石嘉丽, 张晓龙, 闵雷雷, 张婧, 王妍, 沈彦军. 多源土壤水分产品在河北平原的适用性评价[J]. 中国生态农业学报 (中英文), 2022, 30(5): 809−819. DOI: 10.12357/cjea.20210697
引用本文: 石嘉丽, 张晓龙, 闵雷雷, 张婧, 王妍, 沈彦军. 多源土壤水分产品在河北平原的适用性评价[J]. 中国生态农业学报 (中英文), 2022, 30(5): 809−819. DOI: 10.12357/cjea.20210697
SHI J L, ZHANG X L, MIN L L, ZHANG J, WANG Y, SHEN Y J. Adaptability evaluation of soil moisture products in the Hebei Plain[J]. Chinese Journal of Eco-Agriculture, 2022, 30(5): 809−819. DOI: 10.12357/cjea.20210697
Citation: SHI J L, ZHANG X L, MIN L L, ZHANG J, WANG Y, SHEN Y J. Adaptability evaluation of soil moisture products in the Hebei Plain[J]. Chinese Journal of Eco-Agriculture, 2022, 30(5): 809−819. DOI: 10.12357/cjea.20210697

多源土壤水分产品在河北平原的适用性评价

Adaptability evaluation of soil moisture products in the Hebei Plain

  • 摘要: 河北平原地处华北平原中部, 是我国重要的粮食产区, 是世界上冬小麦、夏玉米最高产的地区之一。土壤水分作为作物生长的直接水源和基础条件, 对灌溉决策、干旱预报均有重要意义。虽然多源土壤水分产品已获得了长足发展, 但其在河北平原的适用性还缺乏全面的定量评价。本文利用河北平原望都、霸州、威县、栾城4个站点2018年1月至2019年10月的表层10 cm土壤水分实测数据, 通过相关系数、偏差、均方根误差、无偏均方根误差4个指标, 对比分析了SMOS、SMAP、FY3B、ERA-Land、GLDAS、GLEAM等6种土壤水分产品在河北平原典型农田的具体表现。整体而言, 除夏季FY3B存在高估外, 多源土壤水分产品对河北平原不同站点实际土壤含水量有不同程度的低估, 研究时段内各土壤水分产品平均相关系数排序为GLEAM>FY3B>ERA-Land>GLDAS>SMAP>SMOS, 平均无偏均方根误差排序为GLEAM<GLDAS<SMAP<ERA-Land<SMOS<FY3B。具体表现为: 1)同化多源数据的GLEAM、GLDAS、ERA-Land数据精度较好, 平均相关系数较大而平均无偏均方根误差较低。在土壤含水量高的夏季, 模拟数据更接近实测值。2) FY3B数据缺失较多、波动范围较大且平均无偏均方根误差较大, 但与实测数据相关性较好, 平均相关系数为0.43 m3·m−3, 夏季普遍高估土壤含水量, 数据精度较差, 其他季节则低估。3) SMAP整体数据精度高于SMOS, 夏季相关性较高但平均无偏均方根误差较大, 秋季则与之相反, 当实测土壤水介于0.30~0.40 m3∙m−3时表现较好。4) SMOS因射频干扰等原因在各站点表现最差, 各站点平均相关系数仅为0.20 m3·m−3, 偏差均大于0.10 m3∙m−3

     

    Abstract: The Hebei Plain, located in the central part of the North China Plain, is an important grain production area in China and one of the most productive areas worldwide for winter wheat and summer corn. Soil water is foundation of material transportation and energy transmission; and participates in the carbon-water cycle and energy exchange between the land surface and atmosphere. It is also a direct water source and key element of crop growth, which has an important impact on agricultural production, weather forecasting, and drought prediction. Although multisource soil moisture products have been extensively developed and widely utilized, a comprehensive evaluation of the applicability of these products in the Hebei Plain is lacking. Evaluating the applicability of soil moisture products and using them to understand the soil moisture dynamics of the Hebei Plain are of great significance for agricultural production, moisture monitoring, and irrigation decision-making. To compare and analyze the specific performance of the soil moisture products of SMOS, SMAP, FY3B, ERA-Land, GLDAS, and GLEAM in typical farmland in the Hebei Plain, in-situ soil moisture data of surface soil moisture from Wangdu, Bazhou, Weixian, and Luancheng stations in the Hebei Plain from January 2018 to October 2019 were analyzed by considering correlation coefficients, biases, root mean square errors, and unbiased root mean square errors (ubRMSE). Overall, except the data of FY3B in summer, all soil moisture products underestimated the actual soil water contents of different stations in the Hebei Plain. The average correlation coefficient of each soil moisture product during the study period was ranked as GLEAM > FY3B > ERA-Land > GLDAS > SMAP > SMOS, and the average ubRMSE was ranked as GLEAM < GLDAS < SMAP < ERA-Land < SMOS < FY3B. The specific performance of each soil moisture product showed that 1) based on assimilated multi-source data, the accuracies of GLDAS, GLEAM, and ERA-Land were better than those of SMOS and SMAP, with high correlation coefficients and low ubRMSE. The inversion data of GLDAS, GLEAM, and ERA-Land were relatively close to the in-situ data when the water content was high in summer. 2) Many missing data and large fluctuation ranges were found in the FY3B product, but FY3B had a good relationship with the in-situ data with an average correlation coefficient of 0.43 m3∙m−3. The soil water content was generally overestimated in summer and underestimated in the other seasons. The correlation coefficient of FY3B in summer was low, but the opposite was true in autumn. 3) Overall, the data accuracy of SMAP was higher than that of SMOS. The correlation coefficient between SMAP and in-situ data was higher in summer, but the ubRMSE was higher at the same time; however, they had opposite values in autumn. SMAP could capture dynamic changes in soil moisture when the soil moisture content is high. The data accuracy was better when the measured soil water content was between 0.30 m3∙m−3 and 0.40 m3∙m−3. 4) Owing to radio frequency interference and other reasons, SMOS greatly underestimated the soil moisture content and performed the worst at each station. The average correlation coefficient of each station was only 0.20 m3∙m−3, and the biases were all greater than 0.10 m3∙m−3.

     

/

返回文章
返回