Data Analysis of Satellite Images from scratch: A Briefing for Landsat 8+9 Project

I am so glad to announce that Landsat 9 images are finally on track! Since February 2022, the Landsat project "has opened its second eye" to witness our mother earth. It will provide us more frequent and comprehensive information of our mother earth.

But why I am so excited about Landsat 9? Why I said "open its second eye" instead of something like "implant the better eye"? Let us dive into the world of Landsat project together!

Introduction of Landsat Satellite Data

Landsat satellite programme started in the early 1980s, which is a joint NASA/USGS programme. The Earth Resources Technology Satellite was launched in 1972 and renamed Landsat 1. The recent satellite is Landsat 9 launching on 27 September 2021 and providing the open-resource data since 2022.

Landsat satellite programme provides the earth surface periodical detecting and monitoring. The latest operation satellites are Landsat 8 and Landsat 9. The co-existence of two similar satellites aims to reduce the Landsat data gap. The origin combination was Landsat 8 and Landsat 7. Each satellite revisits the same place every 16 days with sun-synchronous orbit, a 705 km altitude at the equator and crossing the equator at 10:11 am (+/- 15 minutes) Mean Local Time (MLT). The combination enables provide satellite images every 8 days. Then since Landsat 9 has launched, it replaces Landsat 7 and joins the same orbit as Landsat 8. Furthermore, they both are with a similar design (Figure 1) and so far radiometrically and geometrically better than earlier generation Landsats.

Narrowing down to the current operating Landsat, the detail of images as output is crucial for researchers and scientists working with land-cover management. The open resources such as Google Earth Engine or Open Data Cube with great cloud storage and GPU computing capacity are sufficient for beginners to manipulate, analyse and illustrate with satellite images. However, if you would like to apply more complicated algorithms, for example, satellite image fusion or some state-of-art geostatistical inference, the basic comprehension of the satellite images is required, c’est-à-dire, Landsat 8 + 9 here. Therefore, let’s dive into an entry-level explanation of Landsat 8 + 9 products.

Figure 1 Timeline of the Landsat program (Image resource: Global Media Studios, NASA)

Summary of the products of Landsat 8 + 9

It is a easy way to summarise the products of Landsat 8+9 from decomposing their product name. Figure 2 shows an example while downloading a Landsat 8+9’s image. Splitting each element by “_”, it would make a good picture for you to understand.

Figure 2 An explanation to a product of Landsat program

Sensor

The initial observatory stores the data from two major sensors, Operational Land Image (OLI) and Thermal Infrared Sensor (TIRS) in the Solid State Recorder (SSR). Then the

Ground Network (GN) receives the data at several station and forward them to EROS center. Afterward, the OLI and TIRS data are merged into single product. The bullet points below describe the characteristics of OLI and TIRS:

  • Operational Land Imager (OLI) : measure in the visible, near infrared, and short wave infrared portions of the spectrum. It produces 9 shortwave spectral bands with a 30 meter spatial resolution except the 15m panchromatic (Pan) band.
  • Thermal Infrared Sensor (TIRS) : measure of the land surface temperature in two thermal bands, storing at band 10 & 11 in Landsat 8 + 9’s product. The resolution of TIRS is 100m. Although it has a lower spatial resolution than the 60m ETM+ Band 6 compared with Landsat 7, the two thermal infrared bands advance the single-band thermal data in Landsat 7.

Processing correction level

The label of processing correction level represents usage of different Geometric Processing Subsystem (GPS). L1 represents the products are in the level-1, which deposit radiometrically and geometrically corrected images. Furthermore, The following abbreviation represents the Geometric Processing Subsystem (GPS). The GPS creates Level 1 geometrically corrected imagery, which can be Level 1 Geometric Systematic (L1GS), and the geometrically corrected products can be systematic terrain-corrected (L1GT) or precision terrain-corrected products (L1TP).

  • Level 1 Precision terrain-corrected (L1TP): The highest-level product among all, which is processed by using sufficient Ground Control Points (GCPs) and a digital elevation model (DEM).
  • Level 1 Systematic Terrain (L1GT): The second-level product, which is the systematic product and that make it possess consistent and sufficient locational accuracy. Therefore the terrain model is applicable to process L1GT’s correction.
  • Level 1 Geometric Systematic (L1GS): The lowest-level product which does not have sufficient location accuracy to apply terrain correction.

