Agriculture as a funding source of ISIS: A GIS and remote sensing analysis

Agriculture is an important source of income for the Islamic State in Syria and Iraq (ISIS), which currently rules over large parts of the breadbaskets of the two countries. It has received limited attention compared to other sources of ISIS revenues such as oil, looting, ransom, foreign donations and various forms of taxation. We estimate winter crops production of wheat and barley in ISIS-controlled areas in both Syria and Iraq for the years 2014–2015 and irrigated summer crops production (cotton) in Northeast Syria. We show that remote sensing can give a credible estimation of agricultural production in the absence of statistics. With evidence from MODIS Aqua and Terra Satellites as well as Landsat imagery, we find that agricultural production in ISIS-controlled Syrian and Iraqi zones has been sustained in 2014 and 2015, despite the detrimental impact of conflict. After a drought in 2014 production was able to capitalize on improved rainfalls in 2015. First indications show that the winter grain harvest of 2016 in Iraqi territories of ISIS was significantly above pre-conflict mean and below pre-conflict mean in its Syrian territories. We also show how water flows along the Euphrates have impacted production. We estimate the revenue that ISIS can derive from wheat and barley production and the likely magnitude of an exportable surplus. Agricultural production gives the group a degree of resilience, although its economy is not sustainable in the longer run and could be affected by military collapse. Taxation of recurrent income streams such as agriculture will become more important for ISIS as its extractive sources of revenues show signs of dwindling. Beside non-grain food imports, agricultural production is crucial for its political legitimacy by ensuring food provision to the broader population. Food security considerations would require a high priority in any post-ISIS reconstruction effort and would need to include the rehabilitation of supply chains for agricultural inputs such as quality seeds and fertilizers.     

Deforestation and land tenure in Mexico: A response to Bonilla-Moheno et al.

Bonilla-Moheno et al. published an interesting paper on land tenure as a determinant of deforestation in Mexico, in Land Use Policy in 2013. In this review, we focus critically on the methodology used by the authors to reach two key conclusions. Firstly, we suggest that their use of coarse resolution MODIS EVI data to assess deforestation rates may have resulted in an erroneous conclusion, that forest area in Mexico is increasing rather than decreasing. This finding is contrary to official data and to other recent scientific assessments. Secondly, we question their conclusion that territories under the tenure of ‘comunidades indigenas’ are less likely to experience deforestation, and more likely to experience expansion of forest area, than those under ‘ejidal’ tenure. We believe that the sampling method, which involved the selection of municipalities with high concentrations of one particular type of tenure, and which eliminated the 60% of municipalities which contain mixed forms of tenure, biased the sample, particularly for the case of comunidades. The comunidades included in the sample are almost all found in a single region of the country, which may represent a special case. In reality comunidades are much more widespread through the country. Moreover we find that the explanation given by the authors for the relative success of comunidades in conserving their forests is not founded in data from any study of process at the local level.     

基于多尺度分割的面向对象分类方法提取冬小麦种植面积

应用面向对象方法,对遥感图像进行多尺度分割,即首先进行大尺度分割,结合NDVI提取植被信息,将图像分为植被和非植被;然后在植被信息类内再进行小尺度分割,利用NDVI并融入几何特征进一步提取冬小麦种植面积及空间分布。在遥感分类的基础上,将线性数据按宽度缓冲,从分类结果中扣除。将扣除结果与地面样方实测数据对比分析。结果表明,监测结果减轻了传统分类方法的椒盐效应,监测结果与验证样方数据比较精度为94.06%。     

基于扩散策略的分布式协作频谱检测

分布式认知网络中,认知用户通过与相邻用户交换检测量信息,达到全网一致,实现对主用户信号的检测。为了提高网络检测量信息的收敛速度,提出了一种基于扩散策略的分布式协作检测算法。在该算法中,将最大最小特征值检测算法的检测量作为交换的初始信息,构造自适应矩阵和融合矩阵作为加权因子对认知节点状态值进行迭代更新,实现全网认知用户检测量信息一致,各认知用户独自根据融合的检测量信息,进行最终的检测判决,确定主用户是否存在。仿真结果表明,该算法在网络收敛速度和检测性能上较共识策略和非合作检测有不同程度的提升。     

