Water on the Web

C. Data and Data Archives and Visualization

Here you will find  information about the existing data and data archive. The well organized Water on the Web data archive provides different types of data: raw data, already organized and aggregated data (lake trends) and environmental data, (such as weather information). The raw data can be downloaded either as a complete dataset or separately for each week. The overview of the weekly datasets shows calibration dates and missing data. The lake trends include the following topics:
  1. Surface trends - these are time course plots of RUSS (Remote underwater sampling station) parameters, averaged for each day and for the upper 3 meters of the water column.
  2. Heat and oxygen budgets - these are time course plots of whole-lake heat and oxygen. The lakes are divided into 3 layers using morphometry (depth contour) data.
  3. Water chemistry - this section contains spreadsheets with all nutrients, major ion, chlorophyll, secchi depth, and other non-RUSS data collected for WOW.
  4. Morphometry - spreadsheets with surface area and volumn data for each 1 m thick layer
  5. Other - a smorgasboard of tables, plots, images and possibly anecdotes.

    6. (see: http://wow.nrri.umn.edu/wow/data/options.html)
The environmental data include weather information, landuse maps, and GIS maps such as maps for bathymetry.

Water on the Web's data archive is integrated in a domain Data that contains data as well as tools for their visualization.  A start page (http://wow.nrri.umn.edu/wow/data.html) allows the user to select one of the six lakes, and the information about that lake they would like to see. The available information about the lakes are portioned into four domains:
 
      - RUSS Data:  raw data, data from the measurement of the RUSS Unit
      - Data Visualization: graphic online tools
      - Environmental Data:  information about environment of the respective lake
      - Lake Trends:  already organized and aggregated data, partly enriched with graphical displays

Below we present the data sets and visualization tools  related to these four domains.

RUSS Data

RUSS Units collect measurements in lakes every four or six hours (depending on the lake). Furthermore, the scientists' boat regularly on the lake and take manual measurements which they use to calibrate the RUSS Units.  This is necessary because algae and bacteria settle down on the RUSS Units and influence the measurement results. Further explanations and information about the degree of accuracy of the measurements can be found under Quality Assurance and Quality Control (QA/QC).  All these measurements are available on the Web. The user can access either all measurements taken since installation of the RUSS Unit (complete archive, separated values format (CSV)) or access data in weekly chunks.  Choosing the option "weekly" gives an overview showing weeks when data were (or were not) collected (see table below).  Moreover we can see the days when the RUSS Unit was calibrated and when additional measurements were done manually (these manual profiles can also be downloaded).  The data are available both as Excel files or in HTML format.

RUSS Data Ice Lake (http://wow.nrri.umn.edu/wow/data/ice/current.html)










Below we show the data in HTML format.
 
 

Weekly RUSS data, HTML Format

(http://wow.nrri.umn.edu/wow/data/ice/russ/ice19991031.html)

The Excel tables contain all variables included in the HTML tables as well as the additional variables:  schedule depth (the depth that they hoped to sample at) and actual depth (the actual depth sampled). These two can differ by up to 0.2 m.  An advantage of the Excel spreadsheets is that they contain prepared graphs, which we further describe in our Chapter on Tools for Data Analysis.
The "complete archive" tables are only minimally different from the "weekly Excel tables": only a column with "scheduled time" was added, which is the time the measurement was scheduled to be taken.
 

