WEBVTT 00:00:10.443 --> 00:00:14.981 When the Next Generation Water Observing System got rolling in 2018, 00:00:15.615 --> 00:00:19.152 the program manager contacted me and asked for ideas 00:00:19.152 --> 00:00:22.956 for a groundwater surface water technology testbed. 00:00:23.156 --> 00:00:27.427 Various types of methods that we could use to measure and identify 00:00:27.427 --> 00:00:32.232 where groundwater and surface water exchange at varied scale over time. 00:00:32.432 --> 00:00:36.536 But it's not an easy thing to disentangle at the landscape scale. 00:00:36.536 --> 00:00:40.740 Thermal methods give us different approaches to to do that scaling. 00:00:40.740 --> 00:00:43.576 And there's different types of temperature measurements that we're interested in. 00:00:43.643 --> 00:00:44.677 We're anomaly hunting. 00:00:44.677 --> 00:00:47.113 We're trying to locate exactly where streams and groundwater 00:00:47.180 --> 00:00:49.249 are connected, where the water is emerging from the ground 00:00:49.516 --> 00:00:51.418 and entering the stream or river. 00:00:51.418 --> 00:00:55.155 Once we find those locations, we can make measurements in the subsurface 00:00:55.455 --> 00:00:58.691 of temperature over time and model those temperatures to 00:00:59.225 --> 00:01:01.261 estimate what the rates of exchange are. 00:01:01.261 --> 00:01:03.196 So how much water is coming out over time? 00:01:03.196 --> 00:01:07.300 How does that respond to dry events, to big precipitation events, 00:01:07.300 --> 00:01:08.768 to high flow events in the river. 00:01:08.768 --> 00:01:12.372 All those affect the seepage rates and how surface and groundwaters are connected 00:01:21.247 --> 00:01:23.883 and then somewhat larger, more integrated scales. 00:01:24.150 --> 00:01:25.985 Are these air-/water-temperature measurements 00:01:25.985 --> 00:01:29.155 where we put a temperature logger in the stream where it's flowing, 00:01:29.155 --> 00:01:32.025 or are we expected to be mixed, and in the local air 00:01:32.025 --> 00:01:34.194 And we compare those records over time. 00:01:34.527 --> 00:01:37.197 There's a few different signal processing tricks we use to then 00:01:37.197 --> 00:01:40.066 infer where the streams and groundwater are connected 00:01:40.066 --> 00:01:42.168 at kind of a kilometer plus scale. 00:01:42.168 --> 00:01:45.505 The combination of thermal infrared mapping with the measurements 00:01:45.505 --> 00:01:49.008 like the air-/water-temperature sensing that can help us focus our efforts 00:01:49.008 --> 00:01:52.545 to specific sections of the watershed that may be most important. 00:01:52.545 --> 00:01:54.881 We really need efficient methods to evaluate 00:01:54.881 --> 00:01:56.850 where to focus our efforts, cause we can't. 00:01:56.850 --> 00:01:58.685 We can't measure everywhere. 00:01:58.685 --> 00:02:01.321 And measuring everywhere would be somewhat redundant. 00:02:01.354 --> 00:02:04.524 So we want to target what's sometimes called ecosystem control points, 00:02:04.858 --> 00:02:07.627 which are the points of most influence to surface water 00:02:07.627 --> 00:02:10.697 quality, to bass flow generation, to water temperature, 00:02:10.930 --> 00:02:12.966 where we can better understand what's going on. 00:02:12.966 --> 00:02:15.001 Traditionally, at USGS, 00:02:15.001 --> 00:02:17.937 stream gages were really only measuring flow. 00:02:17.937 --> 00:02:21.708 So part of NGWOS rationale is we're going to add some core parameters 00:02:21.708 --> 00:02:24.911 to some subset of gages that add value, especially the TIME series 00:02:24.944 --> 00:02:26.079 collecting data over time. 00:02:26.079 --> 00:02:30.416 And that's where this Multi-Scale toolkit of options of measuring temperature 00:02:30.416 --> 00:02:33.620 comes into play, from walking with thermal cameras to finding them 00:02:33.620 --> 00:02:37.957 on drones to airplanes, to even harnessing satellite based thermal infrared imagery 00:02:37.957 --> 00:02:41.594 as we're doing four streams that are 100 meters wide or wider, and for reservoirs. 00:02:42.061 --> 00:02:44.697 So it's really using kind of all the data types that are available 00:02:44.697 --> 00:02:49.369 in conjunction to optimize to the scale and process of interest. 