Stranger Creek Watershed and Landcover Analysis

ES 775 Advanced Image Processing

Bradley D Johnson

9 December 2011
Table of Contents
Abstract Site Overview
Composites Vegetation Index
Boolean Image Cluster Image
Conclusions References


Stranger Creek Watershed is located in northeastern Kansas within portions of Atchison, Jefferson and Leavenworth counties. The United States Department of Agriculture’s Natural Resources Conservation Services (NRCS) and the Kansas Department of Health and Environment (KDHE) have listed this watershed as a top priority for remediation and conservation efforts based on pollution levels measured in Total Maximum Daily Loads (TMDLs) occurring within streams. The TMDL of various pollutants currently exceeds allowable levels set forth by the federal and state governments. Many organizations are engaged in remediation and conservation efforts throughout the watershed although priority focus is currently centered on only a few sub-basins. Digital imaging software and the creation and strengthening of GIS databases has led to a further understanding of Stranger Creek Watershed’s landcover, which in turn has allowed for the development of better landuse practices.

Overview and Site Location

In 2008 the United States Department of Agriculture’s Natural Resource Conservation Service (NRCS) and the Kansas Department of Health and Environment (KDHE) issued a Rapid Watershed Assessment for the Lower Kansas Watershed on the Total Maximum Daily Loads (TMDLs) for the basin. TMDLs refer to the maximum amount of pollutants which can enter a given water body and still meet state and federal water quality standards. Stanger Creek Watershed constitutes a large portion of the larger Lower Kansas Watershed and has been listed by the NRCS and KDHE as a High Priority for conservation and remediation efforts due to several impairments including high levels of bacteria (fecal coliform from animal feeding operations), Total Suspended Load (TSS) from stream-bank erosion, phosphorous, atrazine, copper, lead, zinc, and for impaired biological conditions (NRCS and KHDE 2008). The impaired conditions of the Stranger Creek Watershed (including Nine Mile Creek) have resulted in conservation and restoration actions taken by various groups in the immediate region and statewide. Among the most active groups in the region are the Leavenworth County Conservation District and the Kansas Alliance for Wetlands and Streams. Among the tasks performed by these organizations are: identifying sites of current animal feeding operations (AFOs) and potential stream bank stabilization sites; TMDL and general water quality research; remediation and conservation projects; geographic information system (GIS) database construction; and community education and outreach.

Stranger Creek Watershed is found within portions of Atchison, Leavenworth and Jefferson counties in Kansas and occupies an area of approximately 112,792 hectares (~278,714 acres). A drainage divide on the eastern edge of Jefferson County marks the western limit of the watershed while a similar divided located in east-central Leavenworth County marks the eastern edge of the Stranger Creek Watershed. Within the Stranger Creek Watershed, the Leavenworth County Conservation District and Kansas Alliance for Wetlands and Streams are currently focused on the southwest portion of the basin in a sub-basin known as Nine Mile Creek Watershed. Nine Mile Creek flows into Stranger Creek at Linwood, Kansas and from this confluence, Stranger Creek runs into the nearby Kansas River (Kaw) just south of Linwood. Together, Nine Mile and Stranger Creeks drain a significant portion of the region and pollution from these streams flow directly into the Kansas River and then into the nearby Missouri River. The combination of heavy concentrations of animal feeding operations and stream bank erosion sites (leading to higher levels of TSS, turbidity and bio-impairment) and proximity to the Kansas River have led to priority status of the US Environmental Protection Agency’s (EPA) hydrologic unit code 12 (HUCs) 102701040405 and 102701040406 within the basin. (See images below)

Image 2: Location of Stranger Creek Watershed within the State of Kansas. Image © B.D. Johnson

Image on left by B.D. Johnson: The sub-basins colored green are the priority focus of conservation and remediation efforts performed by groups like the Leavenworth County Conservation District and Kansas Alliance for Wetlands and Streams. Image on right by B.D. Johnson: Close-up view of the priority sub-basins. Proposed stream bank stabilization sites and Animal Feedling Operations (AFOs) were ploted on map from GPS Waypoints obtained through ground-truthing.

Pivotal to the implementation of quality remediation and conservation practices of a watershed is a strong understanding of the landcover within the basin. Certain types of landcover can promote the mobilization of pollutants and soil erosion while other varieties can slow or largely eliminate the same toxins and mass wasting. Idrisi image processing software allows scientists, researchers and interested parties to use available satellite data to analyze virtually any location on the planet. Manipulation of this data (stored in various GIS databases and clearinghouses around the world) allows for a better understanding of the landcover at any site of interest.

Idrisi image processing software was used to manipulate United States government satellite images of the southwest portion of the Stranger Creek Watershed to gain a further understanding of landcover, including vegetation and drainage patterns within the basin. (See images below)

Note: Satellite Images are obtained by photographing various aspects of the light spectrum. The different levels of the light spectrum are identified through “bands” which are the result of different levels of light reflectivity off of a surface or object. The light reflection off of an object is known as the “spectral signature.” Figure 1 below reveals the various bands of Landsat Thematic Mapper Satellite images (satellite responsible for the images below) and their applications.