Lake
SchedDate
SchedTime
ActDateTime
Sched
Depth
Act 
Depth
Temp
pH
Cond
DO
DOpctSat
Turb
IceLake
19.06.1998
02:00:00
19.06.1998 02:02
1
0,8
20,1
8,7
108
11,3
125
4
IceLake
19.06.1998
02:00:00
19.06.1998 02:05
2
1,9
19,2
8,8
110
12,3
133
4
IceLake
19.06.1998
02:00:00
19.06.1998 02:07
3
3,2
17,6
8,7
112
11,8
124
4
IceLake
19.06.1998
02:00:00
19.06.1998 02:09
4
3,9
16
8,6
113
11,7
119
5
IceLake
19.06.1998
02:00:00
19.06.1998 02:11
5
5
12,1
8,4
117
13,5
126
5
IceLake
19.06.1998
02:00:00
19.06.1998 02:14
6
6,1
8,3
7,8
123
7,4
63
6
IceLake
19.06.1998
02:00:00
19.06.1998 02:16
7
6,9
7,3
7,6
124
4,2
35
7
IceLake
19.06.1998
02:00:00
19.06.1998 02:18
8
8
6,2
7,4
127
0,9
7
5
IceLake
19.06.1998
02:00:00
19.06.1998 02:20
9
9,1
5,6
7,3
129
0,4
3
8
IceLake
19.06.1998
02:00:00
19.06.1998 02:22
10
10
5,3
7,2
132
0,3
2
11
IceLake
19.06.1998
02:00:00
19.06.1998 02:25
11
11,1
5,2
7,2
138
0,2
2
9
IceLake
19.06.1998
02:00:00
19.06.1998 02:27
12
12
5,2
7,2
140
0,2
2
7
IceLake
19.06.1998
06:00:00
19.06.1998 06:01
1
1
20
8,6
107
11,2
124
4
IceLake
19.06.1998
06:00:00
19.06.1998 06:05
2
2,1
18,9
8,7
110
12,3
132
4
IceLake
19.06.1998
06:00:00
19.06.1998 06:07
3
3,1
17,7
8,7
110
12
126
4
IceLake
19.06.1998
06:00:00
19.06.1998 06:09
4
3,9
16,3
8,6
112
11,7
120
5
IceLake
19.06.1998
06:00:00
19.06.1998 06:11
5
5
11,4
8,4
120
13
120
5
IceLake
19.06.1998
06:00:00
19.06.1998 06:15
6
6,2
7,4
7,7
126
5,3
44
6
IceLake
19.06.1998
06:00:00
19.06.1998 06:17
7
6,9
6,7
7,5
123
2,2
18
5
IceLake
19.06.1998
06:00:00
19.06.1998 06:18
8
7,9
6
7,3
126
0,5
4
5
IceLake
19.06.1998
06:00:00
19.06.1998 06:20
9
9,1
5,5
7,2
128
0,3
3
9
IceLake
19.06.1998
06:00:00
19.06.1998 06:22
10
10
5,2
7,2
133
0,3
2
10
IceLake
19.06.1998
06:00:00
19.06.1998 06:25
11
11,2
5,1
7,2
139
0,2
2
7
IceLake
19.06.1998
06:00:00
19.06.1998 06:27
12
12
5
7,2
142
0,2
2
7
IceLake
19.06.1998
10:00:00
19.06.1998 10:02
1
0,8
19,9
8,6
108
11,2
123
5
IceLake
19.06.1998
10:00:00
19.06.1998 10:05
2
2
19,1
8,7
110
12,4
135
5
...
...
...
...
...
...
...
...
...
...
...
...

The description page gives short explanations for all 12 variables of the "complete data":
 
   -
    Lake   It is an abbreviation of the site name.
     -    SchedDate   The scheduled date for the reading.
     -    SchedTime   All the readings for a given profile have the same SchedTime and SchedDate.
     -    ActDateTime   The actual date and time the reading was made.
     -    SchedDepth   The depth in meters where the reading is scheduled to occur.
     -    ActDepth   The depth in meters where the reading actually was made.
     -    Temp   The temperature in degrees Celsius.
     -    pH   The pH value.
     -    Cond   The electrical conductivity in microSiemens/cm.
     -    DO   Dissolved oxygen in mg/L.
     -    DOpctSat   Dissolved oxygen % saturation at temperature.
     -    Turb   The turbidity in NTUs.

The last six variables link to explanations which include why values on that variable vary and its potential impact on pollution.

Data Visualization

We describe the profile plotter and the color mapper in the chapter tools for data analysis.

Environmental Data

Two types of information are available here: Weather information and GIPS (Bathymetry) and Landuse maps for the lakes studied. We describe the available data below.  The weather data could be helpful in explaining outliers and other strange behavior of the measurement data. The GIS and Landuse maps could be used to explain the  chemical properties of the different lakes.

Weather Information

Actual weather conditions are provided for all six measurement stations. The stations Ice Lake, Grindstone Lake and Independence Lake have weather summaries for 1998 and 1999.  Because the RUSS Unit was not installed at Grindstone Lake until the middle of 1999, there is no weather give prior to 1999. The data sets are available as  "Excel spreadsheets that summarize major weather events and mean daily air temperature". "Major weather events" are days with more than 0.1 inch precipitation or an average wind speed higher than 10 mph. The data sets contain the following variables: Grindstone Lake and Independence Lake also have measurements of average wind velocity and direction.
All data sets come with two prepared graphs. One of them is the time series of the mean daily air temperature. We show below this graph for Lake Independence in 1998:


(http://wow.nrri.umn.edu/wow/data/independence/spreadsheet/indy_weather1998.xls)

The other prepared graph is a time series of  major weather events (precipitation and wind speed).  Again, we show as an example below the graph for Lake Independence in 1998.   (The graph for Ice Lake includes precipitation only.)


