Climate-Based Information for USAID Missions: Future Projections vs. Historical Data

By Timothy S. Thomas

 

Most scientists studying the impact of climate change on agriculture use climate models that project out to 2050 or beyond – some even going to 2100. Even those focusing more short-term rarely study anything earlier than 2030 – the models just have too little change in that time period for them to produce anything of interest. These climate studies can be of significant help to USAID missions when working with host governments in developing longer-range investment plans in the agricultural and environmental sectors, and can also be of help in assessing climate risk in activities that are meant to have impact for multiple decades.

Yet many climate risk assessments for USAID activities need to assess climate impacts for just a few years into the future, for example, just until the early 2020s. For those assessments, typical climate models and studies are not helpful. In such cases, missions would be better served by looking at climate trends from gridded weather data available from a number of sources.

 

Figure 1. Areas in Nigeria with Statistically Significant Trends in Temperature, 1980-2010, in 0C

 

Source: Author, using daily AgMERRA data.
Notes: Based on mean daily maximum temperature for the warmest month of the year, a proxy for heat stress for crops. For Nigeria, it does not perfectly coincide with the cropping period. However, it is highly correlated with other indicators of temperature change for Nigeria. Results are based on regressions on a time trend at each pixel using 31 years of data. Areas in white did not have a statistically significant time trend at the 10% level using a z-test on the parameter estimate. . Areas outlined in red represent Feed the Future’s Zones of Influence.

 

Analysis of daily or monthly gridded data allows researchers to detect statistically significant climate trends that could be reasonably foreseen to continue at least into the near-term. In a recent visit to USAID in Abuja at the end of May 2017, IFPRI researchers therefore developed maps showing important temperature and precipitation trends between 1980 and 2010.

Figure 1 shows areas that had statistically significant warming over that period based on regression analysis at the pixel level. A few areas exceed 20C in that time, which is a large enough change to cause important yield changes in crops grown. The highest warming is noted in the eastern portion of the country along the border with Cameroon, in the North East zone. But there is also some statistically significant warming in Sokoto state in the northwestern corner of Nigeria.

Similarly, Figure 2 shows areas that had statistically significant changes in annual rainfall between 1980 and 2010. Rainfall has noