Pathfinder Workflows
The Land Change Monitoring, Assessment, and Projection (LCMAP) team is developing a series of pathfinder workflows that demonstrate how to use LCMAP science data products in land cover/land surface change research and applications.
The spatiotemporal resolution of LCMAP science products is unprecedented in land cover and land surface change science. LCMAP Conterminous United States (CONUS) version 1.3 contains 37 years of land cover and land surface change data (10 products), for 422 Landsat ARD tiles covering the entirety of CONUS, at 30 m spatial resolution (5000 x 5000 pixels/tile).
This equates to roughly 3,903,500,000,000 data points in the collection. Adding in LCMAP Hawaii version 1.0 data, there are over 3.94 trillion data points in the LCMAP science product collections.
The voluminous LCMAP archive offers new opportunities for land cover change research and applications, but also presents new challenges for accessing, processing, evaluating, analyzing, and synthesizing this incredible trove of information.
The LCMAP Pathfinder Project acts as a practical guide to the details and complexity of LCMAP science products. It provides direction on how to approach land cover analyses by setting up use case examples and demonstrating how to tackle them in automated, Python-based workflows.
Browse the tutorials listed below to learn how to apply the LCMAP science products to answer land change science questions with Python. Each workflow is meant to be fully transportable, meaning you can apply them to your own region of interest (ROI) for your desired time period. Each workflow includes step-by-step instructions on how to complete the analysis, followed by a fully wrapped function you can “plug and play” into your own workflow. Be sure to sign up for the LCMAP listserv to be notified as new pathfinder workflows, covering new LCMAP products, themes, and use case example questions, are released.
LCMAP Pathfinder Workflows:
Tracking Thematic Land Cover Change
Working with the Annual Land Cover Change Product to answer the question: What are the top four land cover conversions between 1985 and 2021 in the U.S. State of West Virginia?
This workflow demonstrates an approach to tracking annual thematic land cover change for a Land Change Monitoring, Assessment, and Projection (LCMAP) time series. The tutorial shows how to (1) identify LCMAP ARD tiles that intersect an area of interest (AOI), (2) open and mosaic multiple tiles, (3) clip to the AOI, and (4) export as GeoTIFF. The tutorial then shows how to calculate statistics on the amount of change by year and by land cover conversion, and how to visualize and export those statistics.
Visit the code repository for instructions on how to set up and execute the tutorial,or download the Jupyter Notebook (.html) output directly.
Examining Land Cover Confidence Margins
Analyzing the Land Cover Confidence Products to answer the question: What are the five most frequent primary and secondary land cover combinations with low land cover confidence margins in the Central Cornbelt Plains Level 3 Ecoregion?
This workflow demonstrates an approach to analyzing the usefulness of the Land Change Monitoring, Assessment, and Projection (LCMAP) Secondary Land Cover product by using the Primary and Secondary Land Cover Confidence products to calculate the margin between the model confidence in the Primary and Secondary Land Cover assignments. The tutorial shows how to (1) identify LCMAP Analysis Ready Data (ARD) tiles that intersect an area of interest (AOI), (2) open and mosaic multiple tiles, (3) clip to the AOI, and (4) export as GeoTIFF. The tutorial then shows how to calculate the land cover confidence margins, collect statistics on the two land cover combinations (Primary-Secondary) most frequently identified as having low confidence margins, and visualize and export those statistics.
Visit the code repository for instructions on how to set up and execute the tutorial, or download the Jupyter Notebook (.html) output directly.
Comparing NLCD and LCMAP Primary Land Cover
Analyzing the NLCD and LCMAP Land Cover Products to answer the question: What are the similarities and differences in land cover classification labels between LCMAP and NLCD in the Delaware River Basin?
This workflow demonstrates an approach to analyzing the similarities and differences between the Land Change Monitoring, Assessment, and Projection (LCMAP) Primary Land Cover product and the National Land Cover Database (NLCD) Land Cover product. The tutorial shows how to (1) directly access LCMAP and NLCD land cover product mosaics and clip to the boundaries of the Area of Interest (AOI), (2) perform a land cover crosswalk to map NLCD Anderson Level 2 classifications into the LCMAP classification legend, (3) visualize the results, (4) calculate a confusion matrix between the two products and generate overall agreement, and (5) create classified difference maps between the two products. The workflow then shows how to interactively visualize the spatial differences and export those statistics. The final section demonstrates how to automate the workflow to run through the analysis for other years.
Visit the code repository for instructions on how to set up and execute the tutorial, or download the Jupyter Notebook (.html) output directly.
Comparing NLCD and LCMAP Primary Land Cover with Morphological Open Analysis
Analyzing the NLCD and LCMAP Land Cover Products to answer the question: What are the similarities and differences in land cover classification labels between LCMAP and NLCD in the Delaware River Basin if we apply image morphology techniques to remove narrow linear features (i.e. rural roads) from the comparison?
This workflow demonstrates an approach analyzing the similarities and differences between NLCD and LCMAP in the Delaware River Basin in 2001. It builds on the first comparison (Comparing NLCD and LCMAP Primary Land Cover) by using image morphology to remove narrow linear features from both products in order to analyze other occurrences of disagreement. The tutorial shows how to (1) directly access LCMAP and NLCD land cover product mosaics and clip to the boundaries of the Area of Interest (AOI), (2) perform a land cover cross-walk to map NLCD Anderson Level II classifications into the LCMAP classification legend, (3) use image morphology to remove roads in the cross-walked NLCD dataset, (4) visualize the results, (5) calculate a confusion matrix between the two products and generate overall agreement, and (6) create classified difference maps between the two products. The workflow then shows how to interactively visualize the spatial differences and export those statistics. The final section demonstrates how to automate the workflow to run through the analysis for other years.
Visit the code repository for instructions on how to set up and execute the tutorial or download the Jupyter Notebook (.html) output directly.
Setup and Dependencies
It is recommended to use the environment manager Conda to set up a compatible Python environment. Download Conda for your OS here: https://www.anaconda.com/download/. Follow the instructions in the introduction to each pathfinder workflow to set up a Python environment on Linux, MacOS, or Windows.
If you prefer to not install Conda, the same setup and dependencies can be achieved with another package manager, such as pip.
Downloading Data
Each pathfinder workflow assumes that you have downloaded the data needed to execute the tutorial. Users can download and unzip the intersecting LCMAP tile bundles from USGS EarthExplorer.
Looking for an automated way to download LCMAP Data? Checkout the Getting Started with LCMAP in Python Direct Access of LCMAP Mosaics and Accessing LCMAP Data via the EarthExplorer Machine to Machine API tutorials.
Experiencing issues? Contact the USGS EROS Customer Services.