<<

P A P E R The of Observing System: Generic Lessons Learned in the First Seven Years of Operation (2001-2008) AUTHORS ABSTRACT Neal R. Pettigrew The Gulf of Maine Ocean Observing System (GoMOOS) was established in the summer Huijie Xue of 2001 as a prototype real-time observing system that now includes eleven solar powered School of Marine Sciences, buoys with physical and optical sensors, four shore-based long-range HF radar systems University of Maine for surface current measurement, operational circulation and wave models, satellite - James D. Irish servations, inshore nutrient monitoring, and hourly web delivery of data. The observing University of system in the Gulf of Maine (GoM) is one of the most comprehensive and operational of the Integrated Ocean Observing Systems (IOOS) systems that have been established in the Will Perrie to date. It has also been a very successful system, with data returns routinely Bedford Institute of Oceanography, in the 85-95% range. Dartmouth, NS, CA The Gulf of Maine is a harsh operational environment. Winter storms pose severe chal- Collin S. Roesler lenges, including high waves and the build-up of ice on buoy sensors, superstructure, Andrew C. Thomas and solar panels, and in summer its productive waters present severe biofouling problems David W. Townsend that can affect the optical sensors. The periods of most difficult field operations often coin- School of Marine Sciences, cide with periods of greatest data value in terms of marine safety, search and rescue, and University of Maine monitoring biological productivity. The challenges of the Gulf of Maine physical environment were paired with the unex- pected challenges of the funding environment that have been the hallmark of the turn of this 1. Introduction century. Funding for the system has been chronically short and subject to the unpredictable he Gulf of Maine Ocean Observing fluctuations of the congressional appropriations process. The inadequacy and variability System (GoMOOS) is a comprehensive of funding has substantially hampered the operations of many of the Integrated Ocean Tprototype integrated coastal ocean observ- Observing Systems, including GoMOOS, and has hindered technological advancements ing system. It serves online an extensive ar- and maintenance measures. As a result, the design of the GoMOOS infrastructure is little ray of real-time oceanographic and marine improved from that developed almost a decade ago, and it has deteriorated with age, usage, meteorological data and data products to and suboptimal replacement schedules. In the absence of an adequate and reliable funding a broad range of users including scientists, stream, the system is fast approaching the end of its expected operational lifetime. Unless state and federal regulators, the National this trend is reversed, the system will no longer well serve the many citizens, organizations, Weather Service, both the U.S. and Ca- and agencies that have come to rely on the data it provides. nadian Coast Guards, the National Data In this article, we present lessons learned by the scientific and technical groups that Buoy Center, educators, regional natural have been responsible for the data acquisition of GoMOOS. We believe that these lessons resource managers, the Gulf of Maine fish- are generic, rather than peculiar to the GoMOOS system, and that they have value for others ing and maritime industries, local airports who are embarking on similar endeavors. However, it is important to make clear that these and airlines, sailors, and the general public. lessons are from the perspective of the scientists, and that the views of others involved in As perhaps the first comprehensive multi- complementary aspects of GoMOOS, including public outreach, fundraising, and providing disciplinary Integrated Ocean Observing data and products to the more general user community, are not represented here. System (IOOS), GoMOOS has served as a proving ground for new technologies, In addition to the hourly operational data have not only illustrated important as- operational procedures, protocols, and data delivery, GoMOOS provides an pects of seasonal and interannual variability for management structure and govern- archive of data and model output that is of the circulation and physical properties ance. GoMOOS has also proven to be significantly advancing the scientific under- of the Gulf of Maine, in some locations an example of meaningful integration of standing of the Gulf of Maine as a physical the array has provided the first meaningful interdisciplinary sensors, platforms, and and an ecological system. Over the first regional baseline of physical and biological predictive models. seven years of operation, the GoMOOS oceanographic data.

