The models

The models integrated in FAO-MOSAICC are organized in four main components:

  • Climate data processing tools: statistical downscaling and interpolation tools aimed at preparing the data for the crop and the hydrology modelling tools
  • Crop models: simulate crop growth under climate change scenarios, using the data produced by the climate data processing tools
  • Hydrological model: models the hydrology of river basins under climate change scenarios, using the data produced by the climate data processing tools
  • Economic model: simulates the impact of yield variations due to climate change on national economies

These models are are interconnected and were designed to facilitate the data flow from one to the other.

Climate data processing

The modelling system comprises tools to generate time series of bio-climatic variables (minimum and maximum temperature, precipitation, reference evapotranspiration, growing season onset and length of the growing season) from observed weather time series as well as from Global Climate Models (GCM) outputs.

A statistical downscaling tool based on the DAD Portal (Data Access and Downscaling) by the Santander Meteorology Group of the University of Cantabria. This tool has been designed to perform statistical downscaling of coarse climate grids generated by GCMs. The data (precipitation, minimum and maximum temperature) can be downscaled simultenously for a set of weather stations, provided that enough observation records are available for each of them. Several methods keeping spatial consistency are available, as for instance weather typing methods and regression methods. The tool also includes a weather generator to derive time series of the weather variables needed.

A second tool has been developped to interpolate the climate data over the study area using the method AURELHY (Benichou and Le Breton, 1986). AURELHY ("Analyse Utilisant le RElief pour l'Hydrométéorologie") is an interpolation method for meteorological and hydrological data based on the analysis of the topography. The tool consists in a R package, freely available on the R packages repositories.

A routine has been written to compute the reference evapotranspiration using Hargreaves' method (Hargreaves and Samani, 1982) directly from interpolated rasters of minimum and maximum temperatures.

Finally, the modelling system also includes a tool to estimate the onset of the growing season and the length of the growing cycle. This tool called PLD was extracted from AgroMetShell, the FAO crop yield forecasting software, and adapted for the project. The two variables are estimated from the analysis of the time series of precipiration and reference evapotranspiration based on the method of Cochemé and Franquin, 1967.


Related links:


Crop simulation

In FAO-MOSAICC two models are available to produce crop yield projections, both chosen for there simplicity and robustness though their level of sophistication is different. The first one is a crop specific water balance model called WABAL. The model is designed to simulate the soil water balance at the level of the crop using minimal input data: bioclimatic data (precipitation and reference evapotranspiration plus the starting date and the length of the growing season), soil data (soil water holding capacity) and crop parameters (crop factors and length of the crop growth stages). The results consist in the values of a number of crop water balance variables (such as actual evapotranspiration water excess and deficit, water satifaction index etc.) for the growing season and the different growth stages. This model has been extracted from AgroMetShell, the FAO crop yield forecasting software. It can be used for any type of crops.

The second model is called AQUACROP and was also developed at the FAO. This crop model simulates the crop response to water in a more sophisticated way than WABAL. Indeed, AQUACROP distinguishes the soil evaporation and the crop transpiration, simulates the root develpment, the expansion of the canopy as well as the water stresses, and provides biomass production and yield estimates. The effect of the CO2 concentration in the atmosphere is taken into account as well. The following crops are already available: cotton, maize, potato, qinoa, rice, soybean, sugar beet, sunflower, tomato and wheat.

These models have been adapted to work with spatial data.

Related links:


Hydrological Modelling

The modellign system comprises one hydrological model based on a new version of STREAM ("Spatial tools for river basins and environment and analysis of management options"), originated J. Aerts (Free University of Amsterdam). STREAM is a spatially distributed precipitation-runoff model aimed at simulating flow accumulation and discharge rate in large water catchment areas. It has been applied in various basins around the world, as for instance the Rhine, the Ganges/the Brahmaputra, the Yangtze and the Zambeze river basins. The new version of STREAM offers more flexibility in term of input and integrates dams in the hydrological cycle. A calibration module as been incorporated as well.

The hydrological model can have several applications for climate change impact studies in the agricultural sector. At the scale of a watershed, STREAM can be used to estimate the availability of water for irrigation schemes under climate change scenarios. At the scale of a country, the model can be used to evaluate the total actual renewable water resources (TARWR).

Related links:


Economic modelling

FAO-MOSAICC comprises a Dynamic Computable General Equilibrium Model that simulates the evolution of the economy of a given country and the changes induces by variations of crop yields projected under climate change scenarios.

This model, developed in partnership with the Free University of Amsterdam, is inspired by the IFPRI DCGE model (Logfren et al. 2002; Thurlow, 2004). The model allows the user to define a number of activities producing one commodity each one to account for different crops as well as differenciated variations of crop yields across the country. The effect of crop yield variations is simulated using a shift parameter in the activity production funtions. The model provides estimations for all the endogenous variables (e.g. commodity prices, imports, taxes, households income and savings etc.). The effects of changing yields can be assessed by comparing benchmark and "shocked" situations.


Related links:

last updated:  Friday, December 19, 2014