| 灌区种植结构反演优化与土壤盐分空间分布协同解析 |
| |
| 引用本文： | 张敬晓,蔡甲冰,许迪,常宏芳,肖春安.灌区种植结构反演优化与土壤盐分空间分布协同解析[J].农业机械学报,2023,54(6):373-385. |
| |
| 作者姓名： | 张敬晓 蔡甲冰 许迪 常宏芳 肖春安 |
| |
| 作者单位： | 中国水利水电科学研究院；河北水利电力学院 |
| |
| 基金项目： | 国家自然科学基金项目(51979286)和“科技兴蒙”专项(NMKJXM202004) |
| |
| 摘 要： | 种植结构与土壤盐分的协同程度与发展关系关乎灌区水土生态质量与农业可持续发展，联动灌区种植结构提取与土壤盐分空间分析对于灌区生态环境评价与治理、保障耕地和粮食安全等具有重要意义。本文以内蒙古河套灌区永济灌域为研究区，利用2021—2022年生育期Landsat 8 OLI遥感数据与地面种植结构调查数据，分别构建决策树、支持向量机、随机森林分类模型，通过对比分析遴选出灌域适用的最优模型，准确获取灌域种植结构分布结果，同时进一步结合灌域土壤盐分实测数据及其空间异质特征，对种植结构与土壤盐分的协同关系进行深入探讨与分析。结果表明，3种模型的分类精度由大到小为随机森林、决策树、支持向量机，2021、2022年随机森林分类模型的总体精度、Kappa系数分别为92.81%、0.91,91.64%、0.89,为3种模型中精度最高，故选定随机森林模型作为最优模型；灌域内土壤盐分呈现“北部重，中、南部轻”的空间分布特征，2021、2022年土壤盐分的半方差函数适用于Gaussian模型，土壤盐分空间自相关在“中—强”等级变化；受土壤盐分制约，葵花以北部地带种植为主，玉米、小麦、小麦套种玉米(套种)和瓜菜等...
|
| 关 键 词： | 种植结构 遥感 随机森林 土壤盐分 协同分析 耦合协调度 |
| 收稿时间： | 2023/3/10 0:00:00 |
Optimization of Crop Patterns Inversion and Collaborative Analysis with Soil Salinity Spatial Distribution in Large Irrigation District |
| |
| ZHANG Jingxiao,CAI Jiabing,XU Di,CHANG Hongfang,XIAO Chun''an.Optimization of Crop Patterns Inversion and Collaborative Analysis with Soil Salinity Spatial Distribution in Large Irrigation District[J].Transactions of the Chinese Society of Agricultural Machinery,2023,54(6):373-385. |
| |
| Authors: | ZHANG Jingxiao CAI Jiabing XU Di CHANG Hongfang XIAO Chun'an |
| |
| Abstract: | Linkage of crop patterns with soil salinity will be of great significance for the assessment and management of ecological environment in large irrigation district, as well as be helpful for the protection of cultivated land and food security. To explore the synergic relationship between them, the coupling coordination degree was collaboratively analyzed based on accurate extraction of crop planting information and spatial analysis of soil salinity. Yongji Sub-irrigation Area in Hetao Irrigation District of Inner Mongolia, which had complex crop patterns and severe soil salinization, was selected as the study area. With remote sensing data of Landsat 8 OLI and ground observing data of crop planting survey during the growth period from 2021 to 2022, three classification models were constructed to inverse the crop planting information, which namely were the decision tree (DT), support vector machine (SVM), and random forest (RF), respectively. By comparing the accuracy of the models, an optimal model would be given accompanied by the best result of crop patterns. Combined with the spatial heterogeneity of soil salinity measured from field sampling sites, the synergic relationship between them was further explored quantitatively. Results showed that the classification accuracy of the three models performed as RF> DT> SVM. The overall accuracy and Kappa coefficient of RF model were 92.81%, 0.91 in 2021, and 91.64%, 0.89 in 2022, respectively, which was the biggest among the three models. Therefore, the RF model was ultimately employed as the optimal one to inverse crop patterns in this area. Moreover, soil salinity presented more severe in the northern part than that in the middle and southern parts. The semi-variance function of soil salinity was best fitted by the Gaussian model, and the spatial autocorrelation of soil salinity fluctuated from medium- to strong- level, which indicated that both structural factors and random factors influenced the spatial variation of soil salinity. Crop pattern, as an important factor of random factors, was essential to be further analyzed with soil salinity collaboratively. Two aspects of the collaborative relationship were mainly revealed. On one hand, the spatial heterogeneity of soil salinity determined the spatial characteristics of crop patterns, specifically that sunflower was mainly cultivated in the northern part, while maize, wheat, interplanting, and other crops (e.g. water melon, pepper, tomato, etc.) were mainly distributed in the middle and southern parts. On the other hand, the crops performed different adaptabilities and tolerance to soil salinity, with the average values of soil salt content from big to small as follows: sunflower (0.377% in 2021, and 0.328% in 2022), maize (0.358% in 2021, and 0.319% in 2022), interplanting (0.246% in 2021 and 2022), and wheat (0.259% in 2021, and 0.248% in 2022). As a result, crop patterns interacted with soil salinity in space, jointly determining the sustainable development of agriculture in the irrigation area. In 2021 and 2022, the coupling coordination degree between them was 0.784 and 0.787 in the study area, respectively, which reached a high level. It could be concluded that the development between crop patterns and soil salinity was balanced and coordinated with each other during the observation period. The results would provide some references for optimizing crop planting patterns and improving soil environment in large irrigation district to some extent. |
| |
| Keywords: | crop pattern remote sensing random forest soil salinity collaborative analysis coupling coordination degree |
|
| 点击此处可从《农业机械学报》浏览原始摘要信息 |
| 点击此处可从《农业机械学报》下载免费的PDF全文 |
|
