3 Crop Models Aggregation Global Econ Models Climate Models RCPs & SSPs Aggregate Outputs Equilibrium Prices Regional and Global Model Intercomparisons and Impact Assessments RAPs Regional Econ Models AgMIP IA Framework
RCPs, SSPs and RAPs Representative Ag Pathways economic & social development storylines agricultural technology trends prices and costs of production ag, conservation, other policy
TOA-MD: multi-dimensional assessment of CC impacts & adaptation Systems are being used in heterogeneous populations A system is defined in terms of household, crop, livestock and pond sub-systems Economic, environmental and social indicators Why use TOA? What about other regional models?
There are two parts of TOA-MD simulations: First, the model simulates the proportion of farms that would adopt a new system (system 2), and the proportion that would continue to use the “base” system (system 1) In CC impact assessment, “adopters” are those who gain from CC, “non-adopters” as those who lose from CC Second, based on the adoption rate of the system 2, the TOA-MD model simulates selected economic, environmental and social impact indicators for adopters, non-adopters and the entire population. Farm income; poverty; soil nutrients and SOM; food security; nutrition; health.
This is a complex challenge! (Climate) x (crops + livestock) x (socio-econ factors) To make IA manageable, we carry out different types of simulation experiments – reference scenarios for model evaluation, validation, intercomparison – sensitivity analysis, varying some parameters while holding others constant – “pathway” analysis to explore the range of possible future states of the world IA simulations involve multiple dimensions: – climate (base, future) – production system (current systems, adapted systems) – policy (mitigation, other) – socio-economic conditions (prices, costs of production, farm size, nutrition, etc.) Let us now fix policy and socio-economic parameters, and consider IA accounting for climate and production system changes – then we can re-introduce those dimensions, i.e., replicate the analysis with those factors changed The Climate Impact Assessment Challenge
We can simulate various “experiments” for climate impact assessment, depending on the type of modeling approach and objectives of the analysis: Climate change impact without adaptation – System 1 = base climate, base technology – System 2 = changed climate, base technology Climate change impact with adaptation (“standard” analysis) – System 1 = base climate, base technology – System 2 = changed climate, adapted technology Adoption of adapted technology with climate change: – System 1 = changed climate, base technology – System 2 = changed climate, adapted technology Simulation Experiments for Impact Assessment
Consider the case of CC without adaptation: – system 1 = base climate, base technology – system 2 = changed climate, base technology = v 1 – v 2 measures the difference in income with the base and changed climates – > 0 CC causes a loss – < 0 CC causes a gain So we need to know the spatial distribution of : = 1 - 2 2 = 1 2 + 2 2 - 2 1 2 12 We observe 1 and 1 2, but not 2, 2 2 or 12, so we use climate data + crop models or statistical models to estimate them Example: Using TOA-MD to Quantify Economic Impacts of Climate Change
Define: Y 1 has mean 1 and variance 1 2 Assume: Y 2 = b Y 1, b = Y 2 /Y 1 = b + b , i.i.d.(0,1) (true?) Two cases:matched vs un-matched data (observed & simulated) Matched: Y 2 = b Y 1 Un-matched:mean of Y 2 is 2 = b 1 2 2 = b 2 1 2 + b 2 ( 1 2 + 1 2 ) 12 = b 1 / 2 A Random Proportional Yield Model to Construct System 2
Goal: use observed data from system 1 plus crop simulations to project yield distributions for system 2 A = actual crop yield, B = simulated crop yield with current climate, C = simulated crop yield with changed climate, R = C/B, R = mean of R etc. Sources of variation: soils, weather management – how to incorporate management variation? Mean bias: A > Y – biases in R = C/B causes a bias in estimate of R – note R C / B Variance bias: B and C A causes bias in R (var of a ratio) Role of Crop Models in CC Impact Assessment
We will use the TOA-MD model setup from Claessens et al. (2012 Ag Systems) to simulate impacts of CC without adaptation on the Machakos farming system: maize: using the Crop Model Team estimates of climate impacts on yields beans, mixed subsistence, dairy: 20% average reduction in productivity, no change in variance of net returns irrigated vegetables: no change in mean or variance Example: Maize Yields in Machakos, Kenya
Results: Observed and DSSAT Crop Yields Claessens et al. 2012 use data for Machakos DSSAT simulations from Thornton et al. 2010 Ag Systems which predicted R = 0.74.
Machakos production activities and system characterization under climate change (Claessens et al. 2012 Ag Systems) CC without adaptation
Sensitivity Analysis: Mean Relative Maize Yield
Impacts by Strata and Aggregated Note: mean relative maize yield = 0.79, between system correlation = 0.9 Stratum 1 = subsistence farms, no dairy or irrigation Stratum 2 = mixed crop-livestock with dairy Stratum 3 = irrigated veges and mixed crop-livestock
CC Impacts for Socio-Economic Scenarios (RAPs) with Low (1) and High (2) Challenges to Adaptation RAP1 = low challenges to adaptation; more commercially-oriented farms with 50% more land allocated to maize, mean relative maize yield = 1, net returns SD reduced 20%, higher maize and dairy prices, 20% increase in farm size, 50% increase in off-farm income RAP2 = high challenges to adaptation; farms maintain subsistence orientation with minimal adaptation to CC, higher maize and dairy prices, 20% increase in production cost, 20% reduction in farm size
Goals for the Workshop – Identify & describe regional systems – Identify regional data & issues – Implement climate-crop-TOA-MD applications for Machakos case study – Economists: review TOA-MD BLM, prepare regional case studies – Plan for RAPs workshops/design Implement IA for regional case studies Prepare strategy for full regional implementation