Fall 2008 Volume 42, Number 3 91 GoMOOS is an integrated ocean ob- trient delivery via the deep inflow of nutri- Maine is the northern coastal geographi- serving system that can be thought of as ent-rich slope waters through the Northeast cal limit of several temperate species as consisting of four major subsystems: the Channel, which cuts between Georges well as the southern coastal limit of several data acquisition subsystem; the data han- Bank and Browns Bank, and strong tidal boreal species (Sinclair et al., 1991). These dling, processing, and archiving subsystem; mixing that effectively mixes the nutrients circumstances make the GoM an area the subsystem of numerical nowcast and up into the lighted (euphotic) zone where that is biologically sensitive to the climate forecast models; and a web-based data dis- they are available to fuel phytoplankton change signals since modest changes in tribution/presentation subsystem. The ac- blooms (Townsend et al., 2006). temperature can result in large swings of quisition system includes a real-time buoy Under modern climate conditions, the the species composition (ranging from array, an array of land-based long-range Gulf of Maine is a of strong physical, phytoplankton species to commercial fish Coastal Ocean Dynamics Applications chemical, and biological gradients. During species) within the Gulf. Thus the GoM Radar (CODAR) installations, a satellite summer there is a strong contrast in upper can be expected to exhibit early ecological data receiving station, and an inshore water properties from warmer, fresher, and effects of climate change. nutrient sampling system in Casco . lower nutrients in the southwest to colder, The GoM circulation is generally cy- The data handling system includes real- saltier, and higher nutrients in the northeast clonic (Bigelow, 1927; Brooks, 1985), and time QA/QC algorithms, data processing, (e.g., Pettigrew et al., 2005a, www.gomoos. its shelf have a complex, variable, calibration, and data archives that include org). There is also a clear southwest-to- and interconnected coastal-current system sensor inventories, deployment histories northeast increase in tidal amplitude, and that is best developed in the summer season and calibration records in addition to the thus also in the degree of vertical mixing. (Brooks, 1985; Pettigrew et al., 2005a). A data records themselves. The numerical A surface temperature front, which trends schematic diagram of the near-surface sum- modeling systems consist of an operational offshore from mid-coastal Maine in the mer circulation is shown in Figure 1, which real-time application of the Princeton vicinity of , often develops in depicts a pair of cyclonic (anti-clockwise) Ocean Model (POM) for circulation and summer and separates the warmer surface gyres over the basins in the eastern Gulf, hydrographic conditions (Xue et al., 2005), waters of the southwestern region from and a partial separation of the coastal cur- and a high-resolution implementation of the colder waters of the northeastern Gulf rent from the shelf at a mid-coast location WAVEWATCH III wave model (Padilla (Pettigrew et al., 1998). in the vicinity of Penobscot Bay. Even this et al., 2007). The operational circulation The southwest-to-northeast gradients somewhat complex pattern is highly simpli- model uses output from the Eta mesoscale in physical processes and water properties fied in both time and space. atmospheric forecast model run by the in the GoM are factors in the seasonal Recent moored current measurements National Center for Environmental Pre- patterns of phytoplankton blooms and the and hydrographic surveys from the Ecol- diction (NCEP), whereas the wave model species composition. Spring blooms occur ogy and Oceanography of Harmful Algal is driven by wind fields provided by the first in shallower coastal regions and later Blooms (ECOHAB) experiment in the Coupled Ocean/Atmosphere Mesoscale in offshore waters. The phase propagation Gulf of Maine showed marked interan- Prediction System (COAMPS) run by of the blooms is southwest-to-northeast in nual variability in the degree of separation FNMOC (Fleet Numerical Meteorological the coastal zone and generally northeast-to- versus through-flow that occurs along and Oceanographic Center). southwest farther from shore. Summer pat- the coast (Pettigrew et al., 2005a). This terns are a strong reflection of stratification seasonal and interannual variability in the 1.1 The Oceanographic Domain and nutrient limitation over the deeper ba- connectivity of the eastern (EMCC) and of the Gulf of Maine sins and tidal mixing and vertical nutrient western (WMCC) branches of the GoM The GoM is a complex and very pro- flux over shallow and northeastern regions coastal current system is expected to have ductive marine ecosystem. The level of pri- (Thomas et al., 2003). By autumn, cool- far-reaching consequences with regard mary (phytoplankton) production, which ing and overturn stimulate strong blooms, to the transport of nutrients, planktonic forms the base of the marine food chain, starting in the northeast and propagating larvae, harmful algal blooms, and coastal is high relative to other continental shelf to the southeast. pollution (cf. Luerssen et al., 2005). All and marginal sea environments. Fisheries The base of the ecosystem is highly of these fluxes are important factors in de- production in the GoM is also high. In sensitive to the patterns and intensity of termining both short- and long-term vari- particular, , which separates thermally and salinity-driven stratification, ations in the state of the GoM ecosystem. the interior GoM from the northwestern which in the Gulf of Maine are impacted GoMOOS moorings I and E, which are, , is historically one of the significantly by patterns of precipitation, respectively, in the EMCC and WMCC most productive fishing regions in the run off, wind, and heat exchange, and thus (see Figure 1), have revealed a strong sea- world. These high levels of production are very sensitive to climate change. Under sonal component to this connectivity. The believed to be due to a combination of nu- present climatic conditions, the Gulf of two branches generally merge each fall

92 Marine Technology Society Journal Figure 1 Funding constraints dictated that the bu-

Schematic diagram of the summer surface circulation of the Gulf of Maine (Pettigrew et al., 2005a). oys be much cheaper to produce, cheaper to maintain, and small enough to be de- ployed from a relatively small ship. While required to be smaller, cheaper, and lighter than the NDBC buoys, the GoMOOS buoys had to be capable of interfacing with many more sensors, handling an order of magnitude greater data volume, and still withstanding the rigors of the Gulf of Maine winters. The starting point of the GoMOOS design was the buoy described by Irish (1997) that he and others used successfully on Georges Bank. The GoMOOS buoy is a multi-chinned two-meter discus buoy with flotation made of closed-cell Surlyn foam. There is a central water-tight instrument well, made of alu- minum, which houses the buoy electronics including the voltage regulation system, solar storage batteries, and the data-log- ger/controller. The buoy is designed to survive knockdown and compression due to forced submergence, it monitors its own position, and sends an alarm if it detects it is off position or has a leak in the electronics and separate each spring (Pettigrew et al., eastern GoM. The geostrophic adjustment well. The buoy is solar powered, has dual 2005a). The lack of long-term direct cur- processes in response to these persistent cellular/iridium and GOES satellite telem- rent measurements made within the Gulf density contrasts engender the cyclonic etry systems for hourly data telemetry, and of Maine Coastal Current system, and at general circulation pattern of the GoM. The has room for expansion in both its power key inflow and outflow locations near the monitoring of these inflows has long been and electronic systems. Some of the design open boundary, has significantly hindered recognized as requisite to a Regional Coastal details of the GoMOOS buoy system have our understanding of both the physical Ocean Observing System (RCOOS) in the been previously described by Wallinga et al. and biological oceanography of the GoM. GoM, as well as to the understanding of the (2003) and Pettigrew and Roesler (2005). The first seven years of operation of the large interannual and decadal variability The basic GoMOOS buoy platform is GoMOOS array have begun to relieve the that characterize the hydrographic structure moored in two different configurations: a paucity of flow data. and the fisheries yields of the region. compliant tether mooring for the offshore The surface inflow into the GoM of basin and channel buoys, and a slack chain relatively fresh Scotian Shelf water (SSW) mooring for the shelf moorings for water from the Atlantic seaboard of , 2. The Ocean Observing depths of 100 m or less. Both mooring and the deep inflow through the Northeast System styles have been very successful, although Channel of relatively warm, salty, nutrient- 2.1 GoMOOS Real-Time neither has been immune to dragging and rich slope waters (SLW) are the two most Buoy Designs damage associated with the fishing and important inflows into the GoM. The large The concept of the GoMOOS Data marine transportation industries. A sche- buoyancy input of the SSW accounts for Buoy System design was an economical, matic diagram of the GoMOOS offshore more of the annual freshwater budget of the moderate sized, stable, solar-powered plat- elastic-tether mooring system is shown GoM than the combined inflow of all the form with real-time telemetry and some in Figure 2. The advantage of the elastic rivers that drain into its confines. The den- onboard data processing capabilities. The mooring system is a reduced watch circle, sity contrast between these relatively fresh GoMOOS design incorporates many sig- which reduces the chances of draggers surface and intermediate waters with the nificant departures from the buoy designs working between the buoy and its anchor, deep salty slope waters survives the vigorous used previously in the Gulf of Maine by and the spar-like motion reduction of the tidal mixing and winter convection in the the National Data Buoy Center (NDBC). surface float. The disadvantages include the