(http://wow.nrri.umn.edu/wow/data/independence/spreadsheet/indy_weather1998.xls)

The above graph has several deficiencies. First, the time axis is not an adequate continuous axis of time. The first three columns belong to three successive days (June 24th to June 26th), while the fifth column represents July 5th, and so on. This problem occurs if one uses Excel without caution in which case the  x-axes is interpreted by Excel as a category axis. A second problem is that the values of wind speed are drawn as a line graph. The slopes can be misleading because the time intervals between two points differ in their length. Wind speed was measured only on particular days, namely when either it was windy (above a certain limit) or rain fell.  The wind on the days between was definitely lower, and therefore the lines connecting the points are additionally misleading. Anyway, we have reproduced the graph elsewhere and the diagnosis of a trend remains true at least related to the selected points of measurement. One reason of the problem is that EXCEL cannot adequately deal with missing data, another source is how to deal with measurements that are always selected at certain points in time and have to asked how "representative" they are.

GIS Maps and Landuse Maps

Geographic Information System (GIS) maps show the lake contours. At the moment, the site has only Bathymetry maps like that below. The depth of the lake is represented at every point of the lake surface. The diagram below of Ice Lake shows depth classified into six categories and displayed in a color gradient.   Mathematically speaking we see a contour map of the function b(x, y) which represents the depth b depending on the coordinate x and y of the lake. This function was probably built from real measurements and some interpolation was done. Students may approach the graph as an empirical data graph in the first step.
 
 


(http://wow.nrri.umn.edu/wow/data/ice/gis.html)

The landuse maps (see example below) are typically color-coded and show how the land surrounding the lake is being used (see key below for usage categories).


(http://wow.nrri.umn.edu/wow/data/minnetonka/maps/landuse.jpg)

Lake Trends

The section "Lake Trends" contains several prepared data graphs based on the RUSS data informing about time trends of several variables.

Many types of data are needed to characterize the ecology of lakes and streams and a variety of techniques are available for best presenting them. This section includes graphs of various parameters plotted over time or versus depth and tabular summaries of other parameters, such as water chemistry, secchi transparency, chlorophyll, etc.

  1. Surface trends - these are time course plots of RUSS parameters, averaged for each day and for the upper 3 meters of the water column.
  2. Heat and oxygen budgets - these are time course plots of whole-lake heat and oxygen. The lakes are divided into 3 layers using morphometry (depth contour) data.
  3. Morphometry - spreadsheets with surface area and volume data for each 1 m thick layer
  4. Other - a smorgasboard of tables, plots, images and possibly anecdotes.
(Taken from data options descriptions)
Surface Trends
The surface trends datasets contain daily means for the surface layer (between 0 and 3 m).  Pre-formulated graphs (and example of which is included below) show daily means minima and maxima for the six variables measured by the RUSS-Unit.


(http://wow.nrri.umn.edu/wow/data/ice/trends.html#do)


Heat and Oxygen budgets
Heat and oygen budget are shown by pre formulated graphs that show the total heat budget and the total dissolved oxygen. The heat budget is measured as the number of calories required to warm water in the lake from 0 °C to its current temperature.  The authors give the following explanations on their graphs:

The RUSS data from Lake Independence is summarized to show the lake's average heat and oxygen content on each day. These are calculated by averaging the amount of heat and oxygen contained in each 1 meter thick layer of the lake over the course of the day and then adding them up from 0-3 m, 3-8 m and 8 meters to bottom. Therefore, the sum of the values for the 3 layers is the total amount in the whole lake on that day.  Limnologists say that these values are morphometrically-, or volume-weighted.
Charts of heat and oxygen content are provided below. These are "stacked-area" charts - showing the contribution of each layer to the total. Below each chart you will find a more detailed description of the calculations involved for determining the heat content and oxygen content. (http://wow.nrri.umn.edu/wow/data/independence/heat.html)

(The vertical lines indicate the calibration dates):


(http://wow.nrri.umn.edu/wow/data/independence/heat.html)


(http://wow.nrri.umn.edu/wow/data/independence/heat.html)

The graphs exemplarily show what is expected from students.
 

An exemplary reconstruction of the meaning of variables

A basic requirement for good graphs is that the variables that are shown are clearly defined for the user of the graph. We will show how difficult and theory laden this can be by clarifying the definition of heat budget. We try to reconstruct it from the available sources of WoW but we will point out some gaps and deficiencies. It should be easier for the user to find definitions of variables.

We also present the reconstruction in our report with great detail, because we want to show the complex definition of variables as one feature of the project and its data.

