Progress 08/15/06 to 08/14/07
Outputs
This initiative is a collaboration of the College of Agriculture and Life Sciences and the Cornell Theory Center, a high-performance computing facility, and involves multiple components. High-Resolution Climate Data: High-res temperature and precipitation fields have been developed through the processing of intensive RUC and RADAR data from the NWS. T data are corrected for elevation and P data are based on correcting errors in radar-based estimates of precipitation fields. Spatial interpolations were developed and have been cross validated. High-resolution climate data (4-5 km grids) are uploaded to SQL Server databases on the CTC web server and accessible through a web interface. Precision Nitrogen Management: The PNM model was developed for precise N management under maize based on soil/crop simulations using hi-resolution climate data. The model was tested against data collected from several field studies. PNM model-based sidedress N recommendations were generated
for 15 climate regions in NY and distributed to farmers. Archived climate data were converted to SQL tables for access by the PNM model. 40-yr PNM model simulations were completed to develop a new dynamic N leaching index. A PNM web interface was developed for farm-specific sidedress N recommendations. The model was also calibrated using field data on N2O emissions from a 2004 experiment. Modeling Weed Competitiveness: Field experiments were conducted on the effects of competition on soil water capture of different weed species. Patterns of early root development were monitored for maize and four weeds and comparative growth was measured in response to variable soil N. A dynamic model of crop-weed competition that utilizes the soil nitrogen model, maize growth model and distributed weather data is being developed. Data Mining: Supervised and unsupervised learning was used to find patterns in the NE Climate Center temp data. Data were prepared using Splus and SQL to run the algorithms.
Analysis of additional climate variables were used to search for spatio-temporal climate phenomena using pattern recognition and data mining techniques. VNIR Spectroscopy: Lab protocols have been developed for using the FieldSpec Pro spectroradiometer. VNIR sensing was used in soil-survey field programs. VNIR sensitivity to chemical concentrations was evaluated, and VNIR sensing was done on thousands of samples from New York, Bangla Desh, and Turkey. Data mining methods, including multiscale wavelet decomposition, MARS and PLS were used. Economics and the Control of Stochastic Growth: Trade-offs in maintaining the appropriate inventory of nitrogen for crop growth, while avoiding excessive nitrogen inventories was analyzed using real-option programming. Two models; one an infinite-horizon model, the other a finite-horizon model, were developed to examine the optimal time to adapt to climate change in terms of irrigation investments. There is a single, critical standard deviation rate
for climate variability which triggers investment.
Impacts
New York is the first state to develop N recommendations based on real-time weather data. A new nitrate leaching index has potential for nationwide adoption by NRCS. The N2O studies have provided insights into the overriding importance of agronomic management impacts on greenhouse gas emissions, and will help in setting policy for mitigating agricultural impact on greenhouse gas emissions. PNM model-based sidedress N recommendations were generated in June 2005, and sent to growers. High-resolution climate data are becoming available for the Northeast USA through a Web-based interface. Since climate/weather is the most important factor affecting crop yield and quality, the efficiency of crop inputs, and their environmental impacts, these data will have widespread application for improved management of agricultural inputs. New statistical methodologies to weather event prediction will also allow for more precise management of inputs in agriculture. This includes the
management of nutrients and pesticides for which this project is developing decision support tools. VNIR hyperspectral sensing has proven to be a very effective method for rapid assessment of a multitude of soil properties, and is likely to become widely adopted as part of soil survey, C sequestration assessment, nutrient management, etc.
Publications
- Berger, A.B., McDonald, A.J., Riha, S.J. 2006. Patterns of early root development for maize and four common weeds as influenced by competitive environment. Functional Ecology (in print.
- van Es, H., Gomes, C, Sellmann, M, and van Es, C. 2006. Spatially-Balanced Designs for Experiments on Autocorrelated Fields. Geoderma (in print).
- Tan, I.Y.S., H. M. van Es, et al. 2006. Nitrous Oxide Losses under Maize Production as Affected by Soil Type, Tillage, Rotation, and Fertilization. Soil&Tillage Research (accepted).
- Sogbedji, J.M., H.M. van Es, J.M. Melkonian, and R.R. Schindelbeck. 2006. Evaluation of the PNM model for simulating drain flow nitrate-N concentrations under manure-fertilized maize. Plant and Soil 282: 343-360.
- H.M. van Es, J.M. Sogbedji, and R.R. Schindelbeck. 2006. Nitrate Leaching under Maize and Grass as Affected by Manure Application Timing and Soil Type. J. Environmental Quality 35:670-679.
|