Portland State University PDXScholar

Physics Faculty Publications and Presentations Physics

2-3-2009 Global Emissions From Wetlands, Rice Paddies, and Lakes

Qianlai Zhuang Purdue University

John M. Melack University of California, Santa Barbara

Sergey Zimov North-East Science Station, Cherski

Katey Marion Walter University of Alaska, Fairbanks

Christopher Lee Butenhoff Portland State University

See next page for additional authors

Follow this and additional works at: https://pdxscholar.library.pdx.edu/phy_fac

Part of the Physics Commons Let us know how access to this document benefits ou.y

Citation Details Zhuang, Q., J. M. Melack, S. Zimov, K. M. Walter, C. L. Butenhoff, and M. A. K. Khalil (2009), Global Methan Emissions From Wetlands, Rice Paddies, and Lakes, Eos Trans. AGU, 90(5), 37, doi:10.1029/ 2009EO050001.

This Article is brought to you for free and open access. It has been accepted for inclusion in Physics Faculty Publications and Presentations by an authorized administrator of PDXScholar. Please contact us if we can make this document more accessible: [email protected]. Authors Qianlai Zhuang, John M. Melack, Sergey Zimov, Katey Marion Walter, Christopher Lee Butenhoff, and M. A. K. Khalil

This article is available at PDXScholar: https://pdxscholar.library.pdx.edu/phy_fac/5 Eos, Vol. 90, No. 5, 3 February 2009

Volume 90 number 5 3 FEBRUARY 2009 EOS, Transactions, American Geophysical Union pages 37–44

accurately quantify global methane emis- Global From sions with these models [e.g., Dlugokencky et al., 2003]. Instrumentation for monitoring methane fluxes and concentrations should Wetlands, Rice Paddies, and Lakes be a priority. Controls on, and processes of, biogenic PAGES 37–38 surveys of vegetation and wetland types methane emissions are not well understood. should be continued as a means to further We suggest the following research priorities: The current concentration of atmospheric develop data on global lake and wetland dis- (1) elucidate the role of degrada- methane is 1774±1.8 parts per billion, and tribution and extent. tion in methane emissions; (2) further exam- it accounts for 18% of total Methane flux measurements, which are ine the effects of the atmospheric deposition radiative forcing [Forster et al., 2007]. Atmo- lacking from a variety of , as of nitrogen and sulfate on methane produc- spheric methane is 22 times more effec- noted below, are needed to improve emis- tion and consumption; (3) investigate the tive, on a per-­unit-­mass basis, than sion estimates and the accuracy of biogeo- microbial community structure and distri- dioxide in absorbing long-­wave radiation chemical models and atmospheric transport bution of those microbes responsible for on a 100-­year time horizon, and it plays an inversion models [e.g., Zhuang and Ree- methane production and consumption; and important role in atmospheric ozone chem- burgh, 2008]. Data acquisition of in situ and (4) investigate the role of different vegetation istry (e.g., in the presence of nitrous oxides, satellite atmospheric concentrations and types in affecting the exchanges of methane tropospheric methane oxidation will lead their profiles should be maintained to more between wetlands and the atmosphere. to the formation of ozone). Wetlands are a large source of , Arc- tic lakes have recently been recognized as a major source [e.g., Walter et al., 2006], and anthropogenic activities—such as rice agriculture—also make a considerable contribution. However, the quantification of methane emissions still has large uncertainties. In this article, we identify some causes for the uncertainty; illustrate the challenges of reduc- ing the uncertainty; and highlight opportu- nities for research from the global perspec- tive and also from the perspective of three principal sources of methane: the Arctic, the Amazon basin, and rice paddies.

