Location: Hydrology and Remote Sensing Laboratory2011 Annual Report
1a. Objectives (from AD-416)
The ability to monitor and predict the ecological dynamics of carbon and nitrogen, agrichemicals, and invasive species over large areas will be critical for maintaining sustainable agricultural systems. Accordingly, this project has four objectives that focus on using remote sensing tools and modeling to scale point observations and field data from the landscape to regional scales. In the first objective, new remote sensing methods will be investigated to quantify soil carbon redistribution, crop residue and carbon dioxide fluxes, and to identify critical process-based variables that will be incorporated into models and decision support systems. Research towards this objective will be focused in three areas: (1) assessing soil organic carbon at field and landscape scales; (2) measurement of crop residues with remote sensing; and (3) estimating regional scale soil carbon sequestration and carbon dioxide fluxes. In the second objective, innovative remote sensing methods will be used to maximize nitrogen use efficiency and crop nitrogen status, which feed into watershed-scale process models used to predict water quality. The specific goals for this objective are: (1) determine remotely sensed leaf chlorophyll content at high spatial resolutions; and (2) determine watershed nutrient budgets and evaluate the use of remote sensing data as inputs into watershed models. The third objective focuses on quantifying field-scale pesticide volatilization and the soil and climatic factors governing those emissions. Recent investigations have shown that pesticide emissions depend largely on weather and soil moisture conditions and this work will also strive to develop a model of pesticide volatilization that reproduces observed emissions and their response to these governing climatic factors. The fourth objective investigates the spread of invasive plant species and uses remote sensing as a cost-effective tool to obtain the distribution of various invasive species. This research will also investigate the utility of image texture analysis techniques for species identification, and use the classified images to test landscape-distribution models.
1b. Approach (from AD-416)
Research planned for this project will be conducted at a range of spatial scales from small fields to regional. Field-scale, intensive studies will be conducted at the Optimizing Production inputs for Economic and Environmental Enhancement (OPE3) site, which has extensive datasets on soils, topography, and surface and subsurface hydrology, and a long-term record of fluxes, weather, and yields. Remote sensing will be used to identify wet and dry areas at the field-scale and will be used to test the effects of soil moisture and temperature on agrichemical behavior using field data and eddy covariance techniques. Additionally, information on carbon and nutrient behavior gleaned from these field-scale studies will be extended to larger scales at two Conservation Effects Assessment Project (CEAP) watersheds: the Choptank River in Maryland and the South Fork of the Iowa River in Central Iowa. While the large watersheds have less data than the intensive site, there are many partners who are providing data to validate models and remote sensing techniques for regional scale applications. Soil organic carbon and crop residues will be measured at laboratory and field scales with high-spectral-resolution sensors. These and other satellite data will be used to test process-based models. Nitrogen status will be assessed for various crops with very-high-resolution imagery (< 1 cm pixel) so that plants and soil can be separated. The maps of nitrogen status and CEAP data will be used to evaluate the Soil and Water Assessment Tool (SWAT) water quality model. Photographs of leaves, plants, and small plots will be used to determine if image texture analysis can be used for invasive species classification. Resulting techniques will be used to plot the distribution of leafy spurge at the landscape scale, which will then be used to test the Weed Invasion Susceptibility Prediction (WISP) model.
3. Progress Report
Partial Least Squared (PLS) regression models were developed for data contained in hyperspectral imagery. These models predicted soil properties including organic carbon and particle size distribution (texture). Aircraft imagery, flux observations and in-situ canopy and soil measurements were collected at the field site in Beltsville, Maryland for evaluating relationships between modeled leaf chlorophyll content and canopy light-use efficiency and used in mapping carbon and water fluxes. Flux observations from ARS collaborators in Bushland, Texas and Ames, Iowa were processed for additional model validation studies. Low-cost true-color and color-infrared digital cameras were compared for different rates of nitrogen fertilization in corn and rye crops in Beltsville, Maryland. Furthermore, nitrogen rate experiments were established at the University of Maryland Eastern Shore. The new remote sensing methods are being compared to currently accepted methods of chlorophyll meter readings, on-the-go sensor NDVI, and canopy reflectance spectra. In collaboration with the U.S. Geological Survey, field data were collected in the Choptank Watershed (Maryland) on the amounts of nitrogen incorporated into winter cover crops and remaining in the soil. In collaboration with the Forest Service information on wetland location and hydro-period was collected in the Tuckahoe sub-basin for use in modeling water quality. Data collection continued for a long-term field-scale experiment in Beltsville, Maryland where atrazine and metolachlor volatilization and surface runoff losses were simultaneously measured. Using an aerial lift, multiple digital camera images, with pixel sizes from 5 cm to 2 m, were collected over the same targets of crops, weeds, a meadow, and illicit drug crops to determine if image texture and pattern recognition software may be used for classification. Reflectance spectra were also acquired for the different targets. Furthermore, collaborations were established with ARS scientists in Cheyenne, Wyoming and Dubois, Idaho to use the camera technologies for determination of range health. The Weed Invasion Susceptibility Prediction (WISP) model was revised using field and remote sensing data obtained during the multidisciplinary program, The Ecological Area-wide Management (TEAM) of Leafy Spurge. Remote sensing and geographic information data were obtained for Theodore Roosevelt National Park, which was another field site for TEAM Leafy Spurge, for model testing.
