基于MODIS-OLI遥感数据融合技术的农田生产力估算

大范围、高精度的农田生产力遥感监测依赖于高时空分辨率的遥感数据,单纯依靠由单一类型传感器数据获取的高时相或者高空间分辨率的遥感数据都不能满足清晰掌握田块尺度上作物生长动态的需求。全球免费提供的空间分辨率250~1 000 m的MODIS数据和空间分辨率30 m的Landsat数据是植被动态监测普遍应用的数据源,针对应用MODIS数据估算的农田生产力空间分辨率较低而Landsat卫星重访周期长的局限性,研究基于空间分辨率30 m的Landsat 8 OLI数据与空间分辨率500 m的MODIS数据,应用时空数据融合技术,融合OLI数据的高清晰空间表达能力与时间间隔8 d的MODIS数据的植被生长时间序列过程的监测能力,获得空间分辨率30 m、时间步长8 d的时间序列数据,利用VPM （Vegetation Photosynthesis Model）模型以宁夏永宁县部分地区为试验区估算该区域的NPP。研究结果表明,融合后所得30 m分辨率的NPP具有良好的空间细节信息,提高了MODIS数据中混合像元上的估算精度,并保留了MODIS数据原始的时间过程信息,以30 m的空间分辨率刻画出作物的生长动态;较单独应用MODIS数据,使用融合数据估算的NPP可更有效检测出高标准农田建设对农田生产力的提升。

Agricultural Productivity Estimation with MODIS-OLI Fusion Data

Large-scale and high-precision of agricultural productivity monitoring depends on remote sensing with high spatio-temporal resolution. Remote sensing data with high spatial resolution or temporal resolution acquired by single type of sensor cannot meet the need of clearly monitoring dynamic crop growth on farmland parcel scale. MODIS data with 250-1 000 m spatial resolutions and Landsat data with 30 m spatial resolution are generally used to monitor vegetation dynamics. To supply the gaps of low spatial resolution of MODIS data and long revisit period of Landsat data, this study used Landsat 8 OLI data with 30 m spatial resolution and MODIS data with 500 m spatial resolution and 8-day temporal resolution as data sources, and adopted data fusion technique to fuse high spatial resolution of OLI data and high temporal resolution of MODIS data. Using this technology, the time-series data with 30 m spatial resolution and 8-day temporal resolution were acquired. We took Yongning County in Ningxia as the study area, and used VPM (Vegetation Photosynthesis Model) to estimate the NPP in this area. Results show that, there was high consistency between fused vegetation indexes and OLI vegetation indexes, the determination coefficient of EVI and LSWI being 0.70 and 0.51, respectively. Fused NPP data with 30 m spatial resolution has better detailed information. This data improves the estimated accuracy of mixed pixels in MODIS images, while retains original time and process information of MODIS data. Fused NPP data was consistent with the NPP obtained with MODIS data in pixels where farmland accounted for more than 30% of the mixed pixel, meanwhile fused NPP data was significantly higher than NPP calculated form MODIS data in pixels where farmland accounted for less than 30% of the mixed pixel, since fused NPP had distinct boundaries while NPP calculated from MODIS had not. Fused NPP data show the growing of crop with 30 m spatial resolution. Compared with studies that use MODIS data with 500 m resolution and MOD17 product with 1 000 m resolution, NPP data estimated by fused data can more effectively detect the promotion of agricultural productivity generated by high standard farmland construction. The difference between regions with high standard farmland construction and the neighbors calculated with MODIS-OLI data, VPN-MODIS data and MOD17 data were 62.66, 39.87 and 2.90 g C/(m²·a), respectively.