with the University of Vermont's 7 00:00:24,060 --> 00:00:26,460 Spacial Analysis Laboratory. 8 00:00:26,460 --> 00:00:29,787 The presentation is titled "Monitoring Forest Health 9 00:00:29,787 --> 00:00:32,657 "With UAS ("drone") Technologies". 10 00:00:33,820 --> 00:00:35,530 And welcome, Adam. 11 00:00:35,530 --> 00:00:36,650 - Great, thanks (indistinct) 12 00:00:36,650 --> 00:00:38,810 and thanks everyone, good morning. 13 00:00:38,810 --> 00:00:41,430 So I'm Adam Zylka. 14 00:00:41,430 --> 00:00:43,060 I'm the UAS team lead 15 00:00:43,060 --> 00:00:45,693 here at the Spatial Analysis Lab at UVM. 16 00:00:46,846 --> 00:00:50,450 A lot of this work also was done by employees 17 00:00:50,450 --> 00:00:51,680 and student employees at the lab 18 00:00:51,680 --> 00:00:54,000 including Maddy Zimmerman, Lauren Cresanti 19 00:00:54,000 --> 00:00:55,033 and Kelly Schulze. 20 00:00:56,604 --> 00:01:00,880 So the issue we saw in Vermont 21 00:01:00,880 --> 00:01:03,550 which there was another presentation yesterday 22 00:01:03,550 --> 00:01:06,160 maybe some of you attended that 23 00:01:06,160 --> 00:01:08,790 was this spring, we saw issues 24 00:01:08,790 --> 00:01:11,040 with Lymantria dispar dispar 25 00:01:11,040 --> 00:01:12,970 formerly known as gypsy moth caterpillars. 26 00:01:12,970 --> 00:01:15,690 And so, these are an invasive moth species 27 00:01:15,690 --> 00:01:18,810 introduced to North America in the late 1860s. 28 00:01:18,810 --> 00:01:20,990 But in their larval stage 29 00:01:20,990 --> 00:01:23,900 they feed on vegetation and foliage 30 00:01:23,900 --> 00:01:26,480 and sort of the extremeness of this 31 00:01:26,480 --> 00:01:29,850 leads to massive defoliation 32 00:01:29,850 --> 00:01:32,310 and degradation of the tree canopy. 33 00:01:32,310 --> 00:01:35,580 And so, since we're geospatial people 34 00:01:35,580 --> 00:01:38,460 at the Spatial Analysis Lab, of course 35 00:01:38,460 --> 00:01:42,390 we were interested based on the impact of the outbreak 36 00:01:42,390 --> 00:01:43,570 this last summer 37 00:01:43,570 --> 00:01:45,770 on what kind of technologies we could use 38 00:01:45,770 --> 00:01:50,090 to sort of monitor the impacts to the forest 39 00:01:50,090 --> 00:01:52,670 and to monitor the regrowth 40 00:01:52,670 --> 00:01:54,870 and recovery of these impacted areas. 41 00:01:54,870 --> 00:01:58,580 And so, this spring, we had an outbreak 42 00:01:58,580 --> 00:02:02,050 the first really major infestation since 1991. 43 00:02:02,050 --> 00:02:06,200 Usually, these moth larva are regulated by a fungus 44 00:02:07,410 --> 00:02:10,830 but that fungus needs sort of wet weather conditions, right? 45 00:02:10,830 --> 00:02:12,010 Spring rains and things 46 00:02:12,010 --> 00:02:14,510 to really to be around. 47 00:02:14,510 --> 00:02:16,900 And so, the dry weather we had this spring 48 00:02:16,900 --> 00:02:18,060 across the northeast 49 00:02:19,100 --> 00:02:21,840 sort of minimized the growth of this fungus. 50 00:02:21,840 --> 00:02:24,670 And as a result, we saw this big outbreak 51 00:02:24,670 --> 00:02:26,253 in the moth population. 52 00:02:27,460 --> 00:02:30,010 Of course, I think across the state and the region 53 00:02:30,010 --> 00:02:32,300 we saw some impacts like this. 54 00:02:32,300 --> 00:02:34,720 But in particular we noticed 55 00:02:34,720 --> 00:02:36,840 areas in Addison County here in Vermont 56 00:02:36,840 --> 00:02:38,380 were hit particularly hard. 57 00:02:38,380 --> 00:02:41,570 And so, for sort of our data collection 58 00:02:41,570 --> 00:02:44,840 we focused on this parcel of land 59 00:02:44,840 --> 00:02:48,250 called Little Hogback Community Forest in Monkton. 60 00:02:48,250 --> 00:02:50,130 Maybe some of you folks have been out there 61 00:02:50,130 --> 00:02:52,300 but it's about 115 acres. 62 00:02:52,300 --> 00:02:55,720 Sort of jointly owned by the 16 shareholders. 63 00:02:55,720 --> 00:02:58,720 The Vermont Land Trust holds a conservation easement there 64 00:02:58,720 --> 00:03:01,620 and they kind of share the remaining rights for timber 65 00:03:01,620 --> 00:03:04,970 and sugaring and things as an LLC with the members. 66 00:03:04,970 --> 00:03:08,210 And so, the community forest and the trails 67 00:03:08,210 --> 00:03:10,370 what they enter on the left in the picture here. 68 00:03:10,370 --> 00:03:11,970 And sort of on the right 69 00:03:11,970 --> 00:03:15,330 is just one image of sort of this impact of defoliation 70 00:03:15,330 --> 00:03:17,993 that we saw here in early June. 71 00:03:20,710 --> 00:03:22,700 Our main data collection area 72 00:03:22,700 --> 00:03:25,503 focused sort of at the southern portion of the site. 73 00:03:27,130 --> 00:03:29,180 And we'll sort of go through the next steps. 