01. Introduction - un-fao/gaezv5 GitHub Wiki

The Agro-Ecological Zones methodology

The quality and availability of land and water resources, together with socio-economic conditions and institutional factors, are essential to assure sustainable food security. In order to optimize the wise use of the land and water resources it is important to determine their agronomic potential. The crop cultivation potential describes the agronomically possible upper limit to produce different crops under given agro-climatic, soil and terrain conditions for specific levels of agricultural inputs and management conditions.

The Agro-Ecological Zones (AEZ) approach determines for each location of the globe the cultivation potentials for about 70 crops, each modelled at three levels of input (high, medium, low) and under rainfed and irrigated conditions.. The methodology adheres to the fundamental principles of land evaluation (FAO, 1976, 1978, 1984, 1993, 2007a). The AEZ concept was originally developed at the Food and Agriculture Organization of the United Nations (FAO) (Higgins et al., 1978). FAO and the International Institute for Applied System Analysis (IIASA) have further developed and applied the AEZ methodology and supporting databases and computer programs.

Global AEZ methodologies and applications were first published in 2000, referred to as GAEZ v1 (FAO/IIASA, 2001; Fischer et al., 2000), followed by several updates GAEZ v2 (Fischer et al., 2002), GAEZ v3 (FAO/IIASA, 2012), and GAEZ v4 (Fischer et al., 2021). Thousands of GAEZ v4 key results are available on a web portal – https://gaez.fao.org/ - for download of global maps and their tabulation to administrative and hydrological units (FAO/IIASA, 2021).

The current GAEZ v5 update includes 2020 baseline data (compared to a baseline of 2010 in v4 and 2000 in v3), and provides a further update of data and extension of the methodology, using updated input data, notably for soils (HWSD v2) (FAO/IIASA, 2023), consolidated land cover of circa 2020, terrain (ALOS), historic climates (AgERA5) and future climate forcing of the most recent Global Circulation Models developed in the CMIP6 modelling experiments for the 6th IPCC assessment report.

Climatic data comprises precipitation, temperature, wind speed, sunshine duration and relative humidity. These parameters are used to compile agronomically meaningful climate resources inventories, including quantified thermal and moisture regimes in space and time. Geo-referenced global climate, soil, terrain and land cover data are combined into a land resources database, which is assembled based on global grids, with a resolution of 30 arc-seconds (about 0.9 km by 0.9 km at the equator) and 5 arc-minutes (about 9 km by 9 km).

Matching procedures to identify crop-specific limitations of prevailing climate, soil and terrain resources and evaluation with simple and robust crop models, under assumed levels of inputs and management conditions, provide maximum potential and agronomically attainable crop yields for basic land resources units. The assessed agricultural production systems are defined by water supply, rain-fed or irrigation system, and levels of inputs and management circumstances. These generic production systems used in the analysis are referred to as Land Utilization Types (LUT).

Attributes specific to each LUT include crop information such as crop parameters (crop growth cycle duration, harvest index, maximum leaf area index, maximum rate of photosynthesis, etc.), cultivation practices and input requirements, and utilization of main produce, crop residues and by-products. For each LUT, the GAEZ procedures are applied for rain-fed and irrigated conditions.

Recent national, regional and global land cover data and land use statistics have been used to produce a global land cover database that quantifies by 30 arc-second grid cell the fractions of land occupied by 11 major land cover categories. Spatial layers of rain-fed and irrigated cropland were calibrated with national and sub-national agricultural statistics of 2019–2021, mainly from FAOSTAT (arable land and land under permanent crops; land equipped with full control irrigation) and selected national agricultural sub-national statistics.

Spatial representation of actual yields and production has been derived through downscaling the annual national average of 2019–2021 agricultural statistics (FAOSTAT), including all food and fiber crops, onto all rain-fed and irrigated cropland areas. Spatially explicit downscaled results are presented as (i) overall crop production value, and (ii) for 33 major commodities in terms of crop area, yield and production. Comparison of simulated potential yields and production with statistically recorded yield and production of crops currently grown provides yield and production gap information for main commodities.

