Usgs Phen Drought Sheet

Brad Reed

1. Remote Sensing & Phenology
SWES 573
Spring 2004

2. Introduction
•Phenology (greek word phaino; to show or to appear) is the
study of periodic biological evebts as influenced by the
environment. It includes the timing of biological events,
particularly in response to climatic changes to the environment.
–Sprouting and flowering of plants; color changes of leaves in autumn,
insect hatches, hibernation, bird migrations
–Certain biological events, such as the time of the start of the growing
season, have a key role in changing land surface/atmosphere boundary
conditions such as surface roughness, albedo, humidity, etc. The
phenology of ecosystems and its connection to climate is a key to
understanding ongoing global change.
–Phenology has historically been studied as direct observations of the
timing of leaf opening, flowering, leaf fall and such events.
•Seasonality is a related term, referring to similar non-biological
events
–Spring break-up of ice; snow melt; dry & wet seasons

3. •Has been a quiet scientific objective, now
becoming more important.
•Only people in biology, ecology, and meteorology
using datasets on this.
•Ground truth observation of biological fluctuating
phenomena (how? - outmoded?)
•Today: modeling of phenologic events;
comparisons with meteorologic parameters, and
correlation attempts with global remote sensing
data sets - important for global change and
climate change analyses.
•A major value of for phenologic data is their
validation value for sesonality models. These
models have gained prominence in global climatic
change models to predict biosphere responses to
climatic parameter changes.

4. Introduction
•Phenology is an interdisciplinary environmental science
•It is integrative (climate, moisture)
•Global change science is stimulating, challenging, and
transforming the discipline of phenology.
•What are the and consequences of variation in timing of
life cycle events

5. Applications
•Biodiversity (resource availability)
•Agriculture, Range, Forestry, and Fisheries
–reproduction, productivity, pests, diseases
–adjust management strategies
•Human health (allergies, diseases, water quality,
etc..)
•Transportation (leaf fall, bird migrations)
•Tourism & recreation

6. Remote Sensing (1)
•The use of satellite imagery provides a unique
vantage point for observing seasonal dynamics of
the landscape.
•Key to understanding large area seasonal
phenomena
•Repeat observations provide a mechanism to
move from plant-specific to regional scale studies
of phenology
•Moderate resolution sensors preferred over high
spatial resolution sensors due to frequency of
coverage and radiometric rectification issues.

7. Remote Sensing (2)
•R.S. tracks integrated greenness of largely
heterogeneous 1-km pixels, rather than single
plants or dominant plant types.
•By analyzing the time-series vegetation index, a
set of algorithms derive phenology metrics such
as onset, end, and duration of growing season.
•The output of these metrics then may be analyzed
to produce products, such as temporal trends in
integrated NDVI values.

8. Remotely-sensed measures
•Greenness (vegetation indices)
•LAI/ FPAR products
•Albedo
•Land surface temperature
•Fires
•Photosynthesis, NPP
•Evapotranspiration

9. The USGS EROS Data Center has developed a data set of
seasonal metrics derived from multitemporal Advanced Very
High Resolution Radiometer (AVHRR) satellite sensor
Normalized Difference Vegetation Index (NDVI) observations.
(http://edc2.usgs.gov/phenological/)

11. QuickTimeª and a
GIF decompressor
are needed to see this picture.
Figure 1 is an animated loop of the NDVI observed over Colorado from
1990 through 1996.
http://geochange.er.usgs.gov/sw/impacts/biology/Phenological-CO/

12. Land characterization from AVHRR
•Trends in the seasonal characteristics become more apparent when analysis is
stratified according to land cover type.
•The post-classification refinement is performed with the aid of digital elevation,
ecoregions data and a collection of other land cover/vegetation reference data
(Brown and others 1993).
•The interpretation is based on extensive use of computer-assisted image
processing tools (Brown and others, in press); however, the classification process
is not completely automated and resembles a traditional manual image
interpretation technique. One hundred fifty-nine seasonal land cover regions
resulted from the analysis of the conterminous United States. The seasonal land
cover regions were then cross-referenced to several different classification
schemes, including an adaptation of the USGS/Anderson Level II scheme
(Anderson and others 1976) (see Figure 4).

