-
Notifications
You must be signed in to change notification settings - Fork 47
Home
This is the beginning of the SEPAL wiki. It's not much. But we'll be working on it...
SEPAL is a cloud computing-based platform for autonomous land monitoring using remotely-sensed data. It is a combination of Google Earth Engine and open source softwares ORFEO Toolbox, GDAL, R, R Studio Server, R Shiny Server, SNAP Toolkit, and OpenForis Geospatial Toolkit. It allows users to access powerful cloud-computing resources to query, access and process satellite data quickly and efficiently for creating advanced analyses.
SEPAL is currently in a BETA testing mode only. It is undergoing constant updates and improvements and has not reached a point of being considered a stable codebase. It works, of course...it's just not something we can call finished.
SEPAL is a project of the Forestry Department of the United Nations Food and Agriculture Organization (FAO) funded by the Government of Norway.
Login…
Login to SEPAL and give it a go!
The essentials…
There is no user manual as of yet…but the initial navigation should be relatively straightforward. However, before you get into it there are a few essential things you should know.
You are working in a secured cloud-based computing environment. In other words, you are not processing anything on your desktop but rather in a range of super-computers, or INSTANCES, located in a computing facility…somewhere. In the current case of SEPAL 2.0, these computers are part of the Amazon cloud. Specifically as of now you will be working on computers located in Ireland.
The image search, compositing, previewing and downloading is handled by Google Earth Engine. The Earth Engine is an amazing system and you can read more about it here: earthengine.google.com
SEPAL 2.0 works on the concept of ‘user budgets’. These budgets refer to a monthly US Dollar amount for processing capacity and storage capacity within the cloud-computing architecture. As BETA users of SEPAL, you have all been given 10 USD monthly for processing and storage. This may not seem like much…but the basic computer, or instance, costs less than 0.03 USD per hour to run! So you get quite a bit of processing for your ‘budget’. You can change the level of instance you are connected to and, depending on the power of the instance, the costs will increase more or less dramatically from the base instance. More on this later.
It should be noted that searching for data, compositing data and previewing the result of the searching and compositing do not count at all towards your monthly user budget!! You can do as much searching and testing of mosaicking as you want without charging your user account for processing time.
So…what to do first? Search!!
When you first login to SEPAL, you are met with 4 choices: Search, Browse, Process and Terminal. Since this is your first time in SEPAL, the Search capability is probably what interests you most.
Click on the Search icon.
You are now presented with some basic parameters to begin your search for data.
Step 1 is to create a mosaic. So...click on 'Mosaic'
The basics here are that you must supply a geographic area and a target date about which to search for data. SEPAL 2.0 includes search for data by country boundary or by a user-drawn polygon. You may specify a target date around which to search for satellite imagery. A target date is the day of the year for which you wish to make a satellite image composite or mosaic. For example…If you know when the best time of year is to get a largely cloud-free picture of your area of interest (AOI), we would suggest using this day as your target date. The default target date is the current date.
When target date and area of interest are selected…
Click on SEARCH…
SEPAL will zoom to your area of interest and populate the area with the LANDSAT data scene boundaries that intersect your AOI. Each scene area will show ‘0’ indicating that you have no data selected yet for each scene.
Next we ‘AUTO-SELECT’ Landsat data to fill the scene area in order to try and make the best image composite/mosaic we can for our ‘target date’ of interest…so…
Click on AUTO-SELECT
Here you are presented with some data selection options.
The slider bar…
Is the target date the most important parameter in your data selection? Or is a cloud-free mosaic the most important thing to obtain? The ‘slider bar’ weighting cloud-cover or target date as the most important parameter will help you automatically choose Landsat data based on that which is most important to you! The default setting is 50/50…which will return the most cloud-free mosaic with pixels from images as close to the target date as possible. You can adjust this to weight cloud-free-ness more importantly than target date or vice-versa.
