First-order impacts of climate change in the agriculture of the Iberian Peninsula

This project is included in the research line on climate change impact study.

PRUDENCE: Prediction of Regional scenarios and uncertainties for defining European Climate change risks and effectsFinanced by the European Commission EVK2-CT-2001-00132Duration: 2002-2004 http://prudence.dmi.dk/

Project Coordinator: Dr Jens C. ChristenssenDanish Meteorological Institute (DKMI) UPM coordinator: Mª Inés MinguezResearchers: Carlos G. Hernandez Diaz-Ambrona Miguel Quemada Federico Sau

PhD: Margarita Ruiz Ramos

Introduction

The uncertainties and sources of variation in projected impacts of climate change on agriculture depend not only on the emission scenarios and climate models used for projecting future climates, but also on the impact models used, and the local soil and climatic conditions of the managed ecosystems under study.

Objectives

The main objective of our work, within the PRUDENCE project, was to evaluate and compare the uncertainties related to predictions of initial, i.e. first-order, agricultural impacts generated by linking crop simulation models using output from a range of global and regional climate models (GCMs and RCMs) projecting climate change over Europe and in particular the Iberian Peninsula between 1961–1990 and 2071–2100 under the IPCC SRES scenarios.

Water use, biomass production and phenological changes were assessed to design adaptation strategies under different optimisation goals.

Methodology

All RCMs used boundary conditions from the atmospheric general circulation model (AGCM)HadAM3 and two of them were also bounded to two other AGCMs. We standardised the analysis to control the sources of variation and uncertainties that were added in the process. The analysis employed crop simulation models (CERES group and CropSyst) that use daily climate data from RCMs and were linked to a regional soil data base. The grids or cells vary among climate models and so individual maps were built for each climate model. Information

was incorporated into a GIS (climate-soil-crop model) to produce maps for a preliminaryevaluation.

Figure 1. (a) Elevation map of the Iberian Peninsula with the PROMES grid. The agricultural regions considered are under 800 m asl. Specified locations correspond to the provinces: 1, 2 Córdoba; 3, Murcia; 4, Albacete; 5, Badajoz; 6, Madrid; 7, 8: León; 9, Zaragoza; 10; Navarra; 11; Lugo; 12, La Coruña. (b) 50 km x 50 km land grid for impact studies derived from HIRHAM and PROMES (vertical grids) over a soil map in Central Spain.1, 2 3Crop models of wheat, barley and maize were used as impact models and yield was used as an indicator that summarised the effects of climate and enabled us to quantify initial impacts and differentiate among regions. The crop simulation models that use daily climate data from RCMs and were linked to a regional soil data base. The grids or cells vary among climate models and so individual maps were built for each climate model. Information was incorporated into a GIS (climate-soil-crop model) to produce maps for a preliminary evaluation.Comparison among RCMs was done through the choice of different crop management options (Figure 1).

The experiments with the CERES crop models and climate from RCMs driven by four GCMs and various emissions scenarios are presented below

GCM IPCCSRES

RCM Crop/management Objectives or tests

A2 All RCMsHadAM3H

B2 PROMES RegCM

Winter cropsWinter and spring wheat: rainfed and irrigated

Summer cropIrrigated maize

- Cold requirements with and without rainfall-linkeduncertainties - Impact on high elevation areas with and without rainfall-linked uncertainties- Effect on summer crop; uncertainties in comparison with autumn sown crops

ECHAM/OPYC4

A2 RCAO Winter and spring wheat, and maize

Comparison with HadAM3H-RCAO

ECHAM/OPYC4

B2 RCAO Winter and spring wheat, and maize

Comparison with HadAM3H-RCAO

ARPÈGE/ OPA

B2 ARPÈGE Winter and spring wheat, and maize

Comparison with HadAM3H-ARPEGE

1, 2 3

45

6

7, 8

9

101112

1, 2 3

45

6

7, 8

9

101112

Figure 2. Relative yield of CERES rainfed winter wheat obtained from A2/control scenarios with HadAM3H boundary conditions from (a) REMO, (b) RegCM, (c) PROMES, (d) HIRHAM and (e) for irrigated winter wheat RegCM.

All RCM-crop model combinations detected crop failures for winter wheat in the South under control (as expected) and future scenarios, and projected yield increases for spring wheat in northern and high altitude areas. Although projected impacts differed among RCMs, similar trends emerged from the mapped distributions produced for relative yields for some regions.

(a) (b)

(c) (d)

Crop failures

A2/current Relative Yield (%)< 5050 - 7070 - 9090 - 110110 - 130130 - 150150 - 200> 200

(e)

RCM-crop model outputs compared favourly to others using European Re-Analysis data(ERA-15), showing the feasibility of using direct daily ouputs from RCM for impact analysis.

Uncertainties were quantified as the standard deviation of the mean obtained for all RCMs in each location (Figure 3) and differed greatly between winter (wheat) and summer (maize) seasons, being lower in the latter.

Figure 3. Mean yield of CERES spring wheat in Albacete, simulated under control, A2 and B2 (only RegCM and PROMES) scenarios from 1, ARPEGE; 2, ETH; 3, GKSS; 4, HadAM3H; 5, HadRM3H; 6, HIRHAM; 7, RegCM; 8, RACMO; 9, REMO; 10, PROMES and 11, RCAO (only A2). Standard deviation is shown by one-sided bars. Smaller yields correspond to rainfed (RSW) and larger to irrigated (ISW) wheat. Significant differences were found for all RCMs except for HIRHAM and HadAM3H between control and A2 SRES. Average yields derived from ERA-15 data are presented with a solid line and SD with a dotted line.

0123456789

1011121314

0 1 2 3 4 5 6 7 8 9 10 11 12

RCMs, GCM, ERA-15

Gra

in Y

ield

(t/h

a)

CONTROL ISW

A2 ISW

B2 ISW

ERA-15 ISW

SD ERA-15 ISW

CONTROL RSW

A2 RSW

B2 RSW

ERA-15 RSW

SD ERA-15 RSW