Fall 2008 Volume 42, Number 3 93 expense and short (~2 year) lifetime of the 2.2 Telemetry System steel mooring cable, rather than dedicated elastic members, and the requirement of The telemetry system of the GoMOOS electrical cables, for communications. Use using an expensive acoustic release system buoy has three functions: to send data from of this system enables the deployment of up that leaves the anchors on the bottom (an submerged instruments to the buoy; to to 100 sensor packages that are inductively expense to the observing system, and a send data from the buoy to the server at the coupled to the mooring cable. Because the potential hazard to the draggers). All buoys University of Maine for processing; and to IM system does not require sensors to be are mechanically identical and carry a sur- allow direct remote communications with plugged into an electrical cable, shallow face sensor payload of dual sonic and me- deployed sensors for trouble-shooting op- sensors can be replaced in water by divers, chanical anemometers (for wind speed and erations. The transmission of data from the and changes in deployment depth can be direction), air temperature, atmospheric submerged sensors to the buoy is handled made by sliding the sensor up or down pressure, and visibility (fog), and a wave by a Sea-Bird Electronics inductive modem the cable. The Sea-Bird IM system has accelerometer. Spectral solar insolation is (IM) system. The system uses the jacketed been very reliable. Failures do occasionally included on about half the buoys. occur (generally caused either by failure of Figure 2 the electrical cable that couples the system to the buoy or by fishing activity damaging Schematic diagram of a GoMOOS Basin Buoy. The buoy scope is provided by an elastic tether near the bot- the jacket on the mooring cable) however tom that reduces the watch circle of the mooring. Subsurface sensors send data up the mooring wire using the data are retrieved from internal instru- inductive modem technology. ment data loggers after buoy recovery. This redundancy maintains the quality of the data archive; however, failure or intermit- tency of the IM system causes a sometimes significant reduction of the real-time func- tionality of the buoy system. The transmission of data from the buoy to the University of Maine is achieved via one of two telemetry systems. The principal system is digital cellular/iridium (the cheaper cellular is chosen in regions of adequate coverage near shore) and the secondary system is a Geostationary Op- erational Environmental Satellite (GOES) transmitter. The use of two different telem- etry systems helps insure that real-time data transmissions are successful even when one or the other is not functioning properly as a result of either environmental conditions or technical difficulties originating with the service provider. The two-way cellular/iridium link has proven to be a vital component of the overall operations of the array. There are numerous occasions when instruments fail to send data because of clock drifts, spontaneous resets of the sampling scheme, or more rarely, problems with the data logger. The ability to call the data logger and “talk through” to the sensor to reinitialize, set clocks, or “reboot” has saved substantial time, effort, and funds by enabling remote “fiddling” rather than having to rely on personnel and ships go- ing to sea to toggle switches and reprogram sensors and data loggers.

94 Marine Technology Society Journal 2.3 The GoMOOS Buoy Array are not covered by NOAA IOOS, and it their previous deployment. Mooring cable The GoMOOS buoy array is shown is not officially part of GoMOOS). Buoys is replaced yearly. This rotation of sensors in Figure 3. The diamonds with alphanu- L and N monitor, respectively, the inflows and buoys, and early replacement of moor- meric labeling show the locations of the into the Gulf of Maine from the Scotian ing components is probably the principal twelve GoMOOS buoy locations within Shelf and the North East Channel that factor that has made the GoMOOS data the GoM. Reference to Figure 1 shows were discussed earlier. Buoy M, located in streams among the most reliable of the that many of the buoys (B, C, E, I, L) are the Jordan Basin, monitors the seasonal IOOS buoy arrays. located within the Gulf of Maine Coastal inventory of the nutrient-rich SLW in the The data quality assurance program, Current System (GMCC) (see Pettigrew interior of the GoM, which contributes performed by the et al., 2005a), which had never been effec- to the high productivity of this important Group at the University of Maine, is a tively monitored “year round” prior to the fisheries region. Bio-optical sensor packages process of continual quasi-real-time evalu- IOOS. These buoys collectively represent are deployed on buoys A, B, D01, D02, E, ation and validation. Detailed histories are the first buoys that have gathered long F, I, M, and N. kept of sensor performance, repair, and time-series data at multiple sites within calibration. Signatures of failure modes the GMCC, and have been in nearly con- 2.4 Buoy Operations for the various sensors and systems have tinuous operation since July of 2001. The Associated with the eleven active Go- been identified over the years of operation nearshore buoys A, C, F and J monitor MOOS buoy sites are twenty two buoys since 2001, and real-time performance is the mouths of four GoM major estuarine and twenty two complete sets of instru- continually evaluated and compared with embayments and buoys D (D01 and D02) mentation. At approximately six-month these failure modes in order to flag suspect are inshore buoys that are deployed in two intervals (Spring and Fall) the entire array data in near real time. With each passing small within (buoy of buoys and instruments are exchanged deployment year the QA/QC procedures D02 was funded under a grant from NSF for a set that has been refurbished, tested, become more skillful and further automat- to Bowdoin College, its operational costs repaired and calibrated in the interim since ed; however, they still require a great deal of attention and judgment by seasoned Figure 3 oceanographic personnel. Ocean Observing assets in the Gulf of Maine. Red diamonds indicate GOMOOS buoys, the blue diamond represents a buoy funded by NSF that has been integrated with GoMOOS, black inverted triangles are NDBC 2.5 The GoMOOS Optics Program buoys, upright triangles are NOAA island meteorological stations, and the orange diamonds are Environment Of all the components of GoMOOS buoys. GoMOOS CODAR stations are shown as red stars. (Color versions of figures available online that could not be considered to have op- at: http://www.mtsociety.org/publications/?fa=online). erational capabilities in 2001, the optics program stands out as perhaps the biggest leap of faith. The most obvious problem with deploying optical sensors for extended periods in the ocean is biofouling. In ad- dition, many of the optical sensors were still in a beta configuration with electronic and mechanical problems that had to be worked out as we went along. Despite early problems with sensor electronics and system integration, the bio-optical data streams are unparalleled in their temporal and spatial coverage and the comprehensiveness of the derived biologically-relevant data products. Because of the high data rates and high data volume of the optical sensors, on-board processing and raw data storage were neces- sary, with telemetry only of statistical prop- erties. Thus the bulk of raw data processing, data product computations, and archiving are performed on the internally-logged raw data post-recovery. This blend of real-time and post-processed products has allowed

