Mitigating risk of weather variability to agriculture through improved field workability predictions

Tomasek, Bradley

Permalink

https://hdl.handle.net/2142/42455 Copy

Description

Title

Mitigating risk of weather variability to agriculture through improved field workability predictions

Author(s)

Tomasek, Bradley

Issue Date

2013-02-03T19:46:12Z

Director of Research (if dissertation) or Advisor (if thesis)

• Davis, Adam S.
• Williams, Martin M.

Department of Study

Crop Sciences

Discipline

Crop Sciences

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

M.S.

Degree Level

Thesis

Keyword(s)

• predicting field workability
• climate change
• agriculture

Abstract

Evolving weather-related risks, driven by climate change, will impose challenges on agricultural production systems at a global scale. However, the spatial heterogeneity of cropping practices, current climate, projected climate changes, and environmental factors (such as soil) will influence these changing risk profiles. These complex interactions mean the magnitude of these changes in risk, or whether they are even positive or negative, is highly variable across space and time. This variability in turn makes issuing broad recommendations for adaptation and mitigation strategies extremely difficult. Changes in management (field operations) timing can be used a tool to manage weather-related risks. At the same time, performing field operations requires the field to be in a workable state. These days for which the field is workable are called field working days (FWDs), and are driven by soil moisture and temperature. Over a certain threshold of soil moisture or below a certain temperature threshold, a field is considered unworkable. Working a field that is too wet may lead to soil compaction, and below a certain temperature the soil may be frozen or be too cold for seed germination. By standardizing volumetric soil moisture by the plastic limit (PL) or field capacity (FC) different soils are taken to have the same thresholds for workability. I compared different models for determining the thresholds for field workability. As a validation dataset, 97 estimated soil moisture time series using 50 weather stations across Illinois were part of a simplified reconstruction of soil moisture conditions across the state from 1959-2010. Three of models from the previous literature were validated using the reconstruction. These models were proposed by theoretical considerations and originally validated using single-site records. A fourth model was developed using maximum-likelihood based logistic regression approach on the field records from three University of Illinois research extension farms. The fifth and final model chose a soil moisture and temperature threshold which minimized the prediction error of reported weekly FWDs from 1959-2010 when applied to the reconstructed soil moisture dataset. All of the models from previous literature, as well as the logistic regression model, systematically over predicted the number of FWDs week-by-week. On the hand, the optimized model had the highest predictive performance in terms of root mean squared error, and eliminated biases. These results indicate that preset and unchangeable theoretical thresholds may introduce considerable prediction bias into FWD predictions, and the spatial scale of the data can greatly influence threshold identification. Given the method for finding field workability thresholds optimized for prediction, the driving question of the remaining research was: how will weather-related risks and field workability in Illinois change as a result of projected climate change? To accomplish this, nine sites (one representing each crop district in Illinois) used in the soil moisture reconstruction were selected. The weather time series from these sites were analyzed to train a stochastic weather generator. Statistical downscaling of a global climate model was performed for each of these nine sites and this downscaling allowed for the weather generator to simulate weather under three different climate change scenarios (B1, A1B, A2) at three time points (training period, mid-century, and end of century). 1,000 years of weather for each station, time period, and scenario were simulated. These simulated series were analyzed for monthly cumulative drought risk, 90th percentile frost dates, and growing degree day (GDD) accumulation (Celsius). 100 years for each site, scenario, and time period were randomly selected and run through an optimized FWD model. Rising temperatures, driven by greenhouse gas emissions, substantially increased the length of the growing season as determined by the mean number of growing degree days as well as the frost dates. By the end of the century, projections indicated a state average season length 3-8 weeks longer than the baseline period with 500-900 additional GDDs. The same rising temperatures associated with the increased season length also drove a substantial increase in estimated potential evapotranspiration. Along with summer precipitation patterns which had little or no change from the baseline period, this increased evapotranspiration lead to more frequent and higher severity drought risks at the state scale. Changes in drought severity and frequency generally rose in parallel with the projected temperature changes across time and scenarios. There were also noticeable differences in risk changes at the crop district scale between climate scenarios and time horizons. Finally, a consistent increase in field workability was projected for late March and early April across districts. However, overall April through May FWDs were often predicted to decrease or remain similar to historic levels. Only one scenario (A1B) showed an increase in overall April through May FWDs, and only for mid-century. This study indicated that projected climate change may have diverse implications within the state of Illinois in terms of weather-related risks. Specifically, projected summer drought risks and risks arising from changing field workability profiles show considerable interactions with both the climate scenario and crop district. This spatial and temporal variability associated with climate change will likely complicate agricultural mitigation and adaptation efforts, even at the state-level.

Graduation Semester

2012-12

Permalink

http://hdl.handle.net/2142/42455

Copyright and License Information

Copyright 2012 Bradley Tomasek

Owning Collections