14. •Figure shows smoothed time-series NDVI data for representative pixels of several
land cover types in Colorado observed on a biweekly basis from 1990 to 1996.
•Between-year variability of the cropland pixel is apparent, with the low values in
1991 and 1994. The shrubland pixel also shows significant variability, with low
peak values in 1990 and 1991.
•The grassland pixel shows relative consistency in the amplitude of the NDVI
values, but exhibits differences in the length of the growing season. The length of
the growing season in 1992 appears to be significantly longer than that illustrated
in 1994.
•The coniferous forest curve exhibits a marked consistency, both in amplitude and
in length of growing season. While the coniferous forest is green throughout the
year, background effects (such as snow cover) affect the seasonal NDVI profile.

15. •Figures below show images depicting four of the seasonal characteristics,
including the time of the start of the growing season, time of maximum
NDVI, duration of growing season, and time-integrated NDVI over a five
year period from 1991 to 1995. Since the calculation of the seasonal
characteristics requires data that both precede and follow the NDVI data
for a single year, 1990 and 1996 are not shown.

16. •The time of the onset of the growing season in Colorado for 1991-1995 is depicted
in Figure 6. Features that regularly stand out are the late onset of growing season
for the coniferous forests, especially in 1995 and the croplands in northeastern
Colorado. Shrublands and grasslands have an earlier onset of greenness. The bar
chart in Figure 10 illustrates the interannual variability in the time of the start of the
growing season over the five years, but does not exhibit any noticeable trend of
earlier or later starts to the growing season.

17. •Key phenological variables
–Start of the growing season
–End of the growing season
–Growing season length
•The integrated phenology of the pixel is
usually addressed in terms of a more general
statement such as SOS, EOS, rather than
first-leaf, etc..
–Phenology of a mosaic of vegetation types

18. Phenological Metrics
•Deriving metrics that describe the phenology (seasonality) of vegetation growth is
key to understanding when changes in the land surface/atmosphere boundary
layer take place.
•Time-series NDVI data track the greenup and senescence cycle of vegetation well
(Figure). The USGS has developed an algorithm to extract key phenological
phenomena from this time-series curve. Our approach is to utilize a delayed
moving average (DMA) as a comparison to the smoothed NDVI time-series. NDVI
data values are compared to the average of the previous (user-defined) n NDVI
observations to identify departures from an established trend (Reed and others,
1994). This departure is labeled as the start of the growing season (SOS).
•The end of the growing season (EOS) is calculated in a similar manner, but with
the moving average running in the opposite direction. Duration of growing season
is the difference between the time of EOS and SOS. The peak of the growing
season is simply the time of the maximum NDVI. Several other metrics can then be
derived, including the rate of greenup (slope from SOS to peak), rate of
senescence (slope from peak to EOS), and total integrated NDVI (area under the
curve). The figure below summarizes graphically these metrics.

19. Deceleration of photosynthesisRate of senescence
Acceleration of photosynthesisRate of greenup
Photosynthetic activity in growing
season
Seasonally integrated NDVI
Maximum level of photosynthetic
activity
Maximum NDVI
Level of photosynthetic activity at
EOS
NDVI at end of growing season
Level of photosynthetic activity at
SOS
NDVI at start of growing season
Time of maximum photosynthesisTime of Maximum Greenness - Julian
Day
Duration of photosynthetic activityDuration of Growing Season
Cessation of measurable
photosynthesis
Time of End of Season (EOS) Julian
Day
Beginning of measurable
photosynthesis
Time of Start of Season (SOS) Julian
Day

P h e n o lo g ic a l
In te rp re ta tio n

P h e n o lo g ic a l Me tric
The full set of metrics and their phenological interpretation are shown above. Note that the
phenological interpretation is not an absolute value of photosynthesis, but the phenological metrics
are surrogates for such values.

20. gzip.file
131 Mb
readmeCessation of
measurable
photosynthesis
(1989-2001)
End of Season
Date
gzip.file
70 Mb
readmeLevel of
photosynthetic
activity at SOS
(1989-2001)
Start of Season
NDVI
gzip.file
129 Mb
readmeBeginning of
measurable
photosynthesis
(1989-2001)
Start of Season
Date
gzip.file
83 Mb
readmeSimulated Net
primary
production
(1989-2001)
Seasonal
Integrated NDVI

Im a g e File

Mo re
In fo
D e s c rip ti
o n

Me tric

N a m e
Metric Data

21. 3 types of phenology approaches
with R.S.
•Threshold-based Phenology
–E.g., NDVI>0.1 - SOS achieved
–Seasonal midpoint NDVI (SMN) - uses midpoint
between minimum and maximum NDVI
–10% distance between max and min.--SOS
•Inflection point Phenology
–2 inflection points, width, peak
–Greenup, maturity, senescence, and dormancy (Zhang et
al., 2003)
•Curve derivative Phenology

