基于遥感抽样的农作物灾害损失评估方法(Ⅰ)——模拟实验研究

受灾面积与受灾等级是农作物灾情评估的重要内容,本文基于遥感辅助抽样调查的方法设计了一套灾情评估流程,实现对受灾面积与等级的快速评估。本文以Landsat-8影像为实验数据,高分一号影像为模拟真值数据,对济南市长清区2013年夏玉米内涝进行了模拟实验研究。实验结果表明,采用回归估计与比率估计两类常用估计量进行外推,总受灾面积反推精度均在90%以上,各等级受灾面积的反推精度均在80%以上,这证明了遥感数据辅助抽样设计的方法应用于作物的灾害评估具有精度可行性;而且通过一次抽样调查可同时完成受灾面积和受灾等级评估,所以该套方法可以提高评估效率,降低评估成本。由于影像数据、反演指标、抽样方法、外推估计量等多种因素会对评估精度造成不确定性,将来仍需通过实证研究对该法进行进一步验证。

Crop loss assessment based on remote sensing and statistical sampling techniques

It is very important to assess degree and area of crop loss accurately and quickly for a variety of purposes. In this paper, a method is designed as to evaluate crop losses by combining remote sensing and statistical sampling techniques. We conducted a simulation experiment to evaluate the losses of water-logging disaster for a summer maize field, by Landsat-8 and GF-1 remote sensing imagery, in Changqing district, Jinan City in 2013. Results indicate that the total affected areas is above 90% and the estimation accuracy of affected areas at different disaster degree is above 80%, which demonstrated that it is available for assessing the areas affected by disaster. And the proposal method reduces the cost of investigation and improves the survey efficiency by estimating the disaster degree and affected areas at the same time. However, assessment accuracy of the proposal method would be affected by imagery data, inversion index, sampling method and estimator. Thus, the practical performance using this method need to be studied in future experiments.