Agriculture

SSCRI research is primarily focused on the SBA Agriculture. Particular task of SSCRI is to support various SBA Agriculture analyses carried out by other consortium members (BOKU, UHH, IIASA) with relevant global data on topography, soil, land cover, land use and agricultural management. SSCRI activities within GEO-BENE project are identification of the best global data sources and the development and implementation of the global database primarily for bio-physical modeling (EPIC model) but also for LC/LU optimalization modeling (FASOM, Forest/Deforestration model). Global database is also supposed to identify the most serious gaps in global data availability for data-model fusion based interpretations defined by Group on Global Earth Observations (GEO) “Developing a Strategy for Global Agricultural Monitoring in the Framework of GEO Workshop Report” document for SBA Agriculture.

Logical data structure and content of the developed global database mirrors three general aspects taken into account in the database designing process:

Requirements of the models (conceptual level) – concept of spatial units defined on the basis of the most stable landscape characteristics (altitude, slope and soil texture) - homogenous response units (HRU) - was adopted as a spatial frame for setting up the base-run and alternative LC/LU scenarios and as an interface for communication of input/output information between bio-physical model EPIC and optimization models. Moreover, concept of elemental simulation units (SimU) based on selection of the areas homogenous with respect to topography (altitude, slope), soil (soil chemical and analytical properties), land cover and land use (cropland or grassland management type and amount of human inputs to the agro-ecosystem) was defined to satisfy the EPIC model input data requirements while applied in geographical context;
Global data availability and quality – data required by EPIC model or other data necessary for setting up the LC/LU scenarios are available in the different quality. Some of the essential data is absolutely lacking and have to be interpreted or estimated from other data sources. The general quality of the available data (spatial and temporal resolution, attribute depth, thematic accuracy) and data validity or confidence vary significantly. This influence mostly the extent to which the input data can be interpreted or enhanced to satisfy the requirements of bio-physical or optimization models and via data interpretation possibilities it influence also the quality of the global database content – spatial, attribute or temporal resolution of the data, confidence level of the information derived from the global database and a relevance of the data to the particular needs (detailed requirements of EPIC model versus broad and general data available);
Spatial and attribute unification of the data coming from various sources - because various thematic global datasets have been developed for various purposes and it differs seriously in spatial representation (if spatial), spatial, attribute or temporal detail, spatial or temporal resolution, geographical extent, etc. the spatial unification was necessary. Based on the least spatial resolution of the essential data inputs required for HRU and SimU definition the global coverage grid of 5 arcmin resolution pixels was selected the basic spatial frame for representation various source data (global digital elevation model, digital soil map and soil analytical properties estimations, land cover and land use data). Global coverage grid of 30 arcmin resolution pixels and country-level administrative units coverage (spatially represented via 5 arcmin resolution grid) serve the supplemental spatial reference for meteorological and agricultural census data, respectively.

Global database has been organized in three separate datasets addressing different aspects of global modeling in agronomy or forestry sector and specific data requirements (both the data type and data organization) of bio-physical model EPIC and LC/LU optimization models:

EPIC input data dataset store the data on topography, soil and climate mandatory for the bio-physical model runs;
Land cover (LC) and land use (LU) statistics store the data on the relevant LC/LU classes for agricultural and forestry modeling and serve the data source for setting up the global base-run LC/LU scenarios for bio-physical and optimization modeling as well as it creates a basis for estimation of possible LU/LC change scenarios;
Agricultural management dataset store the data on crop and crop management calendar (irrigation, fertilization and tillage) and data on irrigation water amounts and fertilization application rates interpreted from available agricultural census data.

Actually, first version of the global database is available for the test simulations. In phase 2 of the GEO-BENE project the most important task will be therefore the refinement of the database to fit all the mandatory requirements of bio-physical and optimization models as well as to get it in the best accordance with general goals of SBA Agriculture activities.

Presentation (PDF)