1. Using Remote Sensing to Monitor Plant Phenology Response to Rain Events in the Santa Catalina Mountains
Katheryn Landau
Arizona Remote Sensing Center
Mentors: Willem J.D. van Leeuwen, Theresa Crimmins, and Michael Crimmins
Arizona Space Grant Consortium Symposium
University of Arizona
17th April 2010
Desert Scrubland - Riparian Scrub – Oak-Pine Forest – Pine Woodland
Abstract
It is important to monitor phenological changes to better understand plant response to climate change as the IPCC has stated that climate change has a discernible impact on plant populations. This study assesses the use of satellite data to monitor changes in the onset of greenness and total plant productivity for dry and wet years for a sky island in the Southwest. Satellite based normalized difference vegetation index (NDVI) was collected for a 10-year period ( ) across a 5-mile, 4,158 ft elevation gradient in the Santa Catalina Mountains near Tucson, Arizona, USA. Results show a strong relationship between greater rainfall, earlier onset of greenness, and total productivity. Satellite data can be used to identify shifts in the timing of plant activity and allow for greater understanding of the long-term impact predicted decreased precipitation will have on plant phenology.
Photo Acknowledgement: David Bertelson

2. Why Study Vegetation Phenology?
Phenology – study of plant life cycle events
Climate change has been shown to have a discernible impact on plant populations
Studies have related ground observations to climate - Crimmins et al 2008, 2009
Remote sensing allows for regional observation of plant response
Climate Change – IPCC reports
Crimmins et al – want to see if remote sensing can be integrated with ground observations to see the same trends/impacts
Satellites - data can be used for to observe identify shifts in the timing of plant activity and allow for greater understanding of the long-term impact predicted decreased precipitation will have on plant phenology.

3. Study Objectives
Explore the relationship between precipitation and vegetation response
Derive phenological metrics from satellite time series Normalized Difference Vegetation Index (NDVI) data for a range of vegetation communities
Assess applicability of satellite data to monitor changes in the start of season (SOS) and total plant productivity for dry (2002) and wet (2006) years
NDVI – signature of plant productivity

4. Study Area Mesic Semi-Arid Pine Forest Oak-Pine Woodland Oak Woodland
Mile 5
Desert Scrub
Mile 4
Scrub Grassland
Mile 3
Mile 2
5 mile trail in the Santa Catalina Mountains
Sky Island
Mountain islands of forests isolated by intervening valleys of grassland or desert
Several biotic communities are represented along the route to Mt. Kimball
Desert scrub at the base of the mountain range to pine forest at the peak.
4,158 ft elevation change
3100 ft to 7,258 ft
Semi-arid to mesic ecoregions
Bi-modal growing season – spring and summer
Riparian Scrub
Coronado
National
Forest
Tucson
Semi-Arid
Mile 1
Photo Acknowledgement: David Bertelson

5.
Dry
Wet
Precipitation – Seasonally averaged precipitation

6. Methods and Analysis
Process satellite derived NDVI and PRISM rainfall time series data for
Use noise reduction and curve fitting software to extract pheno-metrics: TIMESAT - Jönsson & Eklundh, 2004
Correlations between precipitation and pheno-metrics - SOS and productivity
10 year time series
Advanced Very High Resolution Radiometer (AVHRR)
Parameter-elevation Regressions on Independent Slopes Model (PRISM)

7. LI Start t, NDVI LI – Large Integral Time series Savitzky–Golay filter
Been shown to best fit the bi-modal growing seasons seen in the Southwest
Works well with quasi-periodic time series data
SOS –
Productivity (S Integral) – area between under the curve and the baseline

8. Start of Season Linear Regression R2 – coefficient of determination
0<R<1, 0 – cannot be predicted without error, 1 – can be predicted without error
Examples of upper and lower elevations
Mile 1 Vegetation – Desert Scrub
Mostly annual growths
Creosote
Less year-to-year variability in the spring SOS
No statistically significant relationship between rain events and when SOS occurs
5 - Pine forest, oak-pine woodlands
Mostly evergreens
Very slight relationship between rain events and the SOS, stronger with the spring growing season

9. Productivity Productivity – total photosynthetic activity Desert Scrub
Greater productivity occurring in the summer season, corresponds with overall higher precipitation
Not a statistically significant precipitation-productivity relationship at lower miles
Scrub Grassland
Stronger relationship in spring season

10. Study Findings High elevation more productive than lower elevation
Earlier SOS at lower elevation
Vegetation productivity has positive correlation with rain events
Precipitation does not clearly determine vegetation SOS or productivity
Satellite data can be used to monitor vegetation pheno-metric changes
High elevation more productive that lower elevation
Mile 3 showed overall greater productivity for Spring & Summer seasons
Earlier SOS at lower elevation
Spring & Summer SOS started earlier at lower elevation (mile 1) than at higher elevation (mile 5)
Vegetation productivity has positive correlation with rain events
Wetter years (2006) have overall greater productivity in
Dryer years (2002) have less overall productivity
Summer is the wetter season in the Santa Catalina Mountains
Summer consistently shows greater productivity
Precipitation does not clearly determine vegetation SOS or productivity
Precipitation is not the only factor in determining when a season will start or how much photosynthetic activity will occur
Satellite data can be used to monitor vegetation pheno-metric changes
Although cannot see the exact trends, satellites are still able to give regional trends
Able to see the extent of annual change
Using time series satellite data will still allow to see major shifts in green up activity and the extent of plant growth

11. Recommendations Analyze longer time series
Semi-arid environment has high annual precipitation variability
Incorporate more environmental variables
Temperature
Soil
Topography
Analyze longer time series
More years will give more data, increasing likelihood of observing trends
Semi-arid environment has high annual precipitation variability
Incorporate more environmental variables

12. Thank You!
Photo Acknowledgement: David Bertelson

13. Normalized Difference Vegetation Index
The generic normalized difference vegetation index (NDVI):
has provided a method of estimating net primary production over varying biome types (e.g. Lenney et al., 1996), identifying ecoregions (Ramsey et al., 1995), monitoring phenological patterns of the earth’s vegetative surface, and of assessing the length of the growing season and dry-down periods (Huete and Liu, 1994).