基于地方志的1736-1911年华北秋粮丰歉指数序列重建及其与气候变化的关系

粮食作物的收成丰歉与气候要素的关联是研究历史时期气候变化社会影响的重要切入点。利用清代地方志中的收成丰歉记录,以语义差异法对收成等级进行区分,重建了1736-1911年华北地区秋粮丰收和歉收指数序列,定量分析收成变化特征及其与同期温度和降水变化的关系,并对比以往常用的官方收成分数奏报记录,探讨地方志资料在重建历史粮食收成变化中的优势和不足。结果表明：① 地方志中的歉收记录相比丰收记录更为可靠,重建的歉收指数序列可在相当程度上描述总体收成的变化,就宏观趋势而言,19世纪歉收风险相比18世纪显著增加。② 歉收指数序列与10年、5年尺度的温度变化均呈显著负相关关系,反映气候转冷可显著提升歉收风险;在年际尺度上,歉收指数序列与降水呈显著负相关,与旱灾指数序列的相关系数则达到0.71（p < 0.001）,体现出极端旱灾对于华北粮食收成的严重威胁。③ 地方志中的收成记录虽在量化精度上不如奏报,但其歉收记录对于极端减产年份的反映更为准确,两者在实际应用中存在一定的互补性。以上结论有助于改进利用历史文献资料研究过去气候变化影响的技术手段,并深化气候变化对过去农业生产系统影响规律的科学认识。

Reconstruction of poor/bumper autumn harvest index series based on historical chorography and its relationship with climate change in North China from 1736 to 1911

The relationship between grain crop harvest and climate change (temperature and precipitation) has been a major topic in the research on social impact of climate change in the past. With 5099 records on poor and bumper autumn harvests kept in historical chorography, annual harvest grade at county scale is quantified using semantic differential method, and the poor/bumper autumn harvest index series in North China from 1739 to 1911 is reconstructed. The variable characteristics of autumn harvest and its relationship with climate change are analyzed, and the reliability and applicability of historical harvest records from chorography are discussed by comparing them with those of the records from official documents in the Qing Dynasty. Results show that, first, the poor harvest records from chorography are more reliable than the bumper ones, and the poor harvest index series can reflect the fluctuation of overall autumn harvest to some extent. At the centennial scale, the autumn harvest markedly turned worse in the 19th century than in the 18th century. Second, the poor harvest index series are significantly and negatively correlated with temperature change at 10- and 5-year scales. Therefore, the poor harvest is sensitive to temperature decline. At an annual scale, the poor harvest is also significantly and negatively correlated with precipitation change, and the correlation coefficient between the poor harvest and drought index is up to 0.71 (p < 0.001). That is, drought is a larger threat to crop production than flood in North China. Third, compared with the harvest records kept in official documents that are reported to the government by local officials, the records from chorography have some advantages in the reconstruction of historical harvest. The records on poor harvest from chorography are more reliable than those from official documents, thereby allowing the index series to describe extreme events of crop production drop exactly. The two historical data sources can be complementary to each other; however, direct interpolation without data rectification may increase the system errors. This study is expected to contribute to the method improvement in the usage of historical documents in reconstructing the social impacts of climate change and the deepening in scientific knowledge on the impact rules of climate change on the agricultural production in the past.