Color Band Wavelength Applications
Soil/vegetation and deciduous/coniferous forest differentiation, clear-water bathymetry.
Vegetation growth/vigor, sediment estimation, turbid-water bathymetry.
Crop classification, ferric iron detection, ice & snow mapping.
Near infrared
Biomass (vegetation) surveys, water-body delineation.
Mid infrared
Vegetation moisture, snow-cloud differentiation.
Mid infrared
Hydrothermal mapping, rock/soil type discrimination.
Thermal infrared
Thermal mapping, plant stress, urban/ non-urban landuse differentiation.

Figure 1: Spectral bands of the Landsat Thematic Mapper (J.S. Aber's 2009 Adaptation of A.M Mika [1997])


Composite images of a location can be created when spectral bands are combined in various ways. Rearranging bands in different orders (I.e. 123, 245, 357, etc) with the help of imaging software (Idrisi) can highlight different features of a given site. Natural-color, False-color and Infrared are among the most common types of composites. Composites are often helpful in determining landcover types like natural vegetation, agriculture, water features and human structures (as each feature displays a different spectral signature) and changes in landcover over time (temporal analysis). (See images below).

Image 5 and 6: Natural-color composites (Bands 123) of southwest portion of Stranger Creek Watershed. Note the Kansas River (bottom of scene), Interstate 70 (upper-left), Nine Mile Creek (dendritic pattern left-center), Stranger Creek (sinuous/dendritic pattern right-half of image), town of Linwood, KS (Low-center. White area above Kansas River) and confluence of Stranger Creek with Kansas River (below and right of Linwood).

Images 7 and 8: False-color composites (Bands 234) of southwest portion of Stranger Creek Watershed. Active vegetation appears red. Fallow and harvested fields appear gray to green

Vegetation Index

A Normalized Difference Vegetation Index (NDVI) is a multispectral satellite image processing technique which has been in place for many years and is useful in monitoring biomass production, plant growth or vigor and vegetation cover. Red and near-infrared light are the portions of the electromagnetic spectrum responsible for allowing the NDVI process to work. Chlorophyll absorbs large amounts of incoming radiation in the red-light portion of the spectrum whereas the spongy mesophyll leaf structures reflect considerable light in the near-infrared band of the spectrum. The result of these actions is healthy and vigorous vegetation has little red-light reflectance and much near-infrared reflectance, and thereby, high NDVI values. Low values on the NDVI indicate non-vegetated surfaces (rocks, pavement, barren soil, water, snow, ice and clouds) (USGS 2010).

Images 9 and 10: NDVI (Displayed with NDVI256 Palette) for SW portion of Stranger Creek Watershed. Note how the 2011 image appears to have much more active vegetation than the 2000 image. This may be the result of delayed harvest of crops, as all of the new vegetation still appears to be on agriculture fields. In other words, the new vegetation is not necessarily from native species succession of old fields.

Boolean Image

In a Boolean image, pixels have either a value of 1 or 0. Processing images with the Boolean approach allows for the isolating of desired features, like water bodies. Water bodies generally have the same (or very similar) pixel values. By Reassigning (Idrisi function) all water feature pixel values with a single value of “1,” the corresponding image isolates the water from the rest of the landcover (which is given a value of 0) (Aber 2009)

In the Boolean image of the southwest portion of Stranger Creek Watershed, Stranger Creek is now highly visible, as are many small ponds. Nine Mile Creek (and other small creeks within the watershed) during the winter is largely a series of small pools, rather than a continuous stream. This is evidenced by a series of small pond-like features which dot the landscape. Refer to the dendritic pattern on the Natural-color composite (B123 2008) to the left for approximate location of Nine Mile Creek.

Images: Winter of 2008 (B123) and cooresponding Boolean image. Winter image was selected in order to have better representation of small water features which can be blocked-out by leaves during warm months.

Cluster Image

The cluster module of image processing allows for the identification and classification of image cells which have similar values. Numerical values are generally pulled from data contained across all of the bands in a dataset. The cluster module works by grouping together (clustering) cells with similar values (spectral signatures) (Aber 2009).

In the cluster image below, an unsupervised classification was performed. In an unsupervised classification, class identities are not specified prior to performing the cluster and therefore the nature of each class is not known. In order to determine the features of each class following the completion of the cluster, a comparison to the original imagery and knowledge of the ground cover is necessary (Aber 2009).

Image 12: Cluster Image of SW Portion of Stranger Creek Watershed (Fall 2011). Cluster 1= Fallow Fields; Cluster 2=Active Vegetation-Forest; Cluster 3= Non-harvested Fields; Cluster 4= Harvested Fields; Cluster 5= Barren soil; Cluster 6= Water Features


The Stranger Creek Watershed is impaired beyond TMDL standards by various chemical and biological pollutants, as well as excessive sediment loads due to high rates of erosion. Various levels of government, non-profits and several private landowners are involved in watershed restoration and conservation throughout the Stranger Creek Basin. Geographic Information Systems (GIS) and image processing software like Idrisi (and others) have furthered the understanding of landcover and water resources within the watershed and led to further development of GIS resources and improvements in landuse practices.

Image 13: Example of remediation efforts occuring along Stranger Creek in southwest portion of Stranger Creek Watershed. A once severely eroded streambank was planed-over and reinforced with sub-surface rocks. View here shows the native tall grasses and willows which were planted to anchor soils and serve as buffers between fields and streams.

Related sites