Use of Thematic Mapper Imagery to Assess Water Quality, Trophic State, and Macrophyte Distributions in Massachusetts Lakes

By Marcus C. Waldron, Peter A. Steeves, and John T. Finn

Abstract thematic mapper bands 1, 2, 3, and 4 digital num- bers were unsuccessful, primarily because of the During the spring and summer of 1996, extremely low concentrations of chlorophyll in 1997, and 1998, measurements of phytoplankton- the lakes studied, and also because of the highly chlorophyll concentration, Secchi disk transpar- variable dissolved organic carbon concentrations. ency, and color were made at 97 Massachusetts Predictive relations were developed between lakes within 24 hours of Landsat Thematic Secchi disk transparency and phytoplankton- Mapper imaging of the lakes in an effort to assess chlorophyll concentration, and between color water quality and trophic state. Spatial distribu- and dissolved organic carbon concentration. tions of floating, emergent, and submerged macro- Phytoplankton-chlorophyll concentration was phytes were mapped in 49 of the lakes at least inversely correlated with Secchi disk transparency once during the 3-year period. The maps were during all three sampling periods. The relations digitized and used to assign pixels in the thematic were very similar in 1996 and 1997 and indicated mapper images to one of four vegetation cover that 62 to 67 percent of the variability in Secchi classes—open water, 1–50 percent floating-and- disk transparency could be explained by the emergent-vegetation cover, 51–100 percent chlorophyll concentration. Analysis of color and floating-and-emergent-vegetation cover, and sub- dissolved organic carbon concentrations in water merged vegetation at any density. The field data samples collected by U.S. Geological Survey field were collected by teams of U.S. Geological teams in 1996–98 indicated that 91 percent of Survey and Massachusetts Department of Environ- the variance in color in Massachusetts lakes can mental Management staff and by 76 volunteers. be explained by variations in dissolved organic Side-by-side sampling by U.S. Geological Survey carbon. and volunteer field teams resulted in statistically Areas of open-water, submerged vegetation, similar chlorophyll determinations, Secchi disk and two surface-vegetation-cover classes predicted readings, and temperature measurements, but con- from Thematic Mapper images acquired in the current color determinations were not similar, pos- summer of 1996 closely matched the areas sibly due to contamination of sample bottles observed in a set of field observations. However, issued to the volunteers. the same analysis applied to a set of data acquired Attempts to develop predictive relations in the summer of 1997 resulted in somewhat between phytoplankton-chlorophyll concentration, less reliable predictions, and an attempt to predict Secchi disk transparency, lake color, dissolved 1996 vegetation-cover areas using the relations organic carbon, and various combinations of developed in the 1997 analysis was unsuccessful.

Abstract 1

INTRODUCTION (for simplicity, the word “lake” will be used throughout this report to refer to any open body of water) in Accelerated eutrophication due to changing Massachusetts, the costs and logistical problems asso- drainage-basin activities is a significant problem ciated with a statewide lake-quality-monitoring affecting Massachusetts lakes (Massachusetts Water program are substantial. Resources Commission, 1994). This accelerated, or cultural, eutrophication is caused by nutrient-rich The development of satellite resources such as effluents from sewage treatment plants, runoff of fertil- the Landsat Thematic Mapper (TM) and new tech- izers and animal wastes, stormwater runoff from niques for processing and analyzing satellite data offer impervious surfaces, leaching from septic systems, and the potential for augmenting the data-collection and increased soil erosion resulting from construction and resource-evaluation efforts of State environmental other similar activities. Cultural eutrophication can agencies. Landsat images can provide high-resolution lead to excessive growth of aquatic macrophytes, information concerning a number of important limno- increased turbidity, depletion of dissolved oxygen, and logic features, including chlorophyll-a concentration, subsequent loss of fish habitat. Massachusetts lakes are turbidity, color, algal production rates, nutrient concen- especially susceptible to the problem because most trations, and surface-water temperatures (Scarpace and drainage basins are heavily developed and most lakes others, 1979; Verdin, 1985; Raitala, 1986; Shimoda and are subject to multiple uses. In addition, many lakes in others, 1986). The availability of Landsat images Massachusetts were created or enlarged by impounding dating back to the early 1970s allows for the develop- water behind dams, resulting in submerged soils within ment of long-term records of properties related to lake these impoundments that may provide an additional trophic state and can be used to identify trends (Witzig source of nutrients affecting the trophic state of the and Whitehurst, 1981; Lillesand and others, 1983). lakes. Trophic state, the extent of the effect of eutrophi- cation due to nutrient enrichment, has been difficult The U.S. Geological Survey (USGS), in cooper- to quantify in Massachusetts because many lakes ation with the Massachusetts Department of Environ- develop dense beds of aquatic macrophytes in response mental Management (MADEM), has investigated the to eutrophication, and most methods for assessing use of Landsat TM data for Statewide assessment of trophic state are based on the relative abundance of lake quality and trophic state. Measurements of water phytoplankton algae and do not take into account the temperature, Secchi disk transparency, color, and the biomass of macrophytes (Canfield and others, 1983). concentration of phytoplankton chlorophyll were made The recently adopted Massachusetts Policy on in 97 lakes during the summers of 1996, 1997, and Lake and Pond Management advocates a comprehen- 1998, by USGS and MADEM staff and by a team of sive approach to lake eutrophication that integrates trained volunteers recruited by the Massachusetts education, watershed protection, and in-lake manage- Water Watch Partnership (MassWWP). The lake mea- ment in an attempt to reconcile desired uses of surements were timed to coincide with Landsat-5 TM Massachusetts lakes with their ability to support those imaging of the State. During the same period, the mid- uses (Massachusetts Water Resources Commission, to-late-summer distributions of floating, emergent, and 1994). Central to the Massachusetts Policy on Lake and submerged macrophytes were mapped in 62 lakes, Pond Management is the need to assess lake-water again by a combination of professional and volunteer quality at regular intervals and to identify trends field teams. The field data were correlated with data (both negative and positive) in lake trophic state. With extracted from a set of four TM images, each image more than 3,000 named lakes, ponds, and reservoirs representing the eastern two-thirds of the State.

2 Use of Thematic Mapper Imagery to Assess Water Quality, Trophic State, and Macrophyte Distributions in Massachusetts Lakes

The purpose of this report is to demonstrate STUDY METHODS how Landsat TM data may be used to assess the water quality and trophic state of Massachusetts Landsat-5 orbits the earth at an altitude of lakes and to monitor the distributions of aquatic 705 km in a near-polar, sun-synchronous orbit with a macrophytes. The report describes methods of field- 16-day, 233-orbit repeat cycle. The primary imaging data collection and procedures used for acquiring and processing the TM data. Field data collected by instrument on Landsat-5 is the TM, which senses volunteer water-quality monitoring teams are com- reflected light energy in seven spectral bands, three in pared statistically with concurrent measurements the visible range, three in the near- and mid-infrared, made by USGS field teams. Results are presented sepa- and one in the thermal infrared (table 1). The TM sen- rately for TM-based assessment of lake-water quality sors have a spatial resolution of 120-by-120 m for the and trophic state and for TM-based mapping of thermal-infrared band and 30-by-30 m for the other six lake-macrophyte distributions. Data collected during spectral bands. The sensors can distinguish 256 levels the study are available via the World Wide Web at http://ma.water.usgs.gov/lakesandponds/. of brightness (radiance) in each spectral band for each 30-by-30 m or 120-by-120 m picture element (pixel). The authors wish to thank the volunteers and staff of the Massachusetts Water Watch Partnership for The brightness levels are recorded as digital numbers their generous contributions of time and other (DNs) representing the average radiance measured over resources to this project. the ground area corresponding to each pixel.

Table 1. Thematic Mapper spectral bands

Spectral Wavelength range Nominal spectral Principal Application(s) band (micrometers) location

1 0.45–0.52 Blue-green Designed for maximum penetration of water. Used for bathymetric mapping of shallow water bodies. Also used for distinguishing soil from vegetation and deciduous from coniferous trees. 2 0.52–0.60 Green Designed to measure green reflectance peak of vegetation. Useful for assessing plant vigor. 3 0.63–0.69 Red Designed to measure light that is strongly absorbed by chlorophyll. Used for discriminating vegetation types. 4 0.76–0.90 Near infrared Useful for determining vegetation types, vigor, and biomass. Also used for distinguishing shorelines of water bodies. 5 1.55–1.75 Mid-infrared Measures moisture content of soil and vegetation. Penetrates thin clouds. Used to distinguish snow from clouds. 6 10.4–12.5 Thermal infrared Nighttime images are useful for thermal mapping and for estimating soil moisture. 7 2.08–2.35 Mid-infrared Measures absorption by hydroxyl ions in minerals. Used for mapping hydrothermally altered rocks associated with mineral deposits. Also sensitive to vegetation moisture content.

Study Methods 3

Landsat imagery is subdivided into scenes was provided to the volunteers for measuring Secchi based on a Worldwide Reference System (WRS) disk transparency, for collecting and processing consisting of vertical paths and horizontal rows. water samples to be analyzed for color and phytoplank- Each combination of path and row describes a unique ton-chlorophyll concentration, and for mapping 185-by-170-kilometer rectangle of ground-surface distributions of macrophytes in the lakes. area. The State of Massachusetts is represented by The volunteers’ efforts greatly increased the WRS paths 11, 12, and 13, and by rows 30 and 31 amount of water-quality data collected concurrently (fig. 1); however, because adjacent paths overlap by as with TM image acquisition. MADEM and USGS much as 40 percent, most of the State appears in paths field personnel collected data at 94 stations on 65 lakes 12 and 13. If the images are shifted north along the two and volunteers collected data at 68 stations on 48 lakes. paths, then about 90 percent of the State can be imaged Sixteen of the lakes were sampled jointly by USGS and in only two scenes. volunteers for quality assurance purposes. Volunteers Use of TM imagery to assess lake quality and were able to collect data each time the satellite was trophic state requires that predictive relations be devel- overhead from May through September. Most of the oped between measured water-quality characteristics lakes sampled by MADEM and USGS could be sam- and the TM data. Ideally, these relations are based on pled only once due to resource limitations. A list of the measurements made at or close to the time of TM-data study lakes, their locations, and the numbers and kinds acquisition on a large number of lakes exhibiting the of water-quality measurements made during the three range of conditions likely to be encountered in the study periods is presented in table 6 (at back of report). State. The 16-day Landsat-5 orbital repeat cycle pro- In addition, distributions of floating, emergent, and vides about 10 opportunities for image acquisition submerged macrophytes were mapped in 49 lakes at between May 1 and September 30. However, the least once during the study either by volunteers or by number of lakes that could be sampled during each fly- professional field personnel (table 2). over was limited by the small number of State and Selection of the study lakes was determined USGS personnel, boats, and equipment available for partly by the study requirement that the lakes be repre- use in the study. The solution to this problem was to sentative of lakes throughout the State and partly by engage the Massachusetts Water Watch Partnership other circumstances, including the affiliations and (MassWWP), which is affiliated with the University of interests of the volunteers and the program require- Massachusetts and the Massachusetts Water Resources ments of the MADEM. Most of the volunteers live Research Center in Amherst, Mass., to recruit and train close to the lakes they sampled. The MADEM col- volunteers to sample lakes throughout the State. lected data primarily from lakes in State parks, forests, A total of 76 individuals participated as volun- and reservations. Additional lakes were added to the teers during the three spring–summer sampling peri- list to ensure that the full range of trophic and water- ods. Twenty-one volunteers were involved in the first quality conditions were represented in the data set. The sampling period (1996), 61 during the second sampling lakes ranged in surface area from 4 to 696 ha with a period (1997), and 39 during the third sampling period median surface area of 36 ha. Seventy-five percent of (1998). All volunteers were trained in lake-sampling the lakes had surface areas of 81 ha or less. Maximum and sample-processing techniques in a series of depths of the lakes ranged from 2 to 30 m with a hands-on training sessions conducted each spring by median of 7 m. Seventy-five percent of the lakes were MassWWP, MADEM, and USGS staff. Equipment less than 10 m deep.