22. Sensors & Data Smoothing
•Composited data are inherently affected by a number of
phenomena including
–cloud contamination,
–atmospheric perturbations,
–variable illumination and viewing geometry (sun angle and sensor view
angle)
•In the case of NDVI data, these factors usually reduce the values
and thus, compositing could be accomplished using the maximum
value over a specified time period: usually a week, ten days, or two
weeks.
•The maximum value compositing has been shown to increases
data quality, however, bidirectional (mainly view angle) have been
found over partially vegetated surfaces.

23. Sensors & Data Smoothing
•In the MODIS era, new problems arise
–In general, as better sensors and algorithms are developed, new flaws
that could not be seen in previous datasets, appear (more flaws but
better form in the data)
–Atmosphere corrected data allows angular variations to become
pronounced.
–The reflectance product may over-correct atmosphere influences
resulting in higher NDVI values.
–Clouds may cause both false increases in the EVI and false decreases
in both NDVI & EVI.
–Over-corrected, inland water bodies may become “green”, particularly
with NDVI.
•Role of BRDF, role of modeled data.

24. Sensors & Data Smoothing
•While maximum value compositing increases data quality, further
processing-- in effect a smoothing-- of the temporal NDVI signal
must be performed to facilitate some time-series analyses.
•The smoothing algorithm must serve as a rough interpolation
between observations and, in order to remove the effects of the
remaining NDVI-reducing phenomena, upwardly bias the results.

25. Temporal VI smoothing methods
•Best Index Slope Extraction (BISE, Viovy et al.,
1992)
•Compound mean and median filters
•Splines
•Weighted least-squares approach (Swets et al.,
1999)
–Eliminates some of the timeshifts caused by over
generalization of the signal by giving more weight to
locally (in time domain) high values.
•Smoothing algorithm should retain key temporal
features without over generalization, eliminate
spurious downward spikes in the VI, and retain
sustained temporary declines in VI that are
representative of declines in vegetation condition.

26. Example of compound median smoother
•The “smoother” iteratively processes the time-series by applying median filters of various
widths, then applies a "re- roughing" by reintroducing the original NDVI time-series into the
process.
•The upward bias, or NDVI "peak-catching", is applied by reintroducing unsmoothed NDVI
values that are greater than the smoothed values.
•Further work is being conducted that investigates other smoothing approaches, such as
Fourier analysis similar to that used in the FASIR adjustments (Sellers and others 1994) and
other statistical approaches.
•In the Figure, a three-year time series of NDVI is illustrated in the solid line, while the result of
the compound median smoother are shown as a dashed line. The smoother eliminates cloud
contamination (illustrated by the extremely low NDVI value in Year 2), as well as NDVI
reducing perturbations (illustrated during the greenup during Year 1).

27. Seasonal characteristics
•Once the database is smoothed of temporal discontinuities, methods can be applied to extract
a suite of seasonal characteristics from the time-series data set.
•Some of the more important seasonal characteristics that are needed are the time of the start
of the growing season, the time of maximum photosynthetic activity, and the duration of the
growing season.
•Reed and colleagues (1994) developed a methodology to derive a set of 12 seasonal
characteristics from the smoothed NDVI time series that summarizes characteristics of
ecosystem dynamics.
–The methodology involves applying a moving average filter to the time series, which essentially creates
a new time series with a time lag.
–The moving average time-series (MATS) then can serve as a predicted NDVI based on the previous n
(user-defined) observations. When the actual (smoothed) values are greater than the value predicted
by the MATS, then a trend change (start of growing season) is occurring.
–The end of the growing season can be found similarly and the duration can be calculated as the
difference between the two. Other seasonal characteristics such as the time of maximum NDVI and the
time-integrated NDVI (using the value of NDVI at start of season as a baseline) are also important
surrogate measure of ecosystem characteristics.

