AgMIP Regional Integrated Assessment of Agricultural Systems in Nioro, Senegal: Representative Agricultural Pathways, Climate, Crop and Economic Datasets

This paper describes the datasets that were used to implement an AgMIP Regional Integrated Assessment for the Nioro region of Senegal to assess the potential impacts of climate change on the principal agricultural system in the Senegal peanut basin and to assess adaptations of that system to climate change under current as well as future climate and socio-economic conditions. This dataset includes the Representative Agricultural Pathways developed for Nioro from 2000-2050; the climate data that were used to implement crop yield simulations; the data that were used to parameterize the DSSAT and APSIM crop models, including historical climate data and future climate scenarios; and the data that were used to parameterize the Tradeoff Analysis Model for Multi-dimensional Impact Assessment (TOAMD) economic simulation model, as well as simulated model outputs.

1. OVERVIEW AND CONTEXT: The Agricultural Model Intercomparison and Improvement Project (AgMIP; Rosenzweig et al., 2013) developed protocol-based methods for Regional Integrated Assessment (RIA) of agricultural systems Rosenzweig et al. 2016). These methods have been applied by teams of scientists working with regional and national stakeholders across Sub-Saharan Africa and South Asia (Rosenzweig et al. , 2020. This paper describes the datasets that were used to implement the AgMIP RIA methods for the Nioro region of Senegal (MacCarthy et al, 2020). The goal of the RIA is to assess the potential impacts of climate change on the principal agricultural system in the Senegal peanut basin comprised of peanut, millet, maize, and other minor crops and livestock, and to assess adaptations of that system to climate change, under current as well as future climate and socio-economic conditions. This dataset includes:  the Representative Agricultural Pathways (RAPs, Valdivia et al., 2015) developed for Nioro from 2000-2050;  climate data used to implement crop yield simulations (included with crop modeling data);  data used to parameterize two crop models: Agricultural Production Systems sIMulator (APSIM; www.APSIM.info; Holzworth et al., 2014;Keating et al., 2003) and the Decision Support System for Agrotechnology Transfer (DSSAT; www.DSSAT.net; Hoogenboom et al., 2019a, 2019b, Jones et al., 2003 crop models, including historical climate data and future climate scenarios (Ruane et al., 2015a); and  the data used to parameterize the Tradeoff Analysis Model for Multi-dimensional Impact Assessment (TOA-MD; Antle et al., 2014;Antle and Valdivia 2020) economic simulation model.
The analysis is structured around four "core questions" of climate impact assessment (Rosenzweig et al. 2016):  Q1: What is the sensitivity of current agricultural production systems to climate change?  Q2: What are the benefits of adaptation in current agricultural systems?  Q3: What is the impact of climate change on future agricultural production systems?  Q4: What are the benefits of climate change adaptations?
2. REPRESENTATIVE AGRICULTURAL PATHWAYS: The core questions 3 and 4 of the AgMIP RIA methods address climate impacts (Q3) and adaptation (Q4) under future climate and socio-economic conditions. The social-economic component is embodied in Representative Agricultural Pathways (RAPs) that are scenarios co-developed by scientists conducting the assessment and stakeholders who help define narratives for future conditions and quantify the variables needed as "drivers" or "boundary conditions" for the crop, livestock and economic modeling used in the RIA process. An Excel spreadsheet tool called "DevRAP" was used to organize and document the RAPs and the data used to quantify the scenarios. This spreadsheet includes detailed descriptions of the storylines constructed under each pathway describing the magnitude and direction of change of the key institutional and policy, socio-economic, technology and biophysical drivers. Each driver and its change are supported by a storyline that together characterize plausible future conditions. RAPs were developed for two scenarios, the "Green Road" representing a condition in which agricultural policies are focused on sustainable development, and the "Gray Road" in which agricultural growth has little consideration for agricultural sustainability (Valdivia et al., 2020). DevRAP Excel files for both scenarios are included in this dataset ("NIORO_RAP4_Greening the Road_Final.xls" and "NIORO_RAP5_fossil fuel development_Final.xls" in the RAPs folder). Appendix A describes the contents of each DevRAP file, including a detailed description of the variable names and the sheets included in the Excel file.
3. CLIMATE DATA: Historical climate data for Nioro were collected from the Senegal National Meteorology Service, with gaps (notably the 2010 year as well as a small number of short intervals) filled using biasadjusted values drawn from the Agricultural Modeling version of the NASA Modern Era Retrospectiveanalysis for Research and Applications (AgMERRA; Ruane et al., 2015b). Following the AgMIP climate methods (Ruane et al., 2015a), the regional distribution of farm locations was estimated by modifying the station dataset to reflect gradients in the WorldClim climatology (www.worldclim.org), and future scenarios were produced according to the AgMIP enhanced Delta approach. Following methods described in Ruane and McDermid (2017), five representative models were selected from 31 CMIP5 earth system models to represent the middle of the temperature and precipitation change distribution, as well as relatively hot/dry, hot/wet, cool/dry, and cool/wet edges of the distribution. Note that "relatively cool" future scenarios are warmer than present conditions but represent those models that project a smaller increase in temperature. Five models were selected independently for a moderate and high emissions scenario (RCP4.5 and RCP8.5, respectively). The result is a historical dataset and 10 future scenarios (5 climate models x 2 RCPs) for each farm in the Nioro household survey. These climate data were used as input to the crop model simulations and are included with the data described in the next section on "Crop Data".

