Using TM Imagery to Analyze Vegetative Change

by

Daniel Call and CeLena Clough

This webpage was created to satisfy requirements for the Remote Sensing course (ES771) during the fall of 2010 at Emporia State University

Table of Contents
Introduction Profile of Region
Isoclust Analysis NDVI
Summer Comparisons Fall Comparisons
Discussion References


Introduction

According to Jensen, man's exploration of the environment using the Landsat imagery began in 1972. It was at this time that NASA launched the ERTS-1 satellite to test whether unmanned satellites could reliably acquire imagery in a cost effective manner. The satellite's success prompted further developments to remote sensing technology. New instruments designed to capture specific bands of the electromagnetic spectrum were installed in later Landsat mission satellites. Today, these bands can be manipulated using imagery software to reveal information about the Earth's surface and the processes that are continually taking place. (Jensen p. 197)

In this discussion, the use of Landsat Thematic Mapper (TM) imagery was employed to analyze a 30 by 30 kilometer stretch of east-central Kansas from 1985 to 2005. It was originally intended to check the spatial variation in vegetation throughout time by the use of Isoclust Analysis and haze corrected NDVI compositions. In the Isoclust Analysis, the 7 bands of TM imagery are run through a sequence of calculations by which clusters are assigned values based on their responses across the electromagnetic spectrum. By comparing these cluster values to a 345 composite and a 123 composite image, vegetation and other structures can be identified. In this portion of the analysis, six categories were identified and the isolated clusters were assigned to these categories: Water Bodies, Coniferous Woods, Deciduous Woods, Grassland/Prairie, Agricultural Lands and Anthropogenic Structures.

The second portion of this analysis was dedicated to NDVI analysis of the terrain. The Red Band (.63-69 μm) indicates how much green light is being absorbed by green leaves that are actively undergoing photosynthesis. This is compared to the Infrared Band (.75-.9 μm) which is indicative of how much of the infrared spectrum is being reflected or transmitted through the spongy mesophyll. By examining the ratio of energy used versus energy absorbed, imagery specialists can identify the overall health of vegetation in a region by using the following formula:

(Infrared - Red)/(Infrared + Red)

(Jensen, Chapter 11)

Back to Table of Contents

Profile of Region

Vegetation

East-central Kansas is characterized by tall grass prairies. The two main types of native grass are little bluestem and big bluestem grasses. Indian grass, switchgrass, sideoats grama, and eastern gama grasses are also common, although nearly ninety different species of native grass are present in this area. More than 500 species of native broadleaf plants exist in the prairies. Wildflowers also contribute to the vegetation of east-central Kansas. The majority of wooded vegetation grows near creeks and rivers. Some common tree species in the area include white oak, black oak, and post oak (USFWS, 2010). Burning of the tall grass prairie keeps the trees from succeeding the prairie. May and June are the months that have the most vegetation growth. As the temperature increases and precipitation decreases in July, the tall grass prairies begin to die. (Middendorf et al, 2009).

Climate

Temperature and precipitation strongly influence vegetation in east-central Kansas. Eastern Kansas has a temperate climate. The summers are hot and usually humid. Spring and summer seasons have the most precipitation. Rain accounts for the majority of precipitation. Little accumulation of precipitation is from snow (Middendorf et al, 2009).

Land Use

The majority of the upland area is used for cattle grazing. Agriculture is common on flood plains. The production of agriculture on upland surfaces is not common because the soil is shallow and underlying rock makes cultivating the ground nearly impossible (USFWS, 2010).

The climate of the region in graphic form courtesy of the High Plains Regional Climate Center website

Back to Table of Contents

Isoclust Analysis

The area calculation reported for June 1985: Water bodies occupied an area of 32.26 km2, Coniferous forest occupied 113.27 km2, Deciduous forest occupied 225.42 km2, Prairie/Grasslands occupied 113.86 km2, Agricultural lands occupied 291.28 km2 and Anthropogenic structures occupied 125.73 km2 of the land surface.

The area calculation reported for October 1985: Water bodies occupied an area of 35.74 km2, Coniferous forest occupied 209.44 km2, Deciduous forest occupied 35.96 km2, Prairie/Grasslands occupied 78.83 km2, Agricultural lands occupied 464.24 km2 and Anthropogenic structures occupied 77.59 km2 of the land surface.

The area calculation reported for June 1996: Water bodies occupied an area of 53.44 km2, Coniferous forest occupied 101.21 km2, Deciduous forest occupied 195.52 km2, Prairie/Grasslands occupied 114.28 km2, Agricultural lands occupied 382.78 km2 and Anthropogenic structures occupied 54.58 km2 of the land surface.