29. Smoothing
•The input data to extract phenological metrics is time-series advanced very high
resolution radiometer (AVHRR) normalized difference vegetation index (NDVI)
data. NDVI data may be affected by a number of phenomena that contaminate the
signal, including clouds, atmospheric perturbations, and variable illumination and
viewing geometry. Each of these phenomena reduce the NDVI.
•To reduce the contamination of the NDVI signal, we develop a weighted least-
squares linear regression approach to temporally smooth the data (Swets, 1999).
This approach uses a moving temporal window to calculate a regression line. The
window is moved one period at a time, resulting in a family of regression lines
associated with each point; this family of lines is then averaged at each point and
interpolated between points to provide a continuous temporal NDVI signal.
•Also, since the factors that cause contamination usually reduce the NDVI values,
we apply a weighting factor that favors peak points over sloping or valley points. A
final operation assures that all peak NDVI values are retained. The resulting
relationship between the smoothed curve and the original data is statistically
based. The smoothed data may be used to improve applications involving the
analysis of time-series NDVI data, such as land cover classification, seasonal
vegetation characterization, and vegetation monitoring.

30. Smoothing
Swets, D.L., B.C. Reed, J.R. Rowland, S.E. Marko, 1999. A weighted least-squares
approach to temporal smoothing of NDVI. In 1999 ASPRS Annual Conference, From Image
to Information, Portland, Oregon, May 17-21, 1999, Proceedings: Bethesda, Maryland,
American Society for Photogrammetry and Remote Sensing, CD-ROM, 1 disc.

31. Problems
•Ground-based observations of phenology are
subject to inaccuracies due to reliance of
observation skills of the observers & scale.
•Satellite data is confounded by a number of
factors, but it is an objective measure of
environmental dynamics
•One of the difficulties with RS-phenology is
creating an algorithm to handle the wide variety
of real-time -series curves, rather than a model
curves.
–Some biomes have no distinct seasonal signal
(evergreen, desert shrub), then there are multiple
growing curves.

32. References
• Reed, B.C., and E. Bartels, 1999. Phenological Characterization (abs.) in Proceedings of Pecora 14 Conference, Denver,
Colorado, December 1999.
• Schwartz, M.D. and B.C. Reed, 1999. Surface phenology and satellite sensor-derived onset of greenness: an initial comparison.
International Journal of Remote Sensing, Vol. 20, No. 17, pp. 3451-3457.
• Swets, D.L., B.C. Reed, J.R. Rowland, S.E. Marko, 1999. A weighted least-squares approach to temporal smoothing of NDVI. In
1999 ASPRS Annual Conference, From Image to Information, Portland, Oregon, May 17-21, 1999, Proceedings: Bethesda,
Maryland, American Society for Photogrammetry and Remote Sensing, CD-ROM, 1 disc.
• Reed, B.C., 1998. Derivation of phenological metrics. Proceedings of the Northern Great Plains Regional Workshop on Climate
Change and Climate Variability. pp. 47-56.
• Yang, L., B.K. Wylie, L.L. Tieszen, B.C. Reed, 1998. An analysis of relationships among climate forcing and time-integrated
NDVI of grasslands over the U.S. Northern and Central Great Plains. Remote Sensing of Environment, 65: 25-37.
• Reed, B.C., 1997. Applications of the U.S. Geological Survey's global land cover product. Acta Astronautica, Vol. 41, Nos. 4-10,
pp. 671-680.
• Reed, B.C. and K. Sayler, 1997. A method for deriving phenological metrics from satellite data, Colorado 1991-1995. Impact of
Climate Change and Land Use in the Southwestern United States, an electronic workshop.
( http://geochange.er.usgs.gov/sw/impacts/biology/Phenological-CO/)
• Reed, B.C. and Yang, L. 1997. Seasonal Vegetation Characteristics of the United States. Geocarto International 12(2): 65-71.
• Reed, B.C., J.F. Brown, D. VanderZee, T.L. Loveland, J.W. Merchant, D.O. Ohlen, 1994. "Measuring phenological variability
from satellite imagery," Journal of Vegetation Science 5: 703-714.

33. North side of SF Peaks
Between SF Peaks and South
Rim, GC
North Rim, GC
South Rim, GC

35. Rodeo_Chedeski Fire
-0.4
-0.2
0
0.2
0.4
0.6
0.8
2/18/004/18/006/18/008/18/00
10/18/0012/18/00
2/18/014/18/016/18/018/18/01
10/18/0112/18/01
2/18/024/18/026/18/028/18/02
10/18/0212/18/02
2/18/034/18/036/18/038/18/03
10/18/0312/18/03
DOY
VI/ Reflectances
1_km_NDVI
1_km_EVI
1_km_Red
1_km_NIR
1_km_MIR
stress index
Stress index=
(NIR-MIR)/(NIR+MIR)