74 00:03:29,180 --> 00:03:32,050 And so, in terms of technology, right? 75 00:03:32,050 --> 00:03:33,930 When we see issues in the forest 76 00:03:33,930 --> 00:03:36,240 or sort of other environmental issues 77 00:03:36,240 --> 00:03:38,033 or things like disaster response. 78 00:03:38,920 --> 00:03:39,900 In the last 10 years 79 00:03:39,900 --> 00:03:43,200 we've had really rapid development 80 00:03:43,200 --> 00:03:45,070 of drone technology in particular. 81 00:03:45,070 --> 00:03:48,180 And so, what's really great about drones 82 00:03:48,180 --> 00:03:51,910 or unoccupied aircraft systems, UAS 83 00:03:51,910 --> 00:03:53,080 there's many names for these 84 00:03:53,080 --> 00:03:55,330 is that they're really fast 85 00:03:55,330 --> 00:03:58,950 and they're a super cost-effective way to obtain imagery 86 00:03:58,950 --> 00:04:02,360 or sort of other sorts of remotely sensed data 87 00:04:02,360 --> 00:04:04,590 compared to traditional methods, right? 88 00:04:04,590 --> 00:04:06,960 And so, they're very small 89 00:04:06,960 --> 00:04:09,760 they're lightweight and deployable, they're safe. 90 00:04:09,760 --> 00:04:13,740 You maybe aren't able to access some sites by foot 91 00:04:13,740 --> 00:04:16,750 to go and do some field work and scouting. 92 00:04:16,750 --> 00:04:19,820 They're repeatable, so we can go out with a drone 93 00:04:19,820 --> 00:04:21,890 and collect data over the same area every week 94 00:04:21,890 --> 00:04:25,370 or every month or a few times a year 95 00:04:25,370 --> 00:04:27,523 or year over year really easily. 96 00:04:28,796 --> 00:04:30,350 And then we can develop 97 00:04:30,350 --> 00:04:33,150 really rapidly these GIS-ready products 98 00:04:33,150 --> 00:04:35,100 to use in analysis. 99 00:04:35,100 --> 00:04:37,080 And so, that's what's really special 100 00:04:37,080 --> 00:04:38,430 about these technologies. 101 00:04:38,430 --> 00:04:39,450 Obviously, there's a lot of people 102 00:04:39,450 --> 00:04:42,110 who have concerns about drones with privacy 103 00:04:42,110 --> 00:04:43,580 and safety and things 104 00:04:45,010 --> 00:04:48,813 and that's very important as well to consider. 105 00:04:50,814 --> 00:04:53,200 We won't get too deep into the weeds 106 00:04:53,200 --> 00:04:55,440 in sort of the technology portion. 107 00:04:55,440 --> 00:04:57,490 But I think it's interesting to highlight 108 00:04:58,330 --> 00:05:03,110 the types of drone technology that we have at the lab 109 00:05:03,110 --> 00:05:03,980 and we're using for this. 110 00:05:03,980 --> 00:05:08,120 And so, primarily to collect data at the site 111 00:05:08,120 --> 00:05:10,590 we were using a DJI M300. 112 00:05:10,590 --> 00:05:15,230 And so, DJI is sort of the market leader in drones 113 00:05:15,230 --> 00:05:17,530 across the country and globally, right? 114 00:05:17,530 --> 00:05:20,390 So pretty much anytime you see someone flying a drone 115 00:05:20,390 --> 00:05:23,650 it's a pretty good chance it's made by this company. 116 00:05:23,650 --> 00:05:26,120 This is sort of a commercial quad-copter. 117 00:05:26,120 --> 00:05:27,210 It has four rotors. 118 00:05:27,210 --> 00:05:29,580 It's able to carry pretty heavy payloads. 119 00:05:29,580 --> 00:05:33,010 So up to five or six pounds of payload weight 120 00:05:34,230 --> 00:05:37,440 and it has interchangeability for the sensors. 121 00:05:37,440 --> 00:05:41,880 We also used the DJI Mavic 2 Pro which is a bit smaller. 122 00:05:41,880 --> 00:05:46,360 More of a consumer grade type of thing. 123 00:05:46,360 --> 00:05:49,560 And this can collect imagery and video in real time 124 00:05:49,560 --> 00:05:51,990 which is also valuable. 125 00:05:51,990 --> 00:05:54,790 But primarily for this, for our data collection 126 00:05:54,790 --> 00:05:56,770 we went out with a LiDAR sensor. 127 00:05:56,770 --> 00:05:59,253 So this is a Yellowscan Mapper LiDAR sensor. 128 00:06:01,270 --> 00:06:04,640 Basically, the idea with LiDAR, we'll talk about shortly 129 00:06:04,640 --> 00:06:08,230 and then also a MicaSense RedEdge Dual Camera. 130 00:06:08,230 --> 00:06:12,010 So this is a 10-band multi-spectral sensor. 131 00:06:12,010 --> 00:06:16,700 So this gets us from the blue, red, green 132 00:06:16,700 --> 00:06:18,273 into red edge and near-infrared bands 133 00:06:18,273 --> 00:06:19,823 that our eyes can't see. 134 00:06:22,140 --> 00:06:25,870 For those folks who maybe are not familiar with LiDAR. 135 00:06:25,870 --> 00:06:28,620 The key here is that it's an active sensing technology. 136 00:06:28,620 --> 00:06:32,470 So, the sensor actually shoots out a pulse of laser light. 137 00:06:32,470 --> 00:06:35,817 This bounces off objects and it returns back to the sensor. 138 00:06:35,817 --> 00:06:39,590 And so, that's how we can determine elevation data, right? 139 00:06:39,590 --> 00:06:42,910 By the time these pulses take to come back to the sensor. 140 00:06:42,910 --> 00:06:44,930 What's really great about LiDAR technology 141 00:06:44,930 --> 00:06:47,340 is that we can actually get multiple returns 142 00:06:47,340 --> 00:06:49,740 from each laser pulse back. 143 00:06:49,740 --> 00:06:51,970 And so, that allows us to get through vegetation 144 00:06:51,970 --> 00:06:53,550 or through tree canopy 145 00:06:53,550 --> 00:06:55,380 and actually see elevations 146 00:06:55,380 --> 00:06:57,943 of different layers of vegetation. 147 00:06:59,520 --> 00:07:01,180 With the multi-spectral imagery 148 00:07:01,180 --> 00:07:04,660 so here, we can use passive sensors 149 00:07:04,660 --> 00:07:07,250 to capture information about wavelengths of light 150 00:07:07,250 --> 00:07:10,120 that are outside our visible spectrum as humans. 151 00:07:10,120 --> 00:07:13,080 And so, in particular, when we're looking at vegetation 152 00:07:13,080 --> 00:07:15,650 and plant health and forest health 153 00:07:15,650 --> 00:07:17,930 where we might be interested in looking at the red edge 154 00:07:17,930 --> 00:07:20,290 and near-infrared bands of light, right? 155 00:07:20,290 --> 00:07:23,070 And so, really healthy vegetation 156 00:07:23,070 --> 00:07:24,490 has a lot more chlorophyll. 157 00:07:24,490 --> 00:07:26,680 This will reflect more near-infrared light 158 00:07:26,680 --> 00:07:28,390 and green light compared to other way wavelengths. 159 00:07:28,390 --> 00:07:31,050 And so, that's why for us plants look green 160 00:07:31,050 --> 00:07:32,580 when they're healthy 161 00:07:32,580 --> 00:07:34,490 but they absorb more red and blue light. 162 00:07:34,490 --> 00:07:37,850 And so, by comparing differences 163 00:07:37,850 --> 00:07:41,870 in reflectance for the red edge or the near-infrared bands 164 00:07:41,870 --> 00:07:43,370 this can give us a pretty good sense 165 00:07:43,370 --> 00:07:46,623 of sort of is a plant healthy or is it stressed? 166 00:07:47,530 --> 00:07:49,203 How is it doing in general? 167 00:07:51,330 --> 00:07:54,930 For our field work, we went out with the M300. 168 00:07:54,930 --> 00:07:58,210 We collected flights over a few times series. 169 00:07:58,210 --> 00:08:02,670 And so, here is Maddy who's a UAS technician in our lab 170 00:08:02,670 --> 00:08:05,510 flying with our 10-band multi-spectral camera. 171 00:08:05,510 --> 00:08:09,000 And an image of the drone, the LiDAR setup 172 00:08:09,000 --> 00:08:10,143 flying there as well. 173 00:08:11,300 --> 00:08:14,560 We don't need fancy technology necessarily with drones 174 00:08:14,560 --> 00:08:16,840 to get really useful information. 175 00:08:16,840 --> 00:08:18,720 Even just images or video 176 00:08:18,720 --> 00:08:21,520 kind of giving an aerial overview can be really powerful. 177 00:08:21,520 --> 00:08:25,580 So here we see an image of the forest 178 00:08:25,580 --> 00:08:29,690 and sort of that defoliation in middle of June. 179 00:08:29,690 --> 00:08:31,210 Just a week later, right? 180 00:08:31,210 --> 00:08:34,610 We can see really the heavy, heavy impacts 181 00:08:34,610 --> 00:08:36,830 of this defoliation, right? 182 00:08:36,830 --> 00:08:38,990 By early July, you see a lot of the leaves 183 00:08:38,990 --> 00:08:43,040 are sort of starting to bounce back and grow again. 184 00:08:43,040 --> 00:08:46,120 And then by August, we're in pretty good shape 185 00:08:46,120 --> 00:08:49,650 in terms of having our sort of full canopy back, right? 186 00:08:49,650 --> 00:08:54,390 And the fall down in this area was a bit disappointing 187 00:08:54,390 --> 00:08:56,440 with the belief change. 188 00:08:56,440 --> 00:08:57,330 It went pretty quick 189 00:08:57,330 --> 00:09:01,423 but just kind of to give a sense of what it was in October. 190 00:09:03,500 --> 00:09:06,370 What we think is sort of the most interesting here 191 00:09:06,370 --> 00:09:09,829 is looking at mapping data and orthorectified imagery 192 00:09:09,829 --> 00:09:12,140 from these sensors that we use. 193 00:09:12,140 --> 00:09:15,250 And so here, we're looking at NAIP imagery, right? 194 00:09:15,250 --> 00:09:19,600 Just traditional sort of aerial imagery, right? 195 00:09:19,600 --> 00:09:22,730 We can see that section in the forest in the middle. 