+
+
+
+
+
+
+
+
+
@@ -1383,39 +1382,39 @@ 
+
+
+
+![\"Weather](\"data/media/weather_stations_one_project.png\")
\n",
+ "\n",
+ "We have 2 options for how to struture our data.\n",
+ "1. Include all 6 time series in one NetCDF file\n",
+ "2. Create a separate time series for each profile.\n",
+ "\n",
+ "The best practice is to create a separate file for each time series. This might suprise some of you, and the idea of creating dozens or maybe hundreds of files might sound daunting or even ridiculous.\n",
+ "\n",
+ "Firstly, you shouldn't think of your data in isolation. **Your data are a contribution to a much larger network of data**. Imagine that each marker below is a weather station, and the colour is the person or project who is responsible for the data. Imagine you are a looking for data inside the red square. For many data users, the project that the data were collected as part of is irrelevant. Many users want to access data from all the projects and combine them.\n",
+ "\n",
+ "![\"Weather](\"data/media/weather_stations_all.png\")
\n",
+ "\n",
+ "Sure, it is easier for you and the people in your project to group the data by project. But you already have the data and can group them how you like between yourselves. When it comes to publishing data, we need to think about the data users outside of the project. Data, if published correctly, could be used for decades in to the future, perhaps longer in some cases. The majority of your data users might have never heard of your project!\n",
+ "\n",
+ "Creating individual files for each time series has another advantage; each file can have its own set of global attributes! This has many advantages, including:\n",
+ "* If one time series was processed differently, this can be described.\n",
+ "* Each file has its own bounding coordinates. Imagine you go to a data centre or data access portal and you are looking for data on a map. A file with many time series will show up as a box on a map. Without opening the file up, it is difficult to see which locations have been sampled. What if a potential data user is only interested in a certain location within the bounding box? A file that contains a single time series would show up as a point on a map.\n",
+ "* Each time series can be published separately and have its own citation - including different authors if that is desired. If someone uses only one or a few of the files, they can only cite the relevant ones. It is then clear which data they have used in their work. Many good data centres are now able to provide you with a 'parent' page for your entire data collection with its own DOI and recommended citation that one can cite if they are using most or all of the files.\n",
+ "\n",
+ "Some might see it as a hassle to download and open lots of files. Some services are already being developed to allow data to download data from many similar files into a single file. More services will surely be available in the future. When we are publishing data, we should think not only about what is possible now, but what will be possible in the future. \n",
+ "\n",
+ "But for now, in the next few tutorials, we will look at how you can create and access multiple files quickly and easily today using Python!\n",
+ "\n",
+ "Need some convincing? Let's look at another example for vertical profiles (e.g. CTD data)."
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "id": "0851d141",
+ "metadata": {},
+ "source": [
+ "## Multiple profiles\n",
+ "\n",
"Let's import the modules we will use for this tutorial."
]
},
{
"cell_type": "code",
- "execution_count": 2,
+ "execution_count": 1,
"id": "67135b6b",
"metadata": {},
"outputs": [],
@@ -43,18 +84,91 @@
},
{
"cell_type": "markdown",
- "id": "3994ec7f",
+ "id": "e05df89e-437d-4c98-b4cc-58ef297d76a1",
+ "metadata": {},
+ "source": [
+ "Imagine we have the following 4 depth profiles:"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 2,
+ "id": "3e189c74",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "depth = [10,20,30]\n",
+ "temperature_1 = [21.42, 21.21, 20.98]\n",
+ "temperature_2 = [22.08, 21.56, 20.42]\n",
+ "temperature_3 = [22.11, 21.38, 20.12]\n",
+ "temperature_4 = [21.87, 21.01, 19.99]"
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "id": "3b28454d",
"metadata": {},
"source": [
- "## One profile or time series\n",
+ "Many people publish all the data from a single cruise in one file. But we have just learned that this is not the best practice! \n",
"\n",
- "The simplest setup we can have (except a single point) is data with a single dimension. This could be a time series or a vertical profile. Below is some basic code to add an depth dimension, a coordinate variable and data variable (sea water temperature) to a NetCDF file. "
+ "If we wanted to publish all the profiles in one file, we need another dimension to distinguish between the profiles. Time is a sensible choice."
]
},
{
"cell_type": "code",
- "execution_count": 7,
- "id": "77cde279",
+ "execution_count": 3,
+ "id": "6ed61391",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "time = [0,1,2,3]"
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "id": "518de385",
+ "metadata": {},
+ "source": [
+ "But now we need to made a 2D array of our temperature values."
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 4,
+ "id": "f33df992",
+ "metadata": {},
+ "outputs": [
+ {
+ "data": {
+ "text/plain": [
+ "[[21.42, 21.21, 20.98],\n",
+ " [22.08, 21.56, 20.42],\n",
+ " [22.11, 21.38, 20.12],\n",
+ " [21.87, 21.01, 19.99]]"
+ ]
+ },
+ "execution_count": 4,
+ "metadata": {},
+ "output_type": "execute_result"
+ }
+ ],
+ "source": [
+ "temperature_2d = [temperature_1, temperature_2, temperature_3, temperature_4]\n",
+ "temperature_2d"
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "id": "b571ebb4",
+ "metadata": {},
+ "source": [
+ "We can fit the data into our xarray object like this:"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 5,
+ "id": "bb09ff8a",
"metadata": {},
"outputs": [
{
@@ -424,37 +538,39 @@
" fill: currentColor;\n",
"}\n",
"
+
+
+
+YouTube videos will accompany each notebook. When these videos are published, they can be included using the following code:
+from IPython.display import YouTubeVideo
+YouTubeVideo('FGHJhAFf1W0')
+Be sure to change the id of the video!
+ + + + +
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+