Estrada-Peña Agustin, Vatansever Zati, Gargili Aysen, Buzgan Turan
Department of Parasitology, Faculty of Veterinary Medicine, University of Zaragoza, Zaragoza, Spain.
Geospat Health. 2007 Nov;2(1):127-35. doi: 10.4081/gh.2007.261.
In the last few years, Crimean-Congo haemorrhagic fever (CCHF) has been reported as an emerging tickborne disease in Turkey. This paper deals with the preparation of an early warning system, aimed to predict the beginning of the CCHF season in Turkey based on a clear, simple and repeatable remotely-sensed signal. Decadal (mean of 10 days) values of the normalized difference vegetation index (NDVI) at 1 km resolution were used on a set of 952 confirmed, accurately geo-referenced, clinical cases between 2003 and 2006. A prerequisite is that the signal should be observable between 2 and 3 decadals before a given moment of the season to be of value as early warning. Decadals marking the 10th percentile or the 25th quartile in the frequency distribution of case reporting were selected as markers for the beginning of season of risk. Neither raw nor accumulated decadal NDVI signals were able to predict the onset of this season. However, when we defined the NDVI anomaly (NDVIa) as the positive difference between decadal NDVI values and the average for the previous year, this standardized measure gave a homogeneous overview of the changes in the NDVI signal producing a NDVIa slope for the decadals 10 to 13 that was always greater than 0. We conclude that observing this slope over time can be used as an early-warning system as it would predict the build-up of the number of cases 20 days in advance with an accuracy of 82% (10th percentile) or 98% (25th quartile).
在过去几年中，克里米亚-刚果出血热（CCHF）在土耳其被报告为一种新出现的蜱传疾病。本文论述了一个早期预警系统的准备工作，其目的是基于一个清晰、简单且可重复的遥感信号来预测土耳其CCHF季节的开始。在2003年至2006年期间，对952例确诊的、地理定位准确的临床病例，使用了分辨率为1千米的归一化植被指数（NDVI）的十年期（10天平均值）数据。一个前提条件是，该信号应在季节某一特定时刻前2至3个十年期内可观测到，才能作为有价值的早期预警。在病例报告频率分布中标记第10百分位数或第25四分位数的十年期被选作风险季节开始的标志。原始的或累积的十年期NDVI信号均无法预测该季节的开始。然而，当我们将NDVI异常（NDVIa）定义为十年期NDVI值与上一年平均值之间的正差值时，这种标准化度量给出了NDVI信号变化的统一概况，产生了一个十年期10至13的NDVIa斜率，该斜率始终大于0。我们得出结论，随着时间观测这个斜率可被用作一个早期预警系统，因为它能提前20天预测病例数的增加，准确率为82%（第10百分位数）或98%（第25四分位数）。
