How we're advancing climate-smart agriculture: read the 2013 Annual Report

MarkSim 2: Global rainfall generator

MarkSim version 2


MarkSim® is a third-order markov rainfall generator, first released in 2002, that is able to simulate the observed variance of rainfall in both tropical and temperate regions.  It works from a large set of parameters based (in version 1’s case) on a calibration dataset of some 10,000 stations worldwide that have in most cases more than 15 years of daily historical rainfall data.  Climates were clustered into some 700 types, and some of the parameters of the MarkSim model calculated by regression from the cluster most representative of the climate point to be simulated.  Since its release, the tool has been widely used in the agricultural modelling community, and it is now also being used in empirically downscaling the outputs from climate models to produce daily weather data that are characteristic of future climatologies that can then be fed to agricultural models to evaluate possible impacts in the future.  Since 2002, another 30,000 stations of historical daily rainfall have been collected, and version 2 of MarkSim will make use of this greatly expanded calibration data set. 


To develop version 2 of MarkSim, a third-order Markov rainfall generator, via a massively increased calibration data set of daily historical rainfall data that will allow the creation of many more climate clusters globally. 


The first task will be to expand the calibration data set of daily historical rainfall data.  Many thousands of additional data series have been assembled in the last decade, and there are still some web sources to be explored that may yield more stations.  Data will be cleaned and formatted appropriately, and the data set will be documented, as far as this is possible (some of the rainfall data are very old and their provenance is not clear).  For the second task, MarkSim will be refitted to the calibration data set, clustering of climate types will be carried out, and new parameter surfaces produced for the globe.  At the same time, some modifications to the version 1 software will be undertaken, so that daily weather files in some common modelling formats (e.g. DSSAT) can be easily produced.


The main output will be MarkSim v2, which will be available as a web-based tool.  The tool will be documented, as will the calibration data set, as far as possible.


Waen Associates, UK; the International Center for Tropical Agriculture (CIAT)


October 2010 – December 2011