Global Perspective

Accurately mapping the global distribu- tion and dynamics of lakes and wetland types is challenging. Currently, estimates of global wetland areas from ground and sat- ellite instruments vary from 1 to 12 million square kilometers. Some existing estimates do not account for small lakes, resulting in underestimates of lake areas by a factor of Fig. 1. (a) In the Siberian north, the polygonal net of ice wedges is ubiquitous. In -­aged more than 2 [e.g., Walter et al., 2007]. Fur- (1.8 million to 10,000 years before present) (an organic-­rich loess with about 2% car- ther, global characterizations of wetlands bon by mass and with ice content of 50–90% by volume [see Walter et al., 2006]), the tops of and lakes are too coarse to represent signifi- ice wedges are usually situated directly beneath the active layer, so the active layer growth will cant differences in methane emission rates, lead to ice wedge melting and the appearance of the polygonal net of deep channels. Ponding and data on seasonal and interannual wet- of water in these channels causes deep thaw of permafrost, leading to the formation of methane- land inundation derived from satellite sen- ­producing thermokarst lakes. (b) In the Arctic lakes, open holes in early winter lake ice due sors need to be further developed. Satel- to seeps of methane bubbling are visible from airplanes and in TerraSAR-­X satellite imagery. lite data in combination with ground-­level (c) Floodable area (black) for the Amazon basin below the 500-­meter contour, derived from Japanese Earth Resources Satellite 1 (JERS 1) synthetic aperture radar mosaics. Floodable areas are not all inundated simultaneously and may include areas not floodable that are not visible at By Q. Zh u a n g , J. M. Me l a c k , S. Zi m o v , K. M. the scale of the image. These floodable areas are important to methane emissions in the Amazon Wa l t e r , C. L. Bu t e n h o f f , a n d M. A. K. Kh a l i l basin [from Melack and Hess, 2009]. Eos, Vol. 90, No. 5, 3 February 2009

