2014 Temperature Monitoring Project Report

January 13, 2015

Lakes Environmental Association 230 Main Street Bridgton, ME 04009

207-647-8580 lakes@leamaine.org

Project Summary

LEA began using in-lake data loggers to acquire high resolution temperature measurements in 2013. We expanded in 2014 to include 15 basins on 12 lakes and ponds in the Lakes Region of Western Maine. The loggers, also known as HOBO temperature sensors, allow us to obtain important information that was previously out of reach because of the high cost of manual sampling. Using digital loggers to record temperature gives us both a more detailed and longer record of temperature fluctuations. This information will help us better understand the physical structure, water quality, and extent and impact of climate change on the waterbody tested. Most of the lakes tested reached their maximum temperature on July 23rd. Surface temperature patterns were similar across all basins. The date of complete lake mixing varied con- siderably, with shallower lakes destratifying in September and others not fully mixed until No- vember. Differences in yearly stratification were seen in lakes with HOBO sensor chain data from 2013 and 2014. A comparison with routine water testing data confirmed the accuracy of the HOBO sensors throughout the season. A month-by-month comparison of temperature profiles in each lake showed the strongest stratification around the time of maximum temperature, in late July. In addition, August stratification was stronger than June and early July stratification. Shallow sensors showed similar average temperatures between 2013 and 2014.

Deployment of temperature sensors on Trickey Pond.

2 Introduction and Background Because of its role in physical, chemical, and biological processes, temperature is an important and informative lake measurement. In order to get a better idea of temperature patterns in and between lakes, LEA began monitoring lake temperature using in-lake digital data loggers in 2013. These loggers, also known as HOBO sensors, are programmed to record temperature readings every 15 minutes. The sensors are deployed in the spring and the data is stored on the sensors until they are retrieved in the mid to late fall. LEA serves six towns in western Maine, providing comprehensive lake monitoring for 40 lakes and ponds. This comprehensive testing includes measurements of temperature profiles using a handheld “The data collected by these temperature YSI meter. This method is time consuming, resulting sensors provides much greater detail and in at best 8 temperature profiles per year. Using clarity than the traditional method ever could. HOBO sensors there is an initial time investment, Daily temperature fluctuations, brief mixing but once deployed, the sensors record over 12,000 events caused by storms, the date and time of profiles before they are removed in the fall. stratification set up and breakdown, and the This wealth of data provides much greater timing of seasonal high temperatures are all detail and clarity than the traditional method ever valuable and informative events that could. Daily temperature fluctuations, brief mixing traditional sampling can’t measure.” events caused by storms, the date and time of stratification set up and breakdown, and the timing of seasonal high temperatures are all valuable and informative events that traditional sampling can’t measure. The measurements these sensors record allow us to infer the effect of temperature on diverse lake characteristics such as stratification (lake layering), ecology, habitat, and nutrient loading. In addition, comparing temperature data over a number of years allows us to make observations about climate change in our region. During the first season of testing in 2013, four basins on three lakes (Highland Lake, Moose Pond, and 2 sites on Long Lake) contained sensors at 2 meter intervals measuring the entire water column. Nine additional lakes contained one sensor in a shallow (littoral) area. All sensors were attached to a rope held in place by an anchor and a sub-surface buoy. For the second season of testing in 2014, a total of 16 sites were tested (figure 1, next page). Thirteen basins at ten lakes and ponds contained sensors measuring the entire water column (table 1, page 5). Three additional sensors measured shallow temperature on two ponds. The locations of deep water sensors were clearly marked by regulatory-style buoys.

Lake Stratification

Most of the lakes LEA tests become stratified in the summer. This means that the lake separates into distinct layers – the epilimnion, metalimnion and hypolimnion – based on temperature and water density. The top layer, the epilimnion, is the warmest. Somewhere in the middle, there will be a large temperature and density shift. This is known as the metalimnion and it defines the location of the thermocline. The hypolimnion contains the cold- est water and reaches from the bottom of the metalimnion to the bottom of the lake. Each lake’s stratification is unique and is affected by weather, as well as the lake’s size, depth, and shape. Stratification sets up in the Spring and breaks down in the Fall. “Lake turnover” refers to the destratification of a lake, when the water completely mixes and the temperature becomes uniform from top to bottom. 3 Figure 1. Map of temperature sensor sites. Yellow circles indicate shallow sensor placement; green squares show the location of multi-sensor temperature buoys.