Fall 2008 Volume 42, Number 3 95 significant progress not only in algorithm whether the problem is a biofouling or to apply Artificial Neural Networks in order development but also in instrument devel- sensor malfunction. The protocols for real- to fill in missing data values to provide opment and sensor characterization. time and post-recovery calibration have tidally-averaged surface current maps, and Most of the optical sensors used in been described elsewhere (Pettigrew and to work toward the goal of short-term, GoMOOS are configured with combina- Roesler, 2005; Roesler and Boss, 2008). wide area predictions of the GoM surface tions of copper shutters that lie over optical The GoMOOS optics program uses current fields. faces until the instrument turns on, copper two moored optics packages: the “phyto- tape on surfaces surrounding the optical plankton biomass and production” pack- 2.7 Applications of Neural Network heads to prevent macrofaunal growth that ages and the larger “Ocean Color” packages. Models to Sensor Array Data might impede the sensor head (or worse, Each phytoplankton package consists of a Artificial Neural Networks (ANNs) are prevent the shutters from opening), and chlorophyll fluorometer and a ‑4 channel parallel arrangements of simple processing copper tubing on the flow-through instru- downwelling irradiance sensor (used to es- units that are loosely analogous in structure ments. We have found these strategies to timate phytoplankton biomass and spectral and function to a biological central neuron. diminish fouling, even during a 6‑month photosynthetically available radiation). The ANNs are comprised of a large number deployment during the most productive combination of phytoplankton biomass of these neurons in multiple layers. These portion of the year; particularly for the and light measurements are sufficient for neurons connect input and output through algal growth on chlorophyll fluorometers calculating primary production to first a collection of weights and transfer func- (Figure 4) and other flat-faced sensors order (Siegel et al., 2001). The ocean color tions designed to minimize the differences such as colored dissolved organic matter packages have all the instruments of the between predictions and realizations (data). (CDOM) fluorometers and backscattering smaller package, plus an AC9 nine-wave- By adjusting the weights, the ANN can be sensors. We do find some bacterial growth length absorption and attenuation meter, a “trained” to make very accurate predictions. on the optical windows, particularly on backscattering sensor, a CDOM fluorom- ANNs that adjust their weights automati- the absorption-attenuation meter. For this eter, and a 7‑channel upwelling radiance cally can “learn” to make predictions. We reason, it is particularly important that sensor. The details of the moored optics have been quite successful with the nowcast we are able to identify when this fouling program, and its goals have been discussed and forecast of near-surface currents at the occurs in real time and perform cleaning more fully elsewhere (Pettigrew and Roesler, buoy locations (Pettigrew et al., 2005b; using divers when possible as well as look 2005; Pettigrew et al., 2008). Pettigrew and Neville, 2008; Pettigrew et to redundant data products for real-time al., 2008). analysis. One of the important strengths of 2.6 The HF Radar Array Predictions of CODAR data fields are the bio-optical program is the development GoMOOS uses four long-range CO- more problematic than predicting direct of sensor-independent data products. For DAR units to map surface currents over the surface current measurements from the example, on buoys D02, E and I, we have GoM. The GoMOOS long-range CODAR GoMOOS buoy array. The CODAR data the capability of independently estimating array is a potentially exciting element of the have a lower signal-to-noise ratio, and the phytoplankton biomass from four separate observing system capable of widespread, available inputs for the model fluctuate in sensors. Thus when the temporal pattern remote surface-current measurements from time and space. Because the data availabil- of one sensor stops being coherent with a limited number of shore-based radio wave ity was spotty, we were unable to provide the others we can examine, in real time, transceivers. The nominal daytime range of tidal predictions, average currents for the 180 km is almost never realized in the GoM previous tidal cycle, or even previous values Figure 4 region (especially at night), and four units at each location as inputs. In fact, we were Underwater photograph of WET Labs fluorometer, are not capable of providing consistent full limited to wind forcing and a varying col- after 5 months in the water. Antibiofouling copper coverage of the GoM. lection of current vector values from “near- tape has kept algal growth from affecting the opti- Because of the highly variable CODAR est neighbors” in the CODAR field, and cal window even though the sensor was deployed without a shutter. Photo credit: Steve Karpiak radial overlap and vector coverage in the the radial values in regions where vectors GoM, and the very strong tidal variability, were not available because of the absence of it has been challenging to produce use- overlapping signals from adjacent CODAR ful surface current fields. The inability units. In order to cope with the variable to consistently receive data throughout inputs, we developed a weighting system the tidal cycle often reduces the data to that produces a spatially-weighted value a gappy series of realizations that can not for the nearest neighbors so that unique be properly averaged to give a consistent ANNs need not be developed for each of picture of the general circulation patterns. the myriad possible input configurations Because of these limitations, we have begun (Pettigrew et al., 2008). Under favorable

