Picture courtesy of J.S. Aber

Advanced Image Processing

Advanced Image Processing of the Blanca Peak, Colorado

By

Scott Smith

Earth Science Department, Emporia State University

Abstract

This webpage was designed to show differences in snow cover on Blanca Peak in Southern Colorado using Landsat imagery. I used bands 1-5 and 7 to create composite images which show different types of landcover. Images were both taken in October of 2002 and October of 2009. Idrisi software was used to process the images taken from the EROS data center of the USGS. Classification techniques were used to separate land cover and snow cover with the cluster module as well as the isocluster module. I found that in 2002 Blanca Peak had considerably more snow cover than in 2009. However, these images show variations in weather patterns and snow cover which do not directly reflect climate conditions. This technique was successful in finding differences in snow cover over a period of time.



Table of Contents


Introduction

Blanca Peak is located in southern Colorado approximately 20 miles northeast of Alamosa, Colorado in the Sangre de Cristo mountains. The peak ranges in elevation from 13,580 feet to 14,345 which includes 4 of the highest peaks in Colorado. The peak lies next to the San Luis valley to the west as well as the Great Sand Dunes to the north. Blanca peak was formed from uplifted precambrian crystalline rocks which consists of gabbro, mafic diorite, and monzonite. Blanca peak is situated next to the San Luis valley which makes it vulnerable to extreme weather changes. The elevation change from the surface of the valley to the top of the peak is approximately 6,000 feet.

Natural-color composite from 2002 (Left), Natural-color composite from 2009 (Right).
Click on Images to Inlarge.

Climate Data

According to NOAA, Southern Colorado was experiencing "much below normal" temperatures in 2002 for the month of October. On average the departure from normal was approximately 7.5 degrees (F). However, the temperature has been above normal in this region from Novermber 2001 to October 2002. The region also saw above average precipitation in October 2002. From January 2002 to October 2002, southern Colorado saw some of the driest conditions on record. Despite the record drought over the past 2002 year in southern Colorado, this region saw above average precipitation with below normal temperatures in the month of October. This would account for more snow cover on Blanca Peak in October of 2002.

Southern Colorado experienced "much below normal" temperatures in October of 2009. On average, the departure from normal was approximately 4-6 degrees (F). This area also received "much above normal" precipitation for the month of October in 2009. On average, the percent of normal precipitation was approximately 200% for October. However, during the three month period from August to October 2009, southern Colorado experienced normal precipitation. The percent of normal precipitation was approximately 90-100% for the three month period. The southern Colorado region experienced much cooler temperatures as well as above normal precipitation in October of 2009. This does not necessarily reflect climate changes on the satellite images which can mostly be attributed to extreme weather changes in a short period of time. According to the climate data for this region these conditions were quite normal for this time of year (Enloe, 2009).

Back to the table of contents


Process

The images below are an example of a standard false color composite image using band 2, band 3, and band 4. Using Idrisi, I used the Composite module to enter the desired bands. I then selected "linear with saturation points" and "create 24-bit composite with stretched values" (Aber, 2009) The red color in the image represents active vegetation, the tan or brown color represents bare soil or rock, and the white represents snow cover.

False Color Composites from Blanca Peak, Colorado

234 Composite from October 2002 (Left), 234 Composite from October 2009 (Right)


Below are images which show the NDVI (normalized difference vegetation index) values which is the distribution of active vegetation throughout the region. Vegetation is extremely sensitive to short and long term climate changes. In this situation, the green represents (Aber, 2009) active vegetation and the orange and yellow represents inactive vegetation, bare soil, or rock. To create these images I used the VEGINDEX module under Image Processing, and Transformation. I selected bands 3 and bands 4 as well as the NDVI palette to display the correct color scheme. I also used SCALAR to add "1" to the NDVI values. This shifts the value range from zero to two. Then I used SCALAR again to multiply by a constant of 125. Doing this rescales the values from zero to 250. To finish the images I used the CONVERT module to convert from real number to byte-binary form with rounding (Aber, 2009). According to these images, vegetation was much more wide spread in 2009 than in 2002. This would also agree with the climate data mentioned above which explains the drought in southern Colorado in 2002. Precipitation was above average in 2009 which explains the higher active vegetation values.

NDVI Values


To create a cluster image, I used the CLUSTER module under Image Processing and Hard Classifiers. Cluster images are used to identify and classify areas with similar numerical values using all bands of the dataset. Once similar areas are classified, the values can be used to group certain spectral signatures. By grouping different signitures, each group can be classified as a different cluster to determine differences in land cover. Once CLUSTER is selected then select 4 as the "number of files" and enter the original file names (Aber, 2009). These images look different but the cluster values are not important. The separation is the most important aspect of these images. By achieving the separation, it is possible to determine what the values mean and later to separate snow cover and bare rock for instance.

Cluster Analysis

Cluster Image from 2002 (Left), Cluster Image from 2009 (Right)


The images below were created to isolate snow cover and all other land cover. To separate certain values, I used the ASSIGN module under GIS Analysis and Database Query. Then I used the ASSIGN module to separate certain signiture values (Aber, 2009). I separated individual clusters from the images above to distinguish between snow cover and all other land cover values. I used my own color palette design to show the snow cover in 2002 and 2009. Then I used the database query and indicated tabular output and calculated area in square kilometers. These images show the snow cover on Blanca Peak in October of 2002 and October of 2009.

Snow Cover

Area with snow cover from October 2002 (Left), Area with snow cover from October 2009 (Right)

Back to the table of contents


Results

The snow cover images show that Blanca Peak had more snow cover in October of 2002 than in October of 2009. Snow cover in 2002 was calculated as 36 square kilometers and in 2009 was 21 square kilometers. These images reflect local changes in weather patterns and do not directly reflect climate change. The climate data for southern Colorado explains why there is more snow cover in 2002 rather than 2009. Even though southern Colorado experienced a drought in 2002, the month of October had much more precipitation and was much cooler than average. The snow cover image from 2009 depicts normal conditions for Blanca Peak.

2002
Category Square Kilometers
1 36
0 574

2009
Category Square Kilometers
1 21
0 582

Back to the table of contents



References

Aber, J.S., 2009. Advanced Image Processing. ES775 Advanced Image Processing. Emporia State University. 2006
http://academic.emporia.edu/aberjame/es775/syllabus.htm

Earth Resources Observation and Sciences Center, 2009. United States Geological Servey. Earth Explorer. EROS

Enloe, Jesse 2009. State of the Climate. National Oceanic and Atmospheric Administration. National Climate Data Center. NOAA.

Related Links

Emporia State University Earth Science at ESU
Advanced Image Processing NOAA

For more information email ssmith3@emporia.edu.