GIS Trash Map

Clean Jordan Lake gratefully acknowledges REI Co-op for providing a stewardship grant to improve the efficiency and planning of our shoreline trash removal efforts. With this support, we have developed a GIS map of trash, tires and CJL volunteer effort.  Sue King, a graduate student in the Center for Geospatial Analytics at North Carolina State University, carried out the data analysis.  She presented her findings at the 2018 Annual Conference of the NC Water Resources Research Institute.  This link  is to a .ppt file that includes a voice description by Sue of each slide.

Sue started with CJL data of all cleanups dating back to 2009.  The data set consisted of the number of  trash bags, tires and volunteers in 1O5 cleanups conducted by CJL from January 1, 2009 to July 15, 2017.  The shoreline length covered the Haw River Arm from the entrance of Robeson Creek to the dam and both sides of New Hope River Channel from the dam northward to New Hope Overlook at Jordan Lake State Park. Of this 24 mi. stretch, about 17 mi. were cleaned at least once.  This section was selected because the main source of trash and tires found here is from stormwater flows from the Haw River watershed rather than recreational use of the lake.  The spatial variability in intensity of trash and tires and in volunteer effort was defined by dividing the shoreline length into 43 subsections.

Optimized Hot Spot Analysis (OHSA) is a tool within ArcMap that was used to determine significant clustering of high values (hot spot) or low values (cold spot). The Gi_Bin field identifies statistically significant hot and cold spots.  Features in +/- 3 bins reflect statistical significance with a 99 percent confidence and those in the +/- 2 bins, a 95 percent confidence level. Clustering for features in the 0 bin are not statistically significant.

ArcGIS maps have been created for viewing various layers of metrics.  Click here for a beginners guide to displaying layers of information on the ArcGIS Map. Four ArcGIS Maps are available online:

Trash, Tires & Volunteers-All Cleanups in Each Subsection

OHSA All Cleanups

Trash, Tires & Volunteers Excluding First Cleanup in Each Subsection

OHSA Excluding First Cleanup in Each Subsection

The first cleanup in each subsection removed the legacy of trash since the lake was filled to the beginning of volunteer work by Clean Jordan Lake in 2009.  Volunteers returned to  32 of the 43 subsections for additional cleanups.  Including trash and tires removed in the first cleanup each subsection gives the total shoreline loading starting from when the lake was filled up to the date of the last cleanup of each subsection.  The ending date was not the same because rotation of volunteers around to subsections for a second or even more cleanups took several years.

The shoreline length in cleanups varied among subsections.  In some events, the entire subsection was not cleaned; in a few others, only portions of two or more subsection lengths were cleaned.  The total weight of trash and number of tires removed in all cleanups in each subsection were therefore divided by the total shoreline length cleaned to express a grand average of loading intensity.  Bags of trash were converted to pounds by assuming each holds 20 lb.  Volunteer efficiency was expressed as lb of trash and number of tires removed per volunteer-hour.

Below is GIS mapping of total lb of trash removed per 100 ft of shoreline for all cleanups performed in each of 43 subsections.  The circles from smallest to largest represent bin sizes of 6-26.7; >26.7-59; 59-103; and 103-183 lb trash/100 ft.  The number of cleanups in each subsection is not the same because the rotation around the shoreline was not systematic.  Nevertheless, the larger intensities of trash are consistent with sections of shoreline most vulnerable by their position to accumulation from the watershed.  Expand the view (click on + sign) to see the individual subsections more clearly.  Click on a circle to see all data associated with cleanups in the subsection represented.

The accumulation of trash between the first and last cleanups was used to express the loading in each subsection per significant rainfall event.  The number of lake level rises (LLR) occurring during this period was chosen as the indicator of rainfall events.  Although not rigorous, visual inspection of the USGS data for lake level at the Moncure, N.C. gaging station suggests a 1-inch rainfall results in an LLR of about approximately 2 ft.  This was taken as the minimum LLR to measure the effect of rainfall on trash loadings; 47 LLRs of 2 ft or greater occurred from January 1, 2009 to July 15, 2017. 

For subsections with one cleanup subsequent to the first, trash loading intensity per significant rainfall was calculated by dividing the lb of trash per 100 ft of shoreline removed by the number of LLRs that occurred between the first and second cleanups. For subsections with more than two cleanups, the trash loading between the first and last cleanup was divided by the corresponding number of LLRs to give a grand average loading.

If the lb trash/100 ft/LLR remains about the same over the entire database, this would suggest a strong correlation of trash loading with number of rainfall events.  However, scattered values may suggest that the number of LLRs alone does not capture the effect of rainfall.  Two alternative metrics were used to include the effect that rainfall intensity may have on flushing trash off the landscape. Cumulative lake level rise (CLLR) is the total feet of rise cause by all rainfalls between the first and last cleanup. The average height (avg ht) of a LLR is the number of LLRs divided by the cumulative feet of lake level rise.  Higher values of either metric correspond to more intense rainfall events. If either lb trash/100 ft/CLLR or  lb trash/100 ft/ avg ht LLR  remains closer to constant over the database than lb trash/100 ft/LLR, this indicates that rainfall intensity is a more important determinant of trash loading than the number of rainfall events.

Here is an example GIS mapping to show the spatial variability in lb trash/100 ft/LLR.

The range is from 0.6 lb trash/100 ft/LLR to 38.3 lb trash/100 ft/LLR.  Most values were in the range of 1 to 10 lb trash/100 ft/LLR. Nevertheless this is a wide range and suggests that number of LLRs alone does not capture the effect of rainfall on trash loading. The viewer is left to explore on ArcGIS maps whether lb trash/100 ft/CLLR or lb trash/avg ht LLR is a better metric.

An example of OHSA is shown below for tires/100 ft from data for all cleanups.   

The largest circle represents Gi_Bin>2 meaning greater than 95% confidence of a hot spot, i.e., clustering of neighboring subsections with high tire intensity.  The smallest circle is for Gi_Bin >-2 meaning greater than 95% confidence of a cold spot, i.e., clustering of neighboring subsections with low tire intensity. Circles between smallest and largest represent locations that are neither hot nor cold spots.

The hot spots are mainly on the east side of the Haw River Arm on the opposite the entrance of the Haw River Arm to the main body on the east side of the lake. These results reinforce the observation that trash has a strong tendency to deposit in these subsections while being transported by high discharge in the Haw River along from the Haw River Arm toward the dam.

Two composite examples of GIS layers that cannot be viewed together on ArcGIS are given here. Map 1 shows the total number of volunteers and the total lb of trash and number of tires removed per 100 ft  within each of the 43 subsections.  The largest intensity of trash and tires, and correspondingly, the greatest number of volunteers were at subsections along the Haw River Arm and those directly opposite its confluence with the New Hope River Channel on the east side of the lake.

The volunteer-hours (number of volunteers x hours worked) and  lb of trash removed per volunteer-hour for each shoreline subsection are shown in Map 2.  In general, efficiency was lower moving northeast along the New Hope River Channel. Lower trash loads in this direction could explain this trend because bags are filled less quickly when more scattered.