From my study site I have only 6 years of instrumental records of precipitation and temperature. But I have time series data from 14 nearby climate stations (the farther is 40 Km) and I would like to make one for this region. All climate data are monthly. The time series are from 1963 to 2013. The problem is that not all the series are complete, and the available data in the series are from 10 to 50 years. Some series have only the last, intermediate or the first decades. But the mean monthly values from my study site have a high correlation with these series (r=0.70 to 0.97). I tried the Jones & Hulme 1996 method but it works with a common time period of 20-30 years with continuous values, and with the missing years of my time series I can't apply this method.
Any suggestions are useful.
Thanks
a useful paper would be:
Jones, P.D. and Hulme, M. 1996. Calculating regional climatic time series for temperature and precipitation: Methods and illustrations. International Journal of Climatology, 16: 361-377.
Thanks Rob Wilson for your answer, but as I described in my question I have already used this methodology in my analysis, just I would like to know if there is another methods or some R packages to do it.
sorry - did not fully read your Q
You might be able to start with the most complete data-sets and use them to infill the values in the other ones? - assuming the climate is relatively homogeneous over space
no idea if there are R packages for this
yes I did something similar using the median values. Thanks
Seems to me that your study area is quite small, do you have any topographic influence over the temperature or precipitation? Maybe you can just try with a regular average of all the 14 time series. Roudier and Mahe (2010) used the locations with r2 higher than 0.6, which seems to be your case in most of the stations.
The R package 'Climatol' is ideal for solving your problem automatically. Temporarily removed from CRAN, you can download it from http:/www.climatol.eu/ where you can find also a manual explaining its use and methodology.
Juan Rivera my area is small with a hilly topography, but I discarded the climate stations with low correlation, is it the reference "Calcul des pluies et débits classés sur le bassin du Bani (Mali)" ?
José Guijarro, I will try this R package.
Gracias a los dos por su ayuda.
The reference of the paper is: Roudier, P., Mahe, G. 2010. Study of water stress and droughts with indicators using daily data on the Bani River (Niger basin, Mali). Int. J. Climatol. 30: 1689-1705. Good luck!
I am not sure whether you have this available, but in Australia, government sectors have produced gridded or patched point data for problems such as yours. Here is a link to where you can download the patched point data: http://www.longpaddock.qld.gov.au/silo/ . Perhaps you could find the methodology that they use on their website, or contact them for details.
We have faced the same problem in Spain and tried to solve firstly producing a complete "reconstruction of series", by common procedure (selection of neighb by distane and correlation), then complete all the dataset, and finally selecting the most originally data series.The problema is when two very close stations have not common period, but there also ways to solve this point.
See our paper about MOTEDAS dataset in the Int Jr CLimatology,available here on my site
Thanks José, I think you mean this MOPREDAS paper: http://onlinelibrary.wiley.com/doi/10.1002/joc.2115/abstract
I will check it !
You can apply regional correlations like used in these papers
Article Trends and Discontinuities in Regional Rainfall of West and ...
Conference Paper Evolution des relations pluie-débit sur des basins versants du Maroc
Article Analyse par les vecteurs régionaux de la variabilité spatio-...
After primary and secondary validation of your data sets, you will first need to identify the errors and missing values in the data series if you have not done so yet. Such incorrect and missing values can be replaced wherever possible by estimated values (different methods such regression, normal ratio method, distance power method, averaging or linear or polynomial interpolation can be used...of course with some penalties) based on other observations at the same station or at the neighboring stations. After you have been able to deal with the missing values, you may need to conduct some spatial consistency checks of the data.For generation of the climate time series for your region, I believe you may need to carry out several simulations or use some data generators and test how good these are using the appropriate statistical tests in representing your region of interest.
You could still apply the Jones & Hulme 1996 method when the records of common reference period are shorter than 20 years, by simply changing the criteria to no less than 15 years, for example. However, you should calculate percent anomalies of precipitation to form monthly and annual time series for individual sites before you calculate the regional average series. Good luck.
- Badges
- Science topic
- Similar topics
- Books