96 Marine Technology Society Journal conditions, when some neighboring values raphy Data Laboratory, while GoMOOS gov/cofs/Welcome.html). Surface forcing, are available, the vector correlation values provided the motivation and funding to including heat, moisture and momentum between nowcast and observed values vary place data acquisition, processing and de- fluxes, is obtained from the NCEP’s North from 0.8-0.9. An example of the ANN livery into an operational mode. Thus the America Mesoscale (NAM) forecast on 221 nowcast to fill in missing values on a process of incorporating satellite data into AWIPS Grid (http://www.nco.ncep.noaa. CODAR surface current map is shown in the IOOS framework was essentially a tran- gov/pmb/products/nam/). Also specified at Figure 5. Panel A shows the vectors from sition from episodic research funds and ad the coastal boundaries are daily freshwater the CODAR vector data. Spatial gaps in hoc data processing and posting for specific discharges from the St. John, Penobscot, the data are caused by reduced range so research projects, to partial IOOS funding Kennebec, Androscoggin, Saco, and Mer- that radials do not overlap. The map in and a daily operational mode. GoMOOS rimack Rivers based on U.S. Geological Panel B shows considerable improvement funds were also used to modify some of the Survey gauge observations. Satellite derived in coverage using winds, nearest neighbors, extant products and to add scatterometer sea surface temperature is assimilated using and radials as inputs to the ANN model. wind data. The GoMOOS website links a simple optimal interpolation algorithm Although this example shows tremendous directly to the web pages of the University (Xue et al., 2005). improvement, it represents relatively favo- of Maine Satellite Oceanography Data The daily operation includes three rable conditions with regard to the extent Laboratory rather than taking the data and consecutive jobs: pre-processing, model of data gaps. These conditions are not reposting them on the GoMOOS site as is integration, and post-processing. Pre- always met, and further improvements in done with the Buoy and CODAR data. processing downloads the river discharge, the model are required for an operational AVHRR sea surface temperature data from product that produces significant improve- 2.9 Numerical Circulation Model the University of Maine Satellite Data ments over a broader range of conditions. The operational GoMOOS circulation Laboratory, NAM and ROFS forecasts, model (Xue et al., 2005) is based on the which are then interpolated to the Gulf of 2.8 Satellite Data three-dimensional Princeton Ocean Model Maine model grid to prepare the bound- The infrastructure for research-based (Mellor, 2003) in a curvilinear grid with ary conditions. Then the automated daily acquisition and processing of NOAA a variable resolution of 3 to 5 km in the procedure calls for the model integration AVHRR and SeaWIFS ocean color satellite horizontal and 22 sigma levels in the verti- session. A 24-hour nowcast cycle assimilates data at the Satellite Oceanography Labora- cal. It is initialized and forced at the open the satellite SST and updates the initial tory at the University of Maine was already boundaries with 6 major tidal constituents condition for the next 48-hour forecast. in place prior to GoMOOS. Infrastruc- and daily nowcasts of velocity, surface el- Post-processing includes archiving the ture for acquisition and processing of the evation, temperature, and salinity from the model results and updating the graphical MODIS telemetry streams was added after National Center for Environmental Predic- output on the web. the establishment of GoMOOS through tion (NCEP)’s Regional Ocean Forecast The operational GoMOOS circulation research funding of the Satellite Oceanog- System (ROFS - http://polar.wwb.noaa. model has been updated since January Figure 5 A, B

Left plot (A) shows surface current vectors observed by CODAR. Right plot (B) shows the observed data together with the ANN nowcast of the missing vectors. The inputs to the model are: available radial data, weighted-averages of neighboring vector values, and winds (Pettigrew et al., 2008).

(A) (B)

Fall 2008 Volume 42, Number 3 97 2007. It now uses the NAM forecast on approximately weekly, from the surface and of procedures, methods, and opinions 218 AWIPS Grid for meteorological forc- various depths to the bottom. In addition, arise that are consistent with the working ing at the surface, and the forecast from surface water samples are collected approxi- philosophy of the group, and which gen- the NCEP’s Real-Time Ocean Forecast mately daily from a dock at the southwest erally reinforce positive outcomes while System (http://polar.ncep.noaa.gov/ofs/), entrance point to Casco Bay. sequentially eliminating practices that lead which is the NCEP’s current operational to negative outcomes or interactions. forecast system, for the subtidal open 2.12 Shipboard Observations In a complex enterprise like IOOS boundary forcing. At the same time, the Due to funding constraints, there (with a mix of funding agencies, political daily forecast has been extended from 48 has never been a functional shipboard bureaucracies, administrators, scientists, hours to 60 hours. Model output of every observing program as part of GoMOOS. and technicians) no one group has the 3 hours has been archived since 1 Janu- Although the buoy program uses approxi- power to effect all the changes that they be- ary 2001, from which monthly averages mately 24 ship days in a typical operational lieve are needed to improve the system. As have been generated that are accessible year, there has never been a budget to add a result, convergence toward best practices through the web portal Live Access Server a few days for hydrographic surveys. From is noticeably slower than the pace to which (LAS). In addition, daily restart files have the scientific point of view, this practice scientists and technicians are accustomed in been saved incrementally for hindcasts of would certainly seem to be “penny wise classical R&D projects. The frustration that specific events. and pound foolish.” For an additional 10- this situation engenders is undoubtedly 20% in ship time the buoy cruises could similarly experienced by administrators, 2.10 Numerical Wave Model have been augmented to collect important funding agents, and politicians. Each group The GoMOOS wave forecast includes (although not comprehensive) biannual can wind up feeling that their vision, and wave height, direction, duration and peri- hydrographic surveys at stations along the their efforts, are being hampered by others od; as well as a comparison to buoy observa- tracks between the buoy locations and at in control of others aspects of the IOOS tions for validating the forecast predictions. a few other key locations. The decision to process. Below we discuss briefly some The forecast covers a 48 hour period and is not fund shipboard observations is a con- of the lessons learned by the scientists, presented in 3 hour increments. The wave sequence of the overall scarcity of funding engineers, programmers, and technicians forecast, based on the WAVEWATCH III as well as the priorities of a GoMOOS who are responsible for operating the model and driven by wind fields provided administration that strongly favored prod- observing system in the Gulf of Maine. by the COAMPS atmospheric model run ucts easily presented to the general public We hope that these lessons will be of use by FNMOC, is provided by the Bedford in real time. Thus, the limited funds were (or at least comfort) to groups involved in Institute of Oceanography (BIO) in Dart- directed more toward the user interface at similar endeavors. We acknowledge that mouth, Nova Scotia. WAVEWATCH III the expense of the historical content of the our perspective may in some cases be de- is implemented on a system of three nested data archives. It is hard to argue convinc- cidedly different from that of other groups grids. The spatial resolution increases ingly, pro or con, about the wisdom of this that have different roles in the enterprise of from 1.0o in the coarse grid to 0.2o in policy, since it pits the scientific content of ocean observing. the intermediate grid to 0.1o in the fine the archives against the real-time impact of We have structured the list of lessons resolution grid. Details of the grid dimen- the system. In any case, this issue is illus- learned into those that evolved into as sions and resolutions are given by Padilla trative of the tradeoffs that must be made “best practices” from successful opera- et al. (2007). when funds for operating a multiuse system tions (Type I), and those that still need to are severely limited. be implemented to improve the function 2.11 Casco Bay Nutrient Sampling and efficiency of the observing system in Since 2001, water samples collected in the GoM (Type II). Finally, we list some Casco Bay by The Friends of Casco Bay, a 3. Lessons Learned general suggestions about organizational nongovernmental organization, have been In operational oceanography, as in most structures, funding processes, and regional analyzed for NO2+NO3, NH4, Si(OH)4, human endeavor, we learn from successes decision making that we believe are impor- and PO4 at the University of Maine. Home and failures, from hardship and ease, and tant considerations as we look toward the to Maine’s population centers and its largest from discord and harmony. From these future of coastal ocean observing. city (Portland), Casco Bay is prone to the outcomes, individuals and working groups effects of anthropogenic nutrient addi- develop “best practices” aimed at optimiz- Type I Lessons tions, and yet at the same time, it reflects ing success, ease, and harmony. In essence, ■ Academic institutions and academic the influxes of nutrients from the deeper if something works we try to keep doing research groups are capable of ef- offshore waters. Sampling stations are lo- it, and if something fails we try something ficiently and economically operating cated throughout the Bay, and are sampled else. Through this process, a collection ocean observing systems. This lesson

