Harmful Algal Bloom Vulnerability in Michigan
Identifying places where populations in Michigan may be most at risk from harmful algal blooms (HABs).
Background
This project sought to create a measure of population vulnerability to harmful algal blooms (HABs) in Michigan.
It was a statewide extension of an earlier effort for quantifying HAB exposure risk for some communities in Lake Erie’s western basin and the Saginaw Bay basin.
In that project, demographic indicators of HAB sensitivity and adaptive capacity were combined with water quality indicators to generate an overall index of risk among the communities.
You can learn more about that effort here ...
... and view the interactive tool it generated .
In this most recent project, existing statewide indicators of environmental justice were combined with modeled nutrient loading data and satellite images of cyanobacterial loads to explore various prioritizations of HAB risk in Michigan.
The interactive map at right displays recent HABs in Michigan , and is maintained by the Department of Environment, Great Lakes, and Energy (EGLE).
Methods: Socio-econ and Health Data
The first step in creating a HAB vulnerability index was identifying datasets that could map the socio-economic and health characteristics of at-risk populations.
The 2019 U.S. Census Bureau's Community Resilience Estimate was an initial prospect, but its aggregation of demographic categories made it too coarse for drilling down into specific characteristics related to HAB vulnerability, such as the percentage of very young people in an area and the prevalence of asthma.
A more detailed dataset, focused specifically on Michigan, is EGLE's Environmental Justice Screening Tool (MiEJScreen), released earlier in 2022. This dataset combines socio-economic, health, and environmental exposure data to map vulnerable populations at the census tract scale. It also allows users to drill down and view individual components of that vulnerability (such as age demographics and asthma cases), which made it useful for this project. The image at left shows the component datasets of the MiEJScreen.
MiEJScreen's consideration of environmental exposure covered a broad range of areas, from air quality to proximity to hazardous waste sites, but water quality was not a focus, and HABs were not represented at all. This project sought to fill that gap.
EGLE's interactive map at left allows users to explore the MiEJScreen data.
MiEJScreen Overall
The map at left shows MiEJScreen overall percentile values. The higher the value, the greater the vulnerability of residents in a census tract than others in Michigan.
Click on a location to learn more.
MiEJScreen Population Characteristics
The map at left shows the Population Characteristics percentile score (a component of the MiEJScreen overall percentile score). The higher the value, the greater the vulnerability in terms of health and socio-economics of residents in a census tract than others in Michigan.
Click on a location to learn more.
MiEJScreen Socio-economic Factors
Drilling down a bit further into the demographic data, the map at left shows the Socio-economic Factors percentile score (a component of the MiEJScreen population characteristics percentile score). The higher the value, the greater the vulnerability in terms of socio-economics (income, education level, unemployment, minority representation) of residents in a census tract than others in Michigan.
Click on a location to learn more.
MiEJScreen Asthma Cases
Continuing to drill down further into the health data, the map at left shows the asthma-related emergency room visits percentile score (a component of the MiEJScreen socio-economic factor percentile score). The higher the value, the more per-capita asthma-related trips in a census tract than others in Michigan. There are other socio-economic variables that MIEJScreen considers, but asthma prevalence is particularly relevant to an evaluation of risk from HABs, because their airborne byproducts can exacerbate breathing difficulties for those with respiratory conditions.
Click on a location to learn more.
MiEJScreen Population <5 years old
Continuing to drill down further into the demographic data, the map at left shows the young population percentile score (a component of the MiEJScreen socio-economic factor percentile score). The higher the value, the greater the percentage of residents less than 5 years old when compared to others in Michigan. Young children are more susceptible to health complications resulting from HAB exposure.
Click on a location to learn more.
Coupled with asthma, these two variables are the most relevant to the exploring the socio-economic vulnerabilities to HAB exposure. Subsequent sections below will describe how they were combined with water quality indicators to calculate an overall HAB vulnerability index.
Methods: Water Quality Data
The next step in creating a HAB vulnerability index was calculating indices of water quality.