In summary, GAEZ v5 has generated large spatial databases of (i) natural resources endowments relevant for agricultural uses and (ii) assessments of suitability and attainable yields of individual LUT, (iii) harvested area, yields and production of main food and fiber commodities for rain-fed and irrigated cultivated land areas in 2000, 2010 and 2020, and (iv) yield and production gaps. These databases can provide the agronomic backbone for various applications including the quantification of potential land productivity. Geographical layers at 30 arc-seconds used for data aggregation include: (i) gridded maps of the global administrative unit layers updated in 2023 by the UN and (ii) hydrological basin boundaries, based on the spatial units delineated in World Map of Major Hydrological Basins (FAO, 2011). Further, results were aggregated in numerous tables for 2020 major land cover patterns, land protection/exclusion status and of broad agro-ecological zones.

GAEZ v5 data is available from the new GAEZ v5 Data Portal (https://data.apps.fao.org/gaez/). The GAEZ v5 Data Portal has interactive data access facilities, which provide visualization and access to data and information, and offer users various analysis outputs and download options. The Data Portal covers six thematic areas as follows:

Land and Water Resources, including agro-ecological zonation, land cover patterns, soil resources, terrain resources, examples of soil and terrain suitability, protected areas and land with high biodiversity value, and selected socio economic data;
Agro-climatic Resources, including a variety of climatic indicators regarding climate classification, thermal and moisture regimes, and growing period length and conditions;
Agro-climatic Potential Yield for more than 100 crops and crop varieties assessed under different input and management assumptions for historical, current and future climates;
Agro-ecological Suitability and Attainable Yield, providing for about 70 crops, estimates of suitable extents, attainable yields and related attributes of the crop water balance assessed under rain-fed and irrigated conditions for historical, current and future climate;
Actual Yields and Production, giving downscaled historical harvested area, production and yield of 26 main crops/crop groups, and
Yield and Production Gaps, calculated in terms of ratios and differences between actual yield and production and attainable potentials for main crops.

Structure and overview of GAEZ procedures

The suitability of land for the cultivation of a given crop/LUT depends on specific crop requirements as matched with the prevailing agro-climatic and agro-edaphic conditions at a location. GAEZ combines these two components systematically by successively modifying grid-cell specific agro-climatic potential yields according to assessed soil limitations and terrain constraints. This structure allows stepwise review of results. Figure 1-1 presents an overview of the overall GAEZ v5 model structure and data integration. The GAEZ v5 user guide explains where the model outputs are located on the GAEZ v5 data portal. Calculation procedures for establishing crop suitability estimates include five main steps of data processing, namely:

Module I: Climate data analysis and compilation of general agro-climatic indicators for historical, baseline and future climates.
Module II: Crop-specific agro-climatic assessment and water-limited biomass/yield calculation.
Module III: Yield-reductions due to the impacts of agro-climatic risks and constraints of workability, pests and diseases.
Module IV: Crop specific edaphic assessment and yield reductions due to soil and terrain limitations.
Module V: Integration of results from Modules I-IV into crop-specific grid-cell databases. These are used to map by crop, input level and time period the agro-ecological suitability and attainable yields and production.

In addition to estimating crop potentials, two main activities were involved in obtaining grid-cell level harvested area, yield and production of main crops for the period 2019–2021, namely:

Module VI: Joint attribution of area, yield and production of all statistically recorded crops to the rain-fed and irrigated cropland shares of the consolidated land cover database circa 2020.
Module VII: Quantification of yield and production gaps between potential attainable crop yields and downscaled current crop yield statistics for the period 2019–2021, by comparing potential rain-fed and irrigated yields with yields of downscaled statistical production

Figure: Overall structure and data integration of GAEZ v5 (Module I-VII)

fig1_1_res

Module I: agro-climatic data analysis

The main purpose of Module I is the compilation of a geo-referenced climatic resources inventory offering a variety of relevant agro-climatic indicators. These agro-climatic indicators provide a general characterization of land resources and suitability for agricultural uses. Several agro-climatic layers are used as input during the estimation of crop yields and production in Module II, quantification of agro-climatic constraints in Module III, and for estimating agro-ecological suitability and attainable yields in Module V.