Worldwide Reference System

Landsat 8+9’s Observatory follows a sequence of fixed ground tracks defined by the worldwide reference system (WRS-2) instead of WRS. The first three number determines the path. The number adds up along with the range of longitude from small to large. The satellite would cycle back to the same path after 16 days, and hence the Landsat 8 + 9 would cycle the same path in a 8-day cycle; The last three number determine the row, the number adds up along with the range of latitude from large to small.

For example in Taiwan, the certain combination of path and row constructs the whole Taiwan with subtle overlap at the margin of square images. The illustration is made below using the QGIS.

Figure 3 A: 117043 (right middle high); B: 117044 (right middle low); C: 117045 (right bottom) D: 118043 (left middle); F: 117042 (right top); G: 118044 (left bottom); H: 118042(left top).

Data acquired & Data product generated

These two number set follow the calendar of common era (CE). It should be notice that the first set would fix to the image when it was shot, and the second set would possibly change according to the update. Otherwise it would be few days after data acquisition.

Collection & Level

The United States Geological Survey (USGS) has published a explicit table to explain the difference between collection and level. Here is the link:

Comparison of Collection 1 vs. Collection 2 / Level 1 vs. Level 2

Overall:

  1. level 2 adjusted by level 1
  2. collection 2 improved by collection 1

Collection Category (Tier)

Tier system in Landsat is the inventory structure for Landsat products to support rapid and easy identification of suitable scenes for certain usage. The categorisation is based on data quality and level of processing. All scenes in the Landsat archive are assigned to a Collection category. These data are claimed having well-characterised radiometric quality and cross-calibrating among the different Landsat sensors. Here is a bullet point from USGS EROS Archive for the meaning of

Tier 1 (T1) — Contains the highest quality Level-1 Terrain Precision (L1TP) data considered suitable for time-series analysis.

  • Accuracy evaluation is not required: the georegistration is consistent and within prescribed tolerances [<12m root mean square error (RMSE)].

Tier 2 (T2) — Contains L1TP scenes not meeting Tier 1 criteria and all Systematic Terrain (L1GT) and Geometric Systematic (L1GS) scenes

  • Accuracy evaluation is required: users interested in Tier 2 scenes can evaluate the RMSE and other properties to determine suitability for use in their applications and studies.

Real-Time (RT) (Landsat — Contains newly (without evaluation) acquired Landsat 8 scenes.

  • When definitive calibration information becomes available (approximately 14–16 days), these scenes are reprocessed, assigned to the appropriate Tier 1 or Tier 2 category, and removed from the RT Tier.

Reflection

I hope this article gives you some senses while you download a Landsat 8-9's image, and appreciate how magnificent the Landsat project did for us. I am not a professional and don't even have the background of geoscience or astronomy. The process of collecting information and writing this article contributes so much to my understanding. I am looking forward to the potential of this project. Let's leverage "the eye on our earth", nudge the sustainable land use and development, and make our earth a better place!

Reference

  1. Landsat 8 OLI (Operational Land Imager) and TIRS (Thermal Infrared Sensor) 15- to 30- meter multispectral data from Landsat 8. (2018). https://www.usgs.gov/centers/eros/science/usgs-eros-archive-landsat-archives-landsat-8-oli-operational-land-imager-and#overview
  2. Landsat 8 (L8) Data Users Handbook (2019). https://d9-wret.s3.us-west-2.amazonaws.com/assets/palladium/production/s3fs-public/atoms/files/LSDS-1574_L8_Data_Users_Handbook-v5.0.pdf

--

--

--

Programmer | Agriculturist | Environmentalist

Love podcasts or audiobooks? Learn on the go with our new app.

Recommended from Medium

What Job Postings Can Tell Us about Data Product Managers

To Start with K-mean Clustering & Real Usecase

Determining Physicians’ Drug Combination Preferences for Regions in Africa Using the Apriori…

Getting Started With Network Datasets

TU/e PhD graduate develops tool to track down suspects through images

How to create a percentage plot with ggplot2

SENIOR BACK-END DEVELOPER

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
AgriEnvCoder

AgriEnvCoder

Programmer | Agriculturist | Environmentalist

More from Medium

Analyze a statement using QFT

Voices from Ukraine Day XIII

In Memoriam — Leslie Trent

Use of Artificial Intelligence and Machine Learning in Food Technology space