Inter- and intra-annual relationships between water clarity and river loads in the Great Barrier Reef 2002-2013 (NERP TE 4.1, AIMS, sources: NASA, DEHP, DERM, BOM, UQ)

This dataset shows various statistics of photic depth across the Great Barrier Reef (GBR). Data are broken into 35 zones along and across the GBR and photic depth is derived from 11 years of MODIS Aqua data. The data included is:\n1. The statistical strength of correlation between standardized photic depth and freshwater discharges the GBR. \n2. The mean photic depth and the main physical environmental variables that need to be controlled for when assessing how volumes of river freshwater discharges influence photic depth.\n3. Statistics of photic depth controlled to remove the effects of main physical environmental variables (wave height, tidal range) used when assessing how volumes of river freshwater discharges. Data are into dry years (2002 to 2006) or wet years (2007 ¿ 2012).\n\nWater clarity is a key parameter affecting the health of coastal marine systems and their tourism values. We investigated the relationship between volumes of river freshwater discharges of major rivers (from DERM) and the water clarity in 35 zones along and across the GBR waters within the Fitzroy, Whitsundays, Burdekin, Southern and Northern Wet Tropics. For Cape York, water clarity was related to rainfall as a proxy, since river data were incomplete. We used daily 11-years (2002-2013) MODIS-Aqua remote sensing data at 1 km2 resolution, to investigate time scales and processes affecting water clarity in these regions. In all coastal, inshore and lagoonal regions except for Cape York, photic depth was strongly negatively related to the freshwater discharge of the main rivers. The declines started with the onset of river floods, and water clarity typically took 150¿ 260 days until complete recovery. The relationship between photic depth and rivers was strongest in the Northern Wet Tropics, the initiation area of outbreaks of crown-of-thorns starfish, where effects were strong even on the outer shelf. Previous conclusions that river runoff predominantly affects the inshore of the GBR have therefore to be revised for the Central and Northern GBR. The results were used in the setting of regional ecologically relevant targets for fine sediment in the Burnett-Mary and Wet Tropics WQIPs, and will likely be used for other WQIPs.\n\nThe analyses are based on three sets of data: \n1) Daily Modis Aqua satellite data from 2002 - 2013, processed as described previously (Weeks et al. 2012, Logan et al. 2013, Fabricius et al. 2014). \n2) Daily data of freshwater discharge volumes of the main rivers for the same time period, provided by the State of Queensland, Department of Environment and Heritage Protection (DEHP). \n3) For the Normanby River, the discharge station only came online late 2005. Therefore, most of the first four years of daily discharge data for the main river in this region are missing (Stewart and Endeavour Rivers are much smaller than the large Normanby). Also missing are any form of river discharge information for the whole northern half of the region. As an alternative to river discharge data, we used daily rainfall data from the Lockhart River rainfall gauge for Cape York, which is located relatively centrally in this ~400 km long band. Daily rainfall data were obtained from the Australian Bureau of Meteorology (http://www.bom.gov.au/oceanography/projects/abslmp/data/index.shtml).\n\n\nMethod:\n\nWe spatially aggregated the data into 15 zones for the Fitzroy and Whitsundays region, and 5 zones each for the Burdekin, Southern and Northern Wet Tropics, and Cape York regions. For the Whitsundays, Burdekin, Wet Tropics and Cape York, five bands were defined parallel to the coastline:\n- Coastal: 0 ¿ 0.1 fractional units across the GBR\n- Inshore: 0.1 ¿ 0.25 fractional units across the GBR\n- Lagoon: 0.25 ¿ 0.45 fractional units across the GBR\n- Midshelf: 0.45 ¿ 0.65 fractional units across the GBR\n- Outer shelf: 0.65 ¿ 1 fractional units across the GBR\n\nThe Fitzroy region cannot be partitioned up into simple coast-parallel bands, due to its geomorphology around to the Capricorn-Bunkers and Swains complex, and the estuarine Keppel Bay. Consequently, the Fitzroy region was partitioned according to a combination of geomorphological regions and boundary rules (based on distances from coastlines and bioregions) so to reflect its oceanographic and geomorphological characteristics. The Broad Sound was analyzed separately, as its high tidal range and distance from the major Whitsundays and Fitzroy Rivers make this area unrepresentative of the more intensely used and populated areas of the Whitsundays and Fitzroy NRM Regions. The boundaries were chosen to best match those of both the Whitsundays and Fitzroy areas.\n\nThe Cape York and Wet Tropics NRM regions were subdivided into three long-shore bands, with the ¿Cape York¿ band extending to 14.5 degrees latitude (Lizard Island), and a northern Wet Tropics region, split at Cape Grafton south of Cairns), and the southern Wet Tropics to best capture their differences in geomorphology, rainfall, agricultural use patterns, and population outbreak dynamics of crown-of-thorns starfish.\nThe statistical methods to relate photic depth to river discharges are described in Fabricius KE, Logan M, Weeks S, Brodie J (2014) The effects of river run-off on water clarity across the central Great Barrier Reef. Marine Pollution Bulletin 84: 191-200, and in Murray Logan, Katharina Fabricius, Scarla Weeks, Ana Rodriguez, Stephen Lewis and Jon Brodie (2014) NERP Project 4.1: Tracking coastal turbidity over time and demonstrating the effects of river discharge events on regional turbidity in the GBR. NERP Progress Report: Southern and Northern NRM Regions. 63 pp\nPhotic depth: The daily 1 km2 MODIS-Aqua remote sensing data were processed as described by Weeks et al. 2012, Fabricius et al. 2014. Masks were generated to excise optically shallow waters (reefs and very shallow coastal sections of the seabed), and offshore to >200 m bathymetry. As the full gridded daily data series is too large to reside in memory (153,177 grid points per day, over 11 years), it was spatially aggregated into the 35 zones. Data were aggregated to water years (1st October to 30th September) rather than calendar years. Data availability varied greatly between days and months due to cloud cover. To explore temporal differences in photic depth between wet and dry years, the analyses were also performed separately for dry (2002-2006) and wet (2007-2012) years.\n\nPredicted daily tidal amplitudes as a proxy for tidal currents were obtained from the Australian Navy. For each zone, a single tidal location or a set of ¿representative¿ tidal locations was chosen, and the mean tidal range per day was calculated across these locations, to reduce computational exhaustion.\n\nHourly data on wave heights and wave frequencies were obtained from the Queensland State Government, Department of Environment and Heritage Protection (DEHP), from the 4 wave rider buoys available in the study region: Emu Point Buoy for the southern zones, Mackay Buoy for the Whitsunday zones, Townsville Buoy for the Burdekin zone, and Cairns Buoy for the Northern and Southern Wet Tropics. For the Cape York zones, wind data from the Bureau of Meteorology (http://www.bom.gov.au/oceanography/projects/abslmp/data/index.shtml) from Lockhart River were considered more representative than the wave data from the Cairns buoy.\n\nThe analyses was based on daily values and performed separately for each zone. In order to explore the long-term photic depth signals, the data were seasonally detrended and smoothed. Gradient boosted model (GBM) and generalized additive mixed effects models (GAMM) were fitted to remove the effects of tides and wind/waves. The residuals from these GAMM (which thus reflect the photic depth signal after the extraction of wave, tidal and bathymetry signals) were then decomposed to derive the intra-annual trends (i.e., seasonal based on 365.25 day cyclicity) and inter-annual trends in photic depth. Seasonal decomposition was chosen which applies a smoother (typically either a moving average or locally weighted regression smoother) through a time series to separate periodic fluctuations due to cyclical reoccurring influences and long-term trends. Following temporal decomposition, seasonal cycles were re-centered around mean GAMM fitted values, and transformed back into the original photic depth scale via exponentiation.\n\n\nLimitations:\n\nThe analyses only investigated the effects of river runoff on water clarity. This does not indicate that other factors (e.g. coastal developments, dredging) do not additionally affect water clarity; such relationships would have to be investigated separately.\n\n\nFormat:\n\nThis dataset comprises 2 shape files and a csv file:\n- FabriciusAndLoganNerpDataCorrelations.* (142 kb) (dbf, shp and shx files),\n- FabriciusAndLoganNerpDataSummaries.* (142 kb) (dbf, shp and shx files) and\n- FabriciusAndLoganNerpSeasonalStatsDataRound.csv (3 kb).\n\n\nData Dictionary:\n\nFabriciusAndLoganNerpDataCorrelations.shp:\n\nThe shapefile contains a set of polygon zones for 35 zones in the entire. The attributes table contains the strength of the correlation between daily river discharge and daily satellite photic depth, over 11 years. \n\nThe attributes are: \n- SP_ID: shape id\n- Correlatio: correlation value\n\nFabriciusAndLoganNerpDataSummaries.shp:\n\nThe shapefile contains a set of polygon zones for 35 zones in the entire GBR. In each zone, we calculated the mean values of hourly or daily values, over 11 years. \n\nThe attributes are:\n- SP_ID: shape id\n- Photic_dep: photic depth (meters), means over 11 years of daily photic depth values, calculated based on an algorithm developed by Scarla Weeks (UQ) and NASA, (equivalent to Secchi depth)\n- Tidal_rang: tidal range (meters), means over 11 years of tidal range values (difference between highest and lowest sea-level within each day), calculated from tidal predictions of the Australian Navy.\n- Wave_heigh: wave height (meters), means over 11 years of wave height values, calculated from the nearest one of the four coastal DERM Wave Rider Buoys.\n- Wind_speed: wind speed (ms-1), means over 11 years of wind speed values, from the nearest BOM station.\n\nFabriciusAndLoganNerpSeasonalStatsDataRound.csv:\n\nStatistics of photic depth controlled to remove the effects of main physical environmental variables (wave height, tidal range) used when assessing how volumes of river freshwater discharges. Data are broken into 35 zones along and across the GBR as well as into dry years (2002 to 2006) or wet years (2007 ¿ 2012).\n\nThe attributes are:\n- Region: geographical region\n- Zone: Coastal, Inshore, Lagoon, Midshelf, Outershelf\n- Period: Wet or Dry years\n- Maximum: (m) the maximum smoothed photic depth over the year\n- MaxDate: (calendar date) date within a year cycle corresponding to the maximum photic depth, typically middle to end of dry season\n- Minimum: (m) the minimum smoothed photic depth over the year, showing the difference between dry and wet years in some of the inshore zones\n- MinDate: (calendar date) date within a year cycle corresponding to the minimum photic depth, typically middle to end of wet season\n- DeclineTime: (days) duration of time elapsed between the max photic depth and the NEXT minimum photic depth (in the continuous cycle)\n- Decline: (m) absolute difference between max and min photic depth, showing how much photic depth is lost (in absolute terms) between seasons in wet and dry years in some of the inshore zones.\n- PercentDecline: the relative decline expressed as a percentage of the max photic depth (unit: percent)\n- DeclineRate: (m/day) rate of decline\n- RecoveryTime: (days) duration of time elapsed between the min photic depth and the NEXT max photic depth (in the continuous cycle). Note, DeclineTime and RecoveryTime complete the 365(ish) day cycle. Showing how long it takes to re-establish clear water\n- RecoveryRate: (m/day) rate of recovery (Decline/RecoveryTime)\n- Recovery95Date: date within a year cycle corresponding to a recovery of 95% (up to max - decline*0.05)\n- Recovery95Time: (days) duration of time elapsed between the min photic depth and the NEXT 95% recovery in photic depth (in the continuous cycle). Showing how long it takes to re-establish clear water after wet and dry wet seasons\n- Recovery95Rate: (m/day) same as RecoveryRate, yet based on 95% recovery (Decline/Recovery95Time)

