多模式集合模拟气候变化对玉米产量的影响

Multi-model ensemble for simulation of the impact of climate change on maize yield

气候模式驱动作物模型是气候变化影响评估的主要手段。但是,单一气候模式输出和作物模型的结构差异使得研究结果存在不确定性。多模式集合的概率预估可以有效减少研究结果的不确定性。为此,本文利用1981—2009年东北地区海伦、长岭、本溪3地区农业气象站的历史气象资料和玉米作物数据,分别建立了作物统计模型并验证了APSIM机理模型在研究区域的适用性。在此基础上,与CMIP5在RCP4.5情景下的8个全球模式结合,尝试基于多模式集合评估了未来2010—2039年时段和2040—2069年时段气候变化对玉米产量的可能影响(相对于1976—2005年基准时段)。研究结果表明,APSIM模型对玉米生长发育和产量形成有很好的模拟能力。玉米生育期的模拟误差(RMSE)为3~4 d,产量的RMSE为0.6~0.8 t?hm^(-2)。建立的产量统计模型表明,玉米出苗阶段(5月中旬)的温度增加对产量增加有积极作用,而开花到成熟阶段(7月中旬到9月上旬)的温度和降水的增加、光照的不足均不利于产量增加。与1976—2005年基准时段相比,气候因素影响下2010—2039年玉米产量减少3.8%(海伦)~7.4%(本溪),减产的概率为64%(长岭)~73%(本溪);2040—2069年时段减产6.4%(海伦)~10.5%(本溪),减产的概率为74%(海伦)~83%(本溪)。未来2010—2039年时段和2040—2069年时段基于机理模型模拟的产量降低分别为6.6%(海伦)~8.9%(本溪)和9.7%(海伦)~13.7%(本溪),均高于相应时段基于统计模型得到的0.9%(海伦)~6.0%(本溪)和2.0%(长岭)~7.3%(本溪)减产结果。

Climate projections through process-based statistical crop models are important in studying the impacts of climate change on agricultural production.However,extensive assessments have generally relied on single climate with single crop models which have shown large discrepancies in predicted crop yields and estimations uncertainty hardly assessed.The proper understanding of uncertainties associated with such models is essential for effective use of projected results in devising adaptation strategies.Assessing crop yield response to future climate conditions based on an ensemble of possible outcomes from multiple climate projections and crop models could be more reliable than using a single model outcome.To estimate uncertainties associated with the study of the impacts of climate change on crop yield,we used8climate projections by GCMs under RCP4.5in the CMIP5(which represented the uncertainties in the projected climate change)and a statistical process-based crop model(which represented the uncertainties in the different structures or different formulations of physiological processes of crop models).Historical data of crop and meteorological data during1981?2009from agro-meteorological stations of China Meteorological Administration in Hailun,Changling and Benxi in Northeast China were used to establish and evaluate statistical and process-based APSIM(Agricultural Production Systems sIMulator)models,respectively.Then the two crop models were linked with8climate projections to evaluate the impact of climate change on maize yield during2010-2039and2040-2069,using1976-2005as the baseline period.In total,2crop models under8climate projections for a period of30years(a total of480simulations)were generated for both the baseline and two future climate periods.The results showed that APSIM model well simulated the growth and yield of maize.The root mean square error(RMSE)for the growth progress(flowering and maturity)simulation was3-4days and that for the yield simulation was0.6-0.8t?hm?2.The established statistical model suggested that temperature during emergence(mid May)had a positive effect on maize yield.However,the increase of temperature and rainfall,and lack of solar radiation during flowering and grain-filling periods(mid July to early September)had negative impact on increase of maize yield.Compared with1976-2005,the resulting probability distributions indicated that due to climate change,maize yield in2010-2039decreased on average by3.8%(Hailun)?7.4%(Benxi),at a probability of64%(Changling)?73%(Benxi).During2040-2069,maize yield increased by6.4%(Hailun)?10.5%(Benxi),at a probability of74%(Hailun)?83%(Benxi).The simulated yield decrease by the APSIM model was6.6%(Hailun)?8.9%(Benxi)during2010-2039and9.7%(Hailun)?13.7%(Benxi)during2040-2069.These were higher relative to those simulated by the statistical model,which were0.9%(Hailun)?6.0%(Benxi)during2010-2039and then2.0%(Changling)?7.3%(Benxi)during2040-2069.