4 Sampling Methods Thirteen sites on ten lakes were outfitted with a marked, Table 1. Details of HOBO temperature data logger deployment, including lake/pond name, location of sensor string, type of deployment, and number regulatory-style buoy attached by of sensors per string. rope to an anchor (figure 2). The Name Midas # Location Type # sensors HOBO sensors (figure 3) were Back Pond 3199 Main basin Deep 5 attached to the rope at 2 meter in- Hancock Pond 3132 Main basin Deep 9 tervals, beginning 1 meter from Island Pond 3448 Main basin Deep 6 the bottom and ending approxi- Keoka Lake 3416 Main basin Deep 6 mately 1 meter from the top. Each Long Lake 5780 North basin Deep 9 buoy apparatus was deployed at Long Lake 5780 Middle basin Deep 9 the deepest point of the lake or McWain Pond 3418 Main basin Deep 6 basin it monitored. The setup re- Moose Pond 3134 North basin Deep 3 sults in the sensors being located Moose Pond 3134 Middle basin Deep 11 at odd numbered depths through- Moose Pond 3134 South basin Deep 6 out the water column (the shal- Peabody Pond 3374 Western shore Shallow 1 lowest sensor is 1 meter deep, the Peabody Pond 3374 Outlet Shallow 1 next is 3 meters, etc.). Sand Pond 3130 Main basin Deep 7 Two additional ponds, Pea- Stearns Pond 3234 Western shore Shallow 1 body and Stearns, contained tem- Trickey Pond 3382 Main basin Deep 8 perature sensors at shallow loca- Woods Pond 3456 Main basin Deep 4 tions. These were attached to a small mooring buoy and deployed so that they were located approxi- Figure 2. (Left) Diagram mately 1 meter below the surface showing the buoy apparatus of the water, at a total depth of with temperature sensors approximately 3 meters. attached. Temperature sensors were deployed between June 6th and July 9th, 2014 and collected between th th Figure 3. (Below) A HOBO October 30 and November 25 , temperature sensor. 2014. Each sensor was configured to take temperature readings at 15 minute intervals. This results in 96 readings from each sensor every day, and thousands of readings during the course of deployment.

5 Results and Discussion

General Patterns in 2014

Out of a total of 92 sensors deployed, only four were unable to provide data. Two sensors failed due to water intrusion, one was dislodged from the buoy line and lost, and a fourth sensor re- corded faulty data. This section will compare data from a variety of lakes. For a report focusing on temperature data from a specific lake, please see the Lake Informa- tion page on our website, www.mainelakes.org, and click on the lake Individual reports for you want to know more about. each lake can be found There were a few temperature patterns that were common across all basins in 2014, although they varied in intensity. This simi- on the Lake Information larity makes sense because surface temperature is affected strongly page of our website, by the weather, and these lakes and ponds have similar weather pat- www.mainelakes.org terns due to their proximity to one another. Figure 4 (next page) shows the complete dataset for Hancock Pond, with each sensor’s data graphed in a different color. It labels the common patterns seen across most of the lakes. Figure 5 is a more visual representation of the same data from McWain Pond, showing the thermal dynamics of the lake over time. A sharp spike in surface temperature was seen on all basins on July 2nd and/or 3rd. This was, in most cases, the second highest temperature reading for the season. The spike occurred around the time of heavy storms on the 3rd, so it is reasonable to assume the storm caused the subsequent cooling of water temperatures. Another sharp spike in temperature was not seen, though temperatures did gradu- ally increase over the month of July. The data showed that most basins reached their maximum temperature on or around the same date, which was July 23rd. The beginning of seasonal temperature decline across all basins occurred around the 8th and 9th of September. This date marked the beginning of a steady drop in temperature leading to lake turnover. There were two brief warm periods where calm conditions allowed for slight water restratifica- tion, these being between the 27th and 29th of September and the 17th and 18th of October.

6 Results and Discussion

Temperature Spike Temperature Peak Beginning of temperature decline

Slight Restratifica- tion

Lake Turnover

Figure 4. Graph of 2014 temperature data in Hancock Pond from 6/24 to 11/11. Each line represents a different sensor’s data; the top- most line is from the sensor at 1 meter, etc. This graph is representative of data from other lakes with HOBO temperature monitoring.

