IEEE Geoscience and Remote Sensing Magazine - September 2015 - 55

frequent observation of forest cover, phenology and disturbance at 30m spatial resolution.
The predictability of changes in fine resolution synthetic images depends upon the capacity of MODIS to
detect these changes, particularly when they occur in
vegetation structure or stand composition or at sub-pixel ranges [7]. For instance, pixel brightness of fine resolution predictions can only be adjusted at coarse resolution scales. Consequently, our ability to identify change
events is still limited by the extent to which MODIS data
can be used to detect changes at sub-pixel range. Nonetheless, initial applications based on STARFM [15][18]
and its derivatives such as STAARCH [8], [32] have demonstrated the improved ability of this approach to map
changes in vegetation at landscape level scales. Most
forestry based applications have been implemented in
coniferous forest types, which are typical for large parts
of western North America. Further research will need to
investigate the potential of this technique, particularly
in the tropical regions of this world, where changes are
enormous but frequent cloud cover inhibits its timely

May 1 (121)

May 15 (135)

and accurate mapping with sufficient spatial and temporal detail.
B. Crop Monitoring
Crop progress and condition can be affected by climate
changes, local weather, environment changes and human
activities. Monitoring crop condition can help us to understand the impacts of these changes. Over the past three decades, both corn and soybean in the U.S. have been planted
increasingly earlier in the spring. Corn planting dates have
advanced by 10 days and soybean by 12 days from 1981 to
2005 [1]. This earlier planting has been accompanied by a
longer growing season, both of which contribute to yield
increases. In addition to the climate and weather changes,
crop phenology trends also reflect changing farm management strategies and development of new cold- and droughthardy genotypes [33]. High spatial and temporal resolution
information is required to map the changes at field scales.
Using the STARFM data fusion approach, Gao et al.
[20] built Landsat NDVI time-series at daily frequency
for central Iowa. Fig. 6 shows the NDVI trends computed

June 1 (152)

June 15 (166)

July 1 (182)
93°20 'W

N

August 1 (213)

August 15 (227)

Corn
Soybean
Other Hay
Developed
D. Forest
Grassland
93°25 'W

September 1 (244)
0.0

0.2

September 15 (258)
0.4

0.6

October 1 (274)
0.8

0

5

km

42°20 'N

42°20 'N

July 15 (196)

42°25 'N

42°25 'N

93°25 'W

93°20 'W

2011 Cropland Data Layer (CDL)

1.0

Figure 6. Daily Landsat NDVI (shown every half month) computed from MODIS-Landsat STARFM fusion of red and NIR band data from

May 1 to October 1, 2011 over central Iowa. The Cropland Data Layer (CDL) for the same area for 2011 is shown in the lower right corner
(modified from Gao et al., 2013).
september 2015

ieee Geoscience and remote sensing magazine

55
Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - September 2015