4 Use of Thematic Mapper Imagery to Assess Water Quality, Trophic State, and Macrophyte Distributions in Massachusetts Lakes

Sampling and Analysis for Phytoplankton-Chlorophyll Water-Quality Characteristics Concentration

Sampling stations were established over the Water samples for phytoplankton-chlorophyll deepest part of each study lake. For lakes with surface determinations were collected by hand in brown plastic areas greater than about 81 ha, or with multiple basins, 1-liter bottles from a depth of about 0.25 m below the as many as six stations were established and monitored surface. The samples were returned on ice to shore separately. Stations either were marked with a buoy or where measured volumes were filtered onto 47 mm were located by aligning two pairs of landmarks on the Watman GF/F glass-fiber filters using a maximum suc- shore spaced at a 90 degree angle relative to the station. Exact locations (latitude and longitude) of stations tion pressure of 0.5 atmospheres. The filters were sampled by USGS field teams were determined by a folded in half and placed in a drying chamber where global positioning system (GPS). All other stations they were air dried at room temperature for 30 minutes. were marked on appropriate USGS 1:25,000-scale The dried filters were wrapped in aluminum foil and topographic sheets and their locations determined with mailed overnight to the Environmental Analytical a digitizer. Laboratory at the University of Massachusetts, Orbital schedules for Landsat-5 were obtained Amherst, for analysis. In the laboratory, the filters for each study period from the Earth Observation were ground in alkalized 90-percent acetone and ana- Satellite Company (EOSAT) in Lanham, Md. Sampling lyzed spectrophotometrically for chlorophyll-a and usually was scheduled for the morning of the flyover to phaeophytin-a concentrations (American Public Health coincide with the 9:45 a.m. equatorial crossing of the satellite, although data collected up to 24 hours before Association and others, 1995). For the purposes of or after image acquisition were considered acceptable. this study, the phytoplankton-chlorophyll concentra- tion was considered to be the sum of the measured Field Observations chlorophyll-a and phaeophytin-a concentrations. Upon arriving at a station, samplers completed a field form (fig. 2) in which they identified the lake and Lake Color the station, and recorded maximum depth, percent cloud cover, barometric pressure, and air temperature. Filtrate produced during field processing of Surface-water temperature was measured either with the chlorophyll samples was transferred to clean, a digital thermometer or with a standard alcohol ther- prelabeled glass or polyethylene bottles, which mometer. Secchi disk transparency was determined were shipped on ice overnight to the University of by lowering a standard 20-centimeter Secchi disk into Massachusetts, Amherst, for analysis of color. Lake the water and noting the exact depth at which it disap- color was determined spectrophotometrically in a peared, then raising the disk and noting the depth at which it reappeared. The Secchi disk transparency was 5-centimeter cell at a wavelength of 425 nm. The recorded as the mean of the two readings to the nearest measured absorbance was converted to platinum-cobalt 0.1 m. Exact times of all field observations were units (PCU) with a standard curve (American Public recorded on the field sheet. Health Association and others, 1995).

Study Methods 5

73°00´ ° 72 30´ 72°00´

WRS PATH 13 ROW 30 VERMONT 11009 NEW HAMPSHIRE 35095 35074 33014 35041 33012 35017 1 33019 3 35035 35090 7 81157

33009 12002 42°30´ 21083 33017 34103

21005 21078 36005

2 6 36173 11 4 8 36082 32013 34051 NEW YORK 21043 36155 21044 12 WRS PATH 13 ROW 31 36150 21011 31044 31004 10 32055 36142 5 31027 41052 41008 42005 1 31052 41014 41016 9 42036 CONNECTICUT 42°00´

INDEX MAP OF EXPLANATION NEW ENGLAND STATES

21043 BASIN BOUNDARY 68˚ STUDY LAKE WITH 70˚ POND AND LAKE IDENTIFICATION SUBBASIN BOUNDARY SYSTEM NUMBER CANADA USA NOMINAL LOCATION OF SATELLITE 46˚ PATH OR ROW BOUNDARY. THE WORLDWIDE REFERENCE SYSTEM MAINE (WRS) NUMBERS ARE SHOWN CANADA 72˚ IN HELVETICA OBLIQUE TYPE USA VERMONT RIVER DRAINAGE BASINS 1. Hudson 11. Nashua 18. North Coastal 22. Cape Cod 44˚

2. Housatonic 12. Blackstone 19. Boston Harbor 23. Islands NEW 3. Deerfield 13. Merrimack a.Mystic 24. Buzzards Bay HAMPSHIRE ATLANTIC OCEAN

4. Westfield 14. Concord b.Neponset 25. Taunton 0 40 80 MILES NEW YORK NEW 5. Farmington a.Assabet c.Weymouth 26. Narragansett Bay 0 40 80 KILOMETERS MASSACHUSETTS 6. Connecticut b.Concord and Wier and Mt. Hope 42˚ 7. Millers and Sudbury 20. Charles Bay Shore RHODE ISLAND 8. Chicopee 15. Shawsheen 21. South Coastal 27. Ten Mile CONNECTICUT 9. Quinebaug 16. Parker a.North and

10. French 17. Ipswich South River NEW YORK b.South Coastal Shore

Figure 1. Locations of study lakes and Landsat-5 Worldwide

6 Use of Thematic Mapper Imagery to Assess Water Quality, Trophic State, and Macrophyte Distributions in Massachusetts Lakes

71°00´ 70°30´

WRS PATH 12 ROW 30

71°30´ 16 NEW HAMPSHIRE 13 91010 84001 ATLANTIC OCEAN

17 93014 81154 15

11 81007 18 82110 71019 71047 HIGH ST. IMPOUNDMENT 82118 82109 19a 82011 82092 81153 82042 82119 71040 Massachusetts 82104 14a 82015 82058 72017 Bay 82029 72052 51027 14b 72125 51118 82061 51024 82120 82003 20

51112 72140 73062 19c 51152 19b 10 12 72008 21a 70°00´ 42064 51179 WRS PATH 12 ROW 31 51172 9 RHODE ISLAND 25 27 62205 94007 70°30´ 95034 21b 95119 95030 WRS PATH 11 ROW 31 62108 95054 95025 26 96091 96039

24 96307 96170 95151 96302 96218 22 26 96012 96198 96333

96155

Nantucket 41°30´ Sound

0 50 MILES 23

0 50 KILOMETERS 23

Reference System paths and rows for Massachusetts.

Study Methods 7

Table 2. Massachusetts lakes for which the distributions of floating, emergent, and submerged aquatic macrophytes were mapped in 1996, 1997, and 1998 for calibration of Landsat-5 Thematic Mapper imagery

[PALIS, Pond and Lake Identification System; X, mapped; blank space, not mapped; ---, no assigned code] 1996 1997 1998 PALIS Lake name Emer- Sub- Emer- Sub- Emer- Sub- code Floating Floating Floating gent merged gent merged gent merged

Althea Lake 84002 X X X Ashmere lake 21005 X X X Bare Hill Pond (USGS) 81007 X X X Bare Hill Pond (USGS) 81007 X X Bearse Pond 96012 X X Big (Benton) Pond 31004 X X X Buckley-Dunton Lake 32013 X X X Charge Pond 95025 X X Chebacco Lake 93014 X X X Coes Reservoir 51024 X X X Cook Pond 51027 X X Curlew Pond 95034 X X X Dudley Pond 82029 X X East Brimfield Reservoir (East) 41014 X X X East Brimfield Reservoir (West) 41014 X X X Fearing Pond 95054 X X X Goose Pond 21043 X X X Greenwater Pond 21044 X X X Heard Pond 82058 X X X High Street Impoundment --- X X X Horn Pond 71019 X X X X X X X X X Mauserts Pond 11009 X X X Merino Lake 42036 X X X Metacomet Lake 34051 X X Onota Lake 21078 X

Dissolved Organic Carbon Analytical Quality Assurance Concentration Twelve sets of duplicate samples were collected Samples for DOC determinations were filtered at various sampling sites during the study and analyzed through 0.45-µm-pore-size silver filters into baked separately for DOC by the USGS National Water Qual- brown-glass bottles using a stainless steel filtration ity Laboratory and the University of Massachusetts system. The samples were then stored on ice prior to Environmental Analytical Laboratory. Differences analysis. DOC determinations were carried out by the between DOC determinations by the two laboratories Environmental Analytical Laboratory at the University ranged from 18 to 40 percent with a mean of 26 per- of Massachusetts, Amherst. Analysis of DOC was cent. Concentration differences among 12 duplicate by wet oxidation with carbon dioxide detection by determinations made by the Environmental Analytical infrared spectroscopy (Fishman and Friedman, 1989). Laboratory ranged from 0 to 18 percent with a mean of

8 Use of Thematic Mapper Imagery to Assess Water Quality, Trophic State, and Macrophyte Distributions in Massachusetts Lakes

Table 2. Massachusetts lakes for which the distributions of floating, emergent, and submerged aquatic macrophytes were mapped in 1996, 1997, and 1998 for calibration of Landsat-5 Thematic Mapper imagery—Continued

1996 1997 1998 PALIS Lake name Emer- Sub- Emer- Sub- Emer- Sub- code Floating Floating Floating gent merged gent merged gent merged

Patch Reservoir 51118 X X X Pequot Pond 32055 X X X Pontoosuc Lake 21083 Puffer Pond (USGS) 82092 X X X Puffer Pond 82092 X X X X Rocky Pond 95119 X X X Spy Pond 71040 X X Sugden Reservoir 36150 X X X X X Thompson Pond 36155 X X X X X X X X Upper Spectacle Pond 31044 X X X Waban Lake 72125 X X X X X X X X X Walden Pond 82109 X X X Wallum Lake (lower) 51172 X X Warners Pond 82110 X X X Webster Lake 42064 X X Wequaquet Lake 96333 X X X White Pond (Concord) 82118 X X X X X X Whitehall Reservoir (NE) 82120 X X X Whitehall Reservoir (NW) 82120 X X X Whitehall Reservoir (SW) 82120 X X X Winnekeag Lake 81157 X X Winter Pond 71047 X X X X X X X X X Winthrop Lake 72140 X X X X X X York Lake 31052 X X X

7 percent. Color determinations made on the same sets Quality Assurance of of duplicates differed by 4 to 10 percent with a mean of Volunteer Data 6 percent. Twelve sets of duplicate phytoplankton chloro- Measurements of phytoplankton-chlorophyll phyll samples were analyzed during the study period concentration, Secchi disk transparency, color, and water temperature were made simultaneously at 26 sta- by the University of Massachusetts Environmental tions on 16 lakes (table 3) by USGS and volunteer field Analytical Laboratory. The percent difference between teams to determine the reproducibility and reliability duplicates ranged from 7 to 100 and the mean percent of the data. Measurements usually were made from difference was 28. The highest percent differences the same boat. Samples were processed separately by were obtained from samples with extremely low each team and were shipped together to the analytical chlorophyll concentrations (less than 1.0 µg/L). laboratory.

Study Methods 9

USGS-MADEM-MassWWP Lake and Pond FIELD DATA

Name of Lake: PALIS Code: Station No: Lat: Long: Sampled By: Date:

Field Measurements ns =not sampled Time of Measurement Maximum Depth: meters/feet Cloud Cover: % am/pm Barometric Pressure: mmHg/inHg am/pm

Air Temperature: oC/oF am/pm

Surface Water Temperature: oC/oF am/pm Secchi Disk Depth: meters/feet am/pm

Water Samples Collected Date/Time of Sample: Collection Preparation Shipping Vol. Filtered (L) Chlorophyll: Color: DOC:

Remarks

Conversions Multiply by To obtain Feet 0.3048 Meters Gallons 3.7853 Liters Atmospheres 760 mmHg Pounds/sq in 51.715 mmHg

Temperature can be converted to degrees Celsius (oC) from degrees Fahrenheit (oF) by the following equation: oC = (oF-32) / 1.8

Figure 2. Example of field form used in volunteer field-data collection program.

