We have moved!

The bigger, better, brand new DAPA blog is here (link)


Please note this Blog is not updated anymore.

We have moved! -- CLICK HEREe --
Decision and Policy Analysis Research Area – DAPA

DAPA Workshops – Sharing research experience across projects, institutions and themes “Climatic data sources and interpolation” and “Statistical models for crop cultivation”

fotoII.png

Figure 1. Workshop participants

A new series of workshops in the Decision and Policy Analysis Program (DAPA) at CIAT have the aim to share knowledge and experiences among DAPA employees and visiting scientists, help facilitate connections between different research programs and areas at CIAT and with other external institutions, and build capacity for DAPA personnel in the use of relevant analytical tools.

These workshops allow participants to:

  • Communicate between different research groups.
  • Discuss and exchange information about different research projects.
  • Learn new analysis techniques.
  • Propose future research direction.
  • Disseminate information and knowledge about different research lines.

Visiting researcher – Sharon Gourdji

The first set of DAPA workshops, which took place in April 2013, were led by Sharon Gourdji, who visited CIAT for a period of three months thanks to the exchange program Fulbright NEXUS (http://www.state.gov/r/pa/prs/ps/2012/10/199066.htm). Sharon is currently a post-doctoral fellow at Stanford University in the Environmental Earth System Science program and the Center on Food Security & the Environment, and her research focuses on using statistical models to estimate climate change impacts on agriculture, and evaluate associated adaptation options. While working in the DAPA division of CIAT during her visit, she worked on the project “Tortillas on the Roaster”, extending the existing analysis of climate change impacts on bean production in Central America to focus on the adaptation potential for irrigation and new drought-tolerant bean varieties.

The two workshops covered the following themes:

Section one: Climatic data sources and interpolation

  • Sources of climatic data (Global Historical Climatology Network (GHCN) and the Global Surface Summary of the Day (GSOD)): How to identify stations and get the data you need?
  • How to “clean” and merge data from different sources.
  • Interpolation of climate data: temperature vs. precipitation.
  • Interpolation methods (e.g. angular distance weighting, splines).
  • Using WorldClim averages as a reference for spatial variability in climate.
  • Validation of interpolation.

Section two: Crop statistical models

  • Statistical models relating crop yields to climatic variables.
  • Example study with international wheat trial database.
  • Importance of visualizing data.
  • Selecting variables for the model that optimally explain variability in yield.
  • Fixed Effects.
  • Interactions between variables and nonlinear relationships.
  • Interpreting results (e.g. how to derive temperature-yield curves across multiple terms in the regression).

Each workshop included practical exercises corresponding to the themes above. Sample scripts in R were provided by Sharon, which can also be used as a reference for other analyses by the workshop participants.

In the first workshop, participants modified R scripts to download weather data from the GSSD, and then test climatic data interpolation in Manizales (Colombia) using data records from nearby weather stations. Figure 2 shows the input stations and the results of the interpolation for daily maximum temperatures in Manizales in 2011.

Figure2.1. Interpolation of daily maximum temperature data for the weather station of Manizales (Colombia) using climate information from the 10 nearest stations. Figure2.2. Validation of interpolated maximum temperature for the station Manizales (Black points are interpolated values, blue points are actual climate values measured at this site, and the green line is the multi-annual average (1950-2000) from WorldClim).

Figure2.1. Interpolation of daily maximum temperature data for the weather station of Manizales (Colombia) using climate information from the 10 nearest stations. Figure2.2. Validation of interpolated maximum temperature for the station Manizales (Black points are interpolated values, blue points are actual climate values measured at this site, and the green line is the multi-annual average (1950-2000) from WorldClim).

 

In the second workshop, participants experimented with a wheat trial database from the International Center for the Improvement of Maize & Wheat (CIMMYT), along with interpolated weather data, that were analyzed extensively in the publication: Gourdji, S.M., K.L. Mathews, M. Reynolds, J. Crossa & D.B. Lobell, 2013. An assessment of wheat yield sensitivity and breeding gains in hot environments, Proc. R. Soc. B, 280, doi: 10.1098/rspb.2012.2190.

Figure 3. Inferred yield response to temperature from regression model for three growth stages (veg = vegetative; rep = reproductive; GF = grain filling), with the response curves fitted separately for high and low vapor pressure deficit (VPD) trials. The curves have been normalized to equal 0 at 12®C. The line thickness corresponds to the significance of the slope (i.e. thin: NS, medium: p≤ 0.1, thick: p≤ 0.05, where the p-values are from a two-sided t-test).

Figure 3. Inferred yield response to temperature from regression model for three growth stages (veg = vegetative; rep = reproductive; GF = grain filling), with the response curves fitted separately for high and low vapor pressure deficit (VPD) trials. The curves have been normalized to equal 0 at 12®C. The line thickness corresponds to the significance of the slope (i.e. thin: NS, medium: p≤ 0.1, thick: p≤ 0.05, where the p-values are from a two-sided t-test).

Figure 3 shows the results of the wheat yield statistical model, with inferred yield response to temperature for three growth stages under both dry (high vapor pressure deficit, or VPD) and humid (low VPD) conditions.

See below for links to the presentations, R scripts and datasets used by Sharon for each workshop.

Finally, the workshops concluded with the presentations of several young researchers from DAPA about their ongoing research projects related to the workshop themes, along with discussion and interaction regarding these projects. Below are the names of the speakers and titles of the presentations.

Carlos Navarro: Improved WorldClim baselines using data from weather stations in Colombia.

Camilo Barrios: Estimation of missing climatic data using Artificial Neural Networks.

David Arango: Functional Geostatistics for historical climate data.

Jagath S Kularatne: Latent variables to predict crop adaptation under climate change and develop a test based on latent variables: A case study on common bean.

Christian Bunn: The utility of an agro-ecological niche model for developing future scenarios of coffee production under future climate change.

In the following link you can see the presentations, R scripts and datasets used by Sharon for each workshop:

http://www.dropbox.com/sh/ngx8cotqcoffgfq/Trv14ybkrz

 


 

 


 

 


 

Related Posts Plugin for WordPress, Blogger...
Share this:
about CIAT

If you could answer these three short questions, that would be really appreciated http://dapa.ciat.cgiar.org/we-want-to-know-our-readers/

Our Latest Presentations