本数据集呈现了大堡礁（Great Barrier Reef, GBR）全域各区域的透光深度（photic depth）相关统计量。数据按横跨与纵贯大堡礁的35个分区进行划分，透光深度数据源自11年的MODIS Aqua遥感观测数据。所包含的数据如下：
1. 标准化透光深度与大堡礁区域淡水径流量之间的相关统计强度
2. 平均透光深度，以及在评估河流淡水径流量如何影响透光深度时所需控制的主要物理环境变量
3. 经控制以消除主要物理环境变量（波高、潮差）影响的透光深度统计量，用于评估河流淡水径流量的影响；数据按枯水年（2002年至2006年）与丰水年（2007年至2012年）进行分组。

水体透明度是影响近海海洋生态系统健康及其旅游价值的关键参数。本研究针对菲茨罗伊（Fitzroy）、降灵群岛（Whitsundays）、伯德金（Burdekin）、南部与北部湿热带地区内，横跨与纵贯大堡礁海域的35个分区，探究了主要河流（数据源自DERM）的淡水径流量与水体透明度之间的关联。而就约克角（Cape York）而言，由于河流径流量数据不全，本研究以降雨量作为替代指标来表征水体透明度。本研究采用空间分辨率为1 km²的2002年至2013年每日MODIS-Aqua遥感数据，探究了上述区域内影响水体透明度的时间尺度与过程机制。

除约克角外，所有沿岸、近岸与泻湖区域的透光深度均与主要河流的淡水径流量呈显著负相关。水体透明度下降始于河流洪水暴发之时，且通常需要150至260天才能完全恢复。透光深度与河流径流量的关联在北部湿热带区域最为显著——该区域是长棘海星（crown-of-thorns starfish）暴发的起始区域，即便在外陆架区域也能观测到强烈的影响。因此，此前关于"河流径流主要影响大堡礁近岸区域"的结论，针对大堡礁中部与北部区域而言需要进行修正。本研究结果被用于制定伯内特-玛丽（Burnett-Mary）与湿热带地区的区域生态相关细沉积物质量目标（WQIPs），且有望应用于其他水质改善计划（WQIPs）。

本分析基于三类数据集：
1）2002年至2013年的每日MODIS-Aqua卫星数据，数据处理流程参照此前研究（Weeks等，2012；Logan等，2013；Fabricius等，2014）
2）同期主要河流的每日淡水径流量数据，由昆士兰州环境与遗产保护部（DEHP）提供
3）针对诺曼比河（Normanby River），其径流量监测站于2005年底才投入运行，因此该区域主要河流前四年的每日径流量数据大多缺失（斯图尔特河与恩迪沃弗河的径流量远小于诺曼比河）；同时该区域北部过半区域均无任何河流径流量数据。为此，本研究以约克角区域洛克哈特河雨量站的每日降雨量数据作为河流径流量的替代指标——该雨量站位于这片约400公里长区域的相对中心位置。每日降雨量数据源自澳大利亚气象局（http://www.bom.gov.au/oceanography/projects/abslmp/data/index.shtml）。

## 研究方法
本研究将数据按空间聚合为35个分区：菲茨罗伊与降灵群岛区域划分为15个分区，伯德金、南部湿热带、北部湿热带以及约克角区域各划分为5个分区。针对降灵群岛、伯德金、湿热带与约克角区域，按平行于海岸线的方向划分为5个带：
- 沿岸带（Coastal）：大堡礁全域跨度的0~0.1比例区间
- 近岸带（Inshore）：大堡礁全域跨度的0.1~0.25比例区间
- 泻湖带（Lagoon）：大堡礁全域跨度的0.25~0.45比例区间
- 中陆架带（Midshelf）：大堡礁全域跨度的0.45~0.65比例区间
- 外陆架带（Outer shelf）：大堡礁全域跨度的0.65~1比例区间