10 Use of Thematic Mapper Imagery to Assess Water Quality, Trophic State, and Macrophyte Distributions in Massachusetts Lakes

Table 3. Lakes sampled by Massachusetts Water Watch Partnership volunteers concurrently with U.S. Geological Survey staff for chlorophyll concentration, Secchi disk transparency, color, and water temperature in 1997 and 1998

[Raw data available on the Internet at http://water.usgs.gov. PALIS, Pond and Lake Identification System; USGS, U.S. Geological Survey; X, water samples collected or measurements made concurrently; --, not measured] Water Chlorophyll Secchi Color Sampling PALIS Temperature Lake name station code Sampling USGS WWP USGS WWP USGS WWP USGS WWP date

Long Pond 1 62108 8-27-97 X X -- -- X X X X Long Pond 2 62108 8-27-97 X X -- -- X X X X Long Pond 3 62108 8-27-97 X X -- -- X X X X Long Pond 4 62108 8-27-97 X X -- -- X X X X Long Pond 5 62108 8-27-97 X X -- -- X X X X Long Pond Deep 62108 8-27-97 -- -- X X ------hole Lower Naukeag Lake 1 35041 8-04-98 X X X X -- -- X X Onota Lake 1 21078 6-01-98 X X X X X X X X Onota Lake 3 21078 6-01-98 X X X X X X X X Pontoosuc Lake 1 21083 6-01-98 X X X X X X X X Stearns Mill Pond 1 82104 7-12-98 X X -- -- X X X X Upper Naukeag lake 1 35090 8-04-98 X X X X X X X X Wallum Lake 1 51172 6-10-98 X X X X X X X X Watatic Lake 1 35095 8-04-98 X X X X X X X X Webster Lake 1 42064 6-10-98 X X X X X X X X Webster Lake 2 42064 6-10-98 X X X X X X X X White Pond (Concord) 1 82118 8-13-98 X X X X X X X X White Pond (Hudson/Stow) 1 82119 7-12-98 X X X X X X X X Whitins Reservoir 1 51179 6-10-98 X X X X X X X X Whitins Reservoir 2 51179 6-10-98 X X X X X X X X Willet Pond 1 73062 7-28-98 X X X X X X X X Winnekeag Lake 1 81157 8-04-98 X X X X X X X X

Comparisons of measurement results obtained (fig. 3C). Differences ranged from 2 to 200 percent. by the two sampling teams are presented in figures 3A The mean percent difference was 72. The through 3B. There was good agreement between large differences may have resulted from inadequate volunteer and USGS Secchi disk transparency determi- cleaning of the volunteer sample containers. All USGS nations (fig. 3A). The percent difference between samples were submitted to the laboratory in baked the two sets of measurements ranged from 0 to 24 glass bottles while volunteer samples were submitted and the mean percent difference was 8. The two sets of in polyethylene bottles that had been used in a previous phytoplankton-chlorophyll determinations were more investigation. variable (fig. 3B). Percent differences ranged from 7 to 54 and the mean was 24 percent. At concentrations Comparisons of water temperature measure- greater than 7 µg/L, the USGS samples consistently ments were reasonably good (fig. 3D), given the fact yielded 2–3 µg/L more chlorophyll than did the that different types of measuring devices were used. volunteer samples. Percent differences between USGS and volunteer Agreement between color measurements made measurements ranges from 0 to 17 with a mean of on USGS and volunteer water samples was not good 3 percent.

Study Methods 11

8 A. C. 100 7

6 80

5

60 4

3 40 IN PLATINUM-COBALT UNITS IN PLATINUM-COBALT

BY VOLUNTEERS, IN METERS VOLUNTEERS, BY 2 20 LAKE COLOR IN VLOUNTEER SAMPLES, LAKE COLOR IN

SECCCHI DISK TRANSPARENCY MEASURED TRANSPARENCY SECCCHI DISK 1

0 0 012345678 0 20406080100 SECCHI DISK TRANSPARENCY MEASURED LAKE COLOR IN U.S. GEOLOGICAL SURVEY BY U.S. GEOLOGICAL SURVEY STAFF, IN METERS SAMPLES, IN PLATINUM-COBALT UNITS

18 28 B. D. 16 27

26 14

25 12 24 10 23 8 22 6

IN DEGREES CELSIUS 21

4 20 SAMPLES, IN MICROGRAMS PER LITER IN MICROGRAMS SAMPLES, 2 19 CHLOROPHYLL CONCENTRATIONS IN VOLUNTEER IN CONCENTRATIONS CHLOROPHYLL WATER TEMPERATURE MEASURED BY VOLUNTEERS, MEASURED BY TEMPERATURE WATER 0 18 0 2 4 6 8 10 12 14 16 18 18 19 20 21 22 23 24 25 26 27 28 CHLOROPHYLL CONCENTRATIONS IN U.S. GEOLOGICAL SURVEY STAFF, IN DEGREES CELSIUS SURVEY SAMPLES, IN MICROGRAMS PER LITER WATER TEMPERATURE MEASURED BY U.S. GEOLOGICAL

Figure 3. Relations between volunteer and U.S. Geological Survey staff measurements of (A) Secchi disk transparency; (B) phytoplankton-chlorophyll concentration; (C) color; and (D) surface-water temperature. (Dotted line is line of one-to-one correspondence of paired measurements.)

12 Use of Thematic Mapper Imagery to Assess Water Quality, Trophic State, and Macrophyte Distributions in Massachusetts Lakes

THEMATIC MAPPER-BASED the reflected-IR band (TM4). In the absence of scatter- ASSESSMENT OF WATER ing due to haze and other atmospheric irregularities, the QUALITY AND TROPHIC STATE intercepts of the regression lines should pass through the origin (Wilkie and Finn, 1996). In all four scenes, At the end of each spring–summer study period, the regression lines intercepted the TM4 axes at some the available TM images were examined and ranked positive value, indicating the need for correction. The according to their degree of atmospheric interference band was adjusted by the amount that the intercept due to haze and cloud cover, and the amount of lake- shifted from the origin (Wilkie and Finn, 1996). The water-quality data available for correlation with pixel corrected TM values, plus the corresponding lake- brightness values. On this basis, four scenes were pur- water-quality data used in the analysis, are presented in chased from the USGS Earth Resources Observation table 7 (at back of report). Systems (EROS) Data Center (EDC) in Sioux Falls, South Dakota. Scene identification codes and other descriptive information are presented in table 4. Secchi Disk Transparency and Each scene comprises picture elements (pixels) repre- Phytoplankton-Chlorophyll senting either 30-by-30 m (for visible and reflected- Concentration infrared (IR) wave bands) or 120-by-120 m (for the thermal-IR wave band) ground-resolution cells. This Phytoplankton-chlorophyll concentration was study used data from visible TM wave bands 1 (TM1, inversely correlated with Secchi disk transparency 0.45–0.52 µm), 2 (TM2, 0.52–0.60 µm), and 3 (TM3, during all three sampling periods (fig. 4). The regres- 0.63–0.69 µm), and reflected-IR wave band 4 (TM4, sion equations for the relations in 1996 and 1997 were 0.76-0.90 µm). Data for each pixel consist of digital similar, and indicated that 62 to 67 percent of the vari- numbers (DNs) ranging from 0 to 255 that represent ability in Secchi disk transparency could be explained the recorded intensity of reflected radiation in one of by the chlorophyll concentration. The unexplained the wave bands. The scenes were radiometrically and variability is due to a combination of sampling and geometrically corrected, rotated, and aligned to state analytical errors, variations in lake color, and the pres- plane coordinates by the EDC. ence of suspensoids other than phytoplankton algae Lake-water-quality data collected within 24 (Goldman and Horne, 1983). hours of acquisition of each TM scene were compiled The relation was shifted significantly in the 1998 and the brightness values for pixels corresponding to dataset. The slope of the regression line was similar to the station locations were extracted from the TM that calculated for the previous two years, but the y- images. Brightness values for the three visible bands intercept was nearly doubled, so that chlorophyll con- (TM1, TM2, and TM3) were then corrected for haze by centrations associated with a given Secchi disk trans- regressing them against the corresponding values for parency increased by an average of 135 percent over the previous two years. The apparent increase may be related to a change in the analytical instrumentation Table 4. Landsat-5 Thematic Mapper scenes used to assess water-quality and trophic state of Massachusetts lakes used in the Environmental Analysis Laboratory that year. [EDC, Earth Resources Observation Systems (EROS) Data Center; WRS, Secchi disk transparency values and phytoplank- Worldwide Reference System] ton-chlorophyll data corresponding to the four TM Image EDC scene WRS WRS scenes were analyzed using simple linear regression to acquisition identification path row date develop relations that could be used to predict the water-quality characteristics from the TM data. LT5012031009620410 7-22-66 012 031.00000 Twenty-eight combinations of haze-corrected DNs for LT5013030009720610 6-23-97 012 030.97174 TM bands 1, 2, 3, and 4 were used as models in the LT5012030009723810 8-26-97 012 030.97000 analysis (table 5). The analytical approach was to plot LT5012031009816110 6-10-98 012 031.98000

Thematic Mapper-Based Assessment of Water Quality and Trophic State 13

5.0 5.0 log(Chlorophyll) = -1.229 log(Secchi) + 2.670 log(Chlorophyll) = -1.345 log(Secchi) + 2.470 2 2 4.0 R = 0.621 4.0 R = 0.676 n = 65, p < 0.0001 n = 156, p < 0.0001 2

3.0 3.0

2.0 2.0

1.0 1.0

0.0 0.0 NATURAL LOGARITHM OF NATURAL NATURAL LOGARITHM OF NATURAL CHLOROPHYLL CONCENTRATION CHLOROPHYLL -1.0 CONCENTRATION CHLOROPHYLL -1.0 A. 1996 B. 1997 -2.0 -2.0 -1.5 -0.5 0.5 1.5 2.5 3.5 -1.5 -0.5 0.5 1.5 2.5 3.5 NATURAL LOGARITHM OF SECCHI DISK TRANSPARENCY NATURAL LOGARITHM OF SECCHI DISK TRANSPARENCY

Figure 4. Relations between Secchi disk transparency and phytoplankton-chlorophyll concentration in Massachusetts lakes in (A) 1996; (B) 1997; and (C) 1998.

the natural logarithms of the Secchi disk transparencies researchers have reported on the ability of the TM to versus the phytoplankton-chlorophyll concentrations resolve differences in these parameters. More recently, for a given scene to determine if the expected inverse Khorram and others (1991) and Baban (1993, 1997) relation existed between the two datasets. Obvious out- reported successful correlation of lake chlorophyll con- liers were discarded and the remaining natural-log- centration and Secchi disk transparency with TM data transformed lake data were regressed against each of using the same methods and TM models as this study. the TM models listed in table 5. No relations were The most important difference between these observed that could be applied consistently to all the studies and the current one is the extremely low con- scenes and only a few of the models explained more centrations of phytoplankton chlorophyll typically than 60 percent of the variability in either the chloro- found in Massachusetts lakes. Chlorophyll concentra- phyll or the Secchi disk data. tions in the study lakes ranged from 0.5 to 84.5 µg/L. The lack of any predictive relations between the However, the mean chlorophyll concentration was only TM data and phytoplankton-chlorophyll concentration 6.0 µg/L and the median concentration was 3.1 µg/L. or Secchi disk transparency was surprising given the Median chlorophyll concentrations in lakes studied by long history of successful use of TM data to predict the Lillesand and others (1983) generally were much water quality and trophic state of inland waters. Begin- higher, often ranging from 30 to more than 100 µg/L. ning with the work of Lathrop and Lillesand (1986), Values of that magnitude were observed in fewer than who used some of the earliest available TM data to 5 of the 97 lakes included in this study. In addition, assess chlorophyll concentration, Secchi disk transpar- color of the study lakes ranged from less than 1 to ency, turbidity, and the concentration of suspended 547 PCU, with a mean of 91 and a median of 49 PCU. solids in Green Bay and central Lake Michigan, The wide variation in color may have introduced

14 Use of Thematic Mapper Imagery to Assess Water Quality, Trophic State, and Macrophyte Distributions in Massachusetts Lakes

5.0 additional variability into the relations between C. 1998 chlorophyll concentration, Secchi disk trans- 4.0 parency, and the TM data. In any case, it appears that eutrophication of Massachusetts 3.0 lakes frequently is manifested more by prolif- eration of macrophytes than it is by growth of phytoplankton. 2.0

1.0 Lake Color and Dissolved Organic Carbon Concentration 0.0

NATURAL LOGARITHM OF NATURAL log(Chlorophyll) = -1.140 log(Secchi) + 3.936 Analysis of color and dissolved organic 2

CHLOROPHYLL CONCENTRATION CHLOROPHYLL -1.0 R = 0.537 carbon (DOC) concentrations in water n = 192, p < 0.0001 2 samples collected by USGS field teams in -2.0 -1.5 -0.5 0.5 1.5 2.5 3.5 1996–98 indicated that color in Massachusetts NATURAL LOGARITHM OF SECCHI DISK TRANSPARENCY lakes largely is due to DOC. The relation, shown in figure 5, was highly significant with 2 Figure 4. Relations between Secchi disk transparency and R = 0.914. Samples collected by volunteers phytoplankton-chlorophyll concentration in Massachusetts lakes in produced such variable results that no attempt (A) 1996; (B) 1997; and (C) 1998—Continued. was made to correlate them with TM data.

