Pilot Study

Pilot Study on the Use of PROMISE Climate Data in a Crop Model : 

Pilot Study on the Use of PROMISE Climate Data in a Crop Model Type and origin and of climate data Daily, at 2m (Tmax, Tmin, Rs, Hum, wind, rain) 1950-79 (obs + sim), 2010-39 (sim) Senegal (6 points sim, 13 stations obs) Preliminary simulations with ARPEGE (MF Bordeaux) Station data from Cirad-Agrhymet data base H Syabuddin, JC Combres, JF Royer, M Dingkuhn

Slide 2: 

Crop Model Preliminary version of SARRA-H Calibrated for peanut using CERAAS data (Senegal) Photoperiodism inactivated, crop duration = f(Temp) Sowing date sensitive to rainfall (farmer ’s criteria) Water and Rs limited growth & yield Rooting depth limited by wetting front

Slide 3: 

Georeference for Climate Data Saint Louis Dakar Ziguinchor Bakel Bambey Matam Kolda

Slide 4: 

Evaluation of ARPEGE Climate Simulations Simulations for Senegal in 1950-79 show… A strong under-estimation of annual rainfall due to an inaccurate positioning of the Inter-Tropical Convergence Zone (ITCZ) An under-estimation of the N-S climatic gradient An over-estimation of the E-W climatic gradient (coast-to-continent) A strong under-estimation of diurnal temperature amplitudes => To permit meaningful test runs of SARRA-H, a latitudinal (north) shift of simulated climate by 2 degrees was performed

Slide 5: 

Annual rainfall 1950-79 Wrong positioning of ITCZ by about 2° => under-estimation of rainfall

Slide 6: 

Annual rainfall 1950-79 Comparison of measured and simulated data after north-shift of simulated climate by 2°

Slide 7: 

Bakel (continental climate) Observations: Simulated rainy season longer « Slow start » of rainy season causes risks of failure of crop establishment Intra-annual(seasonal) rainfall distribution1950-79 Comparison of measured and simulated data after north-shift of simulated climate by 2° Observed Simulated

Slide 8: 

Rainfall intensity distribution (daily cumulatives) 1950-79 Measured and simulated data after N-shift of simulated climate by 2° Over-estimation of small rains (1-3 mm), under-estimation of big rains (> 10mm), ca. factor 2 Delay in sowing, smaller fraction of useful precipitation (E!), wetting front remains shallow (rooting depth!)

Slide 9: 

Problem: ARPEGE over-estimates rain-days by factor 1,5 to 3 Frequency of rain-days 1950-79 Measured and simulated data after N-shift of simulated climate by 2°

Slide 10: 

SARRA Water Balance: Atmospheric Demand and Soil Reserve Root front Wetting front Rain Sowing Stock ET(pot)=1 Kc 2 compartments simulated ET(max)=Kc * ET(pot)

Slide 11: 

Small rain event: Evaporation Moderate rain event: Stock, Transpiration Big rain event: Runoff, Drainage Partitioning of Precipitation at the Plot Level

Slide 12: 

Air Temperature 1950-79 Measured and simulated data after N-shift of simulated climate by 2° Maximum temperatures OK Strong over-estimation of minimum temperatures => Over-estimation of daily mean temperatures by 4 to 5 °C => Under-estimation of diurnal temperature amplitudes => Simulated crop duration too short

Slide 13: 

Mean simulated grain yields 1950-79 and 2010-39 (preliminary) Thiès Dakar Louga Saint Louis 20 50 80 1950-79 2010-39 Probability for yields to fall below… (%) Causes of yield under-estimation: Stress thru delayed sowing; Short crop duration (high Tmin) Causes of yield decrease: Short crop duration (rising T); Rs lower by 2-3 MJ/d in Sept/Oct

Conclusion 1: ARPEGE climate : 

Conclusion 1: ARPEGE climate Latitude of ITCZ wrong by 2° Tmin too high, (Tmax-Tmin) too low Rainfall intensity distribution very different from station data for 1950-79 Too many small rains (1-3 mm) Too few big rains (>10 mm) => problem of scale? Predicted climate change for 2010-39 More rains in Sept/Oct (favorable) Less Rs in Sept/Oct (unfavorable)

Slide 15: 

Conclusion 2: Test simulations for peanut Simulation results primarily reflect distortions, brought about by… rainfall intensity distribution (effect of large pixel size?) delay in sowing date, resulting in terminal stress Increased proportion of water evaporated Wetting and root front remain shallow (sensitivity of drought spells) High Tmin and low (Tmax-Tmin) Short crop duration High respiration rate (yield reduction) No scenario evaluations yet for climate change

Slide 16: 

PerspectivesAdapting crop model to climate data, or vice versa? Adapting the crop model Would require de-sensitising yield to soil water stock (=> fixed assumptions on useful fraction of rains) Loss of sensitivity to « Sahel » characteristics Adapting the climate simulations Smaller pixels ? (200 x 200 km is pretty course anyway for regional forecasting!) Transforming the climate files for station type intensity distributions of rains Need for parameters -- how to estimate change? Who will do it and when? (end of project!)