Indirect carbon emissions of enhanced weathering in Malaysia

Formal problem definition

The modelling problem, based on a previous work^[29], is formally defined as follows:

• Given the state of technology in an economy that quantifies the input and output flows of resources and wastes defined by an EEIO system;

• Given the resource requirements and waste streams of a new economic sector meant for CDR;

• Given the baseline economic and environmental performance of an economy;

• Given the targeted CDR scale;

The objective is to determine the economic and environmental ripple effects (i.e., changes in sector output and the associated direct GHG emissions) of the large-scale implementation of CDR.

EEIO modelling framework

Leontief^[35] developed IO analysis by using linear equations to model economic networks. Miller and Blair^[36] provide a comprehensive tutorial on this topic. One fundamental assumption in all IO models is that the average technology mix across sectors remains constant and is described by a fixed-recipe Leontief production function. Sensitivity or scenario analysis can be used to relax this limiting assumption. The standard model^[35] is given by

where $$ \vec{\mathrm{y}} $$ is the final demand vector, $$ \vec{\mathrm{x}} $$ is the total output vector, $$ \vec{\mathrm{A}} $$ is the technical coefficient matrix, and $$ \vec{\mathrm{I}} $$ is an identity matrix. The model can be rearranged using the Leontief inverse^[35] and can be expressed as

EEIO is a standard technique for computing carbon^[37] and other environmental footprints^[36]. It relies on augmenting economic data with environmental flows (natural resources and pollutants). EEIO can be used as a stand-alone technique or in hybrid models combined with process-based LCA. For the analysis of a single pollutant such as CO₂, the footprint can be calculated from the sectoral total outputs^[36] using $$ \mathrm{z}=\vec{\mathrm{b}}^{\mathrm{T}} \vec{\mathrm{x}} $$.

Alternatively, it can be calculated from final demands^[36] using

where $$ \vec{\mathrm{b}}^{\mathrm{T}} $$ is the pollutant vector, and z is the total carbon footprint of the system. The elements of $$ \vec{\mathrm{b}}^{\mathrm{T}} $$ are the direct, production-based CO₂ emissions (or lumped GHG emissions expressed as CO₂ equivalents) of the sectors. Diagonalising the final demand vector $$ \vec{\mathrm{y}} $$ allows z to be decomposed into the carbon footprints (i.e., consumption-based emissions) of the different economic sectors. In addition, this basic model can be used to compute the carbon footprint of any specified subset of $$ \vec{\mathrm{y}} $$. The formulation can also be modified to account for the carbon footprint of imported goods, but such extensions are beyond the scope of this work.

Leontief^[30] also proposed an alternative EEIO model by extending the basic model to include a pollutant and an additional pollution control sector, which can be expressed as

where $$ \vec{\mathrm{c}} $$ is the pollution control cost vector, s, is the pollution control scale factor, and δ is the targeted pollution reduction. The additional row in the extended final demand vector corresponds to the residual, or unmitigated, emissions. The new sector acts as a public utility to clean up emissions from other parts of the economy, without attributing costs to polluters. Aviso et al.^[29] applied this model to the control of CO₂ emissions using CDR, and solved it through matrix inversion, which can be expressed as

These calculations can be easily implemented in a Microsoft Excel spreadsheet or other software with equation-solving capabilities.

Data and assumptions

The EEIO model is applied to the Malaysia scenario using data from 2019, the most recent year for which non-pandemic data is available. The 35-sector IO table of the Asian Development Bank^[38] is reduced to eight sectors using the standard approach to facilitate analysis [Table 1]. Procedural details can be found in the textbook of Miller and Blair^[36]. Each column in Table 1 gives the Leontief production function (i.e., purchased inputs per unit of economic output) of the average technology mix used in the sector. This technology mix is assumed to be fixed in the simulations that follow, except where otherwise indicated.

Total sector outputs and final demands, adapted from the Asian Development Bank^[38], are shown in Table 2. These economic data are then combined with GHG emissions data from the United Nations Environment Programme^[39] to derive the direct GHG intensities shown in the last column of Table 2. The emissions data are aggregated to match the sectors in the technical coefficients matrix in Table 1.