196 00:09:22,730 --> 00:09:25,950 We expect it to have quite a bit of canopy cover 197 00:09:25,950 --> 00:09:27,590 during the summer. 198 00:09:27,590 --> 00:09:29,840 But here's what our results looked like in July 199 00:09:29,840 --> 00:09:31,150 from our imagery collection. 200 00:09:31,150 --> 00:09:36,150 And so, you can even see straight through down to the ground 201 00:09:36,760 --> 00:09:37,593 in a lot of places. 202 00:09:37,593 --> 00:09:40,490 You can see the trails and old logging roads and things. 203 00:09:40,490 --> 00:09:42,770 And that's just a really powerful snapshot 204 00:09:42,770 --> 00:09:46,710 to see sort of the impact of this. 205 00:09:46,710 --> 00:09:48,310 When we compare that to August 206 00:09:48,310 --> 00:09:51,960 so we can see that yeah, that the trees 207 00:09:51,960 --> 00:09:52,793 have done a good job. 208 00:09:52,793 --> 00:09:55,340 We have a lot more leaves and canopy coverage, right? 209 00:09:55,340 --> 00:09:58,350 Compared to month to month. 210 00:09:58,350 --> 00:10:01,680 If we take a closer look at some of these datasets 211 00:10:01,680 --> 00:10:03,330 this is from July. 212 00:10:03,330 --> 00:10:04,610 You can see the logging roads 213 00:10:04,610 --> 00:10:07,740 sort of cutting through on the right-hand side. 214 00:10:07,740 --> 00:10:11,270 You can see a lot of the branches and leaves and things 215 00:10:11,270 --> 00:10:13,130 that we might be missing here. 216 00:10:13,130 --> 00:10:16,320 And again, comparing that same section to August 217 00:10:16,320 --> 00:10:20,183 it's quite a difference in terms of growth and recovery. 218 00:10:21,610 --> 00:10:24,000 We can also visualize these datasets 219 00:10:24,000 --> 00:10:26,190 using different bands 220 00:10:26,190 --> 00:10:28,650 than are sort of red, green, blue, true color. 221 00:10:28,650 --> 00:10:32,190 And so, this one, we're looking at the near-infrared band 222 00:10:32,190 --> 00:10:34,880 and the red and green bands instead, right? 223 00:10:34,880 --> 00:10:37,100 And so, here in this case 224 00:10:37,100 --> 00:10:38,950 sort of the darker colors 225 00:10:38,950 --> 00:10:42,680 indicate perhaps a higher chlorophyll content in the leaves. 226 00:10:42,680 --> 00:10:45,870 And so, sort of down in the southern section 227 00:10:45,870 --> 00:10:49,070 we can see that those trees were not impacted 228 00:10:49,070 --> 00:10:51,440 by the moth larva as much. 229 00:10:51,440 --> 00:10:53,090 They're looking quite a bit healthier 230 00:10:53,090 --> 00:10:55,270 compared to sort of the light pink 231 00:10:55,270 --> 00:10:59,110 and other shades of vegetation up north 232 00:10:59,110 --> 00:11:00,800 up on the hill there. 233 00:11:00,800 --> 00:11:02,750 And again, if we compare this to August 234 00:11:04,030 --> 00:11:07,453 the canopy itself is looking much better and healthier. 235 00:11:08,680 --> 00:11:11,700 Just another quick look, a little bit closer here. 236 00:11:11,700 --> 00:11:12,680 And it's really interesting. 237 00:11:12,680 --> 00:11:15,530 You can see sort of differences in specific trees 238 00:11:15,530 --> 00:11:19,360 in terms of this analog for chlorophyll content, right? 239 00:11:19,360 --> 00:11:21,660 Some trees did a lot better than others 240 00:11:21,660 --> 00:11:26,200 or maybe some are in the process of regrowth here in July. 241 00:11:26,200 --> 00:11:27,840 And then we look in August 242 00:11:27,840 --> 00:11:31,963 and we're in much better shape once again. 243 00:11:34,540 --> 00:11:37,380 Next, we'll take a little bit of a look at the LiDAR data. 244 00:11:37,380 --> 00:11:41,510 And so, this is our initial point cloud 245 00:11:41,510 --> 00:11:43,240 that we got from the LiDAR sensor, right? 246 00:11:43,240 --> 00:11:47,850 So this is all just 3D points in space as elevation data. 247 00:11:47,850 --> 00:11:50,670 You can see the sort of those colored lines 248 00:11:50,670 --> 00:11:52,690 above the data set on the right-hand side. 249 00:11:52,690 --> 00:11:55,910 That's actually our flight path trajectory of the drone 250 00:11:55,910 --> 00:11:57,600 to be able to capture this data. 251 00:11:57,600 --> 00:11:58,860 And so, that's pretty typical 252 00:11:58,860 --> 00:12:01,030 that you fly sort of parallel flight lines 253 00:12:01,030 --> 00:12:03,110 and then you're merging all these strips 254 00:12:03,110 --> 00:12:05,163 together using software. 255 00:12:07,650 --> 00:12:10,240 This is the actual point cloud from July. 256 00:12:10,240 --> 00:12:13,850 And so, you can see it's colored by elevation. 