98 Marine Technology Society Journal was not one that needed to be learned decision was made at the onset of to respond and update the opera- by the academic community, but GoMOOS to build two buoys with tional GoMOOS model to make the it was a point of serious discussion two sets of instrumentation for each transition as seamlessly as possible. amongst commercial and govern- buoy monitoring location. This prac- Moreover, we were able to reach an mental entities involved in the IOOS tice allowed us to replace each buoy agreement with the ocean forecast- movement. Colleges and universities, with a newly-prepared buoy every ing group at NCEP to maintain their in part because of their insular nature, six months. Although this schedule previous operational system (ROFS) must perform numerous operational seemed excessive from the point of for about a year so that an overlap functions including operating power view of many of the physical measure- was established to allow us identify plants, fire and police services, radio ments, it was minimal from the point the advantages and disadvantages stations, communication networks, of view of the optical sensors that are between the new and the old systems. and health services as part of their very sensitive to biofouling. In hind- ■ Direct communication between everyday operations. The ability of sight, we believe this protocol to be a operational system developers and academic institutions to add the more primary reason for the unprecedented users can lead to benefits for both “academic” function of operating an success of the GoMOOS buoy array, groups. Dialogue and feedback observing system focused on environ- which has operated in a very harsh between the system developers and mental science and engineering would environment, and yet has consistently users can accelerate the development seem obvious. In each year of its been an IOOS leader in data return of analysis tools, advance scientific operation, GoMOOS’s initial goal of rates, the number and variety of understanding, and improve the 80% data return has been exceeded. sensors deployed, and the percentage system overall. For example, the col- ■ Basing IOOS at academic research of buoys in continuous operation. laboration between the operational institutions is an effective hedge ■ Wide band-width Internet connec- GoMOOS modeling group and against instability and funding ons, parallel software systems, and the Water Resources shortfalls. IOOS, across the country collaborative operations between Authority (MWRA) led to a closer and in the Gulf of Maine region in two satellite receiving stations with examination of the modeled Gulf of particular, has been heavily subsidized overlapping station masks allow both Maine Coastal Current (GMCC) in by academic institutions. Without systems to have backup capability response to Northeast storms during the financial support of the academic in the event of episodic hardware the spring. It also led to the devel- institutions, the observing systems failure. The high cost of satellite data opment of a map-based interactive would have failed years ago as a result receiving stations generally precludes particle tracking module using the of the lack of sufficient funding to redundant, or backup, systems within GoMOOS forecast and the General carry on operations. In the case of an individual observing system. NOAA Operational Modeling Envi- land grant and sea grant colleges and However, station masks are geo- ronment, which is a useful research, universities, this providing of service graphically large. By partnering with management, and educational to the citizens of their states and another station that has similar goals tool (http://rocky.uemcoe.maine. the generation of new knowledge, and operations (in our case Rutgers edu/GoMPOM/cdfs/gnome/web). together with their primary pursuit University), as well as similar soft- ■ Dual telemetry systems increase of education, form the three compo- ware, both systems have backup and the success of real-time data systems. nents of their tripartite mission. In can maintain daily operations and a The real-time mission of IOOS turn, the academic institutions benefit consistent, continuous data stream. dictates that the telemetry functional- from the research and educational ■ Regional forecast models can take ity must be among the most reliable opportunities that accompany IOOS advantage of a higher level of coor- components of the observing system. operations, and from increased vis- dination; not only between model- GoMOOS uses two independent ibility that comes with success in new ers and observers, but also between data telemetry systems (cellular/irid- and highly visible endeavors. Univer- modelers running models of differ- ium and a GOES satellite system) sities are large and diverse entities that ent scales. When we were informed to increase substantially the success- have the ability to take a long view of in 2006 about NCEP’s decision to ful transmission of real-time data. the benefits of programs that are, in switch its operational ocean forecast • Diverse sensors and multiple the short view, a financial burden. model from ROFS to RTOFS, which approaches to deriving data prod- ■ Duplicate buoys and a six-month provides open boundary conditions ucts is a powerful methodology for servicing schedule keep the system for the operational GoMOOS circu- maintaining high-quality bio-opti- at a high level of performance. The lation model, we had sufficient time cal observations and products. Our