Challenges and Opportunities in the Arctic have large uncertainties, and many tropi- 40 milligrams of methane per hour per cal systems lack measurements of emis- square meter. In part, this range reflects the The carbon-­rich Arctic soils and lakes sions. For the Amazon basin, Melack et al. different ecosystems under which rice is are underlain with either continuous or dis- [2004] combined passive and active micro- grown (e.g., irrigated, rain-­fed, deepwater, continuous permafrost, which is a large res- wave remote sensing of the temporally vary- and so forth), and water management (e.g., ervoir of carbon containing 700–­950 pen- ing extent of inundation and vegetation with continuous versus intermittently flooded), as tagrams of carbon in its top 1–­25 meters field measurements to estimate that meth- longer inundation periods produce higher (Figure 1a) [Zimov et al., 2006]. Currently, ane emissions are 22 teragrams per year. fluxes. It also reflects differences in input methane is released from both thaw lakes Further improvements in these estimates will organic matter, soil type, and crop phenol- and soils (Figure 1b) [Walter et al., 2006]. require field campaigns in undersampled ogy, among other variables. In addition, An increase in permafrost degradation environments including extensive savannas, SAFs from the same field in side-­by-­side and the shoreline erosion of existing lakes high-­elevation wetlands, narrow riparian plots are observed to vary over the emis- along with the formation of new permafrost- zones along streams, and small rivers. sion season by a factor of 2 to 4, evidently ­thawing lakes is expected, and these Another important area in northern owing to field inhomogeneities. Sampling of changes could increase methane emissions South America is the upper Negro River methane flux in paddies using the standard from these lakes by several orders of magni- basin, because it includes an approximately method of three replicates can produce tude [Walter et al., 2007]. 80,000-­square-­kilometer area with season- SAFs that vary by 40–60% [Khalil and Buten- As permafrost thawing develops, ally flooded emergent grasses and sedges, hoff, 2008]. increased linear erosion associated with areas with shrubs or palms, and flooded Paddy emissions are sensitive to farm- thaw lake expansion and migration could forests—­all of which are methanogenic ing practices that can change in response lead to stream channel erosion and lake habitats. Year-­round measurements of dif- to economic and political pressures [Khalil drainage. For instance, in the discontinuous fusive and bubble emissions of methane in and Rasmussen, 1993]. In China, conserva- permafrost zone of West , this process these organic-­rich environments, and the tion has reduced the amount of water used has already begun to lead to a decrease in influences of habitat, water depth, variation for paddy irrigation, while bans on biomass lake area. In West Siberia’s far north, deep, in hydrostatic pressure, dissolved oxygen, burning have returned otherwise burnt crop cold ice wedges comprise more than 80% and temperature, have recently been deter- residue to the field. Manure use has declined of permafrost volume and are often covered mined by Belger [2007]. Similar measure- in favor of inorganic fertilizers. Quantifying with only 0.5 meters of soil. Minor increases ments are needed in the Pantanal in central these changes is vital for determining accu- in summer soil active layer thickness will ini- South America, the ­Llanos de Moxos, largely rate inventories, but it remains challenging. tiate the thaw of ice wedges in permafrost. in eastern Bolivia, the Roraima savannas Many farming practices are not officially Once the melting process begins, the dis- of northern Brazil, and the swamps of low- recorded and are available only through per- appearance of these large ice bodies will land Peru, although many of the wetlands sonal communication. Interviews with farm- result in dramatic changes in land surface in these areas are situated in interfluvial ers have yielded important information, but driven by active thaw erosion, thawing tens regions, making access very difficult. more extensive surveying is required. of meters into permafrost. Methane emissions from Amazonian res- Extrapolating paddy fluxes to larger scales In upland environments, permafrost- ervoirs also have been measured occasion- requires geospatial data on rice production ­thawing ponds will form in the channels ally at the Samuel, Tucuruí, and Curuá-­Una and controlling factors. Current agricultural that appear above thawing ice wedges. reservoirs (located in the central and east- census data are insufficient to meet increas- The bottoms of these ponds remain unfro- ern Amazon basin in Brazil), and more ing resolution demands (<10 kilometers). zen throughout winter, actively producing extensively in the Balbina reservoir (located Process-­based modeling and remote sensing and emitting methane year round. On wet north of Manaus in the central Amazon), are being used to fill some of these gaps. For lowlands, the development of a polygonal and the potential importance of downstream example, satellite-­derived vegetation indices system of channels can increase the drain- degassing in tropical reservoirs has been can locate paddies on the basis of crop phe- age regime and partially dry the landscape, demonstrated [Kemenes et al., 2007]. Fur- nology. These techniques can map not only increasing areas for methane oxidation in ther measurements with improved method- spatial coverage, but also temporal changes. dry surface soils and reducing the net meth- ology are needed to evaluate the extent of Only in recent years have scientists been ane emissions from this environment [e.g., degassing. able to use remote sensing to detect meth- Smith et al., 2005]. Thus, modeling land- The entire lowland Amazon basin— ane emissions. Placing precise constraints scape relief and drainage patterns should be which, because of its large area of metha- on individual sources of methane