257 00:12:13,850 --> 00:12:17,950 So that southern section of the areas is quite low. 258 00:12:17,950 --> 00:12:19,170 As we move up the hills 259 00:12:19,170 --> 00:12:21,900 where the tops of the trees turn red. 260 00:12:21,900 --> 00:12:25,220 Our density for the point cloud to give a sense here 261 00:12:25,220 --> 00:12:29,440 is about 150 points per square meter on the ground. 262 00:12:29,440 --> 00:12:31,700 If we compare this to like the Vermont State LiDAR 263 00:12:31,700 --> 00:12:33,420 that's about 10 points per square meter. 264 00:12:33,420 --> 00:12:37,993 So, quite a magnitude higher density in these datasets. 265 00:12:39,740 --> 00:12:42,610 We can also use the point cloud data 266 00:12:42,610 --> 00:12:44,850 to do some additional analysis 267 00:12:44,850 --> 00:12:47,180 and create some sort of extra files here. 268 00:12:47,180 --> 00:12:52,180 And so, if we just rasterize the top surfaces of everything 269 00:12:53,340 --> 00:12:56,060 this creates a digital surface model or DSM here. 270 00:12:56,060 --> 00:12:59,460 And so, that's just showing us sort of the top-down view 271 00:12:59,460 --> 00:13:02,100 all of the above ground objects like vegetation, right? 272 00:13:02,100 --> 00:13:04,330 Obviously, you can see the tree canopy here. 273 00:13:04,330 --> 00:13:07,133 There's some gaps in the canopy as well. 274 00:13:08,650 --> 00:13:09,950 And compare that to August 275 00:13:09,950 --> 00:13:14,440 where we can see really well 276 00:13:14,440 --> 00:13:19,440 how the foliage has sort of filled in those areas, right? 277 00:13:21,360 --> 00:13:23,140 The other thing we can do with point cloud data 278 00:13:23,140 --> 00:13:24,470 is actually classify this 279 00:13:24,470 --> 00:13:25,810 into the points that are above the ground 280 00:13:25,810 --> 00:13:27,760 or points that are the ground. 281 00:13:27,760 --> 00:13:30,170 And so here, we're looking at the same data. 282 00:13:30,170 --> 00:13:31,990 Anything in red is sort of above ground. 283 00:13:31,990 --> 00:13:34,000 So this would include trees 284 00:13:34,000 --> 00:13:36,290 could include cars or buildings or things like that. 285 00:13:36,290 --> 00:13:37,760 And then the green underneath 286 00:13:37,760 --> 00:13:42,370 is sort of just the ground features that we captured. 287 00:13:42,370 --> 00:13:44,900 And so, if we use those to generate a raster file 288 00:13:44,900 --> 00:13:47,310 here, we have our digital elevation model. 289 00:13:47,310 --> 00:13:49,820 And so, this has removed all the trees 290 00:13:49,820 --> 00:13:51,190 and everything above ground 291 00:13:51,190 --> 00:13:54,060 and we can see sort of down in detail 292 00:13:54,060 --> 00:13:57,590 what the terrain looks like and topography. 293 00:13:57,590 --> 00:13:59,900 You can even see the logging roads in here 294 00:13:59,900 --> 00:14:02,400 and some hiking trails and other different things. 295 00:14:03,380 --> 00:14:06,877 But what's great is when we subtract the DSM and the DEM 296 00:14:07,810 --> 00:14:09,490 this actually just gives us the height 297 00:14:09,490 --> 00:14:11,660 of objects above the ground, right? 298 00:14:11,660 --> 00:14:14,773 And so, this is called a normalized digital surface model. 299 00:14:16,330 --> 00:14:19,380 This can be helpful for doing like tree height analysis 300 00:14:19,380 --> 00:14:21,917 canopy, vegetation analysis and things like that. 301 00:14:21,917 --> 00:14:24,950 And so, again that comparison between July and August 302 00:14:26,200 --> 00:14:30,153 it's pretty interesting to see the regrowth. 303 00:14:31,810 --> 00:14:33,560 If we look at the LiDAR point cloud 304 00:14:33,560 --> 00:14:35,900 and sort of just draw a slice down the middle 305 00:14:35,900 --> 00:14:38,740 and look at it from a profile view 306 00:14:38,740 --> 00:14:40,170 that's what we're looking at here. 307 00:14:40,170 --> 00:14:43,240 And so, this is our profile line from July. 308 00:14:43,240 --> 00:14:46,120 You can see sort of each of these individual trees 309 00:14:46,120 --> 00:14:49,563 and their trunks and a lot of the structure which is great. 310 00:14:51,360 --> 00:14:54,080 From this, you can do things like measure tree heights 311 00:14:54,920 --> 00:14:57,633 measure the width of trunks. 312 00:14:58,740 --> 00:15:02,060 maybe do some mapping of canopy as well. 313 00:15:02,060 --> 00:15:03,730 When we compare this to August 314 00:15:05,100 --> 00:15:07,530 we can sort of go back and forth 315 00:15:07,530 --> 00:15:09,630 between the July and the August. 