Sensors…
Which Landsat sensor(s) data would you like to retrieve? By clicking the sensor indicators off or on (DEFAULT is ON) you can control the data returned to your query.
Date Range…
Can you make the mosaic you want for your AOI from only 1 year of satellite acquisitions? If you can, great! If you need MORE than 1 year to do it…for very cloudy areas, for example…then you can adjust the number of years to search for data on either side of the target date in order to make the best mosaic you can.
Min-Max Number of Scenes…
You can set these parameters to control the minimum or maximum number of Landsat scenes returned from your query. Generally speaking, the best mosaic you can make with the least amount of scenes is better than combining more scenes. These parameters are especially useful when dealing with L7-SLC off data, for example. L7-SLC-off data have scan gaps so are missing data even if they may not contain clouds or any other atmospheric contamination. It might be a good idea then to set the minimum number of scenes to > 1 in order to ensure your query returns. Likewise, to return the most cloud-free mosaic possible with, say, 3 Landsat scenes, set the slider bar to 100% cloud free and the maximum number of scenes to 3. Leave the MAX setting blank to remove the limit on the number of possible acquisitions returned. Mix and match the combinations as you choose!
Click on SUBMIT
You will see that your circles are now filled with numbers!! These numbers represent the acquisitions per scene pre-selected to provide a largely cloud-free mosaic with Landsat data that meet your selection criteria. Numbers are nice, but what you really want is data! So…
Click on PREVIEW
Your composite/mosaic is drawn on-screen…in the 3-band color combination you choose.
If you are HAPPY?
You can click on ‘Retrieve’ and your mosaic will be downloaded to your SEPAL 2.0 workspace!
If you are SAD?
You may auto-select again and change your parameters or you may click on individual scenes within your AOI and adjust the acquisitions manually.
That’s a lot of text for a simple introduction. So…I stop there with that bit.
What else can we do???
Processing
SEPAL 2.0 utility is expanded with user-contributed PROCESSING MODULES available through the ‘Processing’ button (e.g. the wrench).
Check out the kind of processing options available by clicking on ‘the wrench’…
Here we have access to:
RStudio - just general use of RStudio for anything you want to do with RStudio but with the processing power of the computing instance you select… Accuracy Assessment Design - a module to design a map accuracy assessment for raster or vector maps Accuracy Assessment Analysis - a module to analyze the results of a map accuracy assessment SAR Toolkit - a module for query and processing of Synthetic Aperture Radar (SAR) data GEO-processing (BETA) - a module to perform geo-processing on data you have in your workspace Visualize (BETA) - a very very basic module to view what you’ve processed
Each module starts an Amazon instance (basic) if one is not already running and processing time will be charged to your user account. If an Amazon instance is already started, then the modules will be run on the existing instance. It may take a few moments for a processing module to start if a new instance has to also be started simultaneously.
The Terminal
The terminal can be used to start instances, terminate instances and perform LINUX command-line functions on your workspace.
Start a Terminal by clicking on the symbol. You will be given the option to start an Amazon instance (with associated processing power and price per hour) or will be asked to join an existing session if one is available.
To terminate an instance go the terminal and type ‘exit’ at the command prompt. Then hit ‘ENTER’. You will be asked if you want to rejoin the session or ‘TERMINATE’ it. Press ’t’ to terminate and hit enter. The remote instance will shut down and you are FREE!!!
This is all I can include at the moment in the way of an introduction to the SEPAL platform. Please feel free to give it a try and e-mail me with comments or questions at [email protected]. Or…better yet…post your question or suggestion to our support community at http://www.openforis.org/support/. Then we can all learn from each other.
How to get started building data in SEPAL
Change Detection in SEPAL...thanks to the United States Forest Service and the USAID CARPE Program
How to perform a stratified area estimation in SEPAL
A GOFC/GOLD video on how to use SEPAL...
Another GOFC/GOLD video on stratified area estimation in SEPAL