Table 5. Thematic mapper spectral bands and combinations of bands used as models to test for correlations with water-quality and trophic-state data for Massachusetts lakes

[TM, Thematic Mapper; TM1, TM band 1; TM2, TM band 2; TM3, TM band 3; TM4, TM band 4]

TM2 TM4– TM3 TM1() TM4 2 ------TM3 TM4+ TM3

TM1 TM2– TM1 TM2ln() TM1 ln ------TM2 TM2+ TM1

TM1 TM2– TM3 TM3ln() TM2 ln ------TM3 TM2+ TM3

TM2 TM4– TM3 TM4ln() TM3 ln ------ln ------TM3 TM4+ TM3

TM1+ TM2 TM2– TM1 ()TM1 2 ln()TM4 ------ln ------2 TM2+ TM1

TM1 TM1+ TM3 TM2– TM3 ()TM2 2 ------ln ------TM2 2 TM2+ TM3 TM1 TM2+ TM3 ()TM3 2 ------TM2– TM3 TM3 2

Thematic Mapper-Based Assessment of Water Quality and Trophic State 15 30 submerged macrophytes. These vegetation-cover class assignments can then be extended to any lake that is y = 0.0437 x + 1.7106 2 visible in the same TM scene. 25 R = 0.914 n = 47, p < 0.0001

20 Field-Mapping of Macrophyte Distributions

15 During 1996–98, distributions of floating, emer- gent, and submerged macrophytes were mapped in 44 10 Massachusetts lakes, ponds, and reservoirs by USGS and MADEM staff and by volunteers affiliated with the IN MILLIGRAMS PER LITER MassWWP. Twenty-four sets of maps, 12 produced in 1996 1997 1998 5 1996 and 12 produced in 1997 from 19 of the lakes, were used to develop the TM-based mapping proce-

DISSOLVED ORGANIC CARBON CONCENTRATION, DISSOLVED 0 dure. Excessive cloud cover during mid-to-late summer 0 100 200 300 400 500 600 1998 precluded the use of maps produced in that year. COLOR, IN PLATINUM-COBALT UNITS The 19 lakes (table 2) are primarily in the eastern half of Massachusetts and represent the range of lake types Figure 5. Relations between color and dissolved organic in that part of the State. Surface areas of the lakes carbon concentration in Massachusetts lakes. ranged from 7 to 233 ha with a median surface area of 29 ha. Maximum depths ranged from 2 to 16 m with a THEMATIC MAPPER-BASED median depth of 6 m. ASSESSMENT OF MACROPHYTE Field-mapping was conducted in late summer DISTRIBUTIONS after the macrophytes had reached their maximum den- sities but before they began to senesce in early autumn. A method was developed for mapping distribu- For each lake, a set of blank maps (field maps) was pro- tions of macrophytes in lakes, ponds, and reservoirs duced with a 1:24,000-scale (USGS Digital Line using TM images processed with a geographic infor- Graphs) outline of the lakeshore overlain by a lattice of mation system (GIS). The TM-based mapping proce- cells representing the 30-by-30-meter spatial resolution dure consists of manually mapping the distributions of of the TM images. These field maps were used by aquatic macrophyte beds in 10 to 15 representative observers to record the macrophyte distributions. lakes and relating the digitized field-generated maps to Aquatic macrophyte beds were identified and a set of TM images of the lakes. These relations are mapped separately as floating, emergent, or submerged then used to assign pixel-brightness values in the TM growth forms. Floating macrophytes, such as water images to one of four vegetation-cover classes: open lilies (Nuphar sp., Nymphaea sp.) and water shield water (no macrophytes), moderately covered (up to (Brasenia schreberi), commonly are found from the 50 percent) with floating or emergent macrophytes, shoreline inward to depths of between 1 and 3 m. densely covered (51–100 percent) with floating or They may or may not be rooted in the sediments. emergent macrophytes, and covered to any extent with Emergent macrophytes, such as cattails (Typha sp.),

16 Use of Thematic Mapper Imagery to Assess Water Quality, Trophic State, and Macrophyte Distributions in Massachusetts Lakes grasses (Phragmites sp.), rushes (Juncus sp.), sedges about 120 m apart, except in the largest lakes, where (Scirpus sp.), arrow arum (Peltandra virginica), and they were spaced about 200 m apart. Landmarks repre- pickerelweed (Pontederia cordata), typically are rooted sented on the maps were used as control points in locat- and have foliage that extends out of the water. Emer- ing the transects. Sampling points were then located at gent macrophytes generally are found along the edges intervals of 60 to 120 m along each transect, either by of lakes in shallow water rarely exceeding 1 m in direct measurement with a range finder or by estimat- depth. Submerged macrophytes, such as fanwort ing the distance and marking the position relative to the (Cabomba sp.), various pondweeds (Potamogeton sp., 30-by-30-meter cells printed on the map. At each sam- Najas sp.), coontail (Ceratophyllum sp.), watermilfoil pling point, a weighted two-sided rake, 0.46 m in (Myriophyllum sp.), and bladderwort (Utricularia sp.), length, was lowered on a line and dragged along the may occur from the shoreline across the entire lake lake bottom for a distance of about 2 m. The amount of bottom, but rarely extend beyond a depth of about plant material retrieved on the rake relative to that 10 m because of hydrostatic pressure and the limited retrieved in an area with visible submerged vegetation penetration of underwater light. was used to estimate the areal coverage of submerged Mapping of floating and emergent macrophytes macrophytes at that point. A submerged-vegetation consisted of moving slowly along the shoreline in a distribution map was then produced for each lake based boat and recording the locations of the macrophyte on the estimated areal coverages. beds on the field maps. The lattice of 30-by-30 m cells superimposed on the lakeshore outline provided a scale by which observers could judge distances from the Digitization and Processing of shore and accurately mark locations of the beds. The Field Maps maps also indicated the positions of major landmarks The hand-drawn field maps of macrophyte distri- such as roads, dams, and tributary streams, which butions were digitized by scoring the centroid of each provided additional reference points for mapping. 30-by-30-meter cell as one of the six ranges of cover Macrophyte density within the mapped beds was values, based on the mapped locations of the macro- estimated by the observers as (1) open water, (2) sparse phyte beds. The scores for each map were then used to (greater than 0 but less than 25 percent cover), (3) mod- populate the cells of a raster grid corresponding to the erate (greater than 25 percent but less than 50 percent lattice originally plotted on the map. The resulting cover), (4) dense (greater than 50 percent but less than grids were vectorized, clipped into the lake shoreline 75 percent cover), (5) very dense (greater than 75 per- boundaries, and merged into a single data layer for cent cover but less than 100 percent cover), or (6) com- each vegetation-cover type. The three data layers were plete (100 percent cover). Visual comparison of then merged into a single data layer, maintaining the duplicate maps prepared at the same time by indepen- cover values for each vegetation-cover type. dent observers for three lakes in 1998 indicated that Because the emergent vegetation was always these density ranges were large enough to subsume close to the lake shorelines, and because it represented minor differences or errors in the observers' density only a small part of the total covered area of most estimates. lakes, the cover values for floating and emergent Mapping of submerged macrophytes consisted vegetation types were combined into a single surface of establishing multiple transects extending from shore vegetation type. Also, the six original vegetation- to shore across the lakes. Transects usually were spaced cover classes were reduced to four summary classes:

Thematic Mapper-Based Assessment of Macrophyte Distributions 17 (1) open water, (2) 1–50 percent floating-and- boundaries were given a new attribute that differenti- emergent-vegetation cover, (3) 51–100 percent float- ated them from the smaller cells that intersected the ing-and-emergent-vegetation cover, and (4) submerged shorelines. Cells associated with islands in the lakes vegetation at all densities (when not hidden by surface were similarly differentiated. vegetation), when preliminary analyses indicated a potential bias in favor of open water. The result of Cells that did not intersect with lake shorelines combining vegetation-cover classes with small areal were grouped according to their NDVI values. For each distributions into larger summary cover classes was NDVI value, the total areas were determined for the to reduce the influence of the large areal extent of two surface vegetation-cover classes (1–50 percent and open water in many of the field maps on the final 51–100 percent floating and emergent) and for a hybrid assignments of the TM pixel-brightness values. vegetation-cover class consisting of open water and submerged vegetation. The vegetation-cover class Image Interpretation comprising the largest total area of the three was then assigned to that NDVI value. In this way, each NDVI Data in the TM scenes were processed into value in the dataset was associated with one of the ARC/INFO by creating raster grids for TM2, TM3, and three surface-vegetation cover classes or with open TM4 DNs. Grids for individual lakes were generated water. These associations were then used to assign from these three TM-scene raster grids and rectified to the lake grids. The individual lake grids were then vec- vegetations cover classes to cells that intersected the torized, clipped into the lakeshore boundaries, and lake shorelines. merged into a single data layer for each of the three TM The vegetation-cover class assignments for each bands. The three data layers were then merged into a cell were then examined to determine if any should be single data layer maintaining the DNs for each TM changed based on the NDVI values of adjacent cells. band. The effects of atmospheric haze were removed from the data for TM2 and TM3 by subtracting the If a given NDVI value predominated in the eight-cell smallest DNs for each wave band from all the bright- neighborhood surrounding a cell, then that NDVI value ness values for that wave band in the vector grid was added to the cell as an alternative value. Next, all (Wilkie and Finn, 1996). cells with that combination of NDVI value and alterna- For each 30-by-30-meter cell in the data layer, a tive value were selected and assigned the vegetation- normalized difference vegetation index (NDVI; Lille- cover class most frequently associated with the combi- sand and Kiefer, 1994) was calculated using the haze- nation. In this way, some inconsistent assignments aris- corrected DNs for TM3 and TM4 as follows: ing from the limited spatial resolution of the TM data TM4 – TM3 were removed. NDVI = ------TM4 + TM3 To determine areas of submerged vegetation, all This data layer was merged with the data layer cells that were not assigned a surface-vegetation cover containing the four vegetation-cover classes. Inconsis- class in the NDVI analysis were isolated, and a ratio tencies in alignment of the 30-by-30-meter cells in the index was calculated by dividing the haze-corrected two data layers were corrected by bringing the com- DNs for TM2 by those for TM3. The steps performed bined data layer back into raster grid mode and using coordinates for each lake derived from the original TM to assign NDVI values were then repeated on these iso- images to rectify the cells in the vegetation-cover class lated cells, the one difference being that the cells digi- data layer. The combined data layer was vectorized and tized as submerged vegetation were maintained and 30-by-30-meter cells falling entirely within lakeshore included as an option for assignment.