Figure 5. Heat map of McWain Pond showing temperature variation in the water column over time. The Y-axis represents with depth from the surface, with the top of the graph representing the top of the lake. Stratification deepens over time (block of color extends further down) until it breaks down. The uniform color on the right side of the graph indicates lake mixing. 7 Results and Discussion

Lake Turnover

Measurements such as the date of lake turnover (See “Lake Stratification” box), are dependent on an individual lake’s characteristics and will therefore vary more widely than surface temperature patterns. The earliest destratification (“turnover”) occurred on September 12th, with the latest occurring on Novem- ber 3rd. Shallower lakes, such as Woods Pond, and those that are exposed to heavy winds, like Long Lake, mixed the earliest. The smaller and/or deeper lakes destratified later. Interestingly, most lakes turned over during one of three 2-day periods, either in mid-September, mid-October, or the beginning of November. Precipitation records from NOAA Climate Data Online (www.ncdc.noaa.gov) show significant rainfall over the days when destratification occurred on lakes in October and November, and a large precipitation event a few days before the destratification of a number of lakes in mid-September. This rainfall likely weakened stratification, allowing for lake turnover a few days later. Wind speed data were not available, though it is reasonable to assume that winds played a factor in the timing of turnover.

Monthly Temperature Profiles

Monthly HOBO sensor temperature profiles from each basin were also graphed to show how stratification changed over time (figure 6). These profiles showed that late July had the strongest stratification, followed by late August, June, and September. The pattern, which shows the strengthening and weakening of stratification over time, was similar for most of the basins tested.

Figure 6. Monthly temperature profiles from Moose Pond, showing the changes stratification at monthly intervals between June and October. Stratification is strongest in late July, as evidenced by the large temperature difference between the surface (at the top of the graph) and the bottom (bottom of the graph). 8 Results and Discussion

Deep Sensors: Year-On-Year Comparisons

Three basins contained strings of sensors in 2013 and 2014. These were Moose Pond’s main basin and the north and middle basins of Long Lake. In each of these basins, the maximum temperature was higher in 2013 than in 2014, with a clear “spike” in temperature, unlike the more gradual peak in 2014. Each basin mixed completely around the same date in 2013 as in 2014, although in all cases the 2014 turnover was slightly earlier. Both Moose Pond and the north basin of Long Lake had much more pronounced daily temperature fluctuation than in 2013. In 2014, the temperature at 7 meters depth on Moose Pond often fluctuates by 5 oC or more in a single day, whereas in 2013 the difference is rarely more than 2 oC (figure 7, next page). One reason for this difference may be the new buoy setup. Slack in the line allows the buoy to be moved by wind and water currents, changing the sensors’ depths slightly, which then impacts the temperature. In 2013, the buoys were under the surface of the water, which kept the sensor lines relatively straight. Graph- ing daily averages for 2014 (figure 7, bottom) allows for a clearer picture of temperature patterns over the season, although compared to 2013 daily averages (not shown) there is still much more variability in the 2014 data. Both basins of Long Lake showed differences between 2013 and 2014 sampling. The north basin’s bottom temperature was consistently about 0.5 oC cooler in 2014. This may be the result of earlier stratification in 2014, or it could be related to a difference in the location of the temperature sensors from year to year. In the middle basin there was less of a difference between the top and bottom temperatures in 2014 compared to 2013, possibly a result of a later date of stratification. Because the middle basin is exposed to more mixing than other basins, it makes sense that stratification would occur later. This is also evident in the much warmer bottom temperatures commonly seen in this basin. Moose Pond appeared to have deeper stratification in 2014 than 2013, which is corroborated by the water testing data. A sample of epilimnetic (top layer) water known as a core sample, the depth of which usually indicates the position of the thermocline, was on average deeper in 2014 than in 2013. This is also the case, albeit less pronounced, in Long Lake’s middle basin. The average core depth for Long Lake’s north basin was the same in both years.