Fall 2008 Volume 42, Number 3 99 primary bio-optical data product is become ardent supporters of the buoy observing assets into the system. The phytoplankton biomass. We have system. As these communities have GoMOOS buoy array was designed between two and four methods for come to rely upon the data provided, in 2000. At that time it represented assessing this product on any of the the system operators have come to a significant advance over the buoy bio-optical buoys (e.g. chlorophyll rely on fishermen and pilots to keep systems that had previously been fluorescence, chlorophyll absorption, them informed about the condition used for real-time observing in the chlorophyll-specific light attenuation, of the buoys. The maritime and fish- region. Because of a shortage of fund- solar-stimulated chlorophyll fluores- ing industries are among the most ing, no major improvements in the cence). Because of this approach, we enthusiastic users of the buoy data design have been made since opera- can lose 75% of our sensing capabili- and often take the initiative to report tions started. Today the technology ties onboard and still maintain the problems that they observe (broken exists to produce a much more able time series observations. In addition, light, broken solar panels, etc.). system than the one that is pres- this sensor diversity provides the ently deployed in the GoM. The data necessary to identify and correct Type II Lessons CODAR array has similarly suffered for biofouling and sensor drift. ■ Observing systems can not operate from that lack of ongoing invest- ■ Two-way communications with at optimal efficiency when funding is ment. Initial estimates of the range the buoy data logger and sensors year-to-year, late in arriving, and sub- of the CODAR system in the GoM is an indispensable element of the ject to significant and unpredictable were optimistic, and the perform- successful observing system. We fluctuations. Year-to-year operating ance has deteriorated with age. The routinely call the buoys to retrieve budgets for GoMOOS have fluctu- result is that the overlap of the units missed data transmissions, and can ated by a factor of two–four. Under is rarely adequate, and the system often reset sensors that have stopped these chaotic conditions, effective has not achieved operational status. sampling and/or communicating. planning, scheduled infrastructure ■ Test platforms for new technologies ■ Wind sensors are a potentially maintenance, and staff continuity are are a vital element of an evolving lifesaving element of the OOS. extraordinarily difficult to achieve. IOOS program. New advances are Dual sensors with different operat- ■ Scientists responsible for the collec- not made in operational oceanog- ing principals helps assure that these tion and QA/QC of data and must raphy without missteps, equipment potentially lifesaving data are reliably have authority over data released to failure, and lost data. Since the at hand. On GoMOOS buoys we use users. There can be only one au- real-time nature of IOOS is a very both traditional mechanical propeller thoritative data archive, and it must important aspect of the overall mis- and vane sensors (susceptible to inter- be controlled by the scientists who sion, new technologies need to be ruption in freezing-spray conditions) are responsible for the data collec- tested in a way that does not detract and sonic anemometers (suscepti- tion, calibration, and quality control. from the existing service. Once ble to noise during heavy rain). In the case of the Gulf of Maine, an innovation has been tested and ■ Automatic alarms for exceeding a separate nonprofit organization proven on a test mooring, it can be the watch circle or leaks in the (GoMOOS, Inc.) controls the release implemented throughout the array. electronics well are simple to imple- of data to the public and to NDBC. ment and provide an early warn- At various times over the first seven Suggestions for Future IOOS ing of impending trouble. The years of operation, the data being ■ The IOOS funding process needs to GoMOOS buoys send a message released by GoMOOS, Inc. were not be much more stable and predictable. to the University of Maine Physi- the highest quality data available, and To date, it has not been well matched cal Oceanography Group (PhOG) in some cases were corrupted in the to operations, continuity, and techni- server if the buoy exceeds its normal transfer process. Although the correct cal advancement of the observing watch circle, or if moisture is detected data were available at the University systems. Until recently, IOOS has in the electronics/battery well. of Maine website, the majority of been funded primarily through the ■ Although the fishing and piloting users downloaded data either from political process of congressional communities in the GoM were initial- the GoMOOS site, or from NDBC. earmarks. As a direct result of this ly skeptical of the value of the buoys, ■ A fraction of an RCOOS budget process, IOOS funding went prefer- initially complained about the de- (~15%) should be set aside to fund entially to states with representatives ployment locations, and in a few cases research and development, replace- who were powerful in the appropria- dragged or attempted to drag buoys ment of aging sensors and platforms, tions process. This flawed arrange- out of their way, they have since and the incorporation of additional ment has resulted in bizarre funding

100 Marine Technology Society Journal decisions where some states without ■ Although it is often said that data 4. Summary and Conclusions established ocean observing capabili- acquisition, data management, and For the scientists who have made the ties or existing OOS infrastructure modeling are the three pillars of GoMOOS data acquisition and modeling were given several million dollars per IOOS, the funding of modeling programs notable successes, the experience year, while other states with extant activities has been limited. There is has been simultaneously exhilarating and and successful operational systems sometimes the lack of enthusiasm to disheartening. It has been very gratifying went virtually unfunded because the support operational models, especially to have had the opportunity to implement representatives of their states lacked when the funding becomes tight, plans, which had long existed, to instru- sufficient political power to compete because the models are often per- ment the GoM in a comprehensive and for funds. We believe that this unfair ceived as inaccurate and inaccessible sustained manner. The scientific impact of and ineffective funding process has to users. It is thus important for the this system on GoM research is enormous significantly hampered IOOS, and community to establish standards for and its impact will continue for years even has resulted in far less data per dol- operational models and to heighten if the system does not. It was also a rare treat lar being served to system users. public awareness about the usage for many of the scientists to be involved ■ The competitive process for future and limitation of ocean forecasts. in a project that was so highly regarded IOOS funding needs to be fully Continuous investments are needed by the general public, generated so much implemented without political influ- in maintaining operational models to interest, and engendered a devoted group ence remaining as a major factor in meet standards as well as in research of GoMOOS data users. We owe this op- funding decisions. Although there and development that will allow portunity, and the public interest, largely to has been improvement in the funding achievement of improved standards. those who served on the administrative and decisions by having a formal proposal ■ The management structure of the organizational side of GoMOOS. process, it seems that “regional” fac- RCOOS systems should be careful At the same time, the experience has tors are still weighted heavily in the not to put the scientists and engi- been unusually frustrating and dishearten- funding decisions. We believe that neers in the position of having the ing. Funding levels seemed unrelated to proposal merit, scientific justifica- major responsibility for making the excellence and to the successful execution tion, efficiency, past performance, RCOOS function, without having of the project. In addition, the scientists and technical capability should be commensurate authority over the had an ever decreasing role in the funding the overriding factors that deter- setting of technical and budgetary process and the setting of priorities. In a mine funding decisions. Otherwise priorities. The highest priority of good funding climate, the structure of Go- the limited IOOS funds will con- the RCOOS should be the reli- MOOS would have been less problematic, tinue to be used sub-optimally. able measurement and delivery of since the proposal process affords direct ■ Decision making by Regional research quality data to all users. In input from the scientists and a contractual Associations, as the conduits for the times of funding shortages, these obligation to honor the priorities, plans, NOAA competitive IOOS funds, core functions need to be preserved and budgets set forth in the proposal. can be negatively influenced by those even at the expense of higher-level However, the proposals were almost never who continue to view the funds as product development, improved user fully funded, and detailed guidance for a regional entitlement to be divided interfaces, and public relations. structuring the cuts were not (to our knowl- among the participating states and ■ RCOOS systems should resist the edge) provided by NOAA. The end result institutions. We favor starting by es- temptation to favor larger numbers was that the administration and board of tablishing the highest priority regional of “qualitative” data streams over directors restructured the budget and, as outcomes, and then determining research-quality data streams. In some the funding levels dwindled, the scientists those technical groups best able to cases managers and administrators felt powerless to protect the observing produce those outcomes economi- favor data quantity over data qual- infrastructure. cally and reliably. There is a tendency ity since many real-time users are for the process to be derailed by an satisfied with approximate readings. effort to find something that could While this may offer short-term Acknowledgments be done by each of the regional public relations advantages, the long- We thank the entire staff of the Uni- partners. We hope that the NOAA term benefits of the scientific archive versity of Maine Physical Oceanography IOOS office will provide high-level continue to accrue indefinitely. Group, the Ocean Modeling Group, the guidance on allocations; especially Environmental Optics Group, the Satel- in situations where the award is lite Oceanography Data Laboratory, and less than the requested budget.