The USGS SPARROW model simulates nutrient loading at watershed scales , and was run for the tributaries of both the Great Lakes and Mississippi river basins.
Phosphorus and nitrogen loads in surface water are both drivers of HAB formation. Map outputs from SPARROW can identify areas more prone to high nutrient loads in nearby lakes and streams.
Water Quality: SPARROW Total Phosphorus
The map at right shows annual total SPARROW-modeled phosphorus (2012) loading rates at sub-watershed scales in Michigan. Phosphorus in surface water can come from fertilizers (applied on farm fields and suburban lawns), from animal manure (livestock and geese), and sewage treatment.
Click on a location to learn more, including how other sources of phosphorus contributed to the overall load.
Water Quality: SPARROW Total Nitrogen
The map at right shows annual total SPARROW-modeled nitrogen loading rates (2012) at sub-watershed scales in Michigan. Nitrogen sources are similar to ones for phosphorus listed above, but also include atmoshperic deposition.
Click on a location to learn more, including how other sources of nitrogen contributed to the overall load.
Water Quality: Cyanobacteria Loading
Cyanobacteria, sometimes referred to as blue-green algae, are microscopic organisms that are the main cause of HABs in Michigan. EPA's CyanApp allows users to view weekly cyanobacteria loads recorded by satellite.
The map at right shows average weekly loading between April 2016 and July 2022 for Michigan's inland surface waters and Great Lakes coastal areas. Dark blues reflect areas where loads are relatively low, greens slightly higher, yellow higher than green, and red the highest average loads.
Zoom in on the map to explore the cyanobacteria spatial distribution at finer scales.
Methods: Scaling to the Census Tract Level
To combine socio-economic and water quality data into a single index, their different spatial scales had to be resolved.
The catchment-scale SPARROW nutrient data and the pixel-based cyanobacteria loading data was converted to the census tract scale used by MiEJScreen (n=2,765).
The map at left shows an area-weighted calculation of SPARROW-modeled phosphorus loading as a percentile at census-tract scales.
Click on a location to learn more.
This map shows nitrogen loading rates converted to a percentile score.
The north-south gradient in the both the phosphorus and nitrogen maps reflect concentration of agricultural land in the lower portion of Michigan. However, as will be discussed later when reviewing the top 10 most vulnerable census tracts, urban sources of nutrients (e.g. wastewater treatment) can be significant contributors to an individual tract's HAB vulnerability score.
Click on a location to learn more.
To convert the pixel-based cyanobacteria loading rates to the census tract scale required a more complicated geoprocessing approach than the clipping routine used in the SPARROW catchment scale conversion.
For example, the lake at left regularly carries a moderate load of cyanobacteria ( according to CyanApp's database of satellite images ). The pink line represents the boundaries between two census tracts. A traditional geoprocessing clip operation would not associate any cyanobacterial load from the lake with the west census tract, because no pixels intersected with the tract's area. However, if the lake were to experience a significant algal bloom vulnerable residents in the west census tract could still be at risk (especially those along the border).
To ensure that census tract's are not only associated with cyanobacteria loads within their own boundaries, average weekly loads within a 1km buffer of the boundary of each census tract was calculated.
Now, the risk posed by the cyanobacteria load in the lake will be factored into the overall HAB vulnerability score of the west census tract.
The map at left shows cyanobacteria loading percentile scores for Michigan census tracts. Higher percentile scores tend to occur on the coastlines, reflecting detected cyanobacteria within the Great Lakes. However, there are hot spots in the state's interior, which is discussed further in the review of the top 10 vulnerable tracts.
Click on a location to learn more.
HAB Vulnerability Index
The final step was to combine the socio-economic and health data with the water quality indicators to yield an overall index of HAB vulnerability. The goal was to identify areas prone to algal blooms (indicated by satellite measured cyanobacteria loads and modeled nutrient loading) with populations most prone to complications from their development (asthma sufferers and young children).
