Possibility of Improving Three-Band Model for Different Phytoplankton Species in Case II water : Evidences from Three Experiments
Possibility of Improving Three-Band Model for Different Phytoplankton Species in Case II water : Evidences from Three Experiments
复制标题
DOI：
10.11440/rssj.29.653
复制
发表时间：
2009-11
影响因子：
--
通讯作者：
B. Matsushita;Wei Yang;Jin Chen;T. Fukushima
中科院分区：
文献类型：
--
作者：
B. Matsushita;Wei Yang;Jin Chen;T. Fukushima
Dall‘Olmo()发现了三个波段的反射；()第二个波段(Nm)应该在其他领域和时期得到微小的应用。即便如此，基于四个要求的测量：()SIOP的第一个频段对于许多应用来说仍然是一个很大的负担。Nm)对浮游植物的吸收最敏感--最近，一个概念模型被提出，该模型只包含在对浮游植物吸收最敏感的红色和近红外范围内的三个光谱波段的反射率，但具有半光谱(称为三波段模型)，用于较大的Tripton和CDOM的吸收，与第一次在第二类水域中反演Chl-a浓度的模型相同。因此，可用于最小化三个波段的吸收效应的要求是最大限度地最小化Tripton和CDOM的吸收效应；()Tripton和CDOM用于准确估计Chl-a的第三个波段(Nm)效应应最小地受到植物浓度吸收的影响(见部分)。以了解详细信息)。目前，浮游生物、Tripton和CDOM具有相似的后向散射，只有MERIS(介于第一和第二波段之间，可以用于分辨率成像光谱仪)和EO-/Hyperion的光谱波段配置可以最大限度地减小后向散射的影响；()这三个波段符合三波段模型的要求。利用五台独立的(中分辨率成像光谱仪)对MERIS模型的三个波段对应的光谱波段进行了标定和验证。从美国湖泊和水库收集的数据集是三波段模型的斜率和截距，分别位于和之间。结果表明，这三个条带分别存在。在以往的研究中，斜率和截距模型都可以应用于变化较大的Chl-a、Secchi圆盘和水体浊度，而不需要对采集进行调整。显然，在其他地方校准后，和是最关键的模型参数，并在三个波段模型中产生参数。根据Chl-a(r.)的观察，要得到可靠的回归方程，通常需要有足够的样本数据，才能得到令人满意的Chl-a估计值。其中遥感反射率R()在三个光谱集中的数据的平均偏差。致。然而，应该注意的是，每个样品的谱带和Chl-a浓度都在模型中进行了校准和验证，只包括采集的样品。来自美国的湖泊和水库，那里浮游植物的特定固有光学特性(SIOP)是有限的。因此，Chl-a浓度和相应的反射率谱需要在日本两个湖泊采集的TRUM下进行进一步的验证，而中国首先是在各种田间条件下进行的，特别是对于不同的植物-用来检验三波段模型的性能(图)。浮游生物种类(即，具有不同的SIOP)。鹿儿湖位于日本东部，本次研究的主要目的是测试关东平原的实绩。不同浮游植物的表面积为Km，平均为三带模型。M的深度(最大深度。M)，这是实现这一目标的第二个，我们使用三个数据集获得了日本仅次于琵琶湖的最大湖泊。从实地调查、水池实验和模型模拟来看，该湖营养盐负荷高，水体富营养化。共有14种浮游植物处于浅层。虽然平均Chl-a浓度包含在这些数据集中。在过去的三十年里，平均总磷浓度从…下降到mgm
Dall’Olmo() found that a three-band refiectance tion;() the second band( nm) should be miniapplied in other areas and periods. Even so, the measurement based on four requirements:() the first band(of the SIOPs is still a big burden for many applications. nm) should be maximally sensitive to phytoplankton absorp-Recently, a conceptual model only containing refiectance of three spectral bands in the red and near infrared range of the mally sensitive to phytoplankton absorption but has the simispectrum (known as the three-band model) was suggested for lar absorption of tripton and CDOM with that in the first retrieving Chl-a concentrations in case II waters. The band, and thus can be used to minimize the absorption e ects requirement of three bands is to maximally minimize the of tripton and CDOM;() the third band( nm) e ects of tripton and CDOM for accurate estimation of Chl-a should be minimally a ected by the absorption of phytoconcentration (see section. for the details). Currently, plankton, tripton, and CDOM but has the similar backscatteronly the spectral band configuration of MERIS (Medium ing with the first and second bands, and thus can be used to Resolution Imaging Spectrometer) and EO-/Hyperion can minimize the e ect of backscattering;() these three bands meet the requirement of three-band model. The three-band should be corresponding to the spectral band of MERIS model was calibrated and validated using five independent(Medium Resolution Imaging Spectrometer). The and datasets collected from lakes and reservoirs in the US be- are the slope and intercept of the three-band model, respectween and. Results showed that the three-band tively. In the previous study, the slope and intercept were model can be applied for widely varying Chl-a, Secchi disk both determined by regression analysis based on data depth, and turbidity of water bodies without adjustment of collection. Obviously, the and are the most critical the model coe cients after calibration elsewhere, and resulted parameters in the three-band model. To derive reliable regresin a satisfactory Chl-a estimation strongly correlated with sion coe cients () usually requires adequate sample data, observed Chl-a (r.) with an average bias across data in which remote sensing refiectance R () in three spectral sets of. to. However, it should be noted that the bands and Chl-a concentration for each sample are inmodel was calibrated and validated by samples acquired only cluded. from American lakes and reservoirs, where the specific inherent optical properties (SIOPs) of phytoplankton are limited. Chl-a concentration and corresponding refiectance spec-Therefore, further validation should be carried out under a trum collected from two lakes in Japan and China was firstly variety of field conditions, especially for di erent phyto- used to test the performance of three-band model (Fig.). plankton species (ie, with di erent SIOPs). Lake Kasumigaura is located in the eastern part of Japan’s The major objective of this study was to test the perform- Kanto Plain. With a surface area of km and an average ance of three-band model for di erent phytoplankton species. depth of m (maximum depth of. m), it is the second To achieve this objective, we used three datasets obtained largest lake in Japan after Lake Biwa. The lake is considered from field survey, tank experiment, and model simulation, eutrophic because it has high loads of nutrients and because of respectively. Total fourteen phytoplankton species were con- its shallow depth. Although the average Chl-a concentration tained in these datasets. decreased from to mg m during the last three decades, the mean total phosphorus concentration …