18 Use of Thematic Mapper Imagery to Assess Water Quality, Trophic State, and Macrophyte Distributions in Massachusetts Lakes Observed Versus Predicted of submerged vegetation cover area than were observed Macrophyte Distributions (fig. 8), although the RMSE values were similar to those exhibited by the 1996 relations (6.3 ha for open- Satellite images from July 22, 1996, and August water cover and 5.3 ha for submerged vegetation 26, 1997, were used, together with mapped distribu- cover). Lakes in the 1997 data set tended to have larger tions of 1996 and 1997 aquatic-vegetation cover in 19 observed open-water cover areas than those in the 1996 study lakes, to develop predictive models relating the data set (fig. 9). The median observed open-water cover satellite data to the observed macrophyte distributions. area was 11.9 ha in the 1997 data set and 8.2 ha in the Relations developed for the 1996 data were used to pre- 1996 data set. Similarly, observed cover areas for sub- dict distributions in the original 1996 lakes, and rela- merged vegetation were much smaller in the 1997 data tions developed for the 1997 data were used to predict set (median = 0.7 ha) than they were in the 1996 data distributions in the original 1997 lakes. Finally, the set (median = 5.8 ha). relations developed for the 1997 data were tested on the 1996 satellite scene and the predicted results compared Agreement between observed and predicted with observed 1996 macrophyte distributions. cover areas was better for the two floating-and- emergent-vegetation cover classes (the RMSE was 2.5 ha for the 1–50 percent floating-and-emergent 1996 Predictions Based on cover class and 1.1 ha for the 51–100 percent floating- 1996 Interpretations and-emergent cover class) than it was for the open Figure 6 shows the relations between observed water and submerged-vegetation cover classes (fig. 8). (field mapped in summer 1996) and predicted (inter- Because most of the observed areas for these classes preted from 1996 Thematic Mapper data) aquatic- were very small (1.0 to 6.0 ha), however, the errors are macrophyte cover areas in twelve 1996 study lakes significant. for each of the four vegetation-cover classes. For open water, the root-mean-squared-error (RMSE) of 1996 Predictions Based on the prediction was 3.6 ha for observed cover areas 1997 Interpretations ranging from 0 to 39.7 ha. Predicted open-water cover areas tended to be smaller than observed open- An attempt to predict vegetation-cover class water cover areas. This result can been observed in areas in the 1996 study lakes based on interpretations figure 7, which shows maps of observed and predicted developed from the 1997 data set was unsuccessful. aquatic-vegetation cover for Whitehall Reservoir, in Large areas of submerged or floating and emergent Hopkinton, Mass. Agreement between observed and vegetation were interpreted as open water for many predicted cover areas for the other three vegetation- lakes. Consequently, predicted cover areas for sub- cover classes was very good, with RMSE ranging from merged vegetation and the 1–50 percent floating-and- 1.3 ha for 51–100 percent floating-and-emergent-vege- emergent-vegetation class were smaller than the tation cover to 5.7 ha for submerged vegetation cover corresponding observed cover areas and had corre- (fig. 6). spondingly large RMSE values [27.7 and 11.4 ha, respectively (fig. 10)]. The only exceptions were the 1997 Predictions Based on areas predicted for the 51–100 percent floating-and- 1997 Interpretations emergent-vegetation cover class, which produced a RMSE of 1.1 ha over observed (1996) cover values The TM-based maps developed from the 1997 ranging from 0 to 26.8 ha. The large discrepancy in data set did not match the observed 1997 maps as predicted versus observed areas of submerged vegeta- closely as did those developed from the 1996 data set. tion can be seen in figure 11, which maps predicted and The TM-based mapping procedure predicted larger observed aquatic vegetation cover for East Brimfield amounts of open-water cover area and smaller amounts Reservoir in Brimfield and Sturbridge, Mass.

Thematic Mapper-Based Assessment of Macrophyte Distributions 19 Open water Submerged vegetation 50 150

RMSE = 3.6 ha RMSE = 5.7 ha 40 125 100 30 75 20 50

10 25

0 0 0 1020304050 0 25 50 75 100 125 150

1-50% Floating and 51-100% Floating and emergent vegetation emergent vegetation 60 40

50 RMSE = 3.4 ha RMSE = 1.3 ha 30 40

1996 PREDICTED COVER, IN HECTARES 30 20

20 10 10

0 0 0 102030405060 0 10203040

1996 OBSERVED COVER, IN HECTARES

Figure 6. Observed (field-mapped in summer 1996) and predicted (interpreted from July 1996 Thematic Mapper data) areal coverages of four aquatic macrophyte cover classes coverages in 12 Massachusetts lakes.

20 Use of Thematic Mapper Imagery to Assess Water Quality, Trophic State, and Macrophyte Distributions in Massachusetts Lakes aquatic 30 METER BY CELLS WHITEHALL RESERVOIR PREDICTED VEGETATION EXPLANATION 0 PERCENT VEGETATION (OPEN WATER) 1 TO 50 PERCENT SURFACE VEGETATION 51 TO 100 PERCENT SURFACE VEGETATION SUBMERGED VEGETATION

00'

˚

70

ATLANTIC OCEAN 30 METER BY CELLS

Sound

Bay

Nantucket

Massachusetts WHITEHALL RESERVOIR OBSERVED VEGETATION

00'

˚

71

30'

˚

41

00'

˚

42

00'

˚

72 50 MILES 50 KILOMETERS LOCATION MAP

00'

˚

73 0 0

30'

˚ Observed (field-mapped in summer 1996) and predicted (interpreted from July 1996 Thematic Mapper data) areal coverages of four

42 macrophyte cover classes coverages in Whitehall Reservoir, Hopkinton, Massachusetts. Figure 7.

Thematic Mapper-Based Assessment of Macrophyte Distributions 21 Open water Submerged vegetation 50 20

RMSE = 6.3 ha RMSE = 5.3 ha 40 15 30 10 20

5 10

0 0 0 1020304050 0 5 10 15 20

1-50% Floating and 51-100% Floating and emergent vegetation emergent vegetation 16 14 14 RMSE = 2.5 ha 12 RMSE = 1.1 ha 12 10 10 8 8 6 6 PREDICTED COVER, IN HECTARES 4 4 2 2 0 0 0246810121416 02468101214 OBSERVED COVER, IN HECTARES

Figure 8. Observed (field-mapped in summer 1997) and predicted (interpreted from August 1997 Thematic Mapper data) areal coverages of four aquatic macrophyte cover classes coverages in 12 Massachusetts lakes.

22 Use of Thematic Mapper Imagery to Assess Water Quality, Trophic State, and Macrophyte Distributions in Massachusetts Lakes ur aquatic 30 METER BY CELLS THOMPSON POND PREDICTED VEGETATION EXPLANATION 0 PERCENT VEGETATION (OPEN WATER) 1 TO 50 PERCENT SURFACE VEGETATION 51 TO 100 PERCENT SURFACE VEGETATION SUBMERGED VEGETATION

00'

˚

70

ATLANTIC OCEAN 30 METER BY CELLS

Sound

Bay

Nantucket

Massachusetts

00'

˚

71

30'

˚

41

00'

˚

42

00'

˚

72 50 MILES 50 KILOMETERS LOCATION MAP

00'

˚

73 0 0

30'

˚

42 Observed (field-mapped in summer 1997) and predicted (interpreted from August 1997 Thematic Mapper data) areal coverages of fo THOMPSON POND OBSERVED VEGETATION Figure 9. macrophyte cover classes coverages in Thompson Pond, Spencer, Massachusetts.

Thematic Mapper-Based Assessment of Macrophyte Distributions 23 Open water Submerged vegetation 150 150

125 RMSE = 39.7 ha 125 RMSE = 27.7 ha

100 100

75 75

50 50

25 25

0 0 0 25 50 75 100 125 150 0 25 50 75 100 125 150

1-50% Floating and 51-100% Floating and emergent vegetation emergent vegetation 60 30

50 RMSE = 11.4 ha 25 RMSE = 1.1 ha

40 20

PREDICTED COVER, IN HECTARES 30 15

20 10

10 5

0 0 0 102030405060 0 5 10 15 20 25 30 OBSERVED COVER, IN HECTARES

Figure 10. Observed (field-mapped in summer 1996) and predicted (interpreted from August 1997 Thematic Mapper data) areal coverages of four aquatic macrophyte cover classes coverages in 12 Massachusetts lakes.

24 Use of Thematic Mapper Imagery to Assess Water Quality, Trophic State, and Macrophyte Distributions in Massachusetts Lakes ur aquatic 30 METER BY CELLS PREDICTED VEGETATION EAST BRIMFIELD RESERVOIR EXPLANATION 0 PERCENT VEGETATION (OPEN WATER) 1 TO 50 PERCENT SURFACE VEGETATION 51 TO 100 PERCENT SURFACE VEGETATION SUBMERGED VEGETATION 30 METER BY CELLS

00'

˚

ATLANTIC OCEAN70

Sound

Bay

Nantucket

Massachusetts

00'

˚

71

30'

˚

41

00'

˚

42

00'

˚ OBSERVED VEGETATION

72 EAST BRIMFIELD RESERVOIR 50 MILES 50 KILOMETERS LOCATION MAP

00'

˚

73 Observed (field-mapped in summer 1996) and predicted (interpreted from August 1997 Thematic Mapper data) areal coverages of fo 0 0

30'

˚

42 Figure 11. macrophyte cover classes coverages in East Brimfield Reservoir, and Sturbridge, Massachusetts.

Thematic Mapper-Based Assessment of Macrophyte Distributions 25 SUMMARY AND color and DOC concentrations in water samples col- CONCLUSIONS lected by USGS field teams in 1996–98 indicated that most of the color in Massachusetts lakes is due to During the spring and summer of 1996, 1997, DOC. and 1998, measurements of phytoplankton- chlorophyll Areas of open-water, submerged vegetation, and concentration, Secchi disk transparency, and color two surface-vegetation-cover classes predicted from were made at 97 Massachusetts lakes within 24 hours TM images acquired in the summer of 1996 closely of Landsat Thematic Mapper (TM) imaging of the matched the areas observed in a set of field observa- lakes in an effort to use the TM imagery to assess tions. Agreement between observed and predicted 1996 lake-water quality. Spatial distribution of floating, submerged-macrophyte cover areas was at least as emergent, and submerged macrophytes were mapped good as the 56- to 70-percent accuracy reported for in 49 of the lakes at least once during the three-year mapping studies using visual interpretations of aerial period. The maps were digitized and used to assign photographs (Schloesser and others, 1987); however, pixels in the TM images to one of four vegetation the same analysis applied to a set of data acquired in cover classes—open water, 1–50 percent floating-and- the summer of 1997 resulted in somewhat less reliable emergent-vegetation cover, 51–100 percent floating- predictions, and an attempt to predict 1996 vegetation and-emergent-vegetation cover, and submerged vegeta- cover areas using the relations developed in the 1997 tion at any density. Concurrent data collection and sam- analysis was unsuccessful. pling by USGS and trained volunteer field teams Differences in the predictive power of the two resulted in similar chlorophyll determinations, Secchi data sets appear to stem from differences in the relative disk readings, and temperature measurements, but sizes of the vegetation-cover areas used in the initial color determinations were highly variable, possibly calibration of the NDVI values. The ranges of observed due to contamination of sample bottles issued to the areas of the four vegetation-cover classes were similar volunteers. in the 1996 data set. By contrast, open water predomi- Attempts to develop predictive relations between nated in lakes forming the 1997 data set, and the other phytoplankton-chlorophyll concentration, Secchi disk vegetation-cover classes had much smaller and more transparency, lake color, dissolved oxygen concentra- variable ranges. Both the field-mapping and the TM- tion (DOC), and various combinations of TM band 1, imaging processes are subject to error. Locations of the 2, 3, and 4 digital numbers (DNs) were unsuccessful. macrophyte beds indicated on the field maps cannot be The poor relations were primarily the result of the exact, and the TM images are limited by the 30-by-30- extremely low chlorophyll concentrations (median = meter ground resolution of the instrument. Under these 3.1 µg/L) in the lakes studied, and also because of conditions, a preponderance of one type of vegetation the highly variable DOC concentrations as indicated cover class in the calibration data set is likely to result by color values ranging from less than 1 to 547 in more assignments of NDVI values to that cover platinum-cobalt units (PCU). class, simply because locational errors involving that Predictive relations were developed between cover class will tend to occur more frequently. It is also Secchi disk transparency and phytoplankton- possible that the failure of the method to accurately chlorophyll concentration and between color and predict the 1996 macrophyte distributions based on DOC concentration. Phytoplankton-chlorophyll con- interpretations of 1997 data was due in part to this centration was inversely correlated with Secchi disk problem. transparency during all three sampling periods. The Ideally, the method should be applied to a set of regression equations for the relations in 1996 and 1997 mapped lakes and then tested on a second set not used were similar and showed that 62 to 67 percent of the in the initial calibration. This was not possible given variability in chlorophyll concentration could be the limited number of field maps available in the two explained by the Secchi disk transparency. Analysis of data sets and the need for equal areal representation of

26 Use of Thematic Mapper Imagery to Assess Water Quality, Trophic State, and Macrophyte Distributions in Massachusetts Lakes vegetation-cover classes. By careful selection of the Lillesand, T.M., Johnson, W.L., Deuell, R.L., Lindstrom, initial set of lakes to ensure adequate representation of O.M., and Meisner, D.E., 1983, Use of Landsat vegetation cover, it may be possible to use fewer lakes data to predict trophic state of Minnesota lakes: in the calibration process without sacrificing predictive Photogrammetric Engineering and Remote Sensing, power. The calibration data set also could be improved v. 49, no. 2, p. 219–229. by using a global-positioning system to accurately Lillesand, T.M., and Kiefer, R.W., 1994, Remote sensing and locate and map the aquatic macrophyte beds. image interpretation (3rd ed.): New York, John Wiley and Sons, Inc., 750 p. Massachusetts Water Resources Commission, 1994, Policy REFERENCES CITED on lake and pond management: Boston, Massachusetts Water Resources Commission, Division of Water American Public Health Association, American Water Works Resources, 5 p. Association, and Water Pollution Control Commission, Raitala, J.T., 1986, Satellite data in aquatic research—Some 1995, Standard methods for the examination of water ideas for future studies: Symposium on Remote Sensing and wastewater (19th ed.): Washington, D.C., APHA, for Resources Development and Environmental [variously paginated]. Management, Enschede, August, 1986, p. 755–758. Baban, S.M.J., 1993, Detecting water quality parameters in Scarpace, F.L., Holmquist, K.W., and Fisher, L.T., 1979, the Norfolk Broads, U.K., using Landsat imagery: Landsat analysis of lake quality: Photogrammetric International Journal of Remote Sensing, vol. 14, no. 7, Engineering and Remote Sensing, v. 45, no. 5, p. 623– p. 1247–1267. 633. _____1997, Environmental monitoring of estuaries; Schloesser, D.W., Manny, B.A., Brown, C.L., and Jaworski, estimating and mapping various environmental E., 1987, Use of low-altitude aerial photography to indicators in Breydon Water Estuary, U.K., using identify submersed macrophytes, in Color Aerial Landsat TM imagery: Estuarine, Coastal, and Shelf Photography in the Plant Sciences and Related Fields: Science, vol. 44, p. 589–598. Ann Arbor, University of Michigan, Proceedings of the Canfield, D.E., Jr., Langland, K.A., Maceina, M.J., Haller, 10th Biennial Workshop, 1984, p. 19–28. W.T., and Shireman, J.V., 1983, Trophic state Shimoda, H., Etaya, M., Sakata, T., Goda, L., and Stelczer, classification of lakes with aquatic macrophytes: K., 1986, Water quality monitoring of Lake Balaton Canadian Journal of Fisheries and Aquatic Sciences, using LANDSAT MSS data: Symposium on Remote v. 40, p. 1713–1718. Sensing for Resources Development and Environmental Fishman, M.J., and Friedman, L.C., eds., 1989, Methods for Management, Enschede, August, 1986, p. 765–770. determination of inorganic substances in water and U.S. Environmental Protection Agency, 1995, The state of fluvial sediments: U.S. Geological Survey Techniques the New England environment, 1970–1995: Boston, of Water-Resources Investigations, book 5, chap. A1, Mass., U.S. Environmental Protection Agency–New 545 p. England, Office of External Programs, 20 p. Goldman, C.R., and Horne, A.J., 1983, Limnology: New Verdin, J.P., 1985, Monitoring water quality conditions York, McGraw-Hill Book Co., 464 p. in a large western reservoir with Landsat imagery: Khorram, Siamak, Cheshire, Heather, Geraci, Alberto L., Photogrammetric Engineering and Remote Sensing, and LaRosa, Guido, 1991, Water quality mapping of v. 51, p. 343–353. Agusta Bay, Italy, from Landsat-TM data: International Wilkie, D.S., and Finn, J.T., 1996, Remote sensing imagery Journal of Remote Sensing, vol. 12, no. 4, p. 803–808. for natural resources monitoring—A guide for first-time Lathrop, R.G., Jr., and Lillesand, T.M., 1986, Use of users: New York, Columbia University Press, 295 p. Thematic Mapper data to assess water quality in Green Witzig, A.S., and Whitehurst, C.A., 1981, Current use and Bay and central Lake Michigan: Photogrammetric technology of LANDSAT MSS data for lake trophic Engineering and Remote Sensing, vol. 52, no. 5, classification: Water Resources Bulletin, vol. 17, no. 6, p. 671–680. p. 962–970.

References Cited 27 Tables 6 and 7 ature Temper- bration of Landsat-5 bration of Landsat-5 DOC Color Secchi phyll Chloro- concentrations; DOC, dissolved organic carbon organic concentrations; DOC, dissolved a ironmental Management; S.F., State Forest; S.P., State Park; State Park; S.P., State Forest; ironmental Management; S.F., and phaeophytin- Sampled in: of measurements number Total a 1996 1997 1998 Chlorophyll, sum of chlorophyll- Chlorophyll, Town/city Drainage basin Total number of measurements: number of measurements: Total Lakes Sampled by Massachusetts Water Watch Partnership Volunteers Partnership Watch Water Sampled by Massachusetts Lakes code PALIS PALIS cation System. fi station Sampling Continued — PALIS, Pond and Lake Identi Pond and Lake PALIS, Lake name Massachusetts lakes sampled in 1996, 1997, and 1998, numbers of measurements water-quality characteristics made for cali Massachusetts lakes sampled in 1996, 1997, and 1998, numbers of measurements water-quality characteristics made for cali Bearse PondBillington SeaBillington SeaBillington SeaBoon LakeBoon Lake 16 1Boon Lake 3 ReservoirBuffumville 6Carding Mill PondChabacco Lake 96012 Barnstable 1 LBA1 94007Chandler Pond Plymouth 94007 LBA2 PondDudley Plymouth 94007 1 LBA3 Reservoir Meadow Fort 82011 Plymouth Stow/Hudson Pond) (Wales Lake George 82011 42005 1Goose Pond Stow/Hudson Oxford/Charlton 1 1 82011 Stow/Hudson 1Heard Pond Cape Cod 82015 SudburyHigh St. Impoundment South Shore Coastal 1 South Shore CoastalHorn Pond 93014 Assabet (Concord) Hamilton/Essex South Shore CoastalJenkins Pond X French 41016 82042 1 Wales Marlborough/Hudson Assabet (Concord) 72017 XJenkins Pond Boston 1 X Assabet (Concord) X 82029Long Pond Wayland 1 X X Assabet (Concord)Long Pond X North Shore CoastalLong Pond (Concord) Sudbury X -- X 1Long Pond 1 21043 X X Lee/Tyringham XLong Pond 2 Maynard/Acton X 82058 1 6 Wayland X 1 X Quinebaug X X Charles 8 1 1 -- 6 (Concord) Sudbury 71019 X -- 2 96155 -- Woburn Falmouth -- 3 96155 Housatonic -- -- Assabet (Concord) 2 Falmouth 2 X 4 3 2 5 5 62108 X 3 -- 3 -- 2 1 Freetown/Lakeville 2 X 3 62108 (Concord) Sudbury -- 3 Freetown/Lakeville 3 62108 -- -- X 3 -- -- Freetown/Lakeville 3 3 62108 X -- 3 Freetown/Lakeville X Cape Cod Taunton Mystic 62108 5 -- 3 3 Freetown/Lakeville X -- Cape Cod 2 Taunton 7 X X 3 3 Taunton 2 1 X 1 X Taunton 3 -- 3 -- 5 -- Taunton 5 2 10 -- -- 10 3 3 X 3 -- 4 ------X X 4 5 X 4 4 X 12 X X 1 5 X X 6 12 4 8 X 9 4 1 5 5 1 6 11 5 -- -- 7 -- 5 -- -- 5 8 4 -- 5 5 1 -- 4 10 -- 4 3 -- 1 5 -- 10 3 -- 4 6 -- 1 6 -- 6 6 6 PALIS Code: PALIS Table 6. Thematic Mapper imagery concentration; Secchi, Secchi disk transparency; Temperature, surface-water temperature. MADEM, Massachusetts Department of Env surface-water Temperature, concentration; Secchi, Secchi disk transparency; --, not measured or unknown] USGS, U.S. Geological Survey; S.R., State Reservation; Table 6. [ Thematic Mapper imagery

Table 6 31 ature Temper- bration of Landsat-5 DOC Color Secchi phyll Chloro- Continued Sampled in: of measurements number Total 1996 1997 1998 Town/city Drainage basin eld Housatoniceld Housatoniceld Housatoniceld/Lanesboro Housatonic X X X X X X X X 12 X X 5 -- -- 11 11 8 -- 4 -- 18 11 9 6 15 18 8 7 14 7 fi fi fi fi code PALIS PALIS Lakes Sampled by Massachusetts Water Watch Partnership Volunteers— Partnership Watch Water Sampled by Massachusetts Lakes station Sampling Continued — Lake name Massachusetts lakes sampled in 1996, 1997, and 1998, numbers of measurements water-quality characteristics made for cali Onota LakeOnota LakePontoosuc Lake PondPuffer Singletary LakeSingletary Lake 2 PondSnake 1 3Spectacle Pond (Wilbraham) BeachSpectacle Pond (Wilbraham) 1 1 MiddleSpectacle Pond (Sandwich) 21078 2 Pitts PondSpy 1 21083 36142 21078 Pitts Wilbraham 36142 Pitts PondSpy Wilbraham 1Stearns Mill Pond 51152 82092 Sutton/Milbury Sudbury/Maynard Sugden Reservoir 51152 Sutton/Milbury Thompson Pond 96307 Sandwich Thompson Pond 96302 1 South Sandwich LakeUpper Naukeag Chicopee (Concord) Sudbury Chicopee Blackstone 1 LakeWaban North Blackstone LakeWallum 1 X 71040 PondWarners 1 Arlington 6 82104 X LakeWatatic 71040 Cape Cod Sudbury Arlington 36150 X X Spencer X X Cape Cod 36155 2 Spencer 35090 36155 1 Ashburnham Spencer 14 1 X Mystic 1 -- X Mystic 72125 (Concord) Sudbury Wellesly X 51172 14 2 Douglas Chicopee 82110 1 Concord 2 Millers X X X 2 35095 Chicopee 12 -- Ashburnham/Ashby -- Chicopee X X -- -- 6 12 2 -- X -- Millers X 7 2 -- Charles 2 -- X Blackstone X 14 X X 1 Assabet (Concord) -- 2 2 X X -- X X -- 3 4 X 3 X X 1 2 X 15 2 4 X X X 12 -- 7 16 X 8 10 6 X -- X -- 3 -- X -- X 7 X -- 12 11 4 10 5 5 -- 19 13 9 10 13 4 19 -- 1 -- 1 -- -- 11 13 8 21 5 14 11 1 8 11 15 5 13 11 1 10 13 5 11 Long Pond Lake Naukeag Lower Maspenock Lake Maspenock Lake Merino Lake 1Metacomet Lake Deep end Deep holeMiddle Pond BasinMystic Lake 62108 51112 35041 Freetown/Lakeville Onota Lake Hopkinton/Milford Ashburnham 1 Deep hole 51112 Hopkinton/Milford 34051 Taunton Blackstone Belchertown 1 1 42036 Blackstone Millers Dudley 1 96198 Barnstable Connecticut 96218 Barnstable 21078 X Pitts X X X French X X X Cape Cod 11 X X Barnstable -- -- X 11 -- 10 10 -- 11 -- -- 9 9 -- X X 8 1 11 8 X 9 10 X 1 11 13 8 6 10 11 2 ------9 4 3 7 8 3 7 8 4 Thematic Mapper imagery Table 6.