For historic conditions, GAEZ v5 uses input data from the ‘observed’ climate of the AgERA5 dataset (Boogaard et al., 2020; C3S, 2020), which provides daily surface meteorological data from 1979 to present, specifically tailored for agricultural and agro-ecological studies. The use of observed daily data improves the capability of GAEZ to account for extreme events such as the occurrence of frost days, heat waves, and periods of excessive or no rainfall.

For projections of future climates, the GAEZ v5 analysis considers bias-corrected ISIMIP3b data of five climate models (GFDL-ESM4, IPSL-CM6A-LR, MPI-ESM1-2-HR, MRI-ESM2-0 and UKESM1-0-LL) and three scenarios, SSP1-RCP2.6 (ssp126), SSP3-RCP7.0 (ssp370) and SSP5-RCP8.5 (ssp585)- totaling 15 combinations of scenario and climate models - were used to generate time series of climate input data in GAEZ v5 covering the period 2021 to 2099 and for compiling 20-year average climate attributes across four time periods: 2021–2040 (2030s), 2041–2060 (2050s), 2061–2080 (2070s), and 2081–2100 (2090s). Year-by-year simulations and time series analysis with GAEZ Module I are performed from 1980 to 2099, providing in addition to period averages also information on the distribution and variability of agro-climatic indicators within each 20-year period.

Module II: biomass and yield calculation

The main purpose of Module II is the calculation of agro-climatic potential biomass and yield for a wide range of LUTs under various input/management levels and for rain-fed and irrigated conditions. Biomass and yield calculations and the procedures used for the computation of daily crop water balances are based on models and procedures available from various FAO technical reports (Allen et al., 1998; Doorenbos and Kassam, 1979; Doorenbos and Pruitt, 1977; Kassam, 1977; Smith, 1992). Module II consists of two main steps:

Calculation of maximum crop biomass and yield potentials considering only prevailing radiation and temperature conditions, and
Computation of yield losses due to water stress during the crop growth cycle. The estimation is based on rain-fed crop water balances for a range of 8 different levels of soil water holding capacity. Yield estimation for irrigation conditions assumes that irrigation will be scheduled such that no yield-reducing crop water deficits occur during the crop growth cycle.

Revisions of Module II relate to data and validation experiences available in various national AEZ studies. The range and methods of assigning soil water holding capacity classes used in Module II have been enhanced on the basis of soil data available from HWSD v2.1 (FAO/IIASA, 2023).

By moving computed calendars inside the temperature limited Length of Growing Period (LGP) and choosing the crop start date when the yield is highest, GAEZ mimics a "smart farmer" who can best adapt to the prevailing climate conditions. Such optimal crop calendars determine crop yields for rainfed agriculture using an ideal mix of temperature, radiation, and soil moisture available for a crop or LUT in a particular location and year.

Results of Module II include LUT-specific temperature/radiation defined maximum yields, yield reduction factors accounting for sub-optimum thermal conditions, for yield impacts due to crop water deficits, estimated amounts of net irrigation requirements, potential and actual LUT evapotranspiration, the accumulated temperature sums during each LUT crop cycle, and the simulated optimum crop calendars.

Module III: agro-climatic constraints

Agro-climatic constraints cause direct or indirect losses in the yield and quality of produce. The relationships between these constraints with general agro-climatic conditions such as moisture stress and excess air humidity, and risk of early or late frost are varying by location, between agricultural activities as well as using control measures as assumed for different input levels.

