Rene,

NOAA's correlations web page may be useful for you:

http://www.esrl.noaa.gov/psd/data/correlation/

On that web page, you can pick your variables, seasons, and world region, and it will create a map for you on the fly. The map will show, using color, how strong the linear correlation is between the two variables you picked at each location.

You can even enter your own custom variables.

The underlying data sets it uses are the so-called "reanalysis" which are spatially distributed values (i.e., one value per grid cell of a mesh that covers the globe) which are largely based on observations.

Mariza

in comparison you will need to obtain time series data from meteorological stations about the variable and you can correlate it using SPSS. I hope this will help you in your research.

Thanks, Samuel... but there must be dozens of papers already without my redoing the statistical work.

That means, you have to engage Google search for relevant paper

Rene,

NOAA's correlations web page may be useful for you:

http://www.esrl.noaa.gov/psd/data/correlation/

On that web page, you can pick your variables, seasons, and world region, and it will create a map for you on the fly. The map will show, using color, how strong the linear correlation is between the two variables you picked at each location.

You can even enter your own custom variables.

The underlying data sets it uses are the so-called "reanalysis" which are spatially distributed values (i.e., one value per grid cell of a mesh that covers the globe) which are largely based on observations.

Mariza

Dear Mariza & Benigno

many thanks for your answers!

Benigno: I like your text (ever thought of translating it to English?), but is does not seem to have the specific answer to my question or, maybe, my question was not specific enough.

Mariza: I knew the website and there is actually a figure in the paper I am writing that compares a NOAA map with a similar map obtained based on our indicators.

Let me come back to my question. If you have a minute, kindly have a look at a book chapter I prepared long ago: figure 1.1 in http://www.fao.org/docrep/W5183E/w5183e03.htm. This graph was obtained with annual values.

In fact, the correlations change a lot according to many factors, in particular the averaging period (hours to years) and the reference period (the last ten years or the geological/climate change scale). For instance, if you look at this (De Paola F, Giugnia M. 2013.Coupled spatial distribution of rainfall and temperature in USA. Procedia Environmental Sciences 19:178–187. doi: 10.1016/j.proenv.2013.06.020) or the work of geologists, there is a positive association between rainfall and temperature. Depending on the time scale, some variables are relevant or irrelevant (e.g. CO2).

My time scale is the crop monitoring time scale, i.e. basically 10-15 years and growing seasons or dekads. It happens that, at that scale, the link between rainfall and temperature is usually negative or, at least, the link between rainfall anomalies and temperature anomalies is negative, which has lots of implications. For instance, if we have a drought situation, not only will temperature be higher than "normal", but ETP proportionally increase even more, creating a positive feed back. In fact, ETP depends on T with an exponent larger than 1 among others because of the T^4 in the blackbody radiation and because of the exponential saturation vapour pressure dependence on T!

Just for the fun, I attach a figure showing some correlations as a function of latitude between October and April (P=PAR, R=Rain).

What I was looking for is a systematic study of those variations, and I admit I haven't found any so far... except for radiation, because lots of people have tried to find a simplified method to estimate radiation from other weather variables.

Again, many thanks to both of you.

R

Very good I also endorse