| Title Forest Health of Eldorado County | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Author Josh Cook American River College, Geography 350: Data Acquisition in GIS; Fall 2009 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Abstract Through the use of GIS and remote sensing, an analysis of the current health of Eldorado forest will be assessed. An NDVI (Normalized Difference Vegetation Index), tasseled caps, change detection, and unsupervised classification will be used to utilize the near infrared band that Landsat TM provides. The analysis will show changes made to vegetation with both cause and effect and provide the current health of Eldorado forests. | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Introduction It has seemed that during the summer months, in the state of California that we have become familiar with the term Fire Season. In the past few years it has become apparent that the incidence of fire has been on the rise. Due to the fires destructive nature measures must be taken to prevent these fires from consuming homes, taking tax payers money, and (most importantly) taking lives. The Angora Fire burned around 3,100 acres, destroyed 254 residential homes, and 75 commercial buildings. The total cost of the Angora fire was around $11.7 million. (Calfire) One way to help prevent this destruction and monitor our forest is through Remote Sensing. Remote Sensing has the ability to see over a vast area as well as the ability (in some applications) to see what our eyes cannot. With Landsat TM, the sensor that is attached has the ability to see the normal blue, green, and red wave lengths (which is bands one, two, and three respectively on the sensor) that our eyes can pick up, but Landsat TM also has the ability sense wave lengths in the electromagnetic spectrum that our eyes cannot see. The other bands on the Landsat TM sensor consist of: near infrared or NIR (band 4), mid infrared (bands 5 and 7), and far infrared (band 6) (umn.edu).” The near infrared band that Landsat provide gives an analysis a great tool to work with in monitoring vegetation health. The normal eye can detect if a plant is under stress by the color that it reflects back in shades of green. The greener the vegetation is, the healthier it will be. In the visible range of light, plants tend to absorbed more light than is reflected out. In the near infrared spectrum more is reflected off vegetation giving an analysis more degrees of change inside the vegetation. (nasa.gov) The near infrared light of vegetation will produce a deep red appearance if it is healthy and if it not the pigment will become lighter in color. Using the band of near infrared we can see a clearer picture of what the current stage of health a forest is in. Using methods such as NDVI (Normalized Difference Vegetation Index), change detection, classification of an image through its spectral signatures, and tasseled caps we can create an evaluation and action plan to create a healthier forest and reduce wildfires. Using these functions of analysis we can view the current conditions of Eldorado County’s forest. . | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Background In past studies, a series of Landsat imagery was analyzed to compare a particular location in Vermont to determine a change in forest growth. (Vogelmann) Using different time framed: Landsat TM, Landsat MSS, and Landsat EMS+ they were normalized and compared to generate a comparison of forest health over the years. Reflectance of Infrared and Near Infrared bands were then analyzed. After the comparison, the image of Vermont land cover was then “color coated (Vogelmann isprs.org)” to interpret change in reflectance over the years. The reflective wave lengths that were analyzed were created through the Infrared and Near Infrared spectrum, and have the ability to detect levels of chlorophyll. High levels of chlorophyll equals healthy growth and will show up on an image as dark red. Unhealthy growth will do the opposite, showing up lighter in color. Using the Landsat imagery and the classification process of color coating the data changes of the mid elevation around Vermont has been detected at a five percent decrease from the years of 1976 to 1999. Using Infrared band imagery has been proven helpful in the understanding of our forest health. This article has given the reader a greater idea of how Landsat imagery is used and how year to year comparisons of the data can be used to determine patterns in forest health. If continual studies using this method and other methods, we can determine the rate of change and the possible problems associated with the change. | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Methods The methods used to determine the current conditions on Eldorado County were to do a year to year comparison using Landsat TM data. The data consists of the eight Landsat TM imagery all in the month of July with displayed bands four, three, and two respectively. A process of NDVI, Unsupervised classification, tasseled cap, change detection, and a digital camera equipped with a near infrared filter where used to make an assessment of the health of the counties vegetation. The use of ERDAS imagine was used to develop the NDVI, tasseled caps, unsupervised classification, and change detection. The digital images of near infrared pictures were taken with an Olympus 8 megapixel camera, (model number fe-370) and a filter consisting of four blue lenses of theatrical gel. (kipkay) | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
The NDVI measures the near infrared reflectance through an algorithm provided from ERDAS. The darker the green in the image implies healthy vegetation the lighter the color gets, indicates stressed vegetation or low levels of near infrared signatures. (nasa.gov)
|
| Tasseled cap creates composite values of: brightness, greenness, and wetness. After running through the program in ERDAS, an analyst can figure out how much of each composite is present in an image. For vegetation health the wetness band can be used to see if enough moisture is present to create a healthy forest. The image above shows the wetness band in blue. (Watkins)
| 
| Unsupervised classification was ran with ERDAS and created twenty signature classifications based on pixel value. Areas that are marked are of low near infrared reflectance.
| 
| Change detection was ran through ERDAS with a calculated ratio of band four from two separate images. The calculations were ran to detect increases and decreases of ten percent or greater in the near infrared spectrum. The green indicates an increase and the red indicates a decrease.
| 
| The near infrared filter was used to observe test sights on the ground with the aid of the near infrared spectrum to help further indicate vegetation health.
| Results
| 
| With the NDIV, the resulting images can show that there has been a rise in near infrared reflection from the years of 2005 to 2009. In the 2005 image, shows that there was a wider range of lighter color of green then are present in the 2009 image.
| 
| In the tasseled cap images from the years of 2005 to 2009, darker colors of blue are shown in the 2009 image. The darker color of blue indicates that there was a greater amount of moisture leading to healthier vegetation in 2009 in comparison to 2005.
| 
| The change detection shows that there was a rise in near infrared reflectance (indicated with green) in July of 2009 compared to 2005. This confirms the NDVI and tasseled cap findings, indicating a rise in vegetation health.
| 
| A comparison was ran using July of 1989 data and July of 2009 data. Darker greens in the 2009 data are more distinguished than in the 1989 data.
| 
| 1989 data using tasseled caps indicate that less moisture was present than in the 2009 image. More yellows are present and less blue can be detected in the 1989 image.
| 
| Change detection was ran to evaluate the difference of our forest health in the years and months of July 1989 and July 2009. The 1989 data and the 2009 data were ran through an algorithim to identify changes in the image. The following increase in near infrared is indicated with in green. The area that has been circled in red shows an area of vegetation that went through a forest fire in the past but regained health in the recovery process.
| 
| An unsupervised classification of July 2009 data was ran to detect similar spectral signatures of low near infrared signals. There were twenty classifications ran, and through trial and error the proper signals where highlighted in yellow. Most of the yellow highlights show areas of clear cut around the Stumpy Meadows, Union Valley and Ice House area. Some areas around Ice house needed to have further evaluation.
| 
| 
|
| 
| To help determine what was interpreted as low NIR signal an up close analysis was needed.
| 
| Healthy vegetation is shown through the NIR lens. This picture was taken at Stumpy Meadows (just passed Georgetown) showing deep red color of infrared indicating healthy vegetation.
| 
| This image was taken on Ice House road (Weber Mill lookout) just off highway 50. This area is a young forest due to a fire that took place from Union Valley and down past Ice House. Through other images that were taken of young growth they contained similar pigment in color. The undergrowth on the other hand shows dead or dying vegetation that is prone to fire. The dead or dying vegetation shows up as a light pink. The low NIR levels in the unclassified image (in the Ice House road test area) show that the low levels are from dry undergrowth.
| Analysis The analysis of the project determined that Eldorado county forests have been in the stages of re-growth and increases in health from year to year. Through areas that have been burned in the past, set rise to healthier vegetation. The tasseled caps show direct results in why vegetation can and does fluctuate in health from year to year. Moisture is one of the key elements in forest health. The dryer the conditions, the higher the risk for forest fires and the overall health of the forest.
| Conclusions Controlling wild fires is a hard task to accomplish but with the right tools provided we can create a healthier forest. These tools can identify the decaying forested areas and create a process of selected logging to thin out the forest to prevent fires from spreading and allow younger and healthier growth to rise. Controlled burning may also be another process to consider such as the US Forest Service is doing in the fall of 2009. (fs.fed.us) The Forest Service is underway, performing controlled burns of undergrowth that may otherwise contribute to a forest fire in the future. Other applications that would work well with identifying dying vegetation is hyperspectral. Hyperspectral offers a greater detail in the characteristics of vegetation. The image that hyperspectral produces creates a much higher level of data in the electromagnetic spectrum but it is difficult to work with and considering (Landsat for the most part) is free and creates adequate analysis, it would be hard to look towards hyper spectral as an alternative.
| References (1) Interagency Information Cooperative. (2006) Enhancing the access and use of forest resources data in Minnesota. University of MN. http://iic.gis.umn.edu/finfo/land/landsat2.htm (2)Fire Protection. (2007). Angora Fire Incident Information. California Department of Forestry and Fire Protection. http://cdfdata.fire.ca.gov/incidents/incidents_details_info?incident_id=184 (3)Earth Observitory. (2009) Measuring Vegitation (NDVI & EVI). Nasa. http://earthobservatory.nasa.gov/Features/MeasuringVegetation/measuring_vegetation_2.php (4)Vogelmann, James E. (2002). Monitoring Northeastern United States Forest Conditions Using Landsat Data. USGS. |