The map at right combines the MiEJScreen, SPARROW, and CyanApp data to produce that index (the higher the value the more vulnerable the census tract) using the following weights:
percentile cyanobacterial load within 1km): 0.3
percentile SPARROW-modeled P loading rate: 0.1
percentile SPARROW-modeled N loading rate: 0.1
percentile population < 5 years old: 0.25
percentile population with asthma: 0.25
The weights are specified to equally split water quality and socio-economic components, while prioritizing areas of observed cyanobacteria loading over modeled nutrient loads.
Click on a location to see the calculation for a census tract.
While young children and asthma suffers represent the most vulnerable populations to HAB exposure represented in the MiEJScreen dataset, it may still be worthwhile to explore additional calculations of HAB vulnerability using other MiEJScreen indicators.
The map at right shows HAB vulnerability using MiEJScreen's broader population characteristics (which includes asthma, blood lead levels, low infant weight, income, unemployment, housing burden, and linguistic isolation, among others) as indicators of socio-economic vulnerability. This index used the following weights:
percentile cyanobacterial load within 1km): 0.3
percentile SPARROW-modeled P loading rate: 0.1
percentile SPARROW-modeled N loading rate: 0.1
MiEJScreen Population Characteristics Score: 0.5
Click on a location to see the calculation for a census tract.
The map at right shows HAB vulnerability using MiEJScreen's overall vulnerability index, which combines socio-economic and environmental indicators (including air quality and proximity to toxic sites). This index used the following weights:
percentile cyanobacterial load within 1km): 0.3
percentile SPARROW-modeled P loading rate: 0.1
percentile SPARROW-modeled N loading rate: 0.1
MiEJScreen Overall Score: 0.5
Click on a location to see the calculation for a census tract.
Discussion
This section explores the first version of the HAB Vulnerability Index presented above (the one that incorporates asthma likelihood and percentage of children from MiEJScreen) in greater depth. The map at left disiplays the spatial distribution of that index for reference.
At the statewide scale the main discernible spatial patterns in the index is the north-south/southeast line of lower index values (less vulnerability) in the Lower Peninsula, with clusters of higher vulnerability in the southwest, the Thumb, and southeast.
Calculating a cluster statistic ( Getis-Ord Gi* ) reveals the spatial trends more clearly. The red in the map represent hot spot regions (higher vulnerability) while the blue cold spots indicate lower vulnerability.
The drivers of the hot spot in the southwest vary at more local scales.
South of Grand Rapids, the higher concentration of agricultural land yields greater rates of nutrient loading to surface water bodies, which can potentially contribute to the frequency of heightened cyanobacteria levels in lakes, rivers, and streams. Use the buttons below to toggle between the nutrient and cyanobacteria loading percentiles of the HAB Vulnerability Index.
However, in some parts of the region the nutrient loading is mainly from urban sources. For example, west of Kalamazoo SPARROW-modeled estimates of annual phosphorus and nitrogen loading are primarily comprised of wastewater treatment sources. Use the buttons below to toggle between maps of nutrient loading, and click on the dark catchment to see the its source breakdown.
The hot spot classification in Grand Rapids and Kalamazoo is mainly driven by the higher per capita asthma-related ER visits.
Asthma-related ER visits is also a significant contributor to Muskegon's hot-spot classification,
... as are the relatively higher cyanobacteria loads in Muskegon Lake and Lake Michigan.
Saginaw's hot-spot classification combines the drivers of Grand Rapids, Kalamazoo, and Muskegon. Its proximity to Saginaw Bay's high annual cyanobacteria loads, ...
... nutrient loading wastewater treatment (click on catchments to see source breakdown), ...
... and higher rates of asthma-related ER visits per capita.
In Detroit, the hot-spot follows the trends in the previously analyzed urban areas, with higher nutrient loads from wastewater treatment and urban sources, ...
..., but also much higher rates of asthma likelihood.
The cluster map shows regional spatial trends in the HAB Vulnerability Index, but it does not mean that vulnerable census tracts cannot be found inside of the cold spots.