32 Use of Thematic Mapper Imagery to Assess Water Quality, Trophic State, and Macrophyte Distributions in Massachusetts Lakes ature Temper- bration of Landsat-5 DOC Color Secchi phyll Chloro- Continued Sampled in: of measurements number Total 1996 1997 1998 eldeld X X 1 1 -- -- 1 1 1 1 1 1 fi fi Merrimack X 2 -- 1 1 1 Lakes Sampled by USGS and MADEM Staff Lakes Town/city Drainage basin eld S.F.eld S.F. Housatonic eld S.F. Housatonic X Kinderhook X 1 X 1 -- -- 1 1 1 1 -- 1 1 1 1 1 1 fi fi fi S.F. code PALIS PALIS Lakes Sampled by Massachusetts Water Watch Partnership Volunteers— Partnership Watch Water Sampled by Massachusetts Lakes station Sampling Continued — Lake name Massachusetts lakes sampled in 1996, 1997, and 1998, numbers of measurements water-quality characteristics made for cali Buckley-Dunton LakeBuckley-Dunton Deep hole 32013 October Mountian S.F. West Ashmere LakeAsnacomet Pond QuarryBabson Farm Benedict PondBerry Pond 2Big (Benton) Pond 1 1Billington SeaBillington Sea Deep holeBillington Sea 21005 1 36005 --Box Pond Pitts 21011 Hubbardston Deep hole S.F. Beartown lakeBuckley-Dunton 1 Point S.P. Halibut 12002 Pitts 3 31004 1 Otis S.F. 6 Chicopee Housatonic North Shore Coastal 94007 Plymouth 1 94007 X Plymouth 32013 94007 October Mt. S.F. Plymouth 72008 Farmington X Bellingham/Mendon X South Shore Coastal West South Shore Coastal Charles -- X X South Shore Coastal X -- X 2 1 1 -- 1 1 1 1 1 1 1 1 1 1 1 X -- 1 1 1 1 -- 1 1 1 1 1 1 1 -- 1 1 1 1 1 1 1 1 1 Webster LakeWebster LakeWebster LakeWequaquet LakeWequaquet White Pond (Concord) 1White Pond (Hudson/Stow) 2Whitins Reservoir 14 1 1Whitins Reservoir 15 PondWillet lakeWinnekeag 42064 Webster 96333 42064 1 82119 PondWinter Barnstable 82118 Webster 96333 Hudson/Stow Concord 2 LakeWinthrop Barnstable LakeWyola 1 1 51179 Douglas Althea Lake 51179 Douglas Assabet (Concord) 1 Cape Cod French Deep holeAshland Reservoir Cape Cod French (Concord) Sudbury 81157 72140Ashmere Lake 73062 Center X Ashburnham Holliston Walpole/Norwood/Westwood Neponset X X 71047 34103 Winchester 1 1 X Shutesbury Blackstone X X Blackstone X X 1 X Nashua X 10 X Charles X X 82003 84001 X 11 Tyngsboro Dracut, Ashland S.P. Lowel, -- X X X X Mystic X Connecticut 21005 -- X X 11 19 9 Pitts 18 7 12 X -- -- 7 X 3 -- -- X (Concord) Sudbury 12 X X 19 15 2 X -- 10 18 X 2 X X 12 12 -- 16 X 19 X X 14 3 -- 7 13 10 X 10 9 15 4 11 3 12 -- 12 7 10 -- -- 10 10 3 -- 9 1 10 5 11 11 10 2 12 5 11 10 1 10 6 9 1 1 Table 6. Thematic Mapper imagery

Table 6 33 ature Temper- bration of Landsat-5 DOC Color Secchi phyll Chloro- Sampled in: of measurements number Total 1996 1997 1998 Continued eld X 1 -- 1 1 1 eldeld X X 1 1 -- --eld 1 1 1 1 1 -- X 1 ------1 fi fi fi fi Town/city Drainage basin Lakes Sampled by USGS and MADEM Staff— Lakes code PALIS PALIS station Sampling Continued — eld Reservoir 1 41014Area Streeter Point Rec. Quinebaug X 1 1 1 1 1 eld Reservoir 2 41014Area Streeter Point Rec. Quinebaug X 1 1 1 1 1 eld Reservoir 3 41014Area Streeter Point Rec. Quinebaug X 1 1 1 1 1 fi fi fi Lake name Massachusetts lakes sampled in 1996, 1997, and 1998, numbers of measurements water-quality characteristics made for cali Buckley-Dunton LakeBuckley-Dunton Cedar Pond PondCharge Inlet PondCliff PondCollege PondCurlew 32013Denison Lake October Mountian S.F. 1East Brim 1 West 1 1 1 41008 95025 Sturbridge 1 Myles Standish S.F. 96039 95030 S.P. Nickerson Myles Standish S.F. 95034 Buzzards Bay Myles Standish S.F. 35017 S.F. Otter River Buzzards Bay Quinebaug Cape Cod Buzzards Bay X Millers X X X 1 X X 1 -- 1 1 1 1 1 1 -- 1 -- 1 1 1 1 1 1 1 4 1 1 1 1 1 1 1 1 1 1 East Brim East Brim East Lake Waushacum East Lake Fearing PondFlax Pond 1 PondGreenwater Hallockville Pond 1 81153 1 Sterling 1 1 95054 Myles Standish S.F. 21044 October Mt. S.F. 33009 96091 S.F. Trail Mohawk S.P. Nickerson Nashua South Shore Coastal Deer Housatonic X Cape Cod X 1 X -- X 1 -- 1 ------1 1 -- 1 1 1 -- 1 -- -- 1 1 Hallockville PondHopkinton ReservoirJamaica PondLaurel Lake PondLittle Cliff 2 1Long Pond (Rutland)Long PondLong Pond 1 33009 82061Long Pond 1 1 S.F.Trail Mohawk 1 Hopkinton S.P.Long PondLong Pond 72052Long Pond Boston Deer 1 96170 36082 Lake Naukeag Lower 35035 S.P. Nickerson Rutland S.P. Erving S.F. 2 (Concord) Sudbury Mausert's Pond 3Mcleod Pond 4 X 1 62108 5 Freetown/Lakeville 62108 Cape Cod Deep hole Freetown/Lakeville 62108 1 Chicopee Charles Freetown/Lakeville 62108 Millers 62108 35041 Freetown/Lakeville Outlet Freetown/Lakeville Ashburnham Taunton 62108 Taunton Freetown/Lakeville Taunton 11009 33012 1 Taunton S.P. Clarksburg Catamount S.F. Taunton X -- X Taunton X Millers 1 X Hoosic Deer X X X 1 1 X X 1 X 1 -- 1 1 X 2 1 X X 2 -- 1 2 1 1 X 1 -- 1 1 1 2 1 1 1 -- 1 1 1 1 1 1 1 1 -- -- 1 1 1 -- 1 1 -- 1 1 1 1 1 -- -- 1 -- 1 1 1 1 1 1 1 1 1 Table 6. Thematic Mapper imagery

34 Use of Thematic Mapper Imagery to Assess Water Quality, Trophic State, and Macrophyte Distributions in Massachusetts Lakes ature Temper- bration of Landsat-5 DOC Color Secchi phyll Chloro- Sampled in: of measurements number Total 1996 1997 1998 Continued eldeld X X 1 1 -- -- 1 1 1 1 1 1 eld X 1 ------eldeld X 1 X -- 1 1 1 -- 1 ------fi fi fi fi fi Town/city Drainage basin eld Deer eld Housatoniceld Housatonic X X X X 1 1 1 1 1 1 1 1 1 1 eld/Lanesboro Housatonic X X X 1 -- 1 1 1 fi fi fi fi Lakes Sampled by USGS and MADEM Staff— Lakes code 31027 Otis S.F. Farmington X ------1 31027 Otis S.F. Farmington X ------1 PALIS PALIS inlet) inlet) station Sampling Continued — Lake name Massachusetts lakes sampled in 1996, 1997, and 1998, numbers of measurements water-quality characteristics made for cali eld Pond 1 33017 Plain fi Onota LakeOtis ReservoirOtis ReservoirOtis ReservoirOtis ReservoirOtis Reservoir 3 1 2 3 4 (outlet) 21078 31027 5 (north Pitts Otis S.F. 31027 31027 Otis S.F. Otis S.F. 31027 Otis S.F. Farmington Farmington Farmington Farmington X X X X ------1 -- 1 1 1 1 1 1 1 1 1 1 1 North PondOctober Mountain Reservoir Deep holeOnota Lake -- October Mountian S.F. 1 Housatonic 1 33014 Mountain S.F. Savoy 21078 Pitts Deer X 1 -- 1 1 1 Otis Reservoir 6 (south Rico LakeRico Lake PondRocky Sheomet LakeSouth Pond 1 2 1 1 1 -- -- 95119 35074 Massasoit S.P. Myles Standish S.F. S.F. Warwick Massasoit S.P. 33019 Mountain S.F. Savoy South Shore Coastal Taunton X Millers Deer Taunton X 1 X X -- X 1 1 2 1 1 ------1 2 -- -- 2 -- 1 2 1 Pontoosuc Lake 1Stearns Mill PondThompson PondThompson Pond 21083 PondTurner's Pitts 1 1 2 82104 1 Sudbury 36155 Rutland S.P. 36155 Rutland S.P. 95151 S.R. Acushnet Cedar Swamp South Shore Coastal (Concord) Sudbury X Chicopee Chicopee X X X X X 1 1 1 1 1 1 1 1 1 -- 1 -- 1 1 1 1 1 1 1 1 Pequot PondPlain 2 32055 Hampton Ponds S.P. West Pentucket PondPentucket PondPentucket Pequot Pond Deep hole Inlet 91010 Georgetown 1 91010 Georgetown 32055 Hampton Ponds S.P. Parker Parker West X X 1 -- 1 1 -- 1 1 -- 1 -- Table 6. Thematic Mapper imagery

Table 6 35 ature Temper- bration of Landsat-5 DOC Color Secchi phyll Chloro- Sampled in: of measurements number Total 1996 1997 1998 Continued Town/city Drainage basin eld S.F. Farmington X 1 -- 1 1 1 fi Lakes Sampled by USGS and MADEM Staff— Lakes code PALIS PALIS station Sampling Continued — Lake name Massachusetts lakes sampled in 1996, 1997, and 1998, numbers of measurements water-quality characteristics made for cali Turner's PondTurner's PondTurner's LakeUpper Naukeag Upper Spectacle Pond PondWalden 1 Pond 2Walker 1 Pond 3Walker LakeWallum LakeWatatic 35090 PondWatson 95151 Deep hole Ashburnham 31044 S.R. Acushnet Cedar Swamp 95151 Otis S.F. South Shore Coastal LakeWebster 82109 1 S.R. Acushnet Cedar Swamp Pond S.R. Walden South Shore Coastal LakeWebster 3 XWhite Pond (Concord) 1 XWhite Pond (Hudson/Stow) 1Whitins Reservoir 6 1 41052 Millers Concord 1 S.P. Wells 41052 1Whitins Reservoir Sturbridge 51172 2Whalom Pond Farmington Douglas S.F. 35095Whitehall Pond Ashburnham/Ashby 1 62205 82119Whitehall Reservoir 82118 Taunton 1 Hudson/Stow Whitehall Reservoir Concord 42064 2 X 1 Webster 42064 --Whitehall Reservoir X Millers Webster Quinebaug 1 PondWillet 1 2 Quinebaug 51179 Blackstone X 1 lakeWinnekeag Douglas 2 1 51179 LakeYork X Assabet (Concord) 1 Douglas 3 1 Taunton 81154 X (Concord) 82120 Sudbury X 36173 Lunenburg/Leominster 1 Hopkinton S.P. French 1 X 82120 Rutland S.P. 1 1 Hopkinton S.P. French 1 X 1 1 X 82120 1 Nashua X -- Hopkinton S.P. 1 Blackstone X X -- 1 X Blackstone 1 X -- (Concord) Sudbury X 81157 73062 Ashburnham 1 (Concord) Sudbury 1 Walpole/Norwood/Westwood Chicopee X Neponset 1 -- 1 1 X X (Concord) Sudbury 1 1 X X 31052 1 1 X 1 Stans -- X -- X X X -- X -- X 1 1 1 1 X 1 1 X Nashua 1 X 1 1 1 1 1 X 1 X 1 1 X 1 1 -- 1 1 1 1 1 X 1 1 1 1 1 1 1 1 1 -- 1 1 1 1 1 -- -- 1 1 1 1 1 1 1 -- X 1 1 1 1 1 1 1 1 X 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 -- 1 1 1 1 1 1 Thematic Mapper imagery Table 6.

