The MOdelling System for Agricultural Impacts of Climate Change (MOSAICC) is an easily accessible web-based platform which allows users to assess the impacts of climate change on agricultural production. This presentation provides an overview of the platform, including a description of the different models developed to make projections and results from the project countries. © FAO: http://www.fao.org

1. MOSAICC
a Capacity Development Tool for
Multi-disciplinary Assessments of Climate
Change Impacts on Agriculture
to Support Adaptation Planning
Hideki Kanamaru, Renaud Colmant, and Migena Cumani
NRC
16 November, 2015

2. MOSAICC: Modelling System for
Agricultural Impacts of Climate Change
• Need for a tool to facilitate the user experience
by simplifying data processing and simulation
runs
• Transferable, adaptable (capacity development)
• At no cost (freeware)

3. Capacity development tool
• By national experts (ministries, universities,
research institutions)
• Using the country’s own data
• For assessing medium- to long-term climate
change impacts on agriculture
• To aid climate change adaptation planning

4. Multi-disciplinary assessments
Downscaled climate
projections under
various climate
scenarios
Crop yield
projections
under climate
scenarios
Simulation of the
country’s hydrology
and estimation of
water resources
Economic impact
and analysis of
policy response at
national level
Forest
productivity
changes under
climate scenarios
Robustness rather than sophistication (minimum input
data required, simple), flexibility, wide application, open
source

5. • Different needs
of climate data
among modelers
– Hydrology – on
small grids down to
1km, monthly
– Crop – at station or
on grid or by
province, 10-daily
– Economics – by
province, annual
Integration
• Server
• Spatial database
• Web interface

6. Statistical downscaling
of climate projections to station level
For the historical period, establish a statistical relationship
between station obs and large-scale climate (from reanalysis)
-> Apply the statistical model with GCM projections as inputs to
derive future climate at station level, daily scale
Santander Meteorology group,
University of Cantabria

7. % change in precipitation (A1B, BCM2 model) from 1971-1999 to 2011-2040
BCM2 A1B and A2 Tmin projections aggregated to 79 provinces (2011 - 2040 mean)

8. Number of Dry Days (5-consecutive days with <1
mm of daily rainfall) under MPEH5 GCM
Extreme events
2011-2040 vs 1971-2000
Number of Days with Extreme Daily Rainfall
exceeding >= 100 mm of daily rainfall under
MPEH5 GCM
Dry spells Heavy rainfall

9. RCP 4.5 RCP 8.5
CanESM2 15 % 23 %
CNRM-CM5 5 % 10 %
MPI-ESM-MR 10 % 20 %
• Valores de cambios proyectados de precipitación:
Precipitation - Ensamble de 6 (3 ESMs x 2 RCPs)
proyecciones 'plausibles' para Precipitación (promedio de
265 estaciones)
Precipitación (an1) – 265
estaciones

10. Precipitation - Mapas de % de cambios para
precipitación

11. STREAM – hydrological model
• Empirical model of
surface hydrology ---
from rainfall,
temperature,
evapotranspiration, to
the simulation of river
runoff and water
availability in large
river basins.
IVM, Free University of
Amsterdam and WaterInsight

12. Water balance PREC-PET (map) and Discharge
(box plots) for 3 GCMs x 2 emission scenarios
2011-2040

13. Changes in discharge by season and agreement
among 3 GCMs x 2 emission scenarios
2011-2040 vs 1971-2000

14. WABAL
• Crop specific water
balance model
• Initially used in crop
forecasting
(AgroMetShell, FAO)
• Produces various
variables such as the
Water Satisfaction
Index (WSI)

15. AQUACROP
• FAO crop water productivity
model to simulate yield
response to water
• Focuses on water
• Uses canopy cover instead of
leaf area index
• Balances simplicity, accuracy
and robustness
• Planning tool
• Calibrated for cotton, maize,
potato, tomato, wheat, rice,
sugar beet, quinoa, soybean
etc.