由于菲茨罗伊区域的地貌包含摩羯座-邦克群岛与斯温兹群岛复合体，以及河口的凯普尔湾，因此无法按简单的平行海岸线方式进行分区。为此，菲茨罗伊区域结合地貌区域与边界规则（基于海岸线距离与生物区域）进行分区，以反映其海洋学与地貌学特征。布罗德湾（Broad Sound）单独进行分析，因其潮差较大且距离主要的降灵群岛与菲茨罗伊河流域较远，无法代表降灵群岛与菲茨罗伊自然资源管理区（NRM Regions）中开发与人口更为密集的区域。分区边界的设置尽可能匹配降灵群岛与菲茨罗伊区域的原有边界。

约克角与湿热带自然资源管理区进一步划分为3个沿岸带："约克角"带延伸至南纬14.5度（蜥蜴岛区域）；北部湿热带区域以凯恩斯以南的格拉夫顿角为界；南部湿热带区域则单独划分，以充分体现它们在地貌、降雨量、农业利用模式以及长棘海星种群暴发动态方面的差异。

透光深度与河流径流量的关联分析方法详见以下文献：
1. Fabricius KE、Logan M、Weeks S、Brodie J（2014）《河流径流对大堡礁中部区域水体透明度的影响》，《海洋污染通报》（Marine Pollution Bulletin）84卷：191-200页
2. Murray Logan、Katharina Fabricius、Scarla Weeks、Ana Rodriguez、Stephen Lewis、Jon Brodie（2014）《NERP项目4.1：实时追踪沿海浊度变化并验证河流径流事件对大堡礁区域浊度的影响》，《NERP进度报告：南部与北部NRM区域》，共63页

透光深度（photic depth）：每日1 km²分辨率的MODIS-Aqua遥感数据的处理流程参照Weeks等（2012）与Fabricius等（2014）的方法。研究中生成了掩膜层，以剔除光学浅水区（珊瑚礁与极浅的近岸海底区域）以及水深超过200米的远海区域。由于完整的每日网格化数据序列体量过大（11年间每日均有153177个网格点），无法直接载入内存，因此将其空间聚合为前述35个分区。数据按水文年度（10月1日至次年9月30日）而非日历年度进行聚合。由于云量覆盖的影响，每日与每月的数据可用率差异极大。为探究丰水年与枯水年之间透光深度的时间差异，本分析分别针对枯水年（2002年至2006年）与丰水年（2007年至2012年）独立开展。

以每日预测潮汐振幅作为潮流的替代指标，数据源自澳大利亚海军。为降低计算负荷，每个分区选取一个或一组"代表性"潮汐监测点，并基于这些点计算每日平均潮差。

波高与波频的逐小时数据源自昆士兰州环境与遗产保护部（DEHP），来自研究区域内的4个波浪浮标：南部区域采用Emu Point浮标，降灵群岛区域采用麦凯浮标，伯德金区域采用汤斯维尔浮标，北部与南部湿热带区域采用凯恩斯浮标。而就约克角区域而言，采用洛克哈特河澳大利亚气象局站点的风速数据（http://www.bom.gov.au/oceanography/projects/abslmp/data/index.shtml）比凯恩斯浮标的波浪数据更具代表性。

本分析基于每日数据，并按每个分区独立开展。为探究透光深度的长期信号，数据先进行季节去趋势处理并平滑。研究采用梯度提升模型（Gradient Boosted Model, GBM）与广义加性混合效应模型（Generalized Additive Mixed Effects Model, GAMM）来剔除潮汐与风浪的影响。基于上述GAMM得到的残差（即剔除波浪、潮汐与水深信号后的透光深度信号）进一步分解，以获取透光深度的年内趋势（即基于365.25天周期的季节波动）与年际趋势。本研究采用的季节分解方法通过时间序列应用平滑器（通常为移动平均或局部加权回归平滑器），以分离周期性重复影响带来的波动与长期趋势。时间序列分解完成后，季节周期以GAMM拟合值的均值为中心重新校准，并通过指数变换还原至原始透光深度尺度。

