1 C
Munich
Thursday, December 12, 2024

Biases in Measuring Vegetation Greenness with Satellite Imagery

Must read

“Vegetation greeness (often shorted to “greenness”) is a measurement of how much vegetation is in a specific area.

How is greenness measured?

In remote sensing, greenness is measured using indices like the Normalized Difference Vegetation Index (NDVI), which measures the difference between visible and near-infrared light reflected by vegetation. Higher NDVI values indicate healthier, more vigorous plant growth, while lower values suggest less vegetation or stressed plants.

Greenness values are often used as a measurement of plant productivity, ecosystem health, and changes in vegetation over time, often in response to environmental factors such as climate change, land use, or seasonal variations.

Remote sensing, greenness, and bias

Scientists can use remote sensing data from satellite imagery bands in order to determine the “greening” of an area. By comparing satellite imagery taken over time, changes in vegetation in a given area can be calculated.

Biases in Measuring Vegetation Greenness with Satellite ImageryBiases in Measuring Vegetation Greenness with Satellite Imagery
NDVI calculates the “greenness” of vegetation. Photo: Caitlin Dempsey.

NDVI is a number that ranges from -1 to 1, where higher values indicate more lush and healthy vegetation. Scientists often use the maximum NDVI value observed in a year to estimate how green an area is. However, this method relies heavily on the number of observations taken throughout the year.

Satellite imagery from long running Earth observation programs, like the Landsat series of satellites, are popular choices among scientists for running NDVI and greenness measurements.

In areas where fewer images are available, especially earlier in the Landsat record, there is a risk that the maximum NDVI value recorded might not fully capture the peak of the growing season. As more images become available in later years, the chances of capturing the peak greenness increase.

Results of the greenness bias study

A recently published study in the journal Ecography looked at the potential for this bias when comparing greenness over time in the European Alps. The researchers used historical Landsat data from 1984 to 2021 to investigate how the number of satellite observations has changed over time and how this change might affect estimates of vegetation greenness.

They found that in areas with a short growing season and fewer satellite observations, particularly at higher elevations where snow cover persists longer, the NDVI values were more likely to be underestimated in the early years of the Landsat record. As the number of observations increased over time, the chances of capturing the true peak of the growing season improved, leading to higher NDVI values and a potential overestimation of greening trends.

map is based on data from the Moderate Resolution Imaging Spectroradiometer (MODIS) on NASA’s Aqua satellite. The map shows the NDVI anomaly: it contrasts vegetation health from March 29 to April 5, 2016, relative to the long-term average from 2000–2015. Brown areas show where plant growth, or “greenness,” was below normal. Greens indicate vegetation that is more widespread or abundant than normal for the time of year. Grays depict areas where reliable data were not available, usually due to cloud cover. Source: NASAmap is based on data from the Moderate Resolution Imaging Spectroradiometer (MODIS) on NASA’s Aqua satellite. The map shows the NDVI anomaly: it contrasts vegetation health from March 29 to April 5, 2016, relative to the long-term average from 2000–2015. Brown areas show where plant growth, or “greenness,” was below normal. Greens indicate vegetation that is more widespread or abundant than normal for the time of year. Grays depict areas where reliable data were not available, usually due to cloud cover. Source: NASA
map is based on data from the Moderate Resolution Imaging Spectroradiometer (MODIS) on NASA’s Aqua satellite. The map shows the NDVI anomaly: it contrasts vegetation health from March 29 to April 5, 2016, relative to the long-term average from 2000–2015. Brown areas show where plant growth, or “greenness,” was below normal. Greens indicate vegetation that is more widespread or abundant than normal for the time of year. Grays depict areas where reliable data were not available, usually due to cloud cover. Source: NASA

For example, in high-altitude grasslands in the Alps, almost half of the observed increase in NDVI over the study period could be attributed to this bias, rather than an actual increase in vegetation. This finding is critical because it suggests that some of the reported greening trends in the Alps and similar regions might be exaggerated due to the increasing availability of satellite data over time.

Accounting for satellite imagery time series bias when calculating greenness

This study highlights the importance of considering changes in satellite observation frequency when interpreting vegetation greenness trends. In mountainous regions like the European Alps, where snow cover and short growing seasons complicate the measurement of peak greenness, the increasing number of satellite images available over time can lead to an overestimation of greening trends. 

The study

Bayle, A., Gascoin, S., Berner, L. T., & Choler, P. (2024). Landsat-based greening trends in alpine ecosystems are inflated by multidecadal increases in summer observations. Ecographyn/a(n/a), e07394. DOI: 10.1111/ecog.07394

Latest article