Fall 2008 Volume 42, Number 3 101 the Nutrient Analysis Laboratory, whose Pettigrew, N.R. and C. S. Roesler. 2005. Thomas A.C., D.W. Townsend and R. Weath- dedication and determination made this Implementing the Gulf of Maine Ocean Ob- erbee. 2003. Satellite-measured phytoplankton work a success. We gratefully acknowledge serving System (GoMOOS). IEEE proceed- variability in the Gulf of Maine. Cont Shelf funding from ONR, NOAA, the State of ings of ‘05 (), Brest , Res. 23:971-989. Maine, the University of New Hampshire, 1234-1241. Townsend, D.W., A.C. Thomas, L.M. Mayer, the University of Maine, and GoMOOS. Pettigrew, N.R., J.P. Wallinga, F.P. Neville and M. Thomas and J. Quinlan. 2006. Oceanog- We also acknowledge the efforts of the K.R. Schlenker. 2005b. Gulf of Maine Ocean raphy of the Northwest Atlantic Continental GoMOOS staff and the GoMOOS Board Observing System: Current Measurement Shelf. pp. 119-168. In: Robinson, A.R. and of Directors who have worked diligently Approaches in a Prototype Integrated Ocean K.H. Brink (eds.), The Sea, Volume 14. Har- to make these data widely accessible to the Observing System. IEEE Proceedings, Eighth vard University Press. public via www.gomoos.org. Current Measurement Technology Conference Wallinga, J.P., N.R. Pettigrew and J.D. Irish. 2005. 127-131. 2003. The GoMOOS moored Buoy Design. References Pettigrew, N.R. and F. Neville. 2008. Gulf of Proceedings of the IEEE Oceans 2003 confer- Bigelow, H.B. 1927. Physical Oceanography Maine Ocean Observing System (GoMOOS): ence, pp. 2596-2599. Current Measurement in an Integrated Ocean of the Gulf of Maine. Fish Bull. 511-1027. Xue, H., L. Shi, S. Cousins and N. R. Pet- Observing System. Proceedings of the IEEE/ tigrew. 2005. The GoMOOS nowcast/forecast Brooks, D.A. 1985. Vernal circulation in the OES/CMTC 9th Working Conference on system. Cont Shelf Res. 25:2122-2146. Gulf of Maine. J Geophys Res. 90:4687-4705. Current Measurement Technology, 143-150.

Irish, James D. 1997. Elastic Tether Moor- Pettigrew, N.R., C. S. Roesler, F. Neville and ing Technology: Experiences in the Gulf of H.E. Deese. 2008. An ocean-sensor network Maine and on Georges Bank, Sea Technol. in the Gulf of Maine. Lecture notes in Com- 38(10):61-69. puter Science. 1-26, (in press).

Luerssen R.M., A.C. Thomas and J. Hurst. Roesler, C.S. and E. Boss. 2008. In situ 2005. Relationships between satellite-measured measurements of the inherent optical proper- thermal features and Alexandrium-imposed ties (IOPS) and potential for harmful algal toxicity in the Gulf of Maine, Deep Sea Res bloom (HAB) detection and coastal ecosystem Pt II. 52:2656-2673. observations. Chapter 5, pp. 153-206. In: M Mellor, G.L. 2003. Users guide for a three- Babin, C.S. Roesler and J.J. Cullen, (eds.), dimensional, primitive equation, numerical Real-time Coastal Observing Systems for Eco- ocean model, 53 pp., Prog. in Atmos. and system Dynamics and Harmful Algal Blooms. Ocean. Science, Princeton University. UNESCO Series Monographs on Oceano- graphic Methodology. Padilla, R., Perrie, W., Toulany, B. and Smith, P, C. 2007. Intercomparison of third Sinclair, M., S. Wilson and D.V. Subba Rao. generation wave models. Weather Forecast. 1991. Overview of the biological oceanogra- 22(6):1229-1242. phy of the Gulf of Maine. Proceedings of the Gulf of Maine Scientific Workshop, Woods Pettigrew, N.R., D.W. Townsend, H. Xue, J. Hole Massachusetts, 91-111. P. Wallinga, P. J. Brickley and R. D. Hetland. 1998. Observations of the Eastern Maine Siegel, D.A., T. K. Westberry, M.C. O’Brien, Coastal Current and its Offshore Extensions N.B. Nelson, A.F. Michaels, J.R. Morrison, A. in 1994. J Geophys Res. 1039C130;30,623- Scott, E.A. Caporelli, J.C. Sorensen, S. Mari- 30,639. torena, S.A. Garver, E.A. Brody, J. Ubante and M. A. Hammer. 2001. Bio-optical modeling Pettigrew, N.R., J. H. Churchill, C.D. Janzen, of primary production on regional scales: the L.J. Mangum, R.P. Signell, A.C. Thomas, Bermuda BioOptics project. Deep Sea Res II. D.W. Townsend, J.P. Wallinga and H. Xue. 48:1865-1896. 2005a. The kinematic and hydrographic struc- ture of the Gulf of Maine Coastal Current. Deep Sea Res II. 52:2369-2391.

102 Marine Technology Society Journal