Application Of Ecological Internet Of Things In Chongming Dongtan Wetland Field Observation

Coastal wetlands are located in the transitional area between land and ocean.They are one of the most productive and dynamic regions that are constantly changing.Coastal wetlands have great ecological value to scientific research and social service and it is necessary to collect and analyze long-term field data.Traditionally,most of the coastal wetland data are collected through field survey,which is difficult in operation and the data are discrete and hard to manage.Ecological Internet of Things(IoT)is an emerging technology for field data collecting.With a set of advanced wireless sensors installed,the IoT system shows its capacity in real-time monitoring through automatic data filtering and transformation under complex environmental situation.After the project of ecological control of Spartina alterniflora and improvement of birds habitats in Chongming Dongtan,it is important to address the evaluation of ecological status and regulation management in the restoration area.Ecological IoT observation system can collect real-time,continuous,long-term,high quality and multi-realm data.While seeing the static characteristics of wetlands,it can also detect their dynamics and evaluate the ecological status of wetlands quantitatively.In the present study,the application of Ecological IoT in Chongming Dongtan Wetland field observation is focused.Water is one of the most important features of coastal wetlands,which is the core carrier of material circulation and energy flow.Therefore,the status of water plays an important role in guiding the ecological status of wetlands.Herein,water quality was chosen as an example to show the data acquisition,filtering,and applications of IoT system in Chongming Dongtan:(1)Construction and Layout of Ecological IoT in Chongming Dongtan Wetland.Long-term observation system for coastal wetlands has not been established in the Yangtze River estuary.Based on existing ecological observation standards and construction experience in China and abroad,hierarchical structure and expert knowledge method were used to propose the indicator system of“water-soil-air-biology-topography”.The principle of data collecting,technical specifications of sensors and requirements of observation sites are also standardized.The layout of Ecological IoT station is designed according to ecological characteristics and research objectives of Chongming Dongtan Wetland.This researchcan provide a demostation for the construction and promotion of Ecological IoT in coastal wetlands.(2)Data Preprocessing of Ecological IoT Observation System.Outliers always occur due to limitations of measuring methods and harsh environmental conditions during observation,which will pose challenges to data quality and applications.According to the characteristics of water quality data in Chongming Dongtan,outliers were classified into three types: abnormal values,abnormal fluctuation and abnormal events.Using look-up table and multi-indicator time series model,a preprocessing method for outliers of the Ecological IoT system was proposed based on the regularity of residual probabilistic distribution.Through the analysis of 9 indicators,2 abnormal events,0.18%-8.12% abnormal values and abnormal fluctuation were detected.Abnormal values and abnormal fluctuation were removed and abnormal events were retained.Compared with traditional methods,this method can not only ensure the accuracy of outliers detection,but also has better ability in distinguishing abnormal events from sensor problems to reduce false positive data.By analyzing the preprocessed data,we found that the principle of observation methods and seasons will affect the stability of sensors,and human activities in the study area are the primary factor causing abnormal events.(3)Regional Water Quality Indicators Estimation Based on Ecological IoT and Remote Sensing.Ecological IoT can collect high-frequency and long-term time-series data,but its spatial scale is small and cannot represent macro ecological status of wetlands.Taking Chongming Dongtan water quality data as an example,random forest algorithm were used to conduct the regression between in situ data and surface reflectance of ESA(European Space Agency)multi-temporal Sentinel-2 MSI(Multi-Spectral Instrument)remote sensing images,which are in high spatial resolution.Then,indicators of water quality can be monitored in multiple spatial-temporal scale from point to surface.The optimal input band,the number of decision trees and splitting features were selected to construct chlorophyll-a,dissolved oxygen and blue-green algae estimation model.The accuracy verification are good,showing that the method is more suitable for this study than band ratio method.We estimated three water quality indicators in the study area,and their concentration ranges were 11.0-38.0 μg/L for chlorophyll-a,4.4-9.2mg/L for dissolved oxygen,and 4.1-6.4 cells/mL for blue-green algae,respectively.The spatial distribution of these indicators is positively correlated,the change is not obvious during the year and the concentration is low in large water areas and high in small water areas.The application of Ecological IoT in Chongming Dongtan field observation,combined with remote sensing,can consider the homogeneity and heterogeneity temporally and spatially.It can also describe the static characteristics and the dynamics of wetland ecosystem.Ecological IoT is a necessary technology for wetland research,which can evaluate the ecological status of Dongtan wetland through its unique state and processes.