Statistics for Spatial Time Series Analysis of Climate Monitoring Data
This aim of this course is to help you to understand the statistical concepts related to working with spatial time series climate monitoring data.
You may be a beginner in this area, or you may be familiar with spatial time series climate monitoring data, and be able to apply statistical analyses using common software tools. But if you have ever wanted to gain a better understanding of the statistical concepts behind the tools and be better equipped to apply them accurately, then this course is for you.
Who is this course for?
Typical participants of other EUMETSAT training courses are the target audience.
Typical participants may include: postgraduate meteorology students, researchers or staff from a national meteorology service or research centre.
You may have a university degree and will have had some contact with basic statistics during your studies, but you may need a refresher on certain topics and methods.
What will I learn in this course?
The main learning objectives of this course are:
- to be able to do specific statistical analyses of spatial time series of climate data related to the course topics using one or more appropriate (software) tools
- to be able to build an understanding of the analysis steps, decisions and the underlying statistical methods involved in producing these analyses
- to be able to interpret these analyses accurately without confusion or misinterpretation
- to be able to produce reports based on these analyses
What do I need to know to do this course?
We will assume you have some familiarity with climate data from your work or studies, even if your practical use of climate data is limited.
We also assume knowledge of basic statistical concepts such as mean, median, standard deviation, frequency and extreme values in the context of climate data.
If you need a refresher in either of these points, we highly recommend taking the Introduction to Statistical Climatology course from the COMET Program. This is a free, self-paced, online course that is intended to be completed within 2 hours.
Familiarity with a scientific programming language, such as R or Python, is desirable but not a requirement. Complete code examples are provided and well documented in this course.
There are many resources online for an introduction to scientific programming for climatology and meteorology. We recommend the resources from the Coding for Earth Observations (R & Python) EUMETSAT course, which are freely available.
How do I take the course?
This course is free and open to be taken at any time. This course is self-paced, meaning you can start at any time and go through the course at your own pace.
There is no live facilitation as part of the course, but there are ways for you to interact with facilitators and other participants through the course forums.
The course content is designed in a modular way. This means that it is made up of many self-contained sections. You can go through the whole course in order, but you do not need to do this. You may also find that only some of the topics are of interest to you, which is fine. You can also use the content as reference material, to look up a specific concept or topic when you need it.
Course structure
The course is made up of tasks. Each main task involves a practical product that may be commonly produced with spatial time series climate monitoring data.
The main tasks in this course are:
- Provide a climatological review of a time point over a region
- Calculate a Trend
- Calculate the (historical) risk of an event over a region
- Provide a review of the seasonal pattern of an element over a region
Each task guides you through a set of steps. They also refer to reference sections that give more detail on a topic. You will find quizzes throughout the task for you to check your understanding of the content along the way.
Each task also includes practical examples which can be followed, adapted and used. Exercises related to the examples allow you to practice what you have learnt and apply it.
Forums
Use this forum to discuss concepts, ideas or questions related to any aspects of this course.
Task 1: Provide a climatological review of a time point over a region
1. Choosing your data
Part 1 - What data do you need?
Part 2 - Time periods and background reference length
Part 3 - Are all data sources suitable for the background reference?
Part 4 - What time interval (temporal resolution) and spatial resolution do I need?
Part 5 - Background Reference and Review time point consistency
2. Creating the background reference
4. Interpreting an anomaly
5. Reporting the results of a climatological review
Example: Characterising heat wave and drought in Europe for 2018/19
We are using a climatological review of 2018 and 2019 drought and heatwaves in Europe as the example of this task.
This review was produced by EUMETSAT and was published on their website: 2018 and 2019 drought and heatwaves in Europe.
Python-based example (Jupyter Notebook)
In the Jupyter Notebook attached, we have reproduced the analysis of this review using Python and data from EUMETSAT's Drought and Vegetation Monitoring Data Cube.
The Notebook follows the steps presented in this task. It guides you through the steps and explains the Python code used. If you are new to Python or Juptyer Notebooks, we recommend using the resources from the online short course Coding for Earth Observations to get set up.
Once you are set up, read through and run the code in the Jupyter Notebook to reproduce the analyses. Once you have done this, go to the exercises section where you will start adapting the code for other examples.
R-based example (CM SAF R Toolbox)
The CM SAF R Toolbox is an R-based tool for working with CM SAF and other gridded climate data. If you prefer to use a non-code tool the CM SAF R Toolbox has a friendly user interface and does not require any knowledge of R to use. The video below shows how to create monthly anomalies for sunshine duration using the CM SAF R Toolbox.
If you are new to R or the CM SAF R Toolbox, we recommend the resources from the online short course Learn to use an R based toolbox to analyse climate time series to get set up.
Exercises
These exercises are based on the example climatological review from the previous section.
Use the example Jupyter Notebook as your starting point to work on these exercises.
For each exercise, start by making a copy of the example Jupyter Notebook.
You can post your results from the exercises here (link to Padlet board?) to share what you have done with other participants, to get comments and comment on other's results.
Exercise 1: Update the climatological review for the review time point 2021. Compare the anomaly maps with their corresponding maps from 2018 and 2019. How does 2021 compare to 2018 and 2019? Was there evidence of drought and heatwaves in 2021?
Exercise 2: Choose a country in Europe to be the focus of the updated review. Find the coordinates of the boundary box for the country, for example, from here. Update the Jupyter Notebook so that only the data for your country's boundary box is used in the review. When you get the results, does this give you any further insights to the extent of the drought and heatwaves for your chosen country in 2018 and 2019?
Exercise 3: As well as restricting the data to a smaller rectangular area, you can also use a shape file to extract a non-rectangular area of data, for example, a country's border or the land masses. Use the shapefile attached and add the code snippet to your Jupyter Notebook to update the climatological review to only produce anomalies over land for sunshine duration. Do you think this is an improvement of the original maps? What might the advantages and disadvantages of doing this be?
If you have been able to work through these exercises, you should be comfortable producing climatological reviews with this data. You may now wish to adapt the Jupyter Notebook further, to use different elements, or different data sources, to conduct your own reviews. Use the example Notebook as your starting point for this. You may find that other EUMETSAT training courses will help you to do this by introducing you to other tools and data you can use. Feel free to use the forums to pose any questions you have.
End of Task Quiz
Task 2:
