Ensemble Results

[DentonGentry]

Drawdown Model Overview

The Drawdown solution models are, at their core, economic models which estimate the total global and regional demand for each solution and the percentage of that demand each year which might adopt the Drawdown solution. The monetary and emissions impacts of that adoption are then calculated.

The models take a number of data inputs:

• data on overall potential for a given solution. For Energy/Industrial/Transportation solutions this might be a total addressable market. For Land/Agricultural solutions it might be the total amount of suitable land area for a given solution.
  These are two dimensional tabular inputs with the year as the row index and the region (like 'World', 'Latin America', 'OECD90', etc) as the column.

Year	World	OECD90	Eastern Europe	Asia (Sans Japan)	Middle East and Africa	Latin America
2012	21534.000	9741.544	1807.518	7709.775	1620.000	1546.022
2013	22203.000	9733.780	1825.743	8226.093	1647.962	1564.385
2014	22548.299	9695.364	1841.430	8876.448	1673.635	1587.453
2015	23924.947	9729.545	1866.925	9463.181	1692.768	1608.198
2016	24432.121	9784.263	1898.392	9999.656	1727.005	1636.848
2017	25006.885	9858.401	1935.739	10490.368	1776.264	1673.008
2018	25647.047	9950.896	1978.845	10939.596	1840.498	1716.292

• data regarding adoption of the particular solution for each region over time. This is also a two dimensional input.

• Variable Meta Analysis, which provide input for specific factors like the efficiency of coal-fired power plants or the yield of conventional farming practices. These are really single dimensional inputs providing the given factor from one or more sources, though they are provided as CSV files with some additional details. The mean and stdev of all sources are computed for use in the model.

SOURCE	Link	Region	Original Units
Potter, Christopher, et al.		USA	gC/m2/yr
van Minnen 2008		OECD90	Mg C/ha/yr
Dewar and Cannell 1992		OECD90	Mg C/ha/yr
Sathaye et al 2001	https://pubarchive.lbl.go	Asia (Sans Japan)	Mg C/ha/yr
Sathaye et al 2001	https://pubarchive.lbl.go	Middle East and Africa	Mg C/ha/yr
Nosetto et al 2006	http://gea.unsl.edu.ar/	Latin America	Mg C/ha/yr
Redondo-Brenes and Montagnini		Latin America	Mg C/ha/yr

• A collection of single-value parameters which are Referred to as 'AdvancedControls,' because that is the name of the sheet in the Excel model. In the Python models all of the parameters are grouped together into a JSON file. Each solution typically has several such JSON files, called scenarios, representing different levels of optimism or pessimism.

    {
     "name": "PDS-16p2050-Optimum (Book Ed. 1)",
     "solution_category": "replacement",
     "vmas": "VMAs",
     "description": "Optimum Scenario, Based on Greenpeace (2015) Advanced Revolution",
     "report_start_year": 2020,
     "report_end_year": 2050,
     "conv_2014_cost": {
         "value": 2010.0317085196398,
         "statistic": ""
     },
     "conv_first_cost_efficiency_rate": 0.02,
     "conv_lifetime_capacity": {
         "value": 182411.2757676607,
         "statistic": ""
     },
     "conv_avg_annual_use": {
         "value": 4946.8401873420025,
         "statistic": ""
     },

Scenarios

Drawdown solutions typically implement at least three scenarios, comprising different assumptions and inputs:

1. Plausible: the most likely, and least difficult, climate remediations are taken
2. Drawdown: a more aggressive but still reasonable set of climate remediations are taken
3. Optimal: everything which can be done is done, no matter the cost or difficulty

Financial and Emissions results are computed for each of these scenarios.

Ensemble Inputs

We propose to leverage the ability to run many variations of a solution to produce ensemble results, where we modulate the inputs around those specified by the scenario and check how much the result varies. This is often referred to as a sensitivity analysis.

This provides several capabilities:

1. Can provide a robustness factor, to gauge how well the solution is likely to work in the real world where things are not so well controlled as in simulation.
2. Can provide a dataset for use in a Gallery UI, allowing inputs to be varied and the results to be immediately available.

Internal Factors

There is also a need to be able to run an ensemble of results varying around some of the internal factors within the model. The Global Warming Potential of methane is one which has come up as being useful.

[DentonGentry]

Noting for future reference: https://github.com/Project-Platypus/Rhodium may be useful in working on this.