316 00:15:09,630 --> 00:15:11,690 And just gives you a sense 317 00:15:11,690 --> 00:15:15,530 of the difference in like vegetation density 318 00:15:15,530 --> 00:15:18,720 and leaf density that we see between between the month 319 00:15:18,720 --> 00:15:21,150 as the forest sort of bounced back. 320 00:15:21,150 --> 00:15:22,470 In particular, this highlights 321 00:15:22,470 --> 00:15:24,540 I think a lot of undergrowth 322 00:15:24,540 --> 00:15:27,430 that pops up between July and August 323 00:15:27,430 --> 00:15:31,020 as well as the canopy there too. 324 00:15:31,020 --> 00:15:33,390 If we overlay these two from July and August 325 00:15:33,390 --> 00:15:35,280 on top of each other, it's pretty interesting. 326 00:15:35,280 --> 00:15:38,600 So, the green here is our July data. 327 00:15:38,600 --> 00:15:40,640 The red here is our August data 328 00:15:40,640 --> 00:15:43,520 and you can kind of just see what the difference is. 329 00:15:43,520 --> 00:15:45,980 I know we're gonna get to questions here in a second 330 00:15:45,980 --> 00:15:48,890 but we've developed a web map 331 00:15:48,890 --> 00:15:52,170 and we'd encourage you folks to check this out. 332 00:15:52,170 --> 00:15:53,630 With all these different datasets 333 00:15:53,630 --> 00:15:55,910 you can do some swiping and different things 334 00:15:55,910 --> 00:15:59,030 as well as a web scene showing the 3D model point cloud. 335 00:15:59,030 --> 00:16:02,130 So we can put these links in the chat if that'd be useful 336 00:16:02,130 --> 00:16:05,170 or there's a QR code here that you can pull up 337 00:16:05,170 --> 00:16:06,950 if you're on your phone. 338 00:16:06,950 --> 00:16:09,970 And just our special thanks to the shareholders 339 00:16:09,970 --> 00:16:12,203 of the Little Hogback Community Forest here. 340 00:16:21,110 --> 00:16:23,150 - Great, thank you so much. 341 00:16:23,150 --> 00:16:24,460 Yeah, I think it would be awesome 342 00:16:24,460 --> 00:16:26,340 to put those links in the chat 343 00:16:27,300 --> 00:16:30,440 and we can open it up for questions. 344 00:16:30,440 --> 00:16:35,440 I see a hand raised from Jerry Carlson, Alrighty. 345 00:16:35,550 --> 00:16:37,470 - [Jerry] Yeah, hi, thanks Adam. 346 00:16:37,470 --> 00:16:38,610 Can you hear me? 347 00:16:38,610 --> 00:16:39,443 - Yes. 348 00:16:39,443 --> 00:16:40,640 - [Jerry] Yeah, okay, good. 349 00:16:40,640 --> 00:16:43,240 What would be the average cost per acre 350 00:16:43,240 --> 00:16:45,720 for that kind of data to be collected? 351 00:16:45,720 --> 00:16:47,760 And are there weather extremes 352 00:16:47,760 --> 00:16:49,570 that you wanna try to avoid? 353 00:16:49,570 --> 00:16:51,100 - Yeah, good, two good questions. 354 00:16:51,100 --> 00:16:54,114 So could give a sense of sort of the cost 355 00:16:54,114 --> 00:16:57,670 of this whole drone setup. 356 00:16:57,670 --> 00:17:01,400 We're pretty close to six figures for the LiDAR 357 00:17:01,400 --> 00:17:04,300 the multi-spectral imagery and the drone. 358 00:17:04,300 --> 00:17:07,180 And then it takes some processing software 359 00:17:07,180 --> 00:17:09,983 to actually make this usable. 360 00:17:13,470 --> 00:17:16,310 So, per acre, it's hard to say 361 00:17:16,310 --> 00:17:21,310 but maybe it's say between 10 and $30 per acre 362 00:17:22,220 --> 00:17:24,660 if you're sort of billing out this type of work. 363 00:17:24,660 --> 00:17:26,510 That's fairly common in the industry. 364 00:17:27,970 --> 00:17:30,840 Weather extremes, certainly a factor. 365 00:17:30,840 --> 00:17:33,360 So today it's very windy in Vermont. 366 00:17:33,360 --> 00:17:35,853 We had to cancel one of our flight operations. 367 00:17:37,730 --> 00:17:40,340 Snow doesn't work really well with LiDAR 368 00:17:40,340 --> 00:17:42,540 due to the wavelength of the laser as well. 369 00:17:42,540 --> 00:17:47,540 So you sort of wanna be able to avoid like rain 370 00:17:47,900 --> 00:17:49,090 heavy snow on the ground 371 00:17:49,090 --> 00:17:51,350 or high winds and things of that nature. 372 00:17:51,350 --> 00:17:53,000 Yeah, hopefully, that answers it. 373 00:17:59,930 --> 00:18:02,060 - [Jerry] Yeah, thanks, Adam. 374 00:18:02,060 --> 00:18:02,893 That was great. 375 00:18:05,180 --> 00:18:10,180 - Great, Jim, put a message in the chat here too. 376 00:18:13,870 --> 00:18:16,820 And I think that may go also with Jack's question. 377 00:18:16,820 --> 00:18:21,554 So those are sort of about access road 378 00:18:21,554 --> 00:18:24,580 or what do you need to be able to take off? 379 00:18:24,580 --> 00:18:28,110 And then about line of sight challenges 380 00:18:28,110 --> 00:18:29,460 when monitoring in the forest. 381 00:18:29,460 --> 00:18:34,140 And so, what's interesting about flying under regulations 382 00:18:34,140 --> 00:18:36,270 in the US for drones 383 00:18:36,270 --> 00:18:39,130 is that we're limited to visual line of sight. 