add peakvalley notebook in repo instead of using fusets

tutorial/_toc.yml

@@ -56,9 +56,12 @@ parts:
     title: How to exploit data on Pangeo
   - file: part3/advanced_udf
     title: Custom algorithms - UDF (OpenEO) and ufunc (Xarray)
+  - file: part3/peak_valley
+    title: Advanced OpenEO - Peak Valley Case
   - file: part3/scaling_openeo
     title: Scaling with OpenEO
 
+
 - caption: Beyond the workshop
   chapters:
   - file: afterword/resources

tutorial/part2/advanced_workflows.md

@@ -10,6 +10,21 @@
 - What are openEO platform 'credits'
 - Learn about metadata generation in openEO
 
+## Algorithm types
+
+When writing an algorithm in openEO or Pangeo, the first thing to do is to determine the type of algorithm.
+Once you know the type, you can fall back to well-known examples as a template to implement your algorithm.
+
+Using the right method for your type of algorithm also ensures that it can be parallellized and executed 
+in the most efficient way. An algorithm that only requires a single observation for instance, does not need
+to bother with retrieving a full timeseries of observations.
+
+In part 2 of this tutorial, we'll cover a few common types:
+- single observation, single pixel
+- temporal analysis over a single pixel
+
+In part 3, we have an example of an analysis that requires multiple values in both space and time.
+
 
 ## Case 1 Rank composites 
 
@@ -20,6 +35,10 @@ Key takeaways:
 - Use array_apply to 'loop' over data
 - Create masks to selectively load data
 
+Algorithm type: 
+- Spatial neighborhood for cloud proximity score/masking
+- Per pixel temporal analysis for scoring
+
 https://github.com/Open-EO/openeo-community-examples/blob/main/python/RankComposites/rank_composites.ipynb
 
 ## Case 2 Machine learning training and inference
@@ -35,6 +54,8 @@ Key takeaways:
 - Transform a timeseries datacube into features for learning
 - Use ML processes directly from openEO
 
+Algorithm type: per-pixel timeseries analysis
+
 Dynamic landcover: https://docs.openeo.cloud/usecases/landcover/
 
 ## Case 3 ESA WorldWater
@@ -45,6 +66,11 @@ by DHI-GRAS.
 Key takeaways:
 - Use of complex, rule-based classification
 
+Algorithm type: 
+
+- temporal compositing into ARD
+- single observation, single pixel, analysis
+
 Note: the most recent code of this example is still under review in a pull request, so we include the relevant bits inline. 
 
 https://github.com/DHI-GRAS/worldwater-toolbox

tutorial/part3/advanced_udf.ipynb

@@ -7,7 +7,12 @@
     "\n",
     "\n",
     "The openEO user defined functions (UDF's), provide a way to run Pangeo code as part of a larger openEO workflow.\n",
-    "This allows you to get the best of both worlds.\n",
+    "This allows you to get the best of both worlds:\n",
+    "- You can use openEO to get access to a wide variety and full archives of EO data.\n",
+    "- Preprocessing to ARD such as Sentinel-1 backscatter is easily done in openEO\n",
+    "- For very fast interactive debugging of the core of your algorithm, local testing based on Pangeo is faster than basically any cloud based service\n",
+    "- Once your algorithm is ready, you can run it in the cloud without worries thanks to openEO\n",
+    "\n",
     "\n",
     "To understand user defined functions, it is recommended to start here:\n",
     "\n",
@@ -43,6 +48,8 @@
     "\n",
     "## UDF with basic XArray code: peak-valley detection\n",
     "\n",
+    "Algorithm type: single pixel timeseries analysis\n",
+    "\n",
     "This example may require a `pip install fusets`\n",
     "\n",
     "The actual peak-valley algorithm can be found here:\n",
@@ -59,6 +66,8 @@
     "\n",
     "## UDF with machine learning\n",
     "\n",
+    "Algorithm type: spatiotemporal data chunk\n",
+    "\n",
     "Applying deep learning models is one of the most common reason to use UDF's.\n",
     "\n",
     "This example relies on understanding 'apply_neighborhood', which is explained here:\n",