00:02:49.569 --> 00:02:53.506 There's an amazing wealth of value for stream temperature records. 00:02:53.840 --> 00:02:57.377 We use temperature records typically to infer base flow 00:02:57.377 --> 00:03:00.313 dynamics, stream temperature sensitivity variables, 00:03:00.480 --> 00:03:03.383 whether the stream flow is dominated by shallow or deep groundwater. 00:03:03.416 --> 00:03:04.717 But some of those analysis 00:03:04.717 --> 00:03:07.654 we're repurposing data people have collected over decades 00:03:07.654 --> 00:03:11.324 for something like a water quality study or a fish habitat study. 00:03:11.424 --> 00:03:14.460 So there's a growing number of publicly available stream 00:03:14.494 --> 00:03:18.198 temperature records that can be repurposed for any number of reasons. 00:03:18.198 --> 00:03:22.035 And that's because stream temperature is basically the master water quality 00:03:22.035 --> 00:03:25.405 and habitat variable that are range of researchers and water managers. 00:03:25.405 --> 00:03:26.139 Care about. 00:03:26.139 --> 00:03:30.543 And we can also use that exact same information for hydrological processes. 00:03:30.543 --> 00:03:34.314 And then folks that are modeling stream temperature and predicting stream 00:03:34.314 --> 00:03:35.782 temperature dynamics over time 00:03:35.782 --> 00:03:39.352 and into the future under climate change and extreme weather events 00:03:39.419 --> 00:03:42.989 can use to stream temperature records worth collecting for calibration points. 00:03:44.224 --> 00:03:44.791 I started 00:03:44.791 --> 00:03:47.260 using heat as a tracer, as a technique 00:03:47.794 --> 00:03:50.263 to understand where groundwater was 00:03:50.263 --> 00:03:50.697 coming into wetland platform 00:03:58.104 --> 00:04:01.140 coming into wetland platform 00:04:01.641 --> 00:04:04.644 in 2012 by using heat as a tracer, both 00:04:04.644 --> 00:04:08.147 through infrared camera and distributed temperature sensing. 00:04:08.514 --> 00:04:11.351 I was able to pinpoint where 00:04:11.351 --> 00:04:15.455 groundwater discharge was coming into our peatland platform 00:04:15.455 --> 00:04:19.626 through being able to identify locations of groundwater discharge. 00:04:19.892 --> 00:04:22.629 We're able to inform peatland restoration 00:04:22.662 --> 00:04:28.601 and so that they could create that stream through these discharge zones. 00:04:28.635 --> 00:04:32.805 And overall, then promoting more healthy habitat post restoration. 00:04:33.339 --> 00:04:39.012 So I have used heat tracing techniques to inform wetland restoration 00:04:39.045 --> 00:04:43.016 as well as to identify contaminant discharge. 00:04:43.082 --> 00:04:46.519 So locations where we know that the groundwater is contaminated. 00:04:46.519 --> 00:04:51.057 We were interested in whether this stream was receiving 00:04:51.057 --> 00:04:54.193 that contaminated groundwater and using both 00:04:54.427 --> 00:04:58.231 infrared cameras as well as distributed temperature sensing. 00:04:58.264 --> 00:05:02.869 We were able to identify the precise locations of the discharge. 00:05:03.469 --> 00:05:08.741 And then sampled directly to establish whether there was a connection or not. 00:05:09.042 --> 00:05:13.346 In our group, we're very focused on finding current and future 00:05:13.346 --> 00:05:17.550 cold water refugia for native clean water fish species. 00:05:17.884 --> 00:05:21.087 And we've done a lot of work with statistical 00:05:21.087 --> 00:05:21.254 And we've done a lot of work with statistical 00:05:30.163 --> 00:05:32.899 and physical process stream temperature modeling. 00:05:33.466 --> 00:05:36.803 But the one factor that often confounds 00:05:36.803 --> 00:05:39.672 that modeling effort is the role of groundwater. 00:05:39.806 --> 00:05:45.978 Groundwater basically modulates surface water temperature in streams. 00:05:46.045 --> 00:05:49.849 So during the summer, when surface water temperatures warm 00:05:49.849 --> 00:05:54.120 up, groundwater provides cool water into the streams. 00:05:54.120 --> 00:05:57.757 And in the winter, when surface water temperatures cool. 00:05:58.324 --> 00:06:00.259 Groundwater actually warms. 00:06:00.259 --> 00:06:01.794 Surface water temperatures. 