384 00:18:39,130 --> 00:18:41,440 So we always have to be able to see the drone 385 00:18:41,440 --> 00:18:42,620 when it's flying. 386 00:18:42,620 --> 00:18:45,820 And we obviously, also need enough open space 387 00:18:45,820 --> 00:18:48,480 that we can take off and land it safely. 388 00:18:48,480 --> 00:18:51,500 And so, in heavily forested areas 389 00:18:51,500 --> 00:18:53,250 yeah, that can be tricky 390 00:18:53,250 --> 00:18:56,480 and it can take some planning in advance, right? 391 00:18:56,480 --> 00:18:59,100 We do a lot of sort of looking at maps 392 00:18:59,100 --> 00:19:00,490 maybe some scouting in the field 393 00:19:00,490 --> 00:19:05,150 to make sure we have that space. 394 00:19:05,150 --> 00:19:06,590 It's easiest in the fall 395 00:19:06,590 --> 00:19:09,460 if you can sort of see through the empty tree canopy 396 00:19:09,460 --> 00:19:12,400 or in this case, when there's no leaves on the trees 397 00:19:12,400 --> 00:19:14,380 in the middle of the summer. 398 00:19:14,380 --> 00:19:15,620 But that is one of the reasons 399 00:19:15,620 --> 00:19:18,650 we sort of weren't able to cover 400 00:19:18,650 --> 00:19:21,230 the entirety of that community forest. 401 00:19:21,230 --> 00:19:24,680 Because due to the sort of terrain and tree canopy 402 00:19:26,150 --> 00:19:28,100 blocking our line sight from the bottom 403 00:19:29,060 --> 00:19:32,403 we couldn't quite see it as it got on top of the hill. 404 00:19:35,150 --> 00:19:37,550 Alison is asking a question 405 00:19:37,550 --> 00:19:40,150 about sort of yeah, a more expensive computer 406 00:19:40,150 --> 00:19:41,540 and equipment and software. 407 00:19:41,540 --> 00:19:43,300 And that's definitely true. 408 00:19:43,300 --> 00:19:45,940 So we're lucky as a geospatial lab 409 00:19:45,940 --> 00:19:49,210 to have some pretty hefty processing power 410 00:19:49,210 --> 00:19:50,613 on our workstations here. 411 00:19:52,236 --> 00:19:53,800 The LiDAR datasets 412 00:19:53,800 --> 00:19:57,650 are gigabytes and gigabytes of data, right? 413 00:19:57,650 --> 00:19:59,410 And you have to be able to sort of visualize that 414 00:19:59,410 --> 00:20:02,810 and analyze that in 3D to make the best use of it. 415 00:20:02,810 --> 00:20:04,430 In the multi-spectral data 416 00:20:04,430 --> 00:20:08,600 for sort of processing all those individual images together 417 00:20:08,600 --> 00:20:10,810 that can take quite a powerful computer as well. 418 00:20:10,810 --> 00:20:13,180 So it can be challenging to sort of do this in the field 419 00:20:13,180 --> 00:20:14,900 on just the laptop. 420 00:20:14,900 --> 00:20:18,480 So there are sorts of all lots of organizational costs 421 00:20:18,480 --> 00:20:19,770 that they may come up 422 00:20:19,770 --> 00:20:22,713 if you're trying to like use UAS in your group. 423 00:20:27,026 --> 00:20:30,930 Greg's asking about sort of using multi-spectral data 424 00:20:30,930 --> 00:20:33,380 for more quantitative analysis 425 00:20:33,380 --> 00:20:35,130 like estimating percent of trees 426 00:20:35,130 --> 00:20:38,000 that might've been defoliated. 427 00:20:38,000 --> 00:20:41,890 And yes, that's definitely something that we can do. 428 00:20:41,890 --> 00:20:43,700 What's great about these datasets 429 00:20:43,700 --> 00:20:47,540 are we're actually planning to use these for more analysis 430 00:20:47,540 --> 00:20:50,770 and hoping to get back in the spring and summer 431 00:20:50,770 --> 00:20:53,680 to just sort of look over time 432 00:20:53,680 --> 00:20:55,770 to see if the defoliation from last year 433 00:20:55,770 --> 00:21:00,680 impacted the growth of the trees come spring with. 434 00:21:01,754 --> 00:21:03,410 With the multi-spectral data 435 00:21:03,410 --> 00:21:07,130 it's a great way to sort of pull out features 436 00:21:07,130 --> 00:21:12,130 that where we have high chlorophyll content versus low ones. 437 00:21:12,320 --> 00:21:15,430 And as undergraduate student in our lab 438 00:21:15,430 --> 00:21:18,290 is gonna do they're undergrad thesis 439 00:21:18,290 --> 00:21:20,100 sort of diving into these datasets 440 00:21:20,100 --> 00:21:23,690 a little bit more for sort of those extra analysis. 441 00:21:23,690 --> 00:21:24,960 So that should be pretty interesting. 442 00:21:24,960 --> 00:21:27,340 Hopefully, maybe by next summer 443 00:21:27,340 --> 00:21:29,853 we can share some of those results. 444 00:21:39,210 --> 00:21:42,420 Yeah, I'm not sure if I missed any questions 445 00:21:42,420 --> 00:21:44,300 please let me know. 446 00:21:44,300 --> 00:21:47,480 Otherwise, happy to answer a few more here 447 00:21:47,480 --> 00:21:50,650 or stick around a little longer 448 00:21:50,650 --> 00:21:53,953 if folks wanna chat about other pieces. 