Database on Salinity Patterns in Florida Bay

Salinity in Florida Bay is closely related to water management in South Florida. Water management activities over the last century have disrupted the quantity, quality, timing and distribution of freshwater flows into Florida Bay affecting salinity conditions. The main goal of the project presented in this paper is to accumulate all the data available on salinity in Florida Bay into one database and make this data available to the researchers and public via the Internet. This unified data source will give scientists one more tool to monitor the fragile ecosystem of the Everglades and to give better recommendations on water management in this area. The challenge of the project is in database design which will accumulate data collected by different groups of people who apply different methodology of data collection, different measuring equipment and techniques as well as different rules for data recording and formatting. Three major data sources on salinity conditions within Florida Bay are available. The three sources are historical data, temporal Everglades National Park (ENP) data, and spatial US Geological Survey (USGS) data. Historical data contains direct salinity observations from 1936 to present. ENP data include salinity and related parameters, such as rainfall, dissolved oxygen received from an increasing number of stations in the Bay continuously monitored by ENP since the early 1980's. The USGS dataset contains the spatially intensive bimonthly salinity survey records. Besides the aspects of the database design the paper covers implementation and maintenance issues. Also the web application is presented which provides access to the data via the Internet in a convenient and intuitive way without special knowledge of the database query tools. 1. SALINITY DATA AND REQUIREMENTS TO THE DATABASE The salinity record for Florida Bay extends from the beginning of 20th century. The early records are not systematic and are usually found across a diverse literature and many unpublished sources. Despite the fact that these data sources are usually sparse, poorly formatted and are not present in the digital form, they compile a historical dataset which should in some form or another be reflected in the database. The historical dataset is of value. It has information on water salinity prior to the present rapid human development of South Florida that changes salinity patterns in Florida Bay, which in turn affects environmental conditions in the area. In the past three decades with the extensive use of computers to store and process data, a number of salinity studies in Florida Bay resulted in a collection of the datasets which contain data in the digital form according to the well-documented formats. This data is ready to be stored in the database. However, the studies producing the data differ in methodology of collecting data and equipment used. Good example of the differences are NPS (National Park Service) hourly monitoring data, which represent time series from single locations, and USGS (United States Geological Survey) boat survey data which is spatial data with many locations represented by a single value at each location. This makes hard the work to unify the data in one database. * Presented at the Seventh International Conference on Remote Sensing for Marine and Coastal Environments, Miami, Florida, 20-22 May 2002. This research was supported in part by NASA (under grants NAG5-9478, NAGW-4080, NAG5-5095, NAS5-97222, and NAG5-6830), NSF (CDA-9711582, IRI-9409661, HRD-9707076, and ANI-9876409), ONR (N00014-99-1-0952), and the Florida Space Grant Consortium. Many salinity studies provide temperature data as well since the measurement equipment allowed to collect salinity and temperature data simultaneously. It was decided to store the acquired temperature data in the salinity database. It is expected that users of the salinity database will be mostly interested in averaged salinity data over a particular area of Florida Bay and during a certain period of time rather than some particular salinity record. Daily, weekly, monthly, seasonal and annual averages are to be used. Since variations in data density are common from one study to another, data integration may lead to inaccurate results. In order to minimize this effect the following rules have been applied when calculating monthly, seasonal and annual averages: (i) these calculations are based on daily average data; (ii) daily average data are calculated as the average salinity/temperature for each station within each study on a given day, and (iii) in spatial data sets salinity/temperature observations are averaged within each basin on a given day and assigned a station location of the geographic center of the respective basin. We refer to (Robblee et al, 2000) for more information. An extensive search has been performed by the researchers for literature references, published and unpublished, interpretable in terms of salinity conditions in Florida Bay. These references describe observations on salinity and other phenomena related to water quality like freshwater occurrences and fish kills. One of the requirements to the database was to facilitate storage and retrieval of these references. 2. CONCEPTUAL DATABASE SCHEMA We have employed a semantic modeling approach for the database design. It has certain advantages in comparison with other techniques: i. the output database schema design is intuitive and clear for understanding even by non-professionals in the field of databases ii. the semantic schema reflects only the semantics of data to be stored in the database and does not show any technicalities of implementation iii. semantic design can be automatically mapped to the relational schema on the implementation step, tools are available to produce instructions for physical database creation using any popular RDBMS (relational database management system). Every concept defined on the semantic schema is one of the following: a category a set of objects about which information is to be aggregated in the database. Categories are denoted on schema by boxes with their names in uppercase bold inside the box; an attribute a pattern of certain printable data about the objects of a category. Attributes are denoted on schema by text in italic inside corresponding categories; a relation a pattern of relationships between objects of two categories. Relations are denoted by arrows on the schema. Certain other notations appear on the semantic schema like cardinality of the relations and indication of some constraints. We refer to (Rishe, 1992) for further reference on semantic schema designs. We now turn to the description of the semantic schema of salinity database. For convenience purposes we are presenting the database schema comprising of three related subschemas, namely studies, locations and datasets subschemas. A key concept of studies subschema is salinity STUDY (see Figure 1). STUDY is performed by a group of INVESIGATORs. One of INVESTIGATORs is a primary investigator, denoted by the relation pi. INVESTIGATORs are associated with INSTITUTIONs. Relation works links these two categories. Category REFERENCE is a container for published and unpublished papers and other documents related to salinity studies and possibly but not necessarily authored by INVESTIGATORs. This subschema accommodates the situations when some information on the reference is missed. For example, it allows to store information in the database in case, nowadays common, when investigators conduct studies and publish reports on the studies; as well as in case when historical reference is taken from a private diary never being published and not related to any study. Category KEYWORD contains keywords related to references and studies and is used to facilitate efficient search of