Module III computes for each grid cell LUT-specific multipliers corresponding to different types of agro-climatic risks and constraints which are applied to further reduce previously calculated agro-climatic potential yields (i.e., the results of Module II).

This step is carried out in a separate module, termed Module III, to make explicit the climatic effect of limitations due to pests and diseases, and workability constraints and to permit time-effective reprocessing in case new or additional information becomes available. Four groups of agro-climatic constraints are applied, including:

Yield losses because of pests, diseases and weed constraints on crop growth;
Yield losses due to water stress, pest and diseases constraints on yield components and yield formation of produce (e.g., affecting quality of produce);
Yield losses due to workability constraints (e.g., excessive wetness causing difficulties for harvesting and handling of produce), and
Yield losses due to occurrence of early or late frosts.

Agro-climatic constraints are expressed as yield reduction factors according to the different constraints and their severity for each crop/LUT and by level of inputs. Due to paucity of available empirical data, the estimates of constraint ratings have been mostly obtained through expert opinion. Agro-climatic constraints for additional crop LUTs from several FAO and IIASA studies (FAO/IIASA, 2024; Fischer et al., 2024; Fischer et al., 2019) have been added (e.g., sesame, tomato, brachiaria grass, macauba palm). Module III outputs highlight the impact of climate forcing on crop production. Results are presented by crop variety where applicable. For example, for coffee, results are shown for the Arabica and Robusta varieties, and one for the aggregate, i.e., the better of the two varieties in each grid-cell location.

Module IV: agro-edaphic constraints

Module IV estimates yield reductions due to the constraints induced by prevailing soil and terrain-slope conditions. Crop yield impacts resulting from sub-optimum soil and terrain conditions are quantified separately for soils and terrain-slopes. The soil suitability is assessed through crop specific evaluations of seven major agronomic soil qualities estimated from soil profile attributes available in the Harmonized World Soil Database, HWSD v2.1 (FAO/IIASA, 2023). Soil qualities include soil nutrient availability, soil nutrient retention capacity, soil rooting conditions, soil oxygen availability, presence of lime and gypsum, presence of soil salinity and sodicity (sodium) conditions, and soil management/workability constraints. These limitations are estimated on a crop-by-crop basis and are combined into a crop and input specific edaphic suitability rating. Available soil Water Capacity (AWC), an important parameter in the crop water balance, is estimated from physical and chemical soil characteristics, effective soil depth and rooting depth of individual crops.

The output of Module IV comprises of result tables by crop and water source (rain-fed, gravity irrigation, sprinkler irrigation, drip irrigation), listing for each component soil of the soil map units recorded in HWSD v2.1 calculated soil quality indicators and soil unit suitability.

Module V: integration of climatic and edaphic evaluation

Module V executes the final step in the GAEZ crop suitability and land productivity assessment. It incorporates the LUT specific results of the agro-climatic evaluation for biomass and yield calculated in Module II/III for different soil AWC classes and it uses the edaphic suitability produced for each crop/soil/slope combination assessed in Module IV.

The inventories of soil resources and terrain-slope conditions are integrated by ranking all soil types according to likely slope classes of terrain-slope distributions occurring within individual soil map units, i.e., considering simultaneously the slope class distribution of all the grid cells belonging to a particular soil map unit and the genetic characteristics of soil types and the shares of the soil map unit assigned to different soil types, a data pre-processing step of Module V results in an overall consistent distribution of soil-terrain slope combinations by individual soil association map units and 30 arc-sec grid cells (i.e., approximately 0.9 km by 0.9 km at the equator).

The algorithm in Module V steps through the grid cells of the spatial soil association layer of the Harmonized World Soil Database and determines for each grid cell the respective make-up of land units in terms of soil types and slope classes. Each of these component land units is separately assigned the appropriate suitability and yield values and results are accumulated for all elements. Processing of soil and slope distribution information takes place at 30 arc-second grid cells, separately for rain-fed and irrigated conditions. One hundred of these 30 arc-second grid cells produce the aggregate agro-ecological characterization at 5 arc-minutes, the resolution used for storing and providing GAEZ results.