9 Results and Discussion

(Note: this spike was caused by temporary sensor removal and should be ignored)

Figure 7. Moose Pond data from 2013 (top) and 2014 (middle) show much more daily variability in 2014. Averaging each day’s 2014 temperature readings (bottom) dampens this variability and gives a clearer picture of temperature patterns.

10 Results and Discussion

Comparison with Water Testing Data

The data from each chain of temperature sensors were compared to LEA’s temperature data from traditional manual water testing. Routine lake monitoring is done on 24 basins bi-monthly in the summer, and temperature data is collected with a handheld YSI probe at every meter from the top of the lake to its deepest point. The resulting temperature profiles were graphed. In all cases, both the sensor and monitoring profiles followed the same curve. However, in 5 out of the 12 basins compared, there was a discrep- ancy in the depth of the sensors, evidenced by the two curves not being aligned with one another (figure 8). The YSI handheld probe used for routine water testing is attached to a long cable which is marked every meter. Therefore, we know that the water testing depth is accurate as long as the cable remains straight in the water. The temperature sensor buoy placement is based on the known lake depth, however the actual depth of the sensors is difficult to determine and will never be exact due to buoy line movement, anchor shift, water level fluctuation, and wind/current effects. It is likely that in the 5 basins where a dis- crepancy was found that the sensors were actually deeper or shallower than was estimated. When adjusted for depth, the temperatures from the HOBO sensors and manual water testing were always within 3 oC of each other, and most often within 0.5 oC. This level of accuracy is acceptable con- sidering the numerous sources of error associated with the set up, including the accuracy of the HOBO sensors, the calibration of the YSI probe, and the inevitable differences in the depth, location and position of the sensors vs. the probe at the time of testing.

Figure 8. Graph of water testing data (blue line) versus HOBO sensor data (red squares and green triangles). The sensor data fits the water testing data much better if the assumed sensor depth is decreased by one meter. This indicates that the sensors were closer to the surface of the water than assumed.

11 Results and Discussion

Shallow Sensor Data

The two ponds with shallow sensors, Peabody Pond and Stearns Pond, also contained sensors in 2013. In each pond, the 2013 and 2014 averages were very similar (±0.1-0.2 oC) over the June – October period analyzed (figure 9). Both ponds had higher maximum temperatures in 2013. The minimums were similar between 2013 and 2014 in both ponds (±0.1 in Peabody Pond and ±0.7 in Stearns). Both ponds had similar temperature patterns in both years, with a distinct temperature peak in 2013 and a more level period of ele- vated temperatures in 2014 occurring in mid-July. Peabody Pond’s second shallow water sensor, located near the outlet of the pond, had much more variable temperatures than the western shore sensor, though the average temperatures differed by a frac- tion of a degree. This sensor was closer to the water’s surface and in a more open location, giving it more exposure to sunlight than the sensor near the shore and likely causing the higher variability.

Figure 9. Comparison of Stearns Pond shallow temperature data in 2013 (black line) and 2014 (blue line). The maximum temperature was higher in 2013, however the average over the course of deployment was almost identical.

12 Looking Ahead to 2015

Overall, 2014 HOBO sensor temperature monitoring was successful. The new buoy setup worked well and most of the materials can be re-used in 2015. Buoy line slack appeared to be an issue this season, causing excessive variability in temperature readings. Next season we will reduce the amount of slack each line is given. Next summer, LEA plans to continue sampling in all of the basins monitored in 2014. Additionally, we will deploy a new temperature buoy in Keyes Pond. We hope to deploy the buoys earlier in the spring of 2015 if possible. This would allow us to determine the date of the onset of stratification, another important lake measurement. We are also planning to use more weather data, including rainfall, wind speed and wind direction measurements, in our analysis. This data will come from our weather station, which was purchased this year as part of our remote sensing buoy project. Additionally, LEA gratefully acknowledges the help and expertise we have received from Dr. Dan Buckley at the University of Maine, Farmington over the course of this project.

LEA Would Like to Thank...

Five Kezar Ponds Watershed Association Hancock and Sand Ponds Association Island Pond Association Keoka Lake Association McWain Pond Association Moose Pond Association Peabody Pond Association Residents of Woods Pond Trickey Pond Association

An anonymous family foundation

and all of our members

…for making this project possible with their generous support!

13 Lakes Environmental Association 230 Main Street Bridgton, ME 04009

207-647-8580 [email protected]