00:06:01.794 --> 00:06:05.932 And unless you know exactly where groundwater seeps 00:06:05.932 --> 00:06:09.936 are entering streams, it is very, very difficult to account 00:06:09.936 --> 00:06:14.307 for that impact of groundwater and surface water temperatures in our models. 00:06:14.407 --> 00:06:18.678 But because our users, managers, people making fisheries 00:06:18.878 --> 00:06:22.615 and other natural resource decisions on the ground know 00:06:22.749 --> 00:06:26.018 that groundwater has a big impact on surface water temperatures. 00:06:26.319 --> 00:06:29.155 They want to see that we're able to account for it in our models. 00:06:29.322 --> 00:06:32.191 And so the work that Marty and his team are doing in order 00:06:32.191 --> 00:06:35.828 to really get at fine scale groundwater surface 00:06:35.828 --> 00:06:38.698 water interactions are very, very important for us. 00:06:39.365 --> 00:06:42.602 Groundwater is a common source of water to streams. 00:06:42.602 --> 00:06:45.004 And when we think about modeling those streams, if you're not accounting 00:06:52.745 --> 00:06:54.147 And when we think about modeling those streams, if you're not accounting 00:06:54.147 --> 00:06:55.948 for the groundwater, you're going to have predictions 00:06:55.948 --> 00:06:59.419 that are too warm during the summer, and too cold during the winter. 00:06:59.986 --> 00:07:04.991 So many of our temperature models that we work with cover large areas, 00:07:05.158 --> 00:07:06.192 and that means that our models 00:07:06.192 --> 00:07:10.329 can get the broad patterns, but not necessarily all the details. 00:07:10.830 --> 00:07:12.565 Most streams are small streams. 00:07:12.565 --> 00:07:18.004 In fact, in some instances, more than 75% of streams are considered small. 00:07:18.004 --> 00:07:22.241 And so our current work is taking on that challenge, predicting temperatures 00:07:22.241 --> 00:07:25.411 at higher resolution and in smaller streams. 00:07:25.745 --> 00:07:26.813 Small streams are small. 00:07:26.813 --> 00:07:28.548 They're might be hard to get to. 00:07:28.548 --> 00:07:32.752 They're likely on private property, but each one by itself 00:07:32.752 --> 00:07:37.423 is relatively unknown and seemingly unimportant relative to larger rivers. 00:07:38.491 --> 00:07:41.527 We're able to analyze the temperature data to identify 00:07:41.527 --> 00:07:45.698 the primary influence on the temperature, whether that's atmospheric conditions 00:07:45.698 --> 00:07:49.635 like the air temperature or whether that's from groundwater. 00:07:49.635 --> 00:07:53.072 And we can use this to identify types of streams like those influenced 00:07:53.072 --> 00:07:56.576 by groundwater where the model needs to be improved. 00:07:56.576 --> 00:07:57.810 But also machine learning models 00:07:57.810 --> 00:08:00.680 are really good at noticing patterns in the data. 00:08:01.080 --> 00:08:05.384 And as they study those multi-year, high resolution temperature observations, 00:08:05.384 --> 00:08:09.155 they will learn about the important processes and apply that to other streams 00:08:09.155 --> 00:08:11.290 that don't have the high resolution projections. 00:08:11.557 --> 00:08:14.927 Of course, we need to check what the model learned against our understanding 00:08:14.927 --> 00:08:17.296 of those processes and the physics and the ecology. 00:08:18.064 --> 00:08:20.333 But the model can yield some powerful insights. 00:08:20.833 --> 00:08:24.103 The USGS has been measuring stream temperature for qualitative 00:08:24.103 --> 00:08:25.605 and quantitative purposes 00:08:25.605 --> 00:08:28.841 for many decades, but they're realizing that the simple temperature 00:08:28.841 --> 00:08:32.178 measurements made a variety of scales are just super valuable, 00:08:32.178 --> 00:08:36.048 indicating hydrological processes that are otherwise difficult to measure. 00:08:36.282 --> 00:08:37.283 But now we can benefit 00:08:37.283 --> 00:08:40.353 from kind of an explosion in technology from thermal infrared, 00:08:40.553 --> 00:08:43.756 from very affordable, reliable temperature loggers to take 00:08:43.756 --> 00:08:46.726 those principles to the next level and apply them to the next generation. 00:08:46.859 --> 00:08:48.561 Water observing system network.