Cropping activities are among the most critical in causing topsoil erosion, because of their management and the particular cover dynamics of annual crops. For this reason, GAEZ applies in Module V a terrain-slope suitability rating procedure to account for important factors that influence production sustainability. This is achieved through: (i) defining permissible slope ranges for cultivation of various crop/LUTs and setting maximum slope limits; (ii) for slopes within the permissible limits, accounting for likely yield reduction due to loss of fertilizer and topsoil, and (iii) distinguishing among a range of farming practices, from manual cultivation to fully mechanized cultivation. In addition, the terrain-slope suitability rating is varied according to amount and distribution of rainfall, which is quantified in GAEZ by means of the modified Fournier index. Terrain suitability is estimated according to terrain-slope class and location specific rainfall amounts and concentration characteristics. Soil and terrain characteristics are read by 30 arc-second grid-cells for which sub-grid soil and terrain combinations have been quantified in the database. These calculations are crop/LUT specific and are separately performed for three input levels for rain-fed and irrigated water supply systems.

The processing in Module V also accounts for fallow period requirements, which have been established for main crop groups, by level of inputs, and for different climatic conditions. The fallow factors included in GAEZ are expressed as percentage of time during the fallow-cropping cycle the land must be under fallow, foremost to maintain its soil fertility status and breaking pest and disease cycles. In crop summary tabulations produced in Module V, the fallow requirement factors are applied for the estimation of attainable average annual production that can be achieved on a sustainable basis under the assumed level of inputs and management.

Application of the procedures in modules I to V, described above, result in an expected yield and suitability distribution under rain-fed and irrigation conditions by 5 arc-minutes grid-cell and for each crop/LUT and input level. Land suitability results for each crop are stored as six classes: very suitable (VS), suitable (S), moderately suitable (MS), marginally suitable (mS), very marginally suitable (vmS), and not suitable (NS). The processing results in large databases, which are used to derive additional characterizations and aggregations of the land. Examples include the calculation of land extents with cultivation potential by land cover type and protection/exclusion status, quantification of climatic production risks by using historical time series of suitability results, impacts of climate change on crop production potentials, and irrigation water requirements under current and future climates.

Module V results are presented for about 70 crops globally. Unlike Module III results, if more than one variety is modelled for a crop (e.g., wheat, maize, rice, barley, millet, rye, sorghum, yam, biomass sorghum, ….), only the best variety is shown in a location. Beyond using administrative units, additional aggregations of results by hydrological basins (and the intersection of countries and major hydrological basins) were implemented in GAEZ v5, thereby increasing the available options of crop summary and statistical tables.

Module VI: actual yield and production

Agricultural production and land statistics are available at national scale from FAO, but these statistical data do not reflect the spatial heterogeneity of agricultural production systems at finer resolutions within country boundaries. A “downscaling” method is needed for attribution of aggregate national production statistics to individual spatial units (grid cells) by applying formal methods that account for land characteristics, assess possible production options and can use available evidence from observed or inferred geo-spatial information, e.g. remotely sensed land cover, soil, climate and vegetation distribution, population density, etc.

Two main steps were involved in Module VI for obtaining downscaled grid-cell level area, yield and production of main crops:

Compilation of calibrated shares of rain-fed and irrigated cropland by 30 arc-seconds (and aggregation to 5 arc-minutes) grid cell, and
Attribution of crop specific harvested area, yield and production to the rain-fed and irrigated cropland of each grid cell.

Based on six global land cover products from remote sensing (Tubiello et al., 2023), an irrigation map (Mehta et al., 2024) and national land use statistics for cropland and forests reported by FAOSTAT, GAEZ v5 has produced a consolidated land use database circa 2020s consisting of a quantification by 30 arc-second grid cell of the fraction of land occupied by 11 main land cover classes. In step 1 of Module VI the spatial cropland shares were calibrated with national and sub-national agricultural land statistics of 2019–2021 (i.e., the share of land occupied by arable land and land under permanent crops).

