Reworked critical questions and added summary #13
Conversation
jacottemonroe
left a comment
jacottemonroe
left a comment
There was a problem hiding this comment.
I have some questions on what stats knowledge is expected for the course.
| Download the data to your computer and open your preferred R IDE to the directory of this tutorial. | ||
|
|
||
| After downloading the data we begin with visualization. The data consists of all the Sentinel 2 bands at a spatial resolution of 20 m. We will also make use of training polygons for the land cover classification, which will be introduced later. | ||
| After downloading the data we begin with visualization. The data consists of all the Sentinel 2 bands at a spatial resolution of 20 m, meaning that each pixel on the scene corresponds to a ground distance of 20 m by 20 m. We will also make use of training polygons for the land cover classification, which will be introduced later. |
There was a problem hiding this comment.
Added a little more description since students may not be familiar with the term "spatial resolution".
There was a problem hiding this comment.
Yes, and in fact the preferred term is actually "pixel size"! The term "resolution" means the smallest recognisable object. Sentinel-2 resolution is actually 10 m (you can tell apart objects that are 10 m across), but in this example, we use bands aggregated to 20 m pixel size or pixel spacing. Likewise, we could disaggregate it to 5 m pixel size (e.g. using bilinear interpolation), but that won't make the resolution any finer, as you still won't be able to identify any objects that are smaller than 10 m across.
|
|
||
| ```{block, type="alert alert-success"} | ||
| > **Question 2**: What is the advantage of including the NDVI layer in the classification? | ||
| > **Question 2**: What is the advantage of including the NDVI layer in the landcover classification? Hint: For information on NDVI, check out [this source](https://gisgeography.com/ndvi-normalized-difference-vegetation-index/). |
There was a problem hiding this comment.
Added a hint with more information about NDVI to help students understand this new term. This is a pretty open ended question so it should be ok.
There was a problem hiding this comment.
Yes, it should be good. Note that there should be a space in "land cover", though.
|
|
||
| ```{block, type="alert alert-success"} | ||
| > **Question 4**: What effect does this have on the overall accuracy of the results (hint: compare the confusion matrices of the original and new outputs). What effect does leaving this variable out have on the processing time (hint: use `system.time()`)? | ||
| > **Question 4**: What effect does this have on the accuracy of the results? Hint: Compare the overall accuracies (or the confusion matrices) of the original and new outputs. What effect does leaving this variable out have on the processing time? Hint: use `system.time()`? |
There was a problem hiding this comment.
The question assumes that students have worked with accuracies and confusion matrices, which may not be the case. I would make the question more flexible, allowing students to solely focus on the overall accuracy to answer the question if they do not feel comfortable with the confusion matrix. (I just noticed a typo at the end of the question, misplaced question mark.)
There was a problem hiding this comment.
Yes, that's fine. And yes, there's a misplaced question mark :) It's also technically two questions, maybe good to split them?
GreatEmerald
left a comment
GreatEmerald
left a comment
There was a problem hiding this comment.
Thanks, here are some suggestions for further improvement!
| Download the data to your computer and open your preferred R IDE to the directory of this tutorial. | ||
|
|
||
| After downloading the data we begin with visualization. The data consists of all the Sentinel 2 bands at a spatial resolution of 20 m. We will also make use of training polygons for the land cover classification, which will be introduced later. | ||
| After downloading the data we begin with visualization. The data consists of all the Sentinel 2 bands at a spatial resolution of 20 m, meaning that each pixel on the scene corresponds to a ground distance of 20 m by 20 m. We will also make use of training polygons for the land cover classification, which will be introduced later. |
There was a problem hiding this comment.
Yes, and in fact the preferred term is actually "pixel size"! The term "resolution" means the smallest recognisable object. Sentinel-2 resolution is actually 10 m (you can tell apart objects that are 10 m across), but in this example, we use bands aggregated to 20 m pixel size or pixel spacing. Likewise, we could disaggregate it to 5 m pixel size (e.g. using bilinear interpolation), but that won't make the resolution any finer, as you still won't be able to identify any objects that are smaller than 10 m across.
|
|
||
| ```{block, type="alert alert-success"} | ||
| > **Question 2**: What is the advantage of including the NDVI layer in the classification? | ||
| > **Question 2**: What is the advantage of including the NDVI layer in the landcover classification? Hint: For information on NDVI, check out [this source](https://gisgeography.com/ndvi-normalized-difference-vegetation-index/). |
There was a problem hiding this comment.
Yes, it should be good. Note that there should be a space in "land cover", though.
|
|
||
| ```{block, type="alert alert-success"} | ||
| > **Question 4**: What effect does this have on the overall accuracy of the results (hint: compare the confusion matrices of the original and new outputs). What effect does leaving this variable out have on the processing time (hint: use `system.time()`)? | ||
| > **Question 4**: What effect does this have on the accuracy of the results? Hint: Compare the overall accuracies (or the confusion matrices) of the original and new outputs. What effect does leaving this variable out have on the processing time? Hint: use `system.time()`? |
There was a problem hiding this comment.
Yes, that's fine. And yes, there's a misplaced question mark :) It's also technically two questions, maybe good to split them?
| # Today's summary | ||
|
|
||
| We learned about: | ||
| Today you performed a supervised classification, you identified patches and sieve connected cells, and you learned to deal with thematic raster data. Some functions to retain: |
There was a problem hiding this comment.
"retain"? Maybe "remember" would be more usual here.
|
|
||
| * `hist()`: Create a histogram for each layer of a `SpatRaster`. | ||
| * `pairs()`: Create a scatterplot for each pair of layers of a `SpatRaster`. | ||
| * `app()`: Apply a function to all pixels of a `SpatRaster` more efficiently. |
There was a problem hiding this comment.
Also apply a custom function, so perhaps add in parentheses "(custom)"
|
|
||
| ## Training data preparation | ||
|
|
||
| * `extract()`: Retrieve a value for the raster below a polygon. |
There was a problem hiding this comment.
But also point or line, so you can just say "a vector"
|
|
||
| ## Run a model on the data | ||
|
|
||
| * `predict()`: Predict raster values based on a predefined model. |
There was a problem hiding this comment.
Perhaps "pretrained" or "trained" would be more clear here.
|
|
||
| ## Applying a raster sieve by identifying patches | ||
|
|
||
| * `setValues()`: Assign a new value to a raster |
There was a problem hiding this comment.
This is not entirely clear; rather, setValues() sets all values of a raster to a certain value or certain values. This is equivalent to MySpatRaster[] = MyValue.
2 participants
@GreatEmerald @Timmarh