The relatively high cyanobacteria loads of Ford and Belleville lakes near Ypsilanti and Van Buren Township cause them to score high on the HAB Vulnerability Index, while falling within a cold-spot region.
The cluster map also left out this high value census tract around Lake Gogebic in the Upper Peninsula, driven by the lake's high cyanobacteria loads.
The Top 10
The maps in this section highlight the ten census tracts with the highest scores on the HAB Vulnerability Index, including the leading contributors for each score.
These two tracts, each sharing a bank at the mouth of the River Raisin as it enters Lake Erie, have the two highest scores on the index; 92.3 for the southern tract and 89.1 for the northern one. Each of those scores are mainly driven by the tracts' proximity to the high caynobacteria loads that plague the western Lake Erie basin.
These two tracts, each sharing a bank of the St. Joseph River outside of Three Rivers, have the next third and tenth highest scores on the index; 87.5 for the southern tract and 82.6 for the northern one. These tracts both contain high percentiles of children under five years old and asthma likelihood, in addition to high nutrient loads from wastewater treatment and cyanobacteria loads in the St. Joseph River.
This tract in Ypsilanti has the fifth highest index score at 86.5. The main contributors to its score are its percentile of children under five years old and cyanobacteria loads in Ford Lake.
This tract in Port Huron has the next highest index score at 86.4. The main contributors to its score are its percentile of children under five years old and high nutrient loads from wastewater treatment.
This tract in Van Buren Township, near the other high-scoring one in Ypsilanti, has the next highest index score at 86.2. The main contributors to its score are its high nutrient loads from wastewater treatment and the high cyanobacteria loads in Belleville Lake.
This tract in Dearborn, along the River Rouge, has the seventh highest score at 85.1. This tract has a relatively high concentration of children under five and large nutrient loads from wastewater treatment.
This tract in Bay City, near the Saginaw River, has the next highest score at 84.6. Like the previous high-scoring tracts, this one has large nutrient loads from wastewater treatment, in addition to relatively high concentrations of children under five and asthma likelihood.
This Detroit tract has a vulnerability score of 84.3, mainly driven by high concentrations of children under five and asthma likelihood.
Limitations / Uncertainty
SPARROW researchers attempted to account for all sources of water, nutrients, and sediment, but it is always possible that some were missed and unaccounted for in the model outputs.
Most of SPARROW’s model calibration was performed against gage locations with relatively large drainage areas, so modeled outputs of nutrients in smaller tributaries or headwaters are more uncertain.
SPARROW outputs are generated at catchment scales ...
... but, as described above, to compare these outputs to MiEJScreen data those outputs had to be scaled to the census tract level. This required the assumption that SPARROW nutrient loading rates were uniform throughout an individual catchment, ...
... which allowed for an area-weighted value that could be compared to MiEJScreen outputs.
The pixel resolution of CyanApp’s images of cyanobacteria may be too coarse to capture algal blooms in smaller water bodies. Therefore, the overall evaluation of HAB risk may be under-representing the risks posed in these areas.
Download
You can download a GIS feature layer of the HAB vulnerability indices an ArcGIS Pro layer package , a file geodatabase for use in ArcMap, or a shapefile for other GIS desktop software. Detailed metadata for the resource is available here .
Conclusion
The HAB Vulnerability Index provides a metric through which efforts to protect vulnerable populations from HABs can be prioritized. By combining modeled estimates of HAB-feeding nutrient loading using the USGS SPARROW model, observed cyanobacterial loading from EPA CyanApp, and socio-economic data from EGLE's Environmental Justice Screening Tool, this project mapped where populations most vulnerable to a HAB exposure were most likely to encounter one. While there are multiple ways that those vulnerable populations could have been defined, this project focused on young children and those prone to asthma. By coupling that information with indicators of water quality, the HAB Vulnerability Index is a tool that can help raise awareness of the potential health impacts of HABs in the places where that awareness is most needed.