The generation of CDR using EW is treated as a new economic sector with backward linkages to the rest of the economy. The basalt application rate in oil palm plantations is assumed to be 20-50 t/ha/y (tonnes per hectare per year). The cost of CDR is estimated at 77.78-82.35 USD/t of basalt, with fresh basalt specifically mined for EW. Techno-economic data are adapted from Strefler et al.^[8] under the assumption that the CDR rate of basalt is 0.15-0.2 kg CO₂ per kg of basalt^[19]. It is further assumed that MRV (i.e., auditing) and other services (e.g., trading, financial, and consulting) account for 10%-15% of the cost of CDR credits, based on the projection of Mannion et al.^[2]. The coefficients and underlying assumptions are summarised in Table 3. In addition, a 0-10% reduction in fertiliser demand is considered^[10]; the effect of this change on Manufacturing (S3) inputs into Agriculture (S1) is estimated based on national fertiliser consumption statistics^[40]. To gauge the effects of these uncertainties, a full factorial design of experiment (DOE) approach is used for global sensitivity analysis^[41], where results are computed for every possible combination of the best and worst values of the parameter intervals.

Combining data in Tables 1-3 gives the full EEIO model. Two main scenarios aim to reduce GHG emissions from the baseline level of 240 Mt CO₂e/y^[39] if basalt EW is applied to all oil palm plantations in Malaysia. These plantations have a combined area of 5.67 Mha, which is about 17% of the national total land area^[33]. The first scenario assumes that the current average carbon intensity of electricity generation of 0.6 kg CO₂e/kWh^[42] is maintained in the deployment of EW; the second scenario assumes that the carbon intensity is cut by 50%, a reduction consistent with recent decarbonisation projections^[43].

The analysis assumes that most model coefficients remain constant (i.e., the technology mix used remains the same), except as indicated for the scenario and sensitivity analyses. For details, the spreadsheet implementation of the EEIO model is provided in the Supplementary Materials.

Scenario 1: baseline electricity grid carbon intensity

This scenario assumes that additional electricity generation occurs at the current grid carbon intensity of 0.6 kg CO₂e/kWh^[42] to support large-scale EW and its ripple effects. The best-case result reduces total GHG emissions by 11.0% from 240 Mt CO₂e/y^[39] to 213.7 Mt CO₂e/y. The amount of CDR generated is 56.7 Mt CO₂e/y, but approximately 54% of this is offset by GHG emissions penalties, leaving a net reduction of only 26.3 Mt CO₂e/y. This result is comparable to the results reported for the Philippines^[29], where approximately half of the CDR benefit is lost due to emissions penalties from other carbon-intensive parts of the EW supply chain. Figure 1 shows pronounced changes in total sector output of the Mining (S2), Electricity (S4), and Land transport (S6) to support EW supply chain operations, while the increments for the other sectors are negligible.

Figure 1. Percentage increase in sectoral total outputs and incremental direct sectoral emissions for Scenario 1 (authors’ original work, generated using Microsoft Excel).

The CDR credits generated by EW can be used to offset GHG emissions from other parts of the economy, thereby reducing the carbon footprints of various products and services^[2]. However, the substantial increases in basalt quarrying (Mining, S2), electricity generation (primarily for basalt grinding) (Electricity, S4), and EW logistics (Land transport, S6) also result in a cumulative GHG emissions penalty that reduces the climate change mitigation efficacy of the system. Although the incremental emissions from Land transport (S6) are not as substantial as those from Electricity (S4), the large fractional increase over the baseline economic output implies the need for additional capital stock in vehicles and road infrastructure. In addition, there is a small GHG emissions penalty due to the increased electricity requirement to support Manufacturing (S3) to provide fuel and auxiliary inputs. The GHG emissions penalty from Services (S8) for MRV and other supplementary services occurs indirectly (e.g., via additional electricity demand). Smaller high-order effects also arise from other supply chain linkages.

