论文部分内容阅读
为探讨利用国产HJ卫星遥感数据进行农作物种植面积估算及农业气象灾害定量监测评估的可行性,以江汉平原南部县市为研究区,利用2011年6景不同时相的HJ卫星CCD遥感数据,结合野外调查样方点和面积较大的试验基地样本,通过分析研究区主要秋收作物(棉花、一季中稻)不同生育期的光谱特征和归一化植被指数(NDVI)时序变化特征,对分类影像进行序列阈值分割、掩膜处理,最后利用决策树算法成功提取了棉花和一季中稻的种植面积,得到面积精度和样本点精度均大于90%,Kappa系数为0.983 5的结果。只要不受到云和降水的影响,能获取江汉平原区域关键时相(5月中下旬、6月上旬、7月上旬、7月下旬或8月上中旬、10月上旬)HJ卫星CCD影像数据,便能很好地应用于江汉平原棉花和中稻作物提取。
In order to investigate the feasibility of using HJ satellite remote sensing data to estimate crop planting area and quantitatively evaluate and monitor agricultural meteorological disasters, taking southern counties and cities of Jianghan Plain as study area and HJ satellite CCD remote sensing data of 6 different phases in 2011, Field survey sampling sites and large sample bases were used to analyze the spectral characteristics and the normalized NDVI seasonal variation characteristics of main autumn harvest crops (cotton, mid season rice) in the study area to classify the images Sequence threshold segmentation and mask processing. Finally, the planting area of cotton and one-season middle rice was successfully extracted by using the decision tree algorithm. The results showed that both area precision and sample point accuracy were over 90% and Kappa coefficient was 0.983 5. As long as they are not affected by clouds and precipitation, they can acquire the key phases of the Jianghan Plain region (late May, early June, early July, late July or mid-August, early October) HJ satellite CCD image data, Can be well applied to Jianghan Plain cotton and mid-crop extraction.