1. Low-cost color-infrared digital camera for monitoring crops and rangelands. Digital camera technologies are readily available at low cost; however, sensors with near-infrared bands are better for monitoring crops because leaves and canopies are highly reflective at near-infrared wavelengths. Scientists in Beltsville, Maryland developed a method in which the red channel of some digital cameras is modified to respond to near-infrared radiation instead. USDA was awarded a patent for this method in order to help private companies bring low-cost color-infrared sensors to market. The scientists used the new technology to monitor growth of winter cover crops and rangeland health.
2. New hyperspectral remote sensing method for estimating leaf dry matter content. Leaf dry matter content is obscured in remotely sensed data because of liquid water in the leaves absorbs much more of the short-wave infrared radiation emitted by the sun. Using advanced hyperspectral sensors, scientists from George Mason University (Fairfax, Virginia) and ARS scientists in Beltsville, Maryland collaborated on the development of a new index based on absorption spectra of liquid water and dried plant materials. The new index was highly correlated with dry matter content at leaf and canopy scales. Information on leaf dry matter content is required for estimation of carbon sequestration, plant nutrient concentrations, and wildfire fuel moisture content.
3. Herbicide volatilization exceeds herbicide runoff losses. Surface runoff was thought to be the major off-site transport mechanism for herbicide. However, until recently no field investigations monitored both surface runoff and turbulent volatilization fluxes simultaneously. An 8-year, field-scale experiment in Beltsville, Maryland was conducted where herbicide (atrazine and metolachlor) volatilization and surface runoff losses were simultaneously monitored and evaluated. Results demonstrate that regardless of weather conditions, volatilization losses consistently exceeded surface runoff losses. Surprisingly, herbicide volatilization losses were up to 25 times larger than herbicide surface runoff losses. The research will affect USDA and USEPA policy with regard to herbicide behavior and the data will be used to develop or improve pesticide behavior models.
4. Improved data mining approaches for measurement of soil properties by use of infrared spectroscopy. Near-infrared and mid-infrared diffuse reflectance spectroscopy hold great promise for rapid measurement of soil properties such as organic carbon, but ability to extract information from soil spectra is limited by data mining approaches. Mathematical treatment of spectral data is shown to be an important determinant of success. The results indicated that calibration models for soil are quite sensitive to the complexity of the model and the ability of locally weighted regression helped selection of appropriate calibration samples for development of robust calibrations. These findings will improve our ability to develop robust calibrations for soil properties such as organic carbon and should enable better assessment of soil carbon storage in agricultural ecosystems under management to sequester atmospheric carbon in soil.
5. Limited information on spatial distribution of manure applied to agricultural fields can cause significant uncertainty in model prediction of bacteria in runoff. Concerns for microbial safety of surface water facilitate development of predictive models to estimate concentrations of pathogen and indicator organisms from manure-fertilized fields in runoff. Experiments carried out in Beltsville, Maryland indicated that there was high spatial variation of the concentrations of fecal coli-form bacteria in applied manure. Using average bacterial concentrations lead to substantial over-estimates of the amount of bacteria reaching the edge of a field suggesting that inaccurate representation of bacteria concentrations in manure with a small number of samples can substantially distort estimates of bacteria loss from the field in runoff. Therefore better sampling and modeling strategies need to be developed.
6. Low-cost remote sensing for crop nitrogen management using a new chlorophyll index for digital cameras. Leaf chlorophyll concentration is closely related to nitrogen status and the demand for fertilization; however, satellite data are expensive, have coarse spatial resolution, have long turnaround times for information delivery, and are susceptible to clouds. A new index was developed called the Triangular Greenness Index, which has high sensitivity to leaf chlorophyll concentration and is not sensitive to other soil and canopy variables. Digital cameras are readily mounted onto small unmanned aerial vehicles so individual fields can be quickly monitored at high spatial resolution and low cost enabling better nitrogen management decisions to insure crop yields and reducing fertilizer runoff.
7. Improving capacity of developing countries to participate in carbon credit markets. International aid and development funding for agricultural improvement in developing countries may be linked to verifiable credits for carbon sequestration in agricultural soils as influenced by better soil management. Improved ability for within country measurement of soil carbon and modeling the fate of carbon in agricultural production systems can build capacity of developing countries to engage in emerging carbon markets. Evaluation and validation of low technology approaches such as the loss on ignition technique for measurement soil carbon has improved the capabilities of soil analytical laboratories in West Africa to efficiently monitor soil carbon in agricultural systems. Soil carbon modeling efforts has reduced uncertainty concerning sequestration and storage of soil carbon in agricultural production systems in both West Africa and Central Asia. A combined capacity for measurement and modeling soil carbon in agricultural soils will strengthen capacity of developing countries to gain development aid for improved agricultural production and greater food security.
McCarty, G.W., Hively, W.D., Reeves, J.B., Lang, M.W., Lund, E., Weatherbee, O. 2010. Infrared sensors to map soil carbon in agricultural ecosystems. In: Viscarra-Rossel, R., McBratney, A., Minasny, B., editors. Proximal Soil Sensing, Progress in Soil Science Volume 1. New York, NY: Springer Science. 14:165-176.