Real-time Estimates of Hypoxic Water Volume

Chesapeake Bay

HV_summary
Quick Summary

Scientists estimate the amount of hypoxia in the Chesapeake Bay using a metric called the "hypoxic volume," which is the volume of water in the Bay with a dissolved oxygen concentration less than 2 milligrams per liter (mg/L). This volume represents an approximate size of the Dead Zone in the Bay. In the spring, the size of the summer dead zone is forecast based on the amount of nutrients supplied to the Bay. Once or twice a month, boats are used to collect observations and estimate the hypoxic volume. However, both these methods provide very infrequent estimates of hypoxic volume.

The Chesapeake Bay Hypoxia Forecast Model estimates the hypoxic volume every day. The daily hypoxic volume forecasts are used to calculate the total annual hypoxic volume throughout the year. This metric provides a single number that represents the severity of hypoxia in a given year. The model was used to estimate hypoxic volume for each day from 2014 through 2019 for comparison with 2020. These daily estimates are based on complex computer models that continue to be improved; therefore, past estimates may be updated as improvements are made to the models.

realtime_HV_table
2020 Dead Zone Size

The amount of hypoxia in the Bay is expected to increase from spring into summer and then decrease as summer progresses into fall, with hypoxia starting sometime in May. The seasonal forecast for 2020 is for the total amount of hypoxia to be about average, relative to the amount of hypoxia in the past. However, short-term weather will affect the amount of hypoxia in the Bay. Based on the forecast model, 2020 hypoxia started later than in other recent years. Check back to see how the size of the dead zone increases seasonally and how daily weather changes the amount of hypoxia. Notable weather that may impact the amount of hypoxia are very windy days or periods of very calm wind. The image below will be continually updated throughout 2020 based on the daily forecast model.

2020_RT_HV_figure

historic_HV_table
Hypoxic Volume (HV) Metrics for Recent Years for Comparison to 2019 Forecast
Year Maximum Daily HV [km3] Average Summer HV [km3] Hypoxic Duration [days] Total Annual HV [km3 days]
Historical 6.8 to 12.6 2.8 to 6.6 93 to 143 411 to 951
2014 7.7 (10%) 4.9 (6%) 115 625 ± 81
2015 9.9 (13%) 4.6 (6%) 98 587 ± 76
2016 10.7 (14%) 5.1 (7%) 101 664 ± 86
2017 9.9 (13%) 5.3 (7%) 92 657 ± 85
2018 10.4 (13%) 4.8 (6%) 123 645 ± 84
2019 13.1 (17%) 6.3 (8%) 136 826 ± 107

Notes: 1 km3 equals about 400,000 Olympic-sized swimming pools of water. Percents (%) represent the percent of the Bay that was hypoxic based on the volume of the Bay in the forecast model. Historical values can be considered the normal range in conditions from 1985 to 2018. Historical values are based on a 34 year model simulation.

  • Maximum Daily Hypoxic Volume (km3): The maximum volume of Chesapeake Bay water experiencing hypoxic conditions on any given day
  • Average Summer Hypoxic Volume (km3): The average volume of hypoxic water from June through September
  • Hypoxic Duration (days): The number of days in a given year between the first and last day of hypoxic conditions exceeding 2 km3 in volume
  • Total Annual Hypoxic Volume (km3 days): The total amount of hypoxia in the Bay for a given year, calculated by summing the hypoxic volume on each day. Uncertainty in total annual hypoxic volume estimates from cruise-based observed data has been estimated at 13% (Bever et al. 2018, Table 4). The 13% was adopted here as a general estimate of the uncertainty of the total annual hypoxic volume for each year, represented by the plus and minus (±) value.
2020_accuracy_summary
Summary of 2020 Model to Data Comparison (Model Accuracy)

The Maryland Department of Natural Resources and Old Dominion University periodically collect dissolved-oxygen data from the water surface to the seabed as part of the long-term Water Quality Monitoring Program. We use these data to calculate an estimate of the hypoxic volume, which we compare with the hypoxic volume estimated using the model. The data-based and model-based estimates of hypoxic volume will not be the same because different methods are used for each. However, we expect they should be similar and follow a similar seasonal pattern. The image and text below compare the model-based and data-based hypoxic volumes; we periodically update the data-based hypoxic volumes through the summer as more data become available. The black lines above and below the gray dots show the uncertainty in the data-based estimate of hypoxic volume.