449 00:21:58,390 --> 00:22:00,540 - [Jerry] Yeah, Adam, Jerry again. 450 00:22:00,540 --> 00:22:03,260 I'm wondering about a little bit about the evolution 451 00:22:03,260 --> 00:22:04,950 of drone technology. 452 00:22:04,950 --> 00:22:07,880 You're still using rotary wing. 453 00:22:07,880 --> 00:22:10,930 What about fixed wing and some of the efficiencies 454 00:22:10,930 --> 00:22:14,150 in the larger landscapes that we can detect that way? 455 00:22:14,150 --> 00:22:16,380 - Yep, another great question, Jerry. 456 00:22:16,380 --> 00:22:21,380 So, fixed wings are really great 457 00:22:21,630 --> 00:22:24,870 for covering large areas super efficiently. 458 00:22:24,870 --> 00:22:26,460 A lot of times they're sort of limited 459 00:22:26,460 --> 00:22:29,140 by the payload capacity, right? 460 00:22:29,140 --> 00:22:32,090 And like it's much more difficult 461 00:22:32,090 --> 00:22:35,840 to put a sort of a LiDAR sensor on a smaller fixed wing 462 00:22:35,840 --> 00:22:37,373 than it is on a multirotor. 463 00:22:38,210 --> 00:22:42,930 We use fixed wings quite a bit in our operations for mapping 464 00:22:42,930 --> 00:22:44,610 and monitoring, right? 465 00:22:44,610 --> 00:22:46,480 Either with true color imagery 466 00:22:46,480 --> 00:22:50,420 to make maps and elevation data or multi-spectral imagery. 467 00:22:50,420 --> 00:22:51,253 Those are really good. 468 00:22:51,253 --> 00:22:53,470 Let's say if you're doing hundreds 469 00:22:53,470 --> 00:22:55,340 or thousands of acres, right? 470 00:22:55,340 --> 00:22:57,640 That's where sort of that efficiency of the fixed wing 471 00:22:57,640 --> 00:22:58,700 comes into play. 472 00:22:58,700 --> 00:23:01,610 The downsides are you need even more room 473 00:23:01,610 --> 00:23:02,780 to launch and land, right? 474 00:23:02,780 --> 00:23:04,000 You might need a big field 475 00:23:04,000 --> 00:23:08,150 or something else that's close enough 476 00:23:08,150 --> 00:23:10,530 where you can launch and land 477 00:23:10,530 --> 00:23:14,290 that sort of more like an aircraft, right? 478 00:23:14,290 --> 00:23:16,440 Where it's it's maybe hand launched 479 00:23:16,440 --> 00:23:18,520 and it sort of comes in for a belly landing 480 00:23:18,520 --> 00:23:21,050 compared to the multirotor. 481 00:23:21,050 --> 00:23:22,340 That's where the multirotors 482 00:23:22,340 --> 00:23:24,380 could be really good for forestry applications 483 00:23:24,380 --> 00:23:26,930 is they don't need as much space. 484 00:23:26,930 --> 00:23:29,770 If you have a big enough opening in the canopy 485 00:23:29,770 --> 00:23:33,440 you can set up and take off and fly from there. 486 00:23:33,440 --> 00:23:35,410 But they are limited in terms of flight time 487 00:23:35,410 --> 00:23:36,460 and coverage area, right? 488 00:23:36,460 --> 00:23:39,920 Maybe 50 to 100 acres per flight with the multirotor 489 00:23:39,920 --> 00:23:42,130 where maybe you're doing 500 acres per flight 490 00:23:42,130 --> 00:23:43,793 with the fixed wing, for example. 491 00:23:45,210 --> 00:23:47,803 - [Jerry] Well, aren't they better at automated flight? 492 00:23:49,310 --> 00:23:51,590 - I would say they're pretty equivalent. 493 00:23:51,590 --> 00:23:56,590 So this, we flew fully autonomously for both the mapping 494 00:23:57,420 --> 00:23:58,980 and the LiDAR collection, right? 495 00:23:58,980 --> 00:24:00,810 You can just do flight planning 496 00:24:00,810 --> 00:24:02,240 adjust it for the elevation 497 00:24:02,240 --> 00:24:06,193 and even the multirotors will follow the flight lines. 498 00:24:17,000 --> 00:24:21,653 Yeah, are there other folks with some questions or? 499 00:24:28,040 --> 00:24:29,890 Great, well, if that's it 500 00:24:32,620 --> 00:24:35,120 yeah, we appreciate you coming. 501 00:24:35,120 --> 00:24:39,300 It's obviously hard to fit sort of this much 502 00:24:39,300 --> 00:24:40,620 into just 15 minutes 503 00:24:40,620 --> 00:24:42,720 but hopefully, that was interesting. 504 00:24:42,720 --> 00:24:44,930 And certainly feel free to reach out to me 505 00:24:44,930 --> 00:24:48,510 or certainly anyone at the Spatial Analysis Lab here at UVM 506 00:24:49,480 --> 00:24:52,280 if you're interested in maybe some forestry applications 507 00:24:52,280 --> 00:24:57,280 or some testing of different things come next year. 508 00:24:58,930 --> 00:25:00,749 So great, thanks, everyone. 509 00:25:00,749 --> 00:25:04,093 And hope you have a good Friday and a happy holiday.