emissions a priority in predicting the response of Arc- nogenic habitats, is important for methane is difficult because all sources contribute tic lake and wetland methane to permafrost emissions—has an estimated floodable area to the signal, but satellite-­derived measure- thaw. of 0.8 million square kilometers (Figure 1c). ments have put upper level constraints on The key questions in estimating methane The improved topography provided by the rice emissions (e.g., <80 teragrams of meth- emissions from permafrost environments are Shuttle Radar Topography Mission’s inter- ane per year [Frankenberg et al., 2005; see as follows: (1) What is the size of the per- ferometric synthetic aperture radar data, also Xiong et al., 2008]). Further refinements mafrost carbon pool? (2) Which fraction of in combination with runoff models, should to the paddy emission budget will benefit this pool will be converted to methane upon improve the estimation of the extent of the from a longer time series of satellite data and thaw? (3) Which fraction of the permafrost methane-­producing habitats. from regional ground-­based networks closer carbon pool will thaw under anaerobic ver- Measuring and modeling inundation and to paddy emissions. These efforts will be vital sus aerobic conditions? (4) What is the role vegetation dynamics in seasonally flooded to verifying and monitoring national commit- of methane oxidation in controlling methane savannas, palm swamps, riparian zones, ments to greenhouse gas emission treaties. emissions? and wetlands in mountainous terrain are top In conclusion, to accurately quantify research priorities, and flux measurements global methane emissions, the priorities in the Amazon should continue. are improving biogeochemical and atmo- Challenges and Opportunities spheric inversion and transport modeling in the Amazon Basin Challenges and Opportunities approaches and developing high-­resolution in Rice Paddy Emission Studies data on wetlands and lake distribution and Tropical wetlands are a major source of their changes using remote sensing data atmospheric methane. However, regional Seasonally averaged methane flux validated with ground surveys. In addition, estimates of emissions from these wetlands (SAF) from rice paddies ranges from 2 to the acquisition of in situ and satellite data Eos, Vol. 90, No. 5, 3 February 2009 of methane fluxes and concentrations plus ment Report of the Intergovernmental Panel on Walter, K. M., et al. (2006), Methane bubbling from ancillary environmental data is needed for , edited by S. Solomon et al., pp. Siberian thaw lakes as a to many areas. 131–234, Cambridge Univ. Press, Cambridge, U.K. climate warming, , 443, 71–75, doi:10.1038/ Frankenberg, C., et al. (2005), Assessing methane nature05040. Acknowledgments emissions from global space-­borne observations, Walter, K. M., et al. (2007), Methane bubbling from Science, 308, 1010–1014. northern lakes: Present and future contributions This article is a contribution of the Meth- Kemenes, A., B. R. Forsberg, and J. M. Melack to the global methane budget, Philos. Trans. R. ane Working Group of the National Center (2007), Methane release below a hydroelec- Soc., 365(1856), 1657–1676. for Ecological Analysis and Synthesis, Uni- tric dam, Geophys. Res. Lett., 34, L12809, Xiong, X., et al. (2008), Methane plume over South versity of California, Santa Barbara. The doi:10.1029/2007GL029479. Asia during the monsoon season: Satellite obser- research is also supported by U.S. National Khalil, M. A. K., and C. L. Butenhoff (2008), vation and model simulation, Atmos. Chem. Phys. Science Foundation grants ARC-­0554811 and Spatial variability of methane emissions from Discuss., 8, 13,453–13,478. rice fields and implications for experimen- EAR-­0630319 to Q. Zhuang and by NASA’s Zhuang, Q., and W. S. Reeburgh (2008), Introduc- tal design, J. Geophys. Res., 113, G00A09, tion to special section on Synthesis of Recent Large-­Scale Biosphere-­Atmosphere Experi- doi:10.1029/2007JG000517. Terrestrial Methane Emission Studies, J. Geophys. ment in Amazonia program. Khalil, M. A. K., and R. A. Rasumussen (1993), Res., 113, G00A02, doi:10.1029/2008JG000749. Decreasing trend of methane: Unpredictability of Zimov, S. A., E. A. Schuur, and F. S. Chapin (2006), References future concentrations, Chemosphere, 26, 803–814. Climate change: Permafrost and the global car- Melack, J. M., and L. L. Hess (2009), Remote sens- bon budget, Science, 312, 1612–1613, doi:10.1126/ Belger, L. (2007), Fatores que influem na emissão ing of the distribution and extent of wetlands science.1128908. in the Amazon basin, in Amazonian Floodplain de CO2 e CH4 em áreas alagáveis interfluviais do médio Rio Negro, Ph.D. thesis, Univ. Fed. da Ama- Forests: Ecophysiology, Ecology, Biodiversity and Author Information zonas and Inst. Nac. de Pesquisas da Amazônia, Sustainable Management, Ecol. Stud. Ser., edited São Paulo, Brazil. by W. J. Junk and M. Piedade, Springer, New Qianlai Zhuang, Departments of Earth and Atmo- Dlugokencky, E. J., et al. (2003), Atmospheric York, in press. spheric Sciences and Agronomy, Purdue University, methane levels off: Temporary pause or a new Melack, J. M., et al. (2004), Regionalization of West Lafayette, Indiana; ­E-­mail: ­qzhuang@­purdue​ steady state?, Geophys. Res. Lett., 30(19), 1992, methane emissions in the Amazon Basin with .­edu; John M. Melack, University of California, Santa doi:10.1029/ 2003GL018126.​­ microwave remote sensing, Global Change Biol., Barbara; Sergey Zimov, North-­East Science Station, Forster, P., et al. (2007), Changes in atmospheric 10, 530–544. Cherskii, Republic (Yakutia), ; Katey M. constituents and in radiative forcing, in Climate Smith, L. C., Y. Sheng, G. M. MacDonald, and L. D. Walter, University of Alaska Fairbanks; and Christo- Change 2007: The Physical Science Basis—Con- Hinzman (2005), Disappearing Arctic lakes, Sci- pher L. Butenhoff and M. Aslam K. Khalil, Portland tribution of Working Group I to the Fourth Assess- ence, 308, 1429, doi:10.1126/science.1108142. State University, Portland, Oreg.

