基于多源遥感数据融合的运城盆地夏玉米估产研究

YIELD ESTIMATION OF SUMMER MAIZE IN YUNCHENG BASIN BASED ON FUSION OF MULTI-SOURCE REMOTE SENSING DATA

关键词:  估产  夏玉米  多源遥感数据  STNLFFM模型  光能利用率模型

Key words:  yield estimation  summer maize  multi-source remote sensing data  STNLFFM model  light utilization model

摘要:

目的 准确估算区域尺度作物产量是确保粮食安全，发展生态、安全农业的关键。方法 文章以山西省重要的粮食产区——运城盆地为例，利用2020年空间分辨率为10m的Sentinel-2A数据和时间分辨率为1d的MODIS数据，采用STNLFFM（Spatial and Temporal Nonlocal Filter based Fusion Model）算法并结合光能利用率模型进行夏玉米NPP（Net Primary Production）模拟和产量估测。结果 （1）融合后的NDVI数据能较好地继承Sentinel-2A NDVI数据的空间细节，同时能够较好地表达较小地物之间空间差异；（2）STNLFFM NDVI时序曲线能准确地反映夏玉米种植时期的NDVI变化趋势和局部突变信息，比MODIS NDVI时序曲线更符合夏玉米实际生长状况。（3）运城盆地夏玉米种植区累积NPP均值为667.42 gC/m²，其中西北部的涑水河冲湖积平原，受人类活动影响显著，灌溉条件较好，NPP累积量较高，为700~900 gC/m²。（4）东北部山区和北部台塬区受地形破碎程度较高影响，NPP累积量小于500 gC/m²。（5）基于STNLFFM NDVI构建的估产模型精度（R²=0.849，MAPE=5.47%）显著高于基于MODIS NDVI数据的估产模型精度（R²=0.113，MAPE=15.65%），说明利用时空融合技术能够有效提高夏玉米单产估测精度。结论 该文可以为多源遥感数据融合与光能利用率模型协同估产提供了新的思路。

Abstract:

Accurate estimation of crop yields at the regional scale is the key to ensure food security and develop ecological agriculture and safe agriculture. This research took Yuncheng Basin, an important grain-producing area in the Shanxi Province as an example. And based on Sentinel-2A with a spatial resolution of 10 m and MODIS with a time resolution of 1 d, it calculated the Net Primary Production (NPP) and estimate summer maize yield by using a Spatial and Temporal Nonlocal Filter based Fusion model (STNLFFM) and a light utilization model. The results were indicated as follows. (1) The fused NDVI could accurately inherit the spatial details of Sentinel-2A NDVI data and express the spatial differences between smaller features more effectively. (2) Compared with the MODIS NDVI, STNLFFM NDVI curve was consistent with the actual summer maize growth condition, which accurately reflected the NDVI trend and local abrupt change information during the summer maize growth period. (3) The average NPP accumulation of the summer maize plant areas in the Yuncheng Basin was 667.42 gC/m². Affected by human activities and good irrigation conditions, NPP accumulation was high in the alluvial plain of the Sushui River in the northwest of Yuncheng Basin, ranging from 700 to 900 gC·m^-2. (4) The accumulation of NPP was less than 500 gC/m² in the northeastern mountainous area and the northern plateau area, due to the high degree of terrain fragmentation. (5) The accuracy of yield estimation model based on STNLFFM NDVI (R²=0.849, MAPE=5.47%) was significantly higher than that based on MODIS NDVI (R²=0.113, MAPE=15.65%), indicating that the spatial-temporal fusion technology can effectively improve the yield estimation accuracy of summer maize. It concludes that this study provided a new method for collaborative yield estimation by multi-source remote sensing data fusion and light utilization model.