Heat budget is computed using the following formula:

heat content = mass * specific heat * temperature = m * C * delta T . Mass is calculated by mass = volume * density

The calculation of the dissolved oxygen is done in a similar way.  There is an assumption being made here that we think should be pointed out to students.  The materials tell the students that "Water has a specific heat of 1.0 calories per gram per degree Celsius. This means that it takes 1 calorie of heat energy to raise the temperature of a gram of water by 1°C." However, heat content is a relative magnitude related to a certain zero-level of temperature. The variable temperature in the above formula, in essence, is a temperature difference. The authors relate the temperature to 0 ° C as we can also see from the graph. That is why temperature and temperature difference have equal values here, although there is an important conceptual difference.
 
 

The authors say that density = 1 gram/milliliter = 1 kg/liter. The authors do not point out here that density is dependent on the temperature of the water. We think that it is all right to ignore these differences because they will have only a minimal influence on the result. However, it would have been better to mention this dependence and to argue that it is reasonable to ignore these differences. This is important because these differences become significant in other parts of the project, because the structures of the water layers are influenced by this property. The authors could have mentioned the chapters on Density Stratification in the Lake Ecology primer. There, this anomaly of water (maximum density at 4 ° C) is shown by a graph. This anomaly accounts for why fish can survive in frozen lakes.


(http://wow.nrri.umn.edu/wow/under/primer/art/densityvstemp.gif)

As an example, the authors' show the calculation of the heat content of the upper layer of Ice Lake for May 29th, 1998, at 6 a.m. They show the following table beside the above formula.

Table: Heat Budget Surface Layer May 29th, 1998 at 6 a.m.
Layer
Volume
(x 105 m3)
Temperature 
(average of layer)
Heat
(calories per layer)
0-1 meters 1.60 20.80 3.33 x 1012
1-2 meters 1.48 20.75 3.07 x 1012
2-3 meters 1.36 19.55 2.66 x 1012
Total (0-3 m) 4.44  
9.06 x 1012

They do not say how they computed average temperature of a layer, but it appears that they get this by taking the mean of the upper and lower border of the layer.

It is more difficult to understand how the volume of the different layers of the Ice Lake was determined. We find some information about volume under the point "morphometry" which can be found via the data archive. We can read the following:
 

To estimate the area at the top of any layer, use the following formula:
Regression of: Area (hectares) at top of layer vs depth of layer top (meters)

A(z) = (-1.184)*z + 16.58
where area(ha) is for the top of the layer and z is the depth in meters at the middle of the layer
To estimate the volume of any 1 meter thick layer, use the following formula:
Regression of: Volume of 1m layers versus mid-layer depth (z)
V(z) = (-0.113)*z + 1.633
where V(z) is the volume of the layer.

We found no information about how these formulae were derived. It would be an interesting task to interpret these formulae and to analyze assumptions that were made. For the first formula (area) a simple interpretation is that we have a volume like a square stone with a slanting side which has a slope of (-1,184). The result for z=0 is the surface area of the Ice Lake, and it accords with the information of the lake summary table. We did not find any information about the formula of the volume.  Given that the authors did not use this formula, we can surmise that they themselves do not trust it.  We have calculated the layer volume according to this formula and show this in the table by means of an added column. We see that the values differ, although the difference is not very large, between the depth of zero and nine meters. With the following formula you get the same results as in the table:

[AREA (z1) + AREA (z2)]/2 * 1m.
This corresponds to the volume of a 1m high cuboid with a base the same size as the average of the top and the bottom areas of the particular layers.
 
Table 2
Complete depth profiles of area (at top of layers) and layer
volumes for Ice Lake. z1 and z2 refer to the
top and bottom depths of a particular layer.
DEPTH
Z (meters)
AREA
@ top (hectares) 
LAYER FROM
z1 to z2 (m) 
LAYER VOLUME
(x106 m3)
V(z) = (-0.113)*x + 1.633 (x106 m3)
0 16.6 0-1 1.600 1.5765
1 15.4 1-2 1.481
1.4635
2 14.2 2-3 1.362
1.3505
3 13.0 3-4 1.244
1.2375
4 11.8 4-5 1.126
1.1245
5 10.7 5-6 1.008
1.0115
6 9.49 6-7 0.889
0.8985
7 8.30 7-8 0.771
0.7855
8 7.11 8-9 0.652 0.6725
9 5.93 9-10 0.534
0.5595
10 4.75 10-11 0.416
0.4465
11 3.57 11-12 0.297
0.3335
12 2.38 12-13 0.179
0.2205
13 1.20 13-14 0.063
0.1075
14 0.060 14-15 0.004
-0.0055
15 0.029 15-16 0.001
-0.1185
  Total Volume =  11.6 (x105 m3) 
(http://wow.nrri.umn.edu/wow/data/ice/morph.html)

If we trust the values of the table we now can reconstruct the calculation now.