## 局限性
本分析仅探究了河流径流对水体透明度的影响，这并不代表其他因素（如沿海开发、疏浚作业）不会额外影响水体透明度；此类关联需单独开展研究。

## 数据格式
本数据集包含2个矢量文件与1个CSV文件：
1. FabriciusAndLoganNerpDataCorrelations.*（142 kb），包含dbf、shp与shx格式文件
2. FabriciusAndLoganNerpDataSummaries.*（142 kb），包含dbf、shp与shx格式文件
3. FabriciusAndLoganNerpSeasonalStatsDataRound.csv（3 kb）

## 数据字典
### 1. FabriciusAndLoganNerpDataCorrelations.shp
该矢量文件包含覆盖大堡礁全域的35个分区多边形要素。其属性表包含11年间每日河流径流量与每日卫星遥感透光深度之间的相关强度。属性字段包括：
- SP_ID：形状要素ID
- Correlatio：相关系数值

### 2. FabriciusAndLoganNerpDataSummaries.shp
该矢量文件包含覆盖大堡礁全域的35个分区多边形要素。每个分区内计算了11年间逐小时或每日数据的平均值。属性字段包括：
- SP_ID：形状要素ID
- Photic_dep：透光深度（单位：米），为11年间每日透光深度的平均值，基于Scarla Weeks（昆士兰大学）与美国国家航空航天局（NASA）开发的算法计算，等效于塞氏深度（Secchi depth）
- Tidal_rang：潮差（单位：米），为11年间每日潮差（当日最高与最低海平面的差值）的平均值，数据源自澳大利亚海军的潮汐预测结果
- Wave_heigh：波高（单位：米），为11年间逐小时波高的平均值，数据取自研究区域内最近的4个DERM波浪浮标之一
- Wind_speed：风速（单位：m/s），为11年间逐小时风速的平均值，数据取自最近的澳大利亚气象局站点

### 3. FabriciusAndLoganNerpSeasonalStatsDataRound.csv
该CSV文件包含经控制以消除主要物理环境变量（波高、潮差）影响的透光深度统计量，用于评估河流淡水径流量的影响。数据按横跨与纵贯大堡礁的35个分区，以及枯水年（2002年至2006年）与丰水年（2007年至2012年）进行分组。属性字段包括：
- Region：地理区域
- Zone：分区类型，包括沿岸带、近岸带、泻湖带、中陆架带、外陆架带
- Period：年份类型，分为丰水年与枯水年
- Maximum：（单位：米）年度内平滑后透光深度的最大值
- MaxDate：（日历日期）年度周期内对应透光深度最大值的日期，通常为枯水季的中晚期
- Minimum：（单位：米）年度内平滑后透光深度的最小值，部分近岸分区可体现丰水年与枯水年的差异
- MinDate：（日历日期）年度周期内对应透光深度最小值的日期，通常为丰水季的中晚期
- DeclineTime：（单位：天）从透光深度最大值到下一个最小值的持续时间（连续周期内）
- Decline：（单位：米）透光深度最大值与最小值的绝对差值，体现部分近岸分区在丰水年与枯水年季节间的透光深度损失量
- PercentDecline：相对下降幅度，以透光深度最大值的百分比表示（单位：%）
- DeclineRate：（单位：m/天）下降速率
- RecoveryTime：（单位：天）从透光深度最小值到下一个最大值的持续时间（连续周期内）。注：下降时间与恢复时间共同构成约365天的年度周期，体现水体恢复至清澈状态所需的时长
- RecoveryRate：（单位：m/天）恢复速率（下降量/恢复时间）
- Recovery95Date：（日历日期）年度周期内对应透光深度恢复95%的日期（恢复至最大值减去下降量的95%）
- Recovery95Time：（单位：天）从透光深度最小值到下一个95%恢复状态的持续时间（连续周期内），体现丰水年与枯水季后水体恢复至清澈状态所需的时长
- Recovery95Rate：（单位：m/天）与恢复速率计算方式一致，但基于95%恢复量计算（下降量/95%恢复时间）