CROP DATA:
The crop model input and output files contained in this dataset comprise all of the necessary data to replicate simulations described by MacCarthy et al. (2021) using the DSSAT or APSIM crop models and to generate compatible simulations for any crop model that has an AgMIP translator (github.com/orgs/agmip/repositories). The AgMIP data interoperability protocols  were developed to facilitate ensemble crop modeling exercises. These protocols describe the standardized formats and vocabulary for crop modeling input and output data used in this dataset. The AgMIP Regional Integrated Assessment included eight Crop Model Simulation Sets (CMSS), as summarized in Table 1. Each of the future climate simulations (i.e., CM2, CM5, and CM6) included simulations for 10 climate scenarios described in Section 3.  Table 2 describes the data types and file formats included in the archive of crop modeling data. Each CMSS in Table 1 includes data for these file types. Table 3 summarizes the crop modeling data files archived in this dataset. There are 122 files included for each of the three crops grown in the region: maize, millet, and peanut. These files contain all of the necessary data to generate simulations for any crop model which has an AgMIP translator. See also the Excel file: "List_of_Crop_Files.xlsx" included in the dataset for a more complete description of every file included in the crop modeling dataset.
The survey data files are in AgMIP Crop Experiment Binary (ACEB) format. This format is not directly human readable, but consists of gzip archives (www.gzip.org) of JavaScript Object Notation (JSON, www.json.org/json-en.html) files. Survey data files include weather and soils data. ACEB files can be translated to crop model-ready formats for multiple crop models using the AgMIP utility, QuadUI (github.com/agmip/quadui/releases). The ACEBViewer app allows the ACEB files to be viewed directly (github.com/agmip/AcebViewer/releases). All variables in the survey data are based on the ICASA vocabulary (White et al., 2013 and www.tinyurl.com/icasa-mvl). The cultivar files are in crop model-specific formats for DSSAT and APSIM. Refer to the model documentation for definitions of these genetic parameters for the respective models.
Field overlay and seasonal strategy files are two types of AgMIP DOME (Data Overlay for Multi-model Export) files, which allow data recorded in field experiments or farm surveys to be supplemented and modified. These allow incomplete survey data to be used for crop model simulations by providing data that are required by crop models but were not recorded in the survey data. In addition, DOMEs can be used to impose hypothetical scenarios on the data, such as climate change, RAPs, and adaptations. DOMEs are fully described in AgMIP documentation, including syntax and descriptions of built-in functions: agmip.github.io/DOME.html.
Linkage files provide the linkage between survey data and DOME files. The usage and format are described here: agmip.github.io/DOME.html. The ALNK file format is in csv format and can be viewed in any text editor.
For CM0 only, we have included model-ready input files that have been converted from the ACEB, field overlay, and linkage files for both DSSAT and APSIM. These are included in the zip archive files contained in the CM0 folder.
The output (ACMO) files contained herein are for simulations performed with the DSSAT and APSIM crop models. These files are CSV (comma-delimited) format and can be read and interpreted with text editors, spreadsheets, and data analysis software.
In addition to the model input and output files,  Input datasets were configured for each of the four "core questions" described in Table 4. TOA-MD Input files have the following filename structure: TOAin-Location-CQ-GCM-RAP-Adaptation-CropModel.xls where: Location: Nioro CQ: Core Question (1, 2, 3, 4) GCM: Climate scenarios (GCMs) RAP: 4.1 and 4.2 (Green Road RAP, high and low prices); 5.1 and 5.2 (Gray Road Rap, High and Low prices) Adaptation No adaptation; 1=Adaptation package 1; 2=Adaptation package 2 Crop Model: APSIM, DSSAT These input datasets are Excel files with sheets containing metadata and the model parameters estimated from the survey data, the crop model simulations and RAPs. Details of the estimation methods are provided in Rosenzweig et al. (2016).
Key output data from the TOA-MD simulation runs are summarized in a spreadsheet "AgMIP_Nioro-Economic Outputs_04_16_19F.xlsx". This file includes the aggregated results (104 observations corresponding to the total number of simulation sets described in Table 4) and the disaggregated results (416 observations corresponding to the 4 strata times the 104 simulation sets). Table 5   Nioro Output Data_Ag: This sheet contains the aggregated modeling results for all core questions across all scenarios.
 Study Site: This sheet contains the metadata corresponding to the AgMIP Regional Integrated Assessment of Agricultural Systems in Nioro, Senegal case study.
The file, "List_of_Econ_files.xlsx", lists all TOA input and output files contained within the dataset with a short description of each.