36 Use of Thematic Mapper Imagery to Assess Water Quality, Trophic State, and Macrophyte Distributions in Massachusetts Lakes al numbers corresponding al numbers corresponding Haze-corrected TM digital numbers Haze-corrected TM Band 1 TM Band 2 TM Band 3 TM Band 4 Lake color Secchi Secchi parency disk trans- tion phyll Chloro- concentra- ″

′

° July 22, 1996 July June 23, 1997 June Longitude In degrees, minutes, and seconds. PCU, platinum-cobalt units; TM, Thematic Mapper; m, meters; mg/L, micrograms TM, minutes, and seconds. PCU, platinum-cobalt units; In degrees, ″

′

° Latitude Station identifier Latitude and longitude: Continued — cation System. fi PALIS; Pond and Lake Identi Pond and Lake PALIS; Lake-water-quality data collected within 24 hours of Landsat-5 Thematic Mapper imaging the lakes, and haze-corrected digit Lake-water-quality data collected within 24 hours of Landsat-5 Thematic Mapper imaging the lakes, and haze-corrected digit Station name code PALIS Ashland Reservoir PondCollege PondDudley Pond) (Wales Lake George Heard PondHopkinton Reservoir 82003 41016Horn PondJamaica Pond 95030 1Singletary Lake 1 82029Thompson Pond 82061 1 PondTurner's 82058 PondTurner's 1 42 14 PondTurner's 42 03 31 1 71 27 41 PondWalden 72 13 22 71019 1 72052 51152 LakeWallum 41 52 03 36155 42 19 44 4.9 70 39 46 LakeWallum -- 1 1 42 15 20Whitehall Reservoir 71 22 23 1 95151Whitehall Reservoir 71 30 50 2 42 21 13 3.4 2.2 95151Whitehall Reservoir 3.0 71 23 01 -- 95151 LakeWinthrop 1 1.8 82109 42 28 13 3.2 42 19 04 2 42 09 40 85 51172 -- 82120 71 09 30 71 07 12 3 42 18 30 1.5 71 46 50 --Horn Pond 51172 82120 3.0 1 71 58 10 Lake Naukeag Lower 82120 24 1A 5.1 41 40 46 5.6 Maspenock Lake 1.3 1 -- 135 .9 41 40 53 7.1 70 58 40 Maspenock Lake 1B 2 7.4 68 41 05 72140 70 58 32Merino Lake 3 7.1 2.2 3.6 5.6 42 26 20 70 58 26 18.1 42 00 57 176 4.6 3.9 5.6 35041 1.3 71 20 31 42 12 60 5.1 5.7 71 46 12 1 42 00 57 6.6 42 13 53 5.2 7.6 71 34 46 13.6 11.1 51112 -- 71019 71 46 12 9 42 14 03 .4 1 71 34 06 6.2 234 1.3 -- 51112 2.4 5.6 .5 71 34 44 5.2 14.2 4.5 11.1 .4 2.2 10.1 9 Deep end 8.6 1 42 11 23 3.1 14.1 445 6.2 42036 4.2 Basin 8.1 9 71 25 26 4.4 383 4.2 42 12 10 2.9 42 39 52 10 22 10.6 4.9 547 11.2 7.6 71 33 29 2.1 71 57 38 6.6 8.1 1A 62 -- 42 11 29 5.6 12.1 2.4 42 28 13 7 8 11.2 71 33 20 74 18 7.1 1.9 71 09 30 15 7.2 103 2.9 22.1 8.2 5.6 10.6 121 5.2 10.1 2.7 42 03 01 11 2.9 -- 4.1 5.4 10.1 5.6 71 54 02 1.9 11 6.1 13.6 10.2 13 5.2 9.1 6.6 9 -- -- 1.9 2.6 7.9 6.6 5.2 13.2 3.6 30 25 20 12 4.6 6.2 4.1 2.8 34 4.2 33 6.2 23.2 12 22 17.2 4.2 4.2 9 3.6 28.2 185 19.2 11.0 10.0 8 9 8 8 17.2 18.0 10.0 5.2 17.1 13.1 11.0 24.1 15.1 12 11 9 14.1 16 11 11 PALIS code: PALIS Table 7. to the station locations in images to the station locations in images Table 7. [ per liter]

Table 7 37 al numbers corresponding Haze-corrected TM digital numbers Haze-corrected TM Band 1 TM Band 2 TM Band 3 TM Band 4 Lake color Secchi Secchi parency disk trans- tion phyll Chloro- Continued concentra- ″

′

° August 26, 1997 August Longitude June 23, 1997— June ″

′

° Latitude Station identifier Continued — Lake-water-quality data collected within 24 hours of Landsat-5 Thematic Mapper imaging the lakes, and haze-corrected digit Station name code PALIS Merino LakeMiddle Pond PondPentucket PondSpy PondSpy Thompson Pond LakeWaban 42036 PondWalden 96198 LakeWallum 91010 PondWarners 1B PondWarners 1 Deep hole 71040 36155 LakeWebster 71040 42 43 54 LakeWebster 42 03 01 72125 LakeWebster 70 59 36 1A 1 82109 71 54 02White Pond (Concord) 1B 41 40 18 51172Whitins Reservoir 82110 70 24 59 7.9 2 Lake Deep holeWinnekeag 5.6 42 24 32 82110 1 PondWinter 42 17 47 42 24 32 42 26 20 42064 71 09 1.6 1A 82118 .5 71 58 38 42064 71 09 71 20 31 2.8 1B 42 17 14 42064 14.2Bearse Pond 1A 5.7 71 18 35 1 51179Boon Lake 72 3.0 5.1 42 00 57 1.2 1B 42 47 52 29Boon Lake 81157 1.7 2 71 46 12 71 24 08 ReservoirBuffumville 42 47 52 2.2 25.2 1 2.1 -- 1.7Chandler Pond -- 42 03 14 71 24 08 17.2 71047 10.5 1 42 25 40 1.2 42 03 14 216 71 50 56Heard Pond 3.8 14.0 71 23 28 71 50 56Long Pond 279 35.2 216 42 02 15 7.1 11.0 1 42005 96012Long Pond 1.7 5.2 24.2 -- 42 04 20 2.4 71 50 Long Pond 20.1 137 3.0 82011 1.6 15.2 71 46 42 20.0 24.2 42 39 48 1.7 14.1Long Pond 82011 1 16 16.0 18.2 308 71 53 55 1.9 4.0 72017 22 18.2 17 4.2 10.0 4.0 42 27 22 27.1 16.0 LBA1 .9 11 80 LBA2 20.1 71 09 15 11.0 22.2 2.8 3.4 82058 18.2 1 10 11.0 41 40 35 14.1 20.1 62108 42 24 13 42 56 47 20 10 4.4 15 22.2 70 20 00 12.1 11.0 4.0 62108 42 23 47 13 71 30 08 2.3 71 18 17 11.0 1 20.2 14.1 62108 71 29 39 6 13 13 16.2 1A 20.2 62108 11.0 21 42 20 41 6.4 18.1 2.2 5.1 10 2A 3.3 13.1 12.0 45 10.6 71 09 58 11 20.2 3A 12.0 9.0 18.1 42 21 13 5A 20.2 3.1 14 4.5 3.3 41 48 43 54 10 16.1 15.2 71 23 01 2.1 -- 12.0 41 48 06 70 56 39 16.1 13.1 11.0 41 46 43 14 70 56 29 71.1 16.2 10.0 -- 41 48 47 51 11 70 56 -- 17.1 3.1 -- .6 11 70 57 09 15.1 10 2.6 13.1 16.3 9.0 .4 14.3 5.8 8.3 11 -- 12.3 1.9 38 12 -- 10.2 11.1 10 -- 10.2 9.2 40 10.2 -- 17.3 51 11.0 10.0 52 10.0 17.3 8 10.0 71 13.2 7.3 53 12 11 18.2 7.3 10 9 17.0 7.3 7.2 9.5 20.0 7.2 12 8.2 8.0 7.3 17 9.0 9.0 9.8 9 8 8 8 Table 7. to the station locations in images

38 Use of Thematic Mapper Imagery to Assess Water Quality, Trophic State, and Macrophyte Distributions in Massachusetts Lakes al numbers corresponding Haze-corrected TM digital numbers Haze-corrected TM Band 1 TM Band 2 TM Band 3 TM Band 4 Lake color Secchi Secchi parency disk trans- tion phyll Continued Chloro- concentra- ″

′

° June 10, 1998 June Longitude August 26, 1997— August ″

′

° Latitude Station identifier Continued — Lake-water-quality data collected within 24 hours of Landsat-5 Thematic Mapper imaging the lakes, and haze-corrected digit Station name code PALIS Long PondLong PondLong PondLong PondLong PondLong Pond Lake Naukeag Lower Maspenock Lake 62108 Maspenock Lake 62108Merino Lake 62108 1B 62108Middle Pond 35041 1C 62108 PondPuffer 2BThompson Pond 51112 62108 3B 1 LakeUpper Naukeag 51112 41 48 43 5B LakeWaban 41 48 43 70 56 39 Deep end Deep hole Pond 41 48 06Walden 70 56 39 42036 Basin Pond 41 46 43Walden 70 56 29 41 48 42 42 12 10 2.4 42 39 52 96198 Lake 41 48 47Watatic 70 56 35090 70 56 41 -- 71 33 29 36155 71 57 38 LakeWequaquet 82092 1 70 57 09 42 11 29 1.5White Pond (Concord) -- 1 3.7 71 33 20 -- LakeWinnekeag 1 2.1 6.0 1 -- 2.1 1 72125 -- 82109 -- 2.0 42 03 01 3.4 Maspenock Lake 48 2.1 1.8 82109 -- 82118 71 54 02 2 41 40 18 Maspenock Lake 96333 35095 46 42 39 42 17 48Middle PondA 50 Deep hole 2.1 70 24 59 42 24 20 71 55 50 42 26 20 7.3 159 33Mystic Lake Deep hole B 71 58 38 -- 81157 -- 1 71 27 36 -- 42 26 20 Pond 71 20 31Puffer 15 1 -- 7.3 1.9 7.3 71 20 31 -- 42 17 14 26 14.3 84.5 1.7 7.2 8.3 51112 1 7.3 71 18 35 2.1 6.3 51112 .9 8.3 7.2 4.4 42 25 40 1.6 13.3 7.2 41 40 08 10.2 5.3 3.2 42 40 49 Deep end .4 8.0 9.2 71 23 28 -- 8.2 70 20 28 6.2 96198 3.5 71 56 22 Basin -- 6.2 8.0 11.2 3.5 42 39 48 42 12 10 96218 13.0 9.0 -- -- 10.0 132 78 71 53 55 71 33 29 -- 82092 6.1 10.0 9 2.7 9.0 -- 1 10.3 42 11 29 13.0 -- 9.0 1 9 14 6.3 -- 71 33 20 16.3 8 12.3 -- 11 4.5 12.3 2.6 8.8 1 8 9.2 8 1.1 -- 15.3 11 15.3 8 4.1 41 40 18 18.2 7.2 10.2 3.0 8.2 3.1 41 40 32 70 24 59 -- -- 14.3 11.2 9.0 -- 11.2 42 24 20 70 25 02 2.0 21.0 11.0 8.0 8.0 42 71 27 36 1.8 13.3 11.2 10.0 -- 8.3 11 10.3 10.0 5.4 15 33 16.1 13 10 9 8.5 5.6 10.2 12.0 10.2 12 9.3 4.3 9.2 12 8.5 .8 10.0 10.0 5.5 9 3 9.2 10.0 1 6.5 300 13 8.7 15.5 8 10.0 12 13.5 10.7 8.5 10.5 11 11 10.5 12 5.5 16.7 14.7 12.7 16 16 14 Table 7. to the station locations in images

Table 7 39 al numbers corresponding Haze-corrected TM digital numbers Haze-corrected TM Band 1 TM Band 2 TM Band 3 TM Band 4 Lake color Secchi Secchi parency disk trans- tion phyll Chloro- Continued concentra- ″

′

° Longitude June 10, 1998— June ″

′

° Latitude Station identifier Continued — Lake-water-quality data collected within 24 hours of Landsat-5 Thematic Mapper imaging the lakes, and haze-corrected digit Station name code PALIS Spy PondSpy PondSpy Stearns Mill Pond LakeWaban LakeWebster LakeWebster White Pond (Concord)White Pond (Hudson/Stow) 82104 71040Whitins Reservoir 71040Whitins Reservoir 82119 72125 1 82118Whitins Reservoir 42064 SouthWhitins Reservoir North 42064 LakeWinthrop 1 2 1 1A 42 24 22 51179 42 32 10 42 24 32 71 09 30 51179 1B 71 27 01 71 09 51179 1A 42 23 30 42 17 14 42 03 14 51179 6.4 42 25 40 2A 11.6 71 28 49 71 18 35 9.1 71 50 56 42 03 14 71 23 28 1B 72140 71 50 56 3.0 2B -- 42 04 15 1.4 5.7 2.2 2.2 3.5 42 04 21 71 45 42 Deep hole 1.9 71 46 43 42 04 15 30 6.2 2.3 5.2 42 11 23 3.3 42 04 21 98 71 45 42 49 1.5 71 25 26 4.3 71 46 43 1.1 12.5 291 12 10.5 14.5 1.4 4.4 0 6.6 7 1.3 4.0 10.5 6 8.5 10.5 4.6 9.5 7.5 19.5 57 13.5 2.6 3.7 12.7 27 19.5 14.7 12.7 6.5 10.5 6.5 12.5 21 7.5 98 12.5 27 12 10.5 11 12.7 15.7 12.5 9 9.7 8.5 10.7 12.5 9.5 15.7 8.5 11 19 8.5 12.7 12 11 8.5 13.7 6.5 19 12.7 14 13.7 16 9.7 14 16 13 Table 7. to the station locations in images

40 Use of Thematic Mapper Imagery to Assess Water Quality, Trophic State, and Macrophyte Distributions in Massachusetts Lakes