多源卫星遥感秸秆焚烧过火面积动态监测

该文针对秸秆焚烧过火面积动态监测中高时间、高空间分辨率难以同时实现的问题,提出利用多源数据(Landsat8、GF-1、HJ-1A/B)来提升中分辨率卫星遥感的观测频次,并通过叠置分析和面向对象影像分析技术提高面积提取精度。使用该方法对河南省太康县进行了8次单日内全覆盖的秸秆焚烧过火面积提取,并通过变化检测获取各乡镇农田秸秆焚烧的累计过火面积、新增过火面积和新增过火农田翻耕面积的时空变化趋势。经验证,面积提取精度达93.89%以上,秸秆焚烧新增过火面积变化趋势与环保部监测结果基本相符。经分析,秸秆焚烧通常会在农作物大面积收割后的某个时间点开始,由若干个起火点随时序朝某个主方向进行传播蔓延,过火区域会随之出现间歇性的大范围翻耕,二者同时进行,即秸秆焚烧新增过火面积与新增过火农田翻耕面积随时序呈反向波浪状变化。说明相比利用低空间分辨率遥感数据进行广域监测,该方法可得到时效性强且精度更高的过火面积空间分布信息,能揭示出秸秆焚烧现象在县、乡尺度上的变化规律与细节。

Dynamic monitoring of straw burned area using multi-source satellite remote sensing data

Abstract: As a general definition, open field burning is the burning of living and dead vegetation. An annual average amount of 730 Tg biomass was burnt in Asia, out of which 250 Tg came from agricultural burning. Burning straw after harvest was common, and it was a significant seasonal source of air pollution, which should not be ignored in China. In recent years, straw combustion was serious in Henan Province in autumn, where mechanized farming was practiced, for the farmers were more inclined to burn the crop residues. At present, remote sense monitoring is a practical solution for detection and assessment of this burning. Many researchers used MODIS (moderate resolution imaging spectroradiometer) and FY data to monitor the straw combustion, but the spatial resolution of these data was low and cannot satisfy the requirement of high frequency and high precision monitoring. Especially, many mixed pixels exist in MODIS and FY remote sensing data, which aggrandized the difficulties to get the spatial distribution with high frequency and precision. So, effective and quick means were necessary to deal with this key problem. Generally, high frequency satellite observations could inverse the changing process of straw burned areas. In the present study, Landsat8, GF-1 and HJ-1A/B data were used comprehensively to improve the remote sensing spatial resolution, while the overlay analysis and the object-oriented image analysis (OOIA) methods were adopted to extract the straw burned areas in Taikang County. Based on the OOIA, the remote sensing interpretation sign was established through the ground investigation, and the straw burned area was extracted with a multi-term single-day form. Straw burned areas of 8 stages were extracted using the full-coverage remote sensing images. With the changing detection at the township scale, the temporal change trend of cumulative straw burned area, new added straw burned area and new added farmland plowing area after straw burned were calculated. The spatiotemporal spreading trend of straw burning showed that after the maize harvest, straw burned started at a certain point in time after a large area of crop was harvested, and spread from a number of fire points to a main direction with the time. The new added straw burned area changed with a wavy pattern, due to that intermittent large-scale plowing occurred subsequently in the added burned area. The rate of plowing was beyond the rate of straw burned, and the incineration activity tended to end. Compared with field observed data, the calculated area extraction accuracy was above 93.89%, and the calculated change trend of new straw burned area was basically consistent with the monitoring results of the Ministry of Environmental Protection. Experiment results have indicated that the method presented in this study is timely and accurate, which can reveal more details and regularities than traditional large-scale application of low spatial resolution satellite remote sensing data.