The model-based estimate was zero hypoxic volume (no hypoxia) in the mainstem of the Chesapeake Bay at the end of May and beginning of June. This matches well with the data-based estimate of hypoxic volume around the same time. The monitoring data captured hypoxia at a single location in the mainstem, corresponding to the very small data-based estimate of hypoxic volume. Both the model and data show hypoxia starting later in 2020 than in other recent years. The forecast model shows hypoxia increasing from the beginning of June to the end of July, matching with the data-based estimates of hypoxic volume and the summary released by the Maryland DNR.

2020_model_data_figure

2019_synopsis
Synopsis of 2019 Dead Zone Size

Please see the 2019 Chesapeake Bay Dead Zone Report for a summary of hypoxia in 2019 relative to historic conditions and other recent years.

Springtime inflows from the Susquehanna River were high in 2019, resulting in the prediction that 2019 would be the 4th largest July hypoxic volume in the last 20 years. However, summer winds and temperatures also play large roles in the severity of hypoxia. Through mid-July 2019, total annual hypoxic volume was on the high end of 2014 to 2018. Weak winds and high temperatures from the end of June to mid-August allowed hypoxia to increase to higher levels. This is different from 2018, when strong winds reduced the amount of mid-summer hypoxia. In 2019, hypoxia decreased quickly in late August and early September (Hurricane Dorian) as winds increased; however, hypoxia returned with the high temperatures in late September and early October until strong winds mixed the Bay water and ended hypoxia in the mainstem of the Bay for the year. Overall, the total amount of hypoxia in 2019 was estimated to be on the high end of the normal range for 1985 to 2018, and higher than in the recent past. As in 2018, hypoxia also lasted longer than in other recent years.

Even with environmental conditions that favor severe hypoxia (high riverine input to the Bay and light winds), the total amount of hypoxia in 2019 was within the normal range, suggesting nutrient reductions since the 1980s have helped improve water quality in the Bay.

2019_model_data_summary

Summary of 2019 Model to Data Comparison (Model Accuracy)

The Maryland Department of Natural Resources and Old Dominion University periodically collect dissolved-oxygen data from the water surface to the seabed as part of the long-term Water Quality Monitoring Program. They use these data to calculate an estimate of the hypoxic volume, which we compare with the hypoxic volume estimated using the model. The data-based and model-based estimates of hypoxic volume will not be the same because different methods are used for each. However, we expect they should be similar and follow a similar seasonal pattern. The image and text below compare the model-based and data-based hypoxic volumes; we periodically update these through the summer as more data become available. The black lines above and below the gray dots show the uncertainty in the data-based estimate of hypoxic volume.

The model-based estimate of hypoxic volume for May 2019 was about average with recent years. However, the model-based hypoxic volume was lower than the data-based hypoxic volume for this first estimate of the severity of hypoxia. The model-based estimate of hypoxic volume near the beginning of June matched very well with the data-based estimate and shows an increase in hypoxia from late-May to early-June. The model-based hypoxic volume suggests a peak in hypoxia around June 2nd, then a decrease in the amount of hypoxia through mid-June. This decrease in mid-June matches well with the finding from Maryland DNR that there was less hypoxia than expected in late-June, likely as a result of windy conditions. The amount of hypoxia again increased during late-June into July in both the model-estimated and the data-estimated hypoxic volumes. High river inflows of freshwater and nutrients coupled with calm winds and high temperatures in July resulted in the amount of hypoxia being quite large compared to the average from 1985 through 2018. The model forecast and observed data suggest the amount of hypoxia began to decrease following the second July data collection cruise and was absent from the Bay by mid-October. The model-estimated hypoxic volume was within or right near the range of uncertainty in the data-based hypoxic volume for all but the May cruise, providing confidence in the model-estimated hypoxic volumes.

HV_mod_data_2019

2018_HV_summary
Synopsis of 2018 Dead Zone Size

The forecast model underwent a significant improvement in the estimated salinity in early 2019, which also affected the estimates of water temperature, dissolved oxygen and hypoxia. Some of the dead zone sizes will be different between these older results and those presented in the above 2019 realtime dead zone size summary.

