Detecting climate change and its impacts on crop yield in the continental United States

Climate change;Change-point detection;Climate impact;Crop yield

Ge, Yan ; Cai ; Ximing

关键词: Climate change;    Change-point detection;    Climate impact;    Crop yield;   

Others  :  https://www.ideals.illinois.edu/bitstream/handle/2142/34227/Ge_Yan.pdf?sequence=1&isAllowed=y

美国|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

Climatic variables, temperature and precipitation in particular, play critical roles in water resources management.Changes in climate have brought up concerns stationarity in climate, a fundamental assumption used in water resources planning and management, due to the change of climatic time series in mean and/or variance.Detecting of past changes is essential to understand more what have already occurred, the causes, and the associated impacts.In this thesis, a novel change point detection method, Bayesian local posterior density method in windows weighted by Pettitt test, is developed to detect unknown multiple change points in a given time series, and to classify change patterns based on the final posterior probability density.The detection method is then applied to the United States Historical Climate Network (USHCN) with thousands of sites, and change patterns including the change time of monthly, seasonal and annual precipitation, and maximum, average and minimum temperature are examined in detail.It is found that seasonal climatic data sets can reflect most of the patterns detected from monthly data sets; while annual data results in complicated patterns due to over-averaging of the climate variable values.Warming temperature occurred over all the continental U.S. except in the Southeast in spring, summer and fall seasons.In most areas of Southeast and Central regions, winter temperature gradually increased after 1970s.Stations where no change is found over the past100 years in terms of winter and fall precipitation are mainly clustered along the east coast of the continental United States, while no obvious cluster in terms of spring and summer precipitation.The detection results also include when and in what form (abrupt and gradual) the historical data changed.It is found that in a long term, climate may change nonlinearly, such as abruptly changed or piecewise linearly.The impacts of the identified climate changes on crop yield are further assessed.Regression model with climate variables expressed as quadratic function is utilized to model how crop yield responds to the climate since 1970 through various testing scenarios.The impacts of climate change on corn yield vary by region, temperature component (minimum, maximum or average) assessed in different time periods (crop growing period or year), and irrigated and rainfed crops.Minimum temperature has the largest impact on the gross grain over the Continental U.S among those climate variables; while warming of maximum temperature boosted the gross grain corn yield, while warming of average temperature and minimum temperature slowed it.In the Midwest, precipitation change has larger impact on rainfed corn than on irrigated corn.