French wine dataset to mapping the expected harvest value by county

Version Note (2026): This version updates the database using the 2022 Graphic Land Registry (RPG) and the latest available administrative records. It incorporates methodological refinements in the optimization process to better align with the revised results presented in the 2024 research paper.

Methodological Overview

This dataset combines several independent public data sources to estimate the expected harvest value of French vineyards at a detailed geographical scale.

Because detailed vineyard register data are protected by confidentiality rules, only aggregated statistics by county and by appellation are publicly available. The objective is therefore not to recover the confidential vineyard register, but to generate a plausible allocation of vineyard surfaces that is fully consistent with the available public information.

The allocation problem is formulated as a constrained optimization model implemented in SAS using mathematical programming (PROC IML and CALL MILPSOLVE). The model estimates the vineyard surface (s_ac) associated with each authorized combination of appellation a and county c.

The optimization satisfies the following constraints:

• Marginal consistency: the sum of estimated vineyard areas cannot exceed the published totals by appellation and by county.
• Geographical authorization: vineyard surfaces are constrained to zero whenever an appellation is not authorized within a county according to the INAO official database.
• Weighted objective function: the optimization maximizes weighted allocated vineyard surfaces while assigning higher priority to AOP vineyards than to PGI and non-PGI vineyards.

Once vineyard surfaces have been estimated, expected harvest values are calculated by combining:

• the estimated vineyard surface,
• Olympic average yields calculated from historical production statistics,
• official crop insurance reference prices published by the French Ministry of Agriculture.

The resulting dataset provides a spatial representation of expected harvest values suitable for economic analyses, actuarial studies, agricultural insurance applications, and climate-risk assessment.

This version keeps the same optimization framework while updating all input datasets to the most recent publicly available sources. The database now incorporates the 2022 Graphic Land Registry (RPG), updated customs vineyard statistics, revised INAO geographical authorizations, and the latest crop insurance reference prices and yield estimates. Several data harmonization procedures and geographical corrections were also introduced before optimization.

Limitations

The estimated vineyard surfaces are model outputs obtained from aggregated public statistics. They should not be interpreted as confidential administrative records or as the actual vineyard register.

Because several allocations may satisfy the same published constraints, the optimization produces one feasible allocation among several possible solutions. The dataset is therefore intended for statistical, economic and spatial analyses rather than for parcel- or farm-level interpretation.

• Graphic Land Registry (RPG 2022): used to initialize the spatial distribution and identify crop codes.

File Description

• Fwm.csv, Fwm.xlsx, Fwm.json contain one estimated observation for each county–wine combination authorized by INAO. For each observation, the dataset provides the estimated vineyard surface, Olympic average yield, official crop insurance reference price, and the corresponding expected harvest value.