During fall 2018, the Virginia Institute of Marine Science joined with Anchor QEA, LLC, to release their annual retrospective analysis of the severity of hypoxia in the Chesapeake Bay. The Annual Chesapeake Bay Hypoxia Report Card summarizes dissolved oxygen in the Bay as estimated by the team's 3-D, real-time hypoxia forecast model. The modeling team also generated dissolved oxygen statistics for 4 previous years for comparative purposes.

Springtime inflows from the Susquehanna River were high in 2018, resulting in the prediction that 2018 would have an above-average amount of hypoxia. However, wind speed and direction also play a large role in the severity of hypoxia during the summer. During 2018, the total annual hypoxic volume was similar to 2014 and 2017 through mid-July, but larger than in 2015 and 2016. Strong winds in the second half of July reduced the amount of hypoxia to near zero. Hypoxia increased rapidly again in early August and peaked at a higher value in early September than in previous years. Strong winds in September again mixed the Bay water and resulted in a large reduction in the volume of hypoxic water. Overall, the total amount of hypoxia in 2018 was estimated to be similar to 2017, but the seasonal patterns in hypoxia were very different; hypoxia was estimated to start earlier and last longer in 2018 than in recent years (Table below). The lack of hypoxia in late July was very unlike historical dissolved oxygen conditions, but consistent with the report released by the Maryland Department of Natural Resources for the Maryland portion of the Bay.

2018_model_data_summary
Summary of 2018 Model to Data Comparison (Model Accuracy)

The forecast model underwent a significant improvement in the estimated salinity in early 2019, which also affected the estimates of water temperature, dissolved oxygen and hypoxia. Some of the dead zone sizes will be different between these older results and those presented in the above 2019 realtime dead zone size summary.

Periodically throughout the year, Maryland Department of Natural Resources and Old Dominion University use boats to measure dissolved oxygen at many depths and numerous locations throughout the Bay. Many components of the Chesapeake Bay Hypoxia Forecast Model have been compared to these boat-based data to validate the accuracy of the forecast model. These comparisons have included both dissolved oxygen concentration and hypoxic volume. Some of the comparisons were published in scientific papers by Bever et al. (2013) and Irby et al. (2016).

Hypoxic volume is estimated from the boat-based data by interpolating (extrapolating) the measured dissolved oxygen concentration throughout the Bay and calculating the volume of water with dissolved oxygen concentration less than 2 mg/L. However, because the data is collected on different dates and times and is then extrapolated throughout the Bay, the data only provide periodic estimates of the amount of hypoxia. The 2018 hypoxic volumes estimated from the Chesapeake Bay Hypoxia Forecast Model were compared to the periodic data-estimated hypoxic volumes, to validate the seasonal pattern and amount of hypoxia estimated by the model. In this comparison, the model-estimated hypoxic volume is shown on the figure below for every day and the data-estimated hypoxic volume is shown as eight individual estimates.

Both the model and the boat-based data show hypoxia began in May and increased into June with relatively large hypoxic volume during the second data collection episode in June. Throughout July the model-estimated and data-estimated hypoxic volumes decreased to very low volumes and then increased rapidly in August. The model-estimated and data-estimated hypoxic volumes then decreased through September. This demonstrates that the model and data estimated the same seasonal pattern of hypoxic volume in which hypoxia was nearly absent near the end of July and in which:

  • Hypoxia began in May and increased through June
  • Hypoxia decreased during the last third of June to unusually low hypoxic volumes
  • Hypoxia increased rapidly in August and into September before decreasing into October

The model-estimated hypoxic volumes were very similar to the data-estimated hypoxic volumes, which were centered around June 5, July 10, July 26 and August 22. The model-estimated hypoxic volume was marginally lower than the data-estimated hypoxic volume for the remaining four data-collection episodes. This demonstrates that the model-estimated hypoxic volumes were very accurate based on half the data-collection dates and were slightly low for the other half of the dates, and that the model was accurate enough to use for understanding the variations in hypoxia at different times and different places in the mainstem portion of the Chesapeake Bay.

2018_HV_and_data