The results of the global sensitivity analysis for this scenario are summarised in Figure 2, showing the effects of basalt application rate, CO₂ capture rate per unit of basalt, MRV cost, and fertiliser demand reduction on net GHG emissions reduction. Using the DOE approach to global sensitivity, each of these parameters is treated as an experimental factor whose best- and worst-case values are coded on a dimensionless scale as +1 and -1, respectively. For example, a basalt application rate of 50 t/ha/y is coded as +1, while a rate of 20 t/ha/y is coded as -1. The sensitivity to each parameter can be gauged by comparing the collective results at +1 with the collective results at -1. In Figure 2A, the basalt application rate has a pronounced effect on net GHG emissions reduction, while Figure 2B shows that the CO₂ capture rate has a slightly weaker effect. On the other hand, Figure 2C and D shows that MRV cost and fertiliser demand reduction resulting from soil quality enhancement have negligible effects. The results highlight the importance of efforts to decarbonise the electricity grid in Malaysia and to enhance the effectiveness of EW implementation. Scenario 1 may represent the transitional carbon intensity of 0.6 kg CO₂e/kWh towards net-zero emissions by 2050^[43]. A decarbonised electricity grid is demonstrated in Scenario 2.

Figure 2. Results of global sensitivity analysis of net GHG emissions reduction with respect to (A) basalt application rate, (B) CO₂ capture rate per unit of basalt, (C) MRV cost, and (D) fertiliser demand reduction for Scenario 1 (authors’ original work, generated using Microsoft Excel). GHG: Greenhouse gas; MRV: monitoring, reporting, and verification.

Scenario 2: decarbonised electricity grid carbon intensity

This scenario assumes that all electricity generation has a carbon intensity of 0.3 kg CO₂e/kWh, a figure consistent with recently published projections^[43]. This reduced carbon intensity is assumed to apply to electricity supplied to both the background economy and to incremental demand supporting large-scale EW and its ripple effects. As a result, baseline emissions are much lower than those of Scenario 1, at 165.3 Mt CO₂e/y. In Scenario 2, the best-case result reduces total GHG emissions by 23.2% to 126.9 Mt CO₂e/y. The amount of CDR generated is still 56.7 Mt CO₂e/y, but this time only 32% of it is offset by GHG emissions penalties, giving a net GHG emissions reduction of 38.4 Mt CO₂e/y. This improvement highlights the critical synergy between grid decarbonisation and CDR based on EW. Economic ripple effects are summarised in Figure 3, which again shows pronounced changes in total sector output of the Mining (S2), Electricity (S4), and Land transport (S6) to support EW supply chain operations, with negligible incremental outputs in the other sectors. The effects on GHG emissions are similar to those in Scenario 1, except that the penalties for electricity generation are 50% lower.

Figure 3. Results of percentage increase in sector total outputs and incremental direct sectoral emissions for Scenario 2 (authors’ original work, generated using Microsoft Excel).

The results of the global sensitivity analysis for Scenario 2 are summarised in Figure 4, which shows the effects of basalt application rate, CO₂ capture rate per unit of basalt, MRV cost, and fertiliser demand reduction on net GHG emissions reduction. The results follow the same trends as in Scenario 1: basalt application rate [Figure 4A] and CO₂ capture rate [Figure 4B] both have significant effects, while MRV cost [Figure 4C] and fertiliser demand reduction [Figure 4D] have negligible effects.

Figure 4. Results of global sensitivity analysis of net GHG emissions reduction with respect to (A) basalt application rate, (B) CO₂ capture rate per unit of basalt, (C) MRV cost, and (D) fertiliser demand reduction for Scenario 2 (authors’ original work, generated using Microsoft Excel). GHG: Greenhouse gas; MRV: monitoring, reporting, and verification.

The scenario analysis suggests policy implications. There is a strong synergy between electric grid decarbonisation and EW implementation. Scenario 2, which reduces carbon intensity by 50%, results in a greater environmental benefit. The results support the implementation of the Carbon Capture, Utilisation, and Storage Act 2025 in Malaysia alongside renewable energy deployment policies. Other carbon management policies also have the potential to enhance the effects of EW implementation in the country. Process improvements such as optimization of rock grinding to reduce energy use^[44] can also be investigated in subsequent modelling studies.