AgsSAT Multiannual (2017-2021) Sentinel-2 Water Indices Composites

AgsSAT Multiannual (2017-2021) Sentinel-2 Water Indices Composites The 2,815 images available for the state of Aguascalientes, Mexico for the years 2017 to 2021 were processed using the Open Data Cube (ODC) platform [Lewis et al. (2017), Gavin et al. (2018), https://www.opendatacube.org/]. These images correspond to multiple coverages of the region of interest. The images were then used to generate cloud-free annual composites by applying geometric median (geomedian) algorithm, as defined in [Roberts et al. (2017)]. Geomedian algorithm produces a pixel-level summary for every pixel, in this case this means that each summary corresponds to a 10m x 10m region in the territory and its observations throughout a calendar year. All these summary pixels form a 12-band (coastal aerosol, blue, green, red, vegetation red edge 5, vegetation red edge 6, vegetation red edge 7, near-infrared, narrow nir, water vapor, swir1 and swir2) composite of the state of Aguascalientes. Another product called GeoMad was generated, which calculates the robust dispersion statistic called MAD, as defined in [Roberts, D., Dunn, B., & Mueller, N. (2018)]. In the resulting image composite, each of the three-pixel bands represents the variation over three distances: Spectral Distance (smad), Euclidean Distance (emad) and the Bray-Curtis Distance (bcmad). More bands were generated to represent different environmental conditions during the study years (2017-2021), these conditions can be captured by analyzing various combinations of bands, these combinations are also called spectral indices, which allow detecting vegetation, presence of water, urbanization, etc., Finally, 28 indices divided into 4 categories were calculated: Vegetation Indices (Atmospherically Resistant Vegetation Index, Kaufman 1972) (Enhanced Vegetation Index, Huete 2002): (Modified Soil Adjusted Vegetation Index, Qi Et Al. 1994) (Normalized Difference Chlorophyll Index, Mishra & Mishra, 2012) (Normalised Difference Moisture Index, Gao 1996) (Normalized Difference Vegetation Index, Rouse 1973) (Optimized Soil Adjusted Vegetation Index, Rondeaux. 1996) (Simple Ratio Vegetation Index Jordan, C.F.1 969) (Soil Adjusted Vegetation Index, Huete 1988) (Visible Atmospherically Resistant Index, Gittleson 2002) Built-up Indexes (Band Ration For Built-up Area, Waqar 2012) (Built-up Area Extraction Index, Bouzekri 2015) (Built-up Index, He Et Al. 2010) (Index-based Built-up Index, Xu 2008) (New Built-up Index, Jieli Et Al. 2010) (Normalized Difference Built-up Index, Zha 2003) (Normalized Built-up Area Index, Waqar 2012) (Urban Index, Kawamura 1996) Water Indices (Modified Normalized Difference Water Index, Xu 1996) (Normalized Difference Water Index, Mcfeeters 1996) (Water Index, Fisher 2016) Other Indices (Bare Soil Index, Rikimaru Et Al. 2002) (Bare Soil Index, Wanhui 2004) (Burn Area Index, Martin 1998) (Clay Minerals Ratio, Drury 1987) (Ferrous Minerals Ratio, Segal 1982) (Iron Oxide Ratio, Segal 1982) (Normalized Burn Ratio, Lopez Garcia 1991) (Normalised Difference Snow Index, Hall 1995).

AgsSAT多时序（2017-2021）哨兵二号（Sentinel-2）水体指数合成数据集。
本数据集针对墨西哥阿瓜斯卡连特斯州2017至2021年间获取的2815幅影像，采用开放数据立方（Open Data Cube, ODC）平台进行处理[Lewis等，2017；Gavin等，2018；https://www.opendatacube.org/]。这些影像覆盖了研究区域的多轨观测范围。
随后基于这些影像，采用文献[Roberts等，2017]中定义的几何中值（geomedian）算法生成无云年度合成影像。几何中值算法可针对每个像素生成像素级统计汇总；在本数据集中，每一项汇总对应研究区内一处10米×10米的区域及其一整自然年的观测数据。所有这些汇总像素共同构成阿瓜斯卡连特斯州的12波段合成影像，波段分别为：海岸气溶胶波段、蓝光波段、绿光波段、红光波段、植被红边5波段、植被红边6波段、植被红边7波段、近红外波段、窄近红外波段、水汽波段、短波红外1波段（swir1）与短波红外2波段（swir2）。
此外还生成了名为GeoMad的衍生产品，其基于文献[Roberts D, Dunn B, & Mueller N, 2018]中定义的MAD稳健离散统计量进行计算。在生成的影像合成产品中，三个像素波段分别对应三种距离的变异程度：光谱距离（Spectral Distance, smad）、欧氏距离（Euclidean Distance, emad）与布莱-柯蒂斯距离（Bray-Curtis Distance, bcmad）。
研究期间（2017-2021）还生成了更多波段以表征不同环境条件；通过分析波段的多种组合可捕捉这些环境特征，这类波段组合亦被称为光谱指数，可用于识别植被、水体分布与城市化进程等。最终共计算得到28项指数，划分为4大类：
1. 植被指数：抗大气植被指数（Atmospherically Resistant Vegetation Index，Kaufman，1972）、增强型植被指数（Enhanced Vegetation Index，Huete，2002）、修正土壤调整植被指数（Modified Soil Adjusted Vegetation Index，Qi等，1994）、归一化差分叶绿素指数（Normalized Difference Chlorophyll Index，Mishra & Mishra，2012）、归一化差分水分指数（Normalized Difference Moisture Index，Gao，1996）、归一化差分植被指数（Normalized Difference Vegetation Index，Rouse，1973）、优化土壤调整植被指数（Optimized Soil Adjusted Vegetation Index，Rondeaux，1996）、简单比值植被指数（Simple Ratio Vegetation Index，Jordan，1969）、土壤调整植被指数（Soil Adjusted Vegetation Index，Huete，1988）、可见光抗大气指数（Visible Atmospherically Resistant Index，Gittleson，2002）；
2. 建筑指数：建筑区波段比值指数（Band Ratio for Built-up Area，Waqar，2012）、建筑区提取指数（Built-up Area Extraction Index，Bouzekri，2015）、建筑指数（Built-up Index，He等，2010）、基于指数的建筑指数（Index-based Built-up Index，Xu，2008）、新型建筑指数（New Built-up Index，Jieli等，2010）、归一化差分建筑指数（Normalized Difference Built-up Index，Zha，2003）、归一化建筑区指数（Normalized Built-up Area Index，Waqar，2012）、城市指数（Urban Index，Kawamura，1996）；
3. 水体指数：修正归一化差分水体指数（Modified Normalized Difference Water Index，Xu，1996）、归一化差分水体指数（Normalized Difference Water Index，Mcfeeters，1996）、水体指数（Water Index，Fisher，2016）；
4. 其他指数：裸土指数（Bare Soil Index，Rikimaru等，2002）、裸土指数（Bare Soil Index，Wanhui，2004）、烧伤面积指数（Burn Area Index，Martin，1998）、黏土矿物比值指数（Clay Minerals Ratio，Drury，1987）、亚铁矿物比值指数（Ferrous Minerals Ratio，Segal，1982）、氧化铁比值指数（Iron Oxide Ratio，Segal，1982）、归一化燃烧比指数（Normalized Burn Ratio，Lopez Garcia，1991）、归一化差分雪指数（Normalized Difference Snow Index，Hall，1995）。