fix: cement trade too high in some regions

.buildlibrary

@@ -1,4 +1,4 @@
-ValidationKey: '6037668'
+ValidationKey: '6060168'
 AutocreateReadme: yes
 AutocreateCITATION: yes
 AcceptedWarnings:

CITATION.cff

@@ -2,8 +2,8 @@ cff-version: 1.2.0
 message: If you use this software, please cite it using the metadata from this file.
 type: software
 title: 'mrmfa: Input data generation for the REMIND MFA'
-version: 2.9.1
-date-released: '2026-10-22'
+version: 2.9.2
+date-released: '2026-10-28'
 abstract: The mrmfa packages contains data preprocessing for the REMIND MFA.
 authors:
 - family-names: Dürrwächter

DESCRIPTION

@@ -1,8 +1,8 @@
 Type: Package
 Package: mrmfa
 Title: Input data generation for the REMIND MFA
-Version: 2.9.1
-Date: 2026-10-22
+Version: 2.9.2
+Date: 2026-10-28
 Authors@R: c(
     person("Jakob", "Dürrwächter", , "jakobdu@pik-potsdam.de", role = c("aut", "cre")),
     person("Bennet", "Weiss", , "bennet.weiss@pik-potsdam.de", role = "aut"),

R/calcCeTrade.R

@@ -83,18 +83,25 @@ calcCeTrade <- function(subtype, category, HS = "92", target_years = NULL) {
   # Aggregation to avoid intra-regional trade
   # ----------------------------------------------------------------------------
 
-  # construct reference for backcasting based on production/gdp and shipping costs.
-  # Note: reference has region-specific units.
+  # construct reference for backcasting based on production/gdp and shipping costs,
+  # expressed in tonnes of trade.
 
   # cement production data as base reference
   production_reference <- calcOutput("CeCementProduction", aggregate = FALSE, years = target_years)
   reference <- production_reference
   target_years <- getYears(production_reference, as.integer = TRUE)
 
-  # complement base reference with GDP data where in regions without production
+  # complement base reference with GDP data where in regions without production.
+  # GDP is converted to cement-equivalents with the yearly global cement intensity of GDP.
+  # China is excluded as its construction boom (and potential decline) would dominate the global intensity.
   gdp_reference <- calcOutput("CoGDP", aggregate = FALSE, years = target_years)
   zero_prod_regions <- getItems(production_reference[dimSums(production_reference, dim = c(2, 3)) == 0, , ], dim = 1)
-  reference[zero_prod_regions, ] <- gdp_reference[zero_prod_regions, ]
+  intensity_regions <- setdiff(getItems(production_reference, dim = 1), "CHN")
+  cement_intensity <- (
+    dimSums(production_reference[intensity_regions, , ], dim = c(1, 3))
+    / dimSums(gdp_reference[intensity_regions, , ], dim = c(1, 3))
+  )
+  reference[zero_prod_regions, ] <- gdp_reference[zero_prod_regions, ] * cement_intensity
 
   # add global shipping cost to base reference as bulk trade generally got cheaper over time
   shipping_cost <- readSource("OWID", subtype = "shipping_costs")
@@ -115,10 +122,16 @@ calcCeTrade <- function(subtype, category, HS = "92", target_years = NULL) {
   shipping_cost <- toolTimeSpline(shipping_cost, dof = 10)
   reference <- reference / shipping_cost
 
+  # transform reference (t production / shipping cost index) to tonnes of trade:
+  # use the global ratio of BACI trade to the reference over BACI years
+  trade_ratio <- dimSums(x, dim = c(1, 2, 3)) / dimSums(reference[, getYears(x), ], dim = c(1, 2, 3))
+  reference <- reference * trade_ratio
+
   # replace US reference with total clinker + cement trade data from USGS
+  # Note: This slightly overweights the US in custom regional aggregations that combine the US with other countries.
   us_reference <- readSource("USGSDS140", subtype = subtype)["USA", ]
-  us_reference <- toolBackcastByReference(us_reference, production_reference["USA", ], doForecast = FALSE)
-  us_reference <- toolBackcastByReference(us_reference, production_reference["USA", ], doForecast = TRUE)
+  us_reference <- toolBackcastByReference(us_reference, reference["USA", ], doForecast = FALSE)
+  us_reference <- toolBackcastByReference(us_reference, reference["USA", ], doForecast = TRUE)
   us_reference <- us_reference[, target_years, ]
   reference["USA", ] <- us_reference
 

README.md

@@ -1,6 +1,6 @@
 # Input data generation for the REMIND MFA
 
-R package **mrmfa**, version **2.9.1**
+R package **mrmfa**, version **2.9.2**
 
    [![R build status](https://github.com/pik-piam/mrmfa/workflows/check/badge.svg)](https://github.com/pik-piam/mrmfa/actions) [![codecov](https://codecov.io/gh/pik-piam/mrmfa/branch/master/graph/badge.svg)](https://app.codecov.io/gh/pik-piam/mrmfa) 
 
@@ -38,17 +38,17 @@ In case of questions / problems please contact Jakob Dürrwächter <jakobdu@pik-
 
 To cite package **mrmfa** in publications use:
 
-Dürrwächter J, Weiss B, Schweiger L, Benke F, Diez T, Hosak M, Zhang Q (2026). "mrmfa: Input data generation for the REMIND MFA." Version: 2.9.1, <https://github.com/pik-piam/mrmfa>.
+Dürrwächter J, Weiss B, Schweiger L, Benke F, Diez T, Hosak M, Zhang Q (2026). "mrmfa: Input data generation for the REMIND MFA." Version: 2.9.2, <https://github.com/pik-piam/mrmfa>.
 
 A BibTeX entry for LaTeX users is
 
  ```latex
 @Misc{,
   title = {mrmfa: Input data generation for the REMIND MFA},
   author = {Jakob Dürrwächter and Bennet Weiss and Leonie Schweiger and Falk Benke and Tobias Diez and Merlin Jo Hosak and Qianzhi Zhang},
-  date = {2026-10-22},
+  date = {2026-10-28},
   year = {2026},
   url = {https://github.com/pik-piam/mrmfa},
-  note = {Version: 2.9.1},
+  note = {Version: 2.9.2},
 }
 ```