The area calculation reported for October 1996: Water bodies occupied an area of 54.55 km2, Coniferous forest occupied 65.75 km2, Deciduous forest occupied 251.49 km2, Prairie/Grasslands occupied 115.45 km2, Agricultural lands occupied 336.93 km2 and Anthropogenic structures occupied 77.62 km2 of the land surface.

The area calculation reported for June 2005: Water bodies occupied an area of 34.04 km2, Coniferous forest occupied 29.11 km2, Deciduous forest occupied 109.52 km2, Prairie/Grasslands occupied 158.82 km2, Agricultural lands occupied 505.95 km2 and Anthropogenic structures occupied 64.37 km2 of the land surface.

The area calculation reported for November 2005: Water bodies occupied an area of 28.71 km2, Coniferous forest occupied 76.76 km2, Deciduous forest occupied 139.57 km2, Prairie/Grasslands occupied 103.25 km2, Agricultural lands occupied 447.28 km2 and Anthropogenic structures occupied 106.25 km2 of the land surface.

Graphic Form

Back to Table of Contents

NDVI

1985

As can be observed in the June image, vegetation is significantly more active than the October scene, especially in the highland surfaces where the land has not been worked as an agricultural surface.

1996

Similarly, in 1996, the June NDVI has a much more active vegetation profile than October, especially in the undisturbed upland surfaces. A noticeable difference between 1985 and 1996 includes the extent of active healthy vegetation between similar times of the year. During the June NDVI images, '96 appears to have a much higher land area at higher vegetation activity. This increased vegetative activity trend carries over to the comparison of the October images as well.

2005

Areal extent of healthy vegetation can be seen in the June image, and appears to have increased since 1996, continuing the trend of increased vegetative activity. One issue with the comparison of the Autumn scenes is that the two NDVI images are taken of the land surface with a temporal difference of about 3 weeks (Oct 10 versus Nov 4). Unfortunately, there were no suitable October images available for 2005 and this was the only year that had the best available combination of late spring and early fall images in the number of years prior (2003, 4) and following (2006,7) 2005. This makes drawing comparisons between early fall activity in 1996 and 2005 virtually impossible.

Back to Table of Contents

Summer Comparisons

The above graphic is designed to show the % change in terrain across the region from 1985 - 1996, from 1996 - 2005, and across the entire 20 year period. This percentage change is based on the area calculations performed in association with the isoclust analysis and is subject to operator and program error when calculating values to apply to each cluster. It was noted that some cluster values consistently overlapped no matter the number of clusters used in the isoclast analysis. Some clusters that were clearly observed as being present in Junction City (anthropogenic feature just southeast of Millford Lake) were given the same value as open fields to the northwest.

Back to Table of Contents

Fall Comparisons

The Fall calculation of land change is subject to the same drawbacks that affected the early Summer values. Additional drawbacks had to deal with the way Deciduous and Coniferous trees were expressed in the 123 and 345 composite images used to reclass the isoclust image. Prairie and grasslands were similarly confused as having similar values as land that was clearly used for agricultural activity.

Back to Table of Contents

Discussion

Upon comparing the Early Summer images, it appeared that wooded regions were becoming smaller in extent across the region over time and were being replaced by grassland prairies or being converted to an agricultural use. The NDVI images support the idea that vegetative activity is increasing across the area during the early portion of the growing season. The switch to the Early Fall images were a little harder to evaluate trends. The lack of availability of images during the same time periods in the spring and fall of the study years hindered the analysis across the 20 year time period. Further studies should be directed in such a manner as to take advantage of the benefits of a clear image and study area. Additional studies should also take advantage of on-site vegetation assessment with a GPS unit at several ground locations that are difficult to identify in the imagery and isoclust analysis. The resolution of the TM imagery series is acceptable for most uses but needs refinement to acquire more spatially precise data in its imagery before it becomes a first-use tool in precise vegetation analysis.

Back to Table of Contents

References

Jensen, J.R., 2007, Remote Sensing of the Environment: An Earth Resource Perspective Second Edition: Upper Saddle River, N.J., Pearson Prentice Hall, Inc. 592 p.

Middendorf, G., T.A. Becerra, and D. Cline, 2009. “Transition and Resilience in the Kansas Flint Hills.” The Online Journal of Rural Research and Policy 4.3 (2009): 1-28.

U.S. Fish and Wildlife Service. Flint Hills National Wildlife Refuge, 2010. Environmental Assessment Flint Hills Legacy Conservation Area. July 2010.

High Plains Regional Climate Center - Historical Climate Data Summaries - http://www.hprcc.unl.edu/data/historical/ Accessed Dec 9, 2010

Back to Table of Contents