In step 2 the spatial representation of actual yields and production consistent with national and sub-national statistical data around year 2020 (mainly FAOSTAT average of period 2019–2021) has been derived through jointly downscaling the agricultural statistics of all cultivated (food, fodder and fiber) crops onto the spatial rain-fed and irrigated cropland areas identified in the updated land cover dataset. Spatial results are presented as: (i) overall crop production values, and (ii) crop area, yield and production for 33 major commodities.

To achieve consistency of land balances, all recorded food, feed and fiber crops (statistical data derived from FAOSTAT, AQUASTAT and selected national sources) were attributed to the total delineated spatial physical cropland

Module VII: yield and production gaps

Module VII carries out the final modelling step in GAEZ v5 processing. The quantitative yield gap analysis relies on both the results of crop suitability and potential yield analysis produced in Module V and the downscaling of base year agricultural area and production statistics undertaken in Module VI. Apparent yield and production gaps have been estimated by comparing at a spatially detailed level of 5 arc-minutes the potential attainable yields and production (as estimated in GAEZ v5) and the harvested areas, estimated actual yields and production obtained by downscaling statistical data for respectively the years 1999–2001, 2009–2011, and 2019-2021.

Comparisons are presented as achievement ratios (actual/potential) for yields and as absolute differences of potential and actual production. The results of yield gap analysis are stored as GIS raster data at 5 arc-minutes resolution, separately for total cropland, irrigated and rain-fed cropland.

Limitations

The agronomic data, such as the data on environmental requirements for some crops, contain generalizations necessary for global applications. In particular, assumptions on occurrence and severity of some agro-climate related constraints to crop production (used in Module III) would certainly benefit from additional systematic data collection and verification.

Land degradation in its multiple aspects, including crucial elements such as soil degradation (soil erosion, contamination, sealing, compaction, nutrient depletion, and biodiversity loss), vegetation degradation, and water resources decline in quality and quantity, are not or only partially taken into account. They obviously influence sustainable yield and production capacities and a more thorough treatment of these factors would be desirable.

Socioeconomic needs of rapidly increasing and wealthier populations are the main driving force in the allocation of land resources to various kinds of uses, with food production as the primary land use. For rational planning of sustainable agricultural development, a systematic and spatially detailed understanding of farmers’ land-use and socioeconomic considerations and constraints will be crucial. So far, the use of socioeconomic information in global AEZ is limited to the specification of modes and purpose of agricultural production, the quantification of levels of inputs and management, the inclusion of agricultural prices and the consideration of population numbers and distribution.

Agriculture covers, by definition, apart from cropping a wide range of other activities and land uses include agro-forestry, livestock rearing and inland fisheries. The GAEZ v5 assessment does not encompass all these sectors and focuses mostly on the potential for growing crops (for food, fodder, fiber or biofuel feedstock). Nonetheless, the outputs of the model can and have been used as spatial agronomic backbone to support various other applications in agricultural development planning, scenario studies of climate change impacts and adaptation, or for assessing renewable bio-energy production options and deployment.

Land has many important functions. GAEZ outputs emphasize the suitability of land for crop production. The need to plan for more and better food supplies, from less resources and with less environmental impacts, will have to continue with high priority in the next decades. Current GAEZ respects land marked by protection/exclusion status or with recognized biodiversity value by using in Module V an ‘exclusion’ layer compiled from up-to-date and reliable international datasets (see Chapter 2 on GAEZ spatial input data). However, GAEZ currently cannot by itself compare the value of a potential production service in a location with the value of potential other ecosystem services of the land. Integration of supplementary modules to quantify additional ecosystem services within the GAEZ framework seems possible and desirable.