5

My study will involve Landsat OLI, ETM+ and TM scenes for forest biomass change assessment (Time Series Analysis) using different vegetation indices and textures. According to the official site https://landsat.usgs.gov/landsat-collections " Landsat scenes with the highest available data quality are placed into Tier 1 and are considered suitable for time-series analysis. Tier 1 includes Level-1 Precision and Terrain (L1TP) corrected data that have well-characterized radiometry and are inter-calibrated across the different Landsat instruments".

Keeping the above in view I take that I can use them straightforwardly assuming that all three OLI, ETM+ and TM stand at the same line and can be used interchangeably since to me it does not make any sense to do any further calibration when these Surface Reflectance Products have already been pre-processed and made ready for respective use and carry well characterized radiometry across the different Landsat instruments. Kindly make me clear please if I stand correct or still I would need any further correction or adjustment of my data.

7
+50

The three sensors are all slightly different. However the OLI/TIRs setup is a marked departure from the TM/ETM+ sensors. The changes are succintly summarised by Li et al. 2013 as the:

replacing of whisk-broom scanners with two separate push-broom OLI and TIRS scanners, an extended number of spectral bands (two additional bands provided) and narrower bandwidths.

Here is a link to the band locations which is a useful proxy for how much the sensors differ. It should confirm to you that they aren't the same!

Many studies comparing how the TOA/Surface Reflectance values differ between the sensors have been done, so it is worth doing a literature search on the particular indices you're interested in. This is a good example of a recent study comparing ETM+ and OLI, which includes formulas to standardise them.

It is entirely possible that the differences will not be a problem for your study. It really depends on its nature. For example if you're attempting to classify distinct forest types that are relatively homogeneous, then the variation in readings from the sensors and locations of the bands will probably not alter your results overly. If on the other hand you're looking for subtle differences in similar ecosystems then the reliability of these readings has a more significant impact on your results.

Clearly validation is the key here.

  • I agree with RoperMaps that it's best to look for solid proofs before you start performing them. Using these sources 'interchangeably' will not bring comparable result. For example, indices using NIR bands from each imageries will not have similar result. The [wavelength]( landsat.usgs.gov/what-are-band-designations-landsat-satellites) for NIR band in TM is 0.76-0.90µm, in ETM+ its 0.77-0.90µm and in OLI its 0.85 - 0.88µm. The TM and ETM+ both rescales and stores radiance values rescaled into 8-bit DN data, OLI does this in 16-bit reflectance data rather than radiance. – blu_sr Jul 2 '17 at 12:56
3

Although the band specs are different for different Landsat sensors your answer depends on the product you're planning to use. The part you have copied and mentioned the "well-characterized radiometry and are inter-calibrated across the different Landsat instruments" is for Landsat Collection 1 Tier 1 data. The aim of producing Collection-1 and not processing pre-collection data is to ensure a quality standart in order to support time series analysis and provide an easier to use data for customers. You can check https://landsat.usgs.gov/landsat-collections for detailed information and please also check the short videos. Also its mentioned in Landsat Collections - 2016 as:

During Collection 1 processing, each Landsat scene is assigned to a “Tier”. All Landsat data are cross-calibrated (regardless of sensor) across the full collection.

Most of the studies based on the continuity of Landsat sensors, also mentioned by @RoperMaps, were done before 2016 which is performed by using pre-collection L1T data equivalent to Collection-1 L1TP. The inter-calibration across different Landsat sensors is mentioned with Collection-1 data that are produced after 2016. In addition I haven't come across a study of sensor comparison with Collection-1 data.

So, if you are using Collection-1 Tier-1 Surface Reflectance data you don't need to perform any adjustments or conversions in order to use images acquired by different Landsat sensors.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.