压缩感知的发展与应用

压缩感知（Compressed Sensing,CS）是近年来新兴的一种信号获取技术。沿着 CS理论的发展历程介绍了CS理论框架,给出了其严格的数学描述,着重讨论了对原始信号的 重构技术,最后介绍了一些可能的应用。     

西双版纳勐养保护区森林火险空间分布的研究

根据林火的燃烧反应式： Ｃ６ Ｈ１０ Ｏ５ ＋ ６ Ｏ２ 燃点温度 ６ Ｃ Ｏ２ ＋ ５ Ｈ２ Ｏ＋ 热量↑,由反应式得出,林火的要素包括：森林可燃物、氧气和温度。森林可燃物随森林植被类型而异,而氧气普遍存在,燃烧温度可分解为火源和立地干燥度。根据研究地区的实际情况,其火险要素具体包括：森林植被类型、人为火源分布和立地坡向。本项研究,利用遥感和地形图采集火险数据, Ｇ Ｉ Ｓ进行火险集成。最后得出森林火险空间分布,并划分为不燃、难燃、可燃、易燃和强燃 ５ 个等级。研究结果与过去的林火统计材料相比较,基本吻合,但方法更科学合理。该项研究,对防火工作提供决策支持。     

GIS支持下的江河源区土地覆盖遥感监测

利用1986年TM和2000年ETM卫星遥感数据,在GIS的支持下对江河源区土地覆盖变化进行了监测,14年来该区草地、林地等地表自然植被面积减少,湖泊、湿地萎缩,沙地、裸地面积扩张,荒漠化程度加剧;气候变异引起冻土退化,叠加上不合理的人类活动,是导致研究区土地覆盖变化的主要原因。     

基于遥感反演的特大城市颗粒物浓度与 三维绿量相关性研究

颗粒物污染对人类健康和环境的可持续发展具有重要影响,宏观尺度监测空气颗粒物污染与三维绿量的关系对缓解空气污染具有重要意义。既往基于遥感反演的植被覆盖与颗粒物污染的研究主要集中在二维绿量方面,对于三维绿量与AOD代表的颗粒物污染的深入研究很少,旨在从三维绿量缓解颗粒物污染的角度为城乡规划设计提供定量化的参考。以中国特大城市武汉为对象,通过遥感反演大气气溶胶光学厚度(AOD)、landsat 8解译土地覆盖类型并获取比值植被指数RVI,进而根据回归模型获取三维绿量,分别在500m×500m、1km×1km、2km×2km、3km×3km的网格尺度上,探讨AOD与三维绿量的相关关系。结果表明：4种尺度下,都在0.01水平上呈极明显负相关,且3km尺度最显著,进而得到4种网格尺度的相对最优回归模型作为三维绿量与AOD的预测模型,揭示了两者的初步变化规律;进而探析了AOD、三维绿量及两者之间的空间分布规律,发现它们都具有显著的空间集聚特征;最后根据三维绿量形成中度及以上污染的阈值,提出缓解颗粒物污染的三维绿量调控策略。从之前常见的二维绿量对空气污染的缓解进一步扩展到三维绿量和颗粒物污染的缓解研究,旨在从缓解空气污染视角为相关规划设计提供理论参考。     

频谱检测门限与n-out-of-K融合规则的联合优化

为了减少计算复杂度,提出联合优化检测门限λ和n-out-of-K融合规则的 算法。以λ和n为参数建立目标函数,并将参数以二进制形式表示,从而把算法转化为 组合优化问题。接着,采用基于样值修改的互熵优化方法渐次逼近最优的参数。仿真表明, 该算法在获得与已有算法几乎相当的总错误率情况下,可有效降低平均搜索次数,且随着K 的增加搜索次数增加更平缓。     

一种基于RAMP算法的OFDM稀疏信道估计方法

针对OFDM稀疏信道估计需要信道稀疏度作先验条件的不足,将正则化自适应匹配追踪（RAMP ） 用于信道重建,可在信道稀疏度未知的情况下,自适应地调整候选集原子的个数,并利用正 则化过程实现支撑集的二次筛选,逐步扩大支撑集,准确地估计出信道的冲激响应。仿真结 果表明,该方法收敛速度快,估计效果好,有较好的应用价值。