Monitoring Algal Blooms products were obtained for each field collec- tion date from the Land Processes Distrib- in a Southwestern U.S. Reservoir System uted Active Archive Center (LP DAAC; http://​ ­edcdaac.​ usgs­ .​ gov/­ main​­ .​ as­ p). These surface PAGES 38–39 Current Study reflectance images are produced from two bands (645 and 856 nano­meters), each of In recent years, several studies have The Salt River reservoir system comprises which has 250-­meter spatial resolution. Each explored the potential of higher-­resolution four reservoirs that are managed for a com- file is georectified (i.e., pixels are assigned sensor data for monitoring phytoplankton pri- bination of water provision and hydroelec- geographic coordinates) and adjusted for mary production in coastal areas and lakes. tric power generation, as well as for recre- aerosols, atmospheric gases, and high, thin Landsat data have been used to monitor algal ational activities. This study focused on the clouds. blooms [Chang et al., 2004; Vincent et al., 2004], two end-­member reservoirs, Roosevelt Lake The regression analyses of the in situ data and Moderate Resolution Imaging Spectro­ and Saguaro Lake, on the basis that they are and the equivalent satellite reflectance val- radiometer (­MODIS) 250-­meter and Medium the inflow and outflow ecological commu- ues produced different results for each lake. Resolution Imaging Spectrometer (­MERIS) full- nities within the system and may be consid- In both Roosevelt Lake and Saguaro Lake, a ­resolution (300-­meter) bands have been uti- ered as representative of the plankton ecol- fourth-­order polynomial provided the best 2 2 lized to detect cyanobacterial blooms [Reinart ogy of the reservoir system overall. Also, fit (­r Roosevelt = 0.871 and r Saguaro = 0.694) to and Kutser, 2006] as well as to monitor water both reservoirs are large enough to produce the data. The observed differences in perfor- quality [Koponen et al., 2004]. a reasonable sample size of uncontaminated mance between these regression equations Field sampling efforts and MODIS 250-­meter 250-­meter pixels. Roosevelt Lake (111.1°W, were large enough to warrant the use of data are now being combined to develop a 33.7°N) is the top reservoir, with a surface lake-­specific algorithms for the subsequent cost-­effective method for monitoring water area of 9698 hectares. When full, the res- time series analysis. quality in a south­western U.S. reservoir sys- ervoir is 36 kilo­meters long and 76 meters tem. In the Phoenix, Ariz., metropolitan area, deep. In contrast, Saguaro Lake (111.5°W, Satellite-Derived Chlorophyll a Time Series the Salt River reservoirs supply more than 33.6°N), the lowest reservoir, at 465-­meter 3.5 million people, a population expected to elevation, is much smaller, with a reservoir The time series of chlorophyll a estimates rise to more than 6 million by 2030. Given that basin of 2.5 kilo­meters × 1.0 kilometer. for 2007 (363 available images) for Roosevelt reservoir capacities have physical limitations, Monthly field sampling began in Febru- Lake produced 221 valid estimates with pre- maintaining water quality will become critical ary 2007 and was conducted on one transect dicted chlorophyll a values ranging from as the population expands. Potentially nox- on each reservoir. At regular points along 0.32 to 7.13 milli­grams per cubic meter and a ious algal blooms that can release toxins and each transect, measurements of tempera- mean of 3.51 milli­grams per cubic meter (Fig- may affect water quality by modifying taste ture and transparency were taken and water ure 1a). The remaining 142 estimates (39% of and odor have become a major concern in samples were collected for chlorophyll a and the 363 available images) were excluded as recent years. While frequent field sampling microscopic analysis. part of a rudimentary quality check because regimes are expensive, satellite imagery can either high cloud cover obscured the area, be applied cost-­effectively to monitor algal Satellite Analysis excessive cloud cover skewed the atmo- biomass trends remotely, and this informa- spheric correction performed during data pro- tion could provide early warning of blooms The daily MODIS Aqua 250-­meter level 2 cessing, or the image data were corrupted in these reservoirs. georectified (L2G) surface reflectance resulting in poor-­quality pixels.