Vegetation Responses to Seasonal Climatic Variability in Blacksburg, Va

By Logan Sleezer

ES775: Advanced Image Processing

May 2017


Indroduction

Blacksburg, home to Virginia Polytechnic and State University, is a town of 42,620 people as of the 2010 US Census (US Census Bureau, 2010), located in western Virginia. Sitting at an elevation of around 640 meters above sea level (US Climate Data, 2017), Blacksburg represents an interesting climatic transition zone between the Blue Ridge mountains just to the north and west and the lowlands of the Atlantic Coastal Plain to the east. The Koppen climate classification system characterizes Blacksburgs climate as humid continental with a cool summer (Dfb) according to 30-year normals (NCDC, 2017). However, climatic averages cannot tell the whole story as there exists a great deal of seasonal variability in climate in this region from year-to-year. This seasonal variability has implications for landcover patterns, especially seasonal vegetation patterns that can be monitored using satellite imagery.

Methods

Using climatic data from the National Climatic Data Center (NCDC), I identified months and years that represented the range of weather possible in the highly variable autumn and late winter/early spring time periods in Blacksburg. After identifying candidate years that best showed the range of conditions that were possible in these two seasons, I conducted a search for Landsat satellite imagery taken during these time periods using USGS Earth Explorer. After finding suitable, nearly cloud-free imagery for comparison between years, the Landsat multiband image data was downloaded and Idrisi Selva software was used to construct NDVI images to aid in comparison of vegetative characteristics between years. All imagery used in this study was derived from Landsat 4/5 multiband scenes downloaded from Earth Explorer (USGS, 2017). Each NDVI image was also reclassified using Idrisi, with the goal of creating categories to quantitatively compare active vegetation land coverage between years and between seasons. The large Landsat tiles were clipped to a smaller study area that includes the city of Blacksburg near the center of the scene, but also the New River, which runs south to north across the western portion of the scene. The portion of the New River shown in this scene runs through the Blue Ridge Mountains and includes two major reservoirs, Bluestone Reservoir and Claytor Reservoir, which represent the northern and southern extent of the study area respectively.

Results

The first image comparison shows the study area in late October during the consecutive years of 1988 and 1989. The natural color and NDVI image comparisons below show that vegetation, especially the deciduous forest dominated ridges of the Blue Ridge Mountains, have begun to change leaf colors and go dormant in the 1989 image, while remaining active and green in the 1988 image. NDVI (Normalized Difference Vegetation Index) is an image processing method utilizing the ratio of band 3 (red) to band 4 (infrared) to create an image that isolates highly active vegetation due to the highly distinctive pattern of high infrared reflectance and low red reflectance of green plants. Areas with high NDVI values (indicating high levels of vegetation activity) appear dark green while less active areas appear lighter green to yellow/orange for the lowest vegetation activity.

Natural Color

before
after

NDVI

before
after
In general, there are three major factors that affect timing of leaf die-off in the fall. These factors are photoperiod, moisture/precipitation, and temperature. Because these images are only three days apart in consecutive years, the effect of photoperiod on vegetative differences in these two scenes should be negligable. However, weather patterns were drastically different in 1988 and 1989 in Blacksburg. Surprisingly, 1988 was both drier and colder than 1989 as indicated by the table below, but still had more active vegetation in October. While the actual cause of this strange phenomenon is a mystery, abnormally heavy rains in September and October of 1989 could have contributed to early leaf die-off. In a similar study, Aber et al. (2002) found that decreases in NDVI values for forests may lag a year or two behind periods of drought because of the dependence of tree growth on food storage from previous years. This lag effect may be a more likely cause of the early leaf die-off in the warm, wet fall season of 1989 because of the delayed effects of the comparable drought conditions in 1988.

The second image comparison shows the study area in mid-March during the consecutive years of 1999 and 2000. The most obvious difference between the scene in 1999 and in 2000 is the snow cover in 1999, showing up light blue on the natural color image, but absent in 2000. The NDVI image comparison shows the expected dearth of active vegetation in 1999 compared to the same time in 2000, due to the harsher late winter in 1999.

Natural Color

before
after

NDVI

before
after
Heavy March snowfall in 1999 was well above average in this month when considering 30 year climatic normals as shown in the table below. The year of 2000 was much more reflective of a normal snowfall pattern in Blacksburg, with January being the month with the highest snowfall total and a more mild late winter.

Reclassification of NDVI images allowed a means of comparison between the two October scenes and two March scenes in terms of the total area of the study site covered by highly active vegetation. For the purposes of this study, NDVI values greater than 0.5 were considered areas of highly active vegetation. The scene with the highest area covered by highly active vegetation was the October 1988 scene at approximately 4322 square kilometers. The March 2000 scene had the second greatest highly active vegetation area at 802 square kilometers, followed by October 1989 at 649 square kilometers, and March 1999 at 482 square kilometers. The image comparisons below are reclassified NDVI images compared between the same season (October and March) in consecutive years. Categories 3 and 4 in the images below are those that were included in the calculation of total areas covered with highly active vegetation.

Reclassified NDVI Images

before
after
before
after

Conclusions

Seasonal weather patterns are highly variable in the Blacksburg area and can have a profound and sometimes surprising effect on landcover characteristics, especially activity of vegetation. Comparing fall and early winter variability, land coverage by highly active vegetation was most variable in the fall, as the October 1988 scene had over 6 times the highly active vegetation coverage as the October 1989 scene. Late winter variability was less, but still significant in this study as the highly active vegetation coverage in the March 2000 scene was nearly twice the coverage seen in March 1999.

References

Aber, J.S., Wallace, J.W., and Nowak, M.C., 2002. Response of Forest to Climatic Events and Human Management at Fort Leavenworth, Kansas. Current Research in Earth Sciences, Bulletin 248, part 1, p. 1-24.

NCDC, 2017. Data Tools: 1981-2010 Normals. https://www.ncdc.noaa.gov/cdo-web/datatools/normals. Date Accessed: 05/01/2017.

US Census Bureau, 2010. QuickFacks: Blacksburg town, Virginia. https://www.census.gov/quickfacts/table/PST045215/5107784. Date Accessed: 05/01/2017.

US Climate Data, 2017. Map of Blacksburg - Virginia. http://www.usclimatedata.com/map.php?location=USVA0068. Date Accessed: 05/01/2017.

USGS, 2017. EarthExplorer. https://earthexplorer.usgs.gov/. Date Accessed: 04/25/2017.