Adding indices to the R-Instat library

[rdstern]

In principle this should be easy. And we can hopefully start reasonably simply. But then it is likely to expand.

The obvious index is of ENSO. We need to look for a consistent and reliable source, possibly IRI? Ideally it starts early, even before 1951. But it would be good to start quickly, and then possibly improve that index and add more indices later. They will go into a directory in the Climatic part of the R-Instat library. Possibly there becomes a menu item of the Open from Library in the File part of the R-Instat climatic menu. It could be the same as the general dialogue except it starts immediately in the Open From Instat > Climatic directory.

This has a new directory called Indices, and within that is the set of indices that we find.

One source is the IRI library. This gives ENSO (NINO) for the different parts of the Pacific Ocean.

It is on a monthly basis and there is then work that could be done so that years are classified as El Nino, Neutral, La Nina. We may want to add that calculation, and there may be direct sources for that 3-level factor.

Later we should also look for similar indices for elsewhere, particularly the Indian Ocean Dipole, etc.

Then we have to "make it easy" to incorporate these elements in our Markov chain (regression) modelling. Whether that is here, or is part of the Markov Chain issue remains to be seen.

In this part of the library we will also want to give the source, so that users c an use the direct, updated versions if they are connected and wish to do so.

[rdstern]

Here is some information from Caroline Staub:

1. The Multivariate ENSO index (MEI).

General information here: https://climatedataguide.ucar.edu/climate-data/multivariate-enso-index https://www.esrl.noaa.gov/psd/enso/mei/#data

Citations: https://climatedataguide.ucar.edu/climate-data/multivariate-enso-index

Data (scroll down for info): https://www.esrl.noaa.gov/psd/enso/mei/table.html

2. The Indian Dipole Index is here:

General info: http://www.bom.gov.au/climate/iod/

Citation: https://www.nature.com/articles/43854

Data: https://www.esrl.noaa.gov/psd/gcos_wgsp/Timeseries/DMI/

3. The Subtropical Indian Dipole index is here:

Citation: https://www.researchgate.net/publication/232643277_Subtropical_SST_dipole_events_in_the_Southern_Indian_Ocean

General info: http://www.jamstec.go.jp/res/ress/behera/SIOD.html

Data: http://www.jamstec.go.jp/res/ress/behera/iosdindex.html

Let me know if this helps, Caroline

[rdstern]

@Vitalis95 it would be good for you to return to this and I suggest it can also become the key technical chapter in your MSc project. The work on the MSc is extra hours, but sorting out ENSO for R-Instat can also justify some of your working time. As you have found, there are different sources of ENSO information, and you will also wish to use the ENSO in different ways. You can just take the value - monthly and make those values daily to include in the Markov chain models. Then you have a numeric variable. (I am not sure that will be very useful, in the model, but certainly useful to try.)

The reason I am doubtful in the model is that I think it is possible that the numeric value has no effect when it is a modest value and that is most of the time.

The other extreme is to calculate a factor and then perhaps use the factor (el, normal, la), or 5 levels on an annual basis. You should try both approaches and they can be included together in the model.

Different, but perhaps also in the chapter, is to include another variable in the model. An obvious addition is wind direction. And that could now come - from 1950 - from ERA5. And John et al are working on making it easy to include ERA5 data in analyses with R and R-Instat. And these data (wind direction) will then be on an hourly basis. I suggest you then need to get it on a daily basis - and it is circular. Then possibly also include it as a factor and see if the pattern of rainfall events, (and amounts) may be related to wind direction. Of course you will have to analyse the direction data first, and I hope you will find it can (also) become a factor variable (e.g. largely from East, or West) In Western Kenya this would relate to rain from Indian Ocean, or from the lake. They may have different patterns.

We should discuss this also with @jkmusyoka.