16. • Climate change makes
differentiated impacts
on provincial yield; some
positive; others negative
• Yields in rainfed areas
will be more negatively
affected than irrigated
areas, both in the A1B
and A2 scenarios at the
BCM2 and CNCM3
climate models
Rainfed rice yield change
2011-2040 vs 1971-2000

19. Rice yield projection - Peru

20. DCGE
• Dynamic Computable General
Equilibrium model, developed by
IVM, Free University of Amsterdam
• Model the future evolution of the
national economy of a country and
the changes induced by variations
of crop yields under climate
change scenarios.
• Generic, adaptable to local
conditions (production factors,
activities, commodities, consumer
types etc) according to the data
availability
• Requires the assemblage of a social
accounting matrix (SAM)

21. Application of MOSAICC
• Results from MOSAICC form a solid evidence-base
about projected impacts of climate change for
national climate change adaptation planning
– Which regions are more affected than other regions
• by temperature increase or precipitation increase/decrease?
• by crop/forest productivity changes?
• by river flow changes, and irrigation potential?
• Best suited for sub-national scale assessment and
national aggregation. Not for exploring best
adaptation options at local scale, but for identifying
areas/crops/basins that require adaptation
intervention

22. Advantages
• Participatory approach - facilitate
a collaborative environment for
inter disciplinary study
• Nothing to install (web browser)
• Remote access
• Easy data exchange
• Low computing time
• No data format or unit conversion
• Data tracking down the flow

23. Distribution
• Delivered to technical institutions
through:
– Constitution of a working group
– Trainings
– Support to carry out an integrated
impact study
• As a component of a project, or on
its own

24. Implementation of MOSAICC
• EU/FAO programme and TCP in Morocco – all
modules
• AMICAF project in the Philippines, Peru
• AMICAF-SSC in Indonesia, Paraguay – except
for economy module
• CSA and NAP projects in Malawi, Zambia –
climate and crop (MOSAICC-basic)

25. LANDIS-II
•Developed by Portland State University
•LANDIS-II is a forest landscape simulation model. It simulates how ecological
processes including succession, seed dispersal, disturbances, and climate
change affect a forested landscape over time.
Forestry Model Selection

26. LANDIS-II
Uses
• Across large (typically 10,000 - 20,000,000 ha) landscapes.
• Spatial and Temporal Flexibility
– variable time steps for each process
– variable spatial resolution and extent
• Built for Collaboration
– on-line database of extensions
– open-source extensions
– well documented
– flexible model architecture

27. LANDIS-II
PnET-Succession
• Purdue University, USA
• Assumption 1:
– Ecological models built on phenomenological relationships and behavior of the past are
“Not robust enough under novel conditions”
Gustafson, 2013 ; Williams et al., 2007
• Assumption 2:
– Process-based models have
“More robust predictions under novel conditions”
Cuddington et al. 2013; Gustafson, 2013
 PnET process-based model integrated
in LANDIS-II as succession process

29. Distribution

30. Main Inputs
Ecoregions input map:
-Temperature
-Precipitation
-Soil
Climate data (by Ecoregion):
-From downscaled and interpolation
Initial communities:
-Input map
-List species age cohorts by Initial Site Classes
Species parameters:
-Longevity
-Sexual maturity
-Seeding distance
-Foliar characteristics
-Shade and Fire tolerance
Values have already been given to most of the
parameters (applied for categories of species)
Disturbances:
-Harvest
-Fire
-Wind

31. Main Outputs
Spatial annual maps:
- By species (user choice)
- By interest:
• Biomass
• LAI
• Soil water
• Establishment
Graphs and tables :
- For all the species
• Total Biomass
• LAI (m2)
• Establishment
• Soil water
• CC impacts
• Disturbance impacts
• Harvested wood

32. Thank you
• Info:
– Hideki.Kanamaru@fao.org
– Renaud.Colmant@fao.org
– Migena.Cumani@fao.org
– www.fao.org/climatechange/mosaicc
• Partners
Mauro Evangelisti
Servizi Informatici
Numerical Ecology of
Aquatic Systems
AgroMetShell
FAO-MOSAICC is developed in the framework of the EU/FAO Programme
on “Improved Global Governance for Hunger Reduction”