36. Rodeo_Chedeski Fire
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
2/18/004/18/006/18/008/18/00
10/18/0012/18/00
2/18/014/18/016/18/018/18/01
10/18/0112/18/01
2/18/024/18/026/18/028/18/02
10/18/0212/18/02
2/18/034/18/036/18/038/18/03
10/18/0312/18/03
DOY
VI/ Reflectances
1_km_EVI
1_km_Red
1_km_NIR
1_km_MIR
1_km_Blue

37. Rodeo-fire control (west of fire)
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
2/18/004/18/006/18/008/18/00
10/18/0012/18/00
2/18/014/18/016/18/018/18/01
10/18/0112/18/01
2/18/024/18/026/18/028/18/02
10/18/0212/18/02
2/18/034/18/036/18/038/18/03
10/18/0312/18/03
DOY
VI/ reflectances
1_km_NDVI
1_km_EVI
1_km_Red
1_km_NIR
1_km_Blue
1_km_MIR
moisture index

38. Mt Lemmon MODIS
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
2/18/004/18/006/18/008/18/00
10/18/0012/18/00
2/18/014/18/016/18/018/18/01
10/18/0112/18/01
2/18/024/18/026/18/028/18/02
10/18/0212/18/02
2/18/034/18/036/18/038/18/03
10/18/0312/18/03
DOY
VI
1_km_NDVI(3x3)
1_km_EVI (3x3)
1_km_NDVI (9x9)
1_km_EVI (9x9)
1_km_MIR
1_km_Red
1_km_NIR
Moisture index

39. Droughts
•Droughts are normal recurring climatic phenomena that vary
in space, time, and intensity.
–They may affect people and the landscape at local scales for short periods
or cover broad regions and have impacts that are felt for years.
–The spatial and temporal variability and multiple impacts of droughts
provide challenges for mapping and monitoring on all scales.
–A team of researchers from the USGS EROS Data Center, the National
Drought Mitigation Center (University of Nebraska) and the High Plains
Regional Climate Center are developing methods for regional-scale
mapping and monitoring of drought conditions for the conterminous U.S.
–The goal is to deliver timely geo-referenced information about areas
where the vegetation is impacted by drought.

40. •We are integrating
information provided by
satellite-derived phenology
metrics (top) and climate-
based drought indicators
(bottom) to produce a timely
and spatially detailed
drought monitoring product.

41. Phenology and Drought Monitoring Projects
Phenology:
Bradley Reed
[email protected]
http://edc2.usgs.gov/phenological
Drought Monitoring:
Jesslyn Brown
[email protected]
http://edc2.usgs.gov/phenological/drought/
Phenology is the study of the timing of biological events, particularly in response to climatic changes to the environment. Certain biological events, such as the time of the
start of the growing season, have a key role in changing land surface/atmosphere boundary conditions such as surface roughness, albedo, humidity, etc. The phenology of
ecosystems and its connection to climate is a key to understanding ongoing global change.
The use of satellite imagery provides a unique vantage point for observing seasonal dynamics of the landscape. The USGS EROS Data Center has developed a data set of
seasonal metrics derived from multitemporal Advanced Very High Resolution Radiometer (AVHRR) satellite sensor Normalized Difference Vegetation Index (NDVI)
observations for the conterminous U.S. By analyzing the time-series vegetation index (Fig. 1), a set of algorithms derive phenology metrics such as onset, end, and duration of
growing season. The output of these metrics then may be analyzed to produce products, such as temporal trends in integrated NDVI values (Fig. 2)
Droughts are normal recurring climatic phenomena that vary in space, time, and intensity. They may affect people and the landscape at local scales for short periods or cover
broad regions and have impacts that are felt for years. The spatial and temporal variability and multiple impacts of droughts provide challenges for mapping and monitoring on
all scales. A team of researchers from the USGS EROS Data Center, the National Drought Mitigation Center (University of Nebraska) and the High Plains Regional Climate
Center are developing methods for regional-scale mapping and monitoring of drought conditions for the conterminous U.S. The goal is to deliver timely geo-referenced
information about areas where the vegetation is impacted by drought.
We are integrating information provided by satellite-derived phenology metrics (Fig. 3) and climate-based drought indicators (Fig. 4) to produce a timely and spatially detailed
drought monitoring product. Research and methods for Drought Monitoring are developed in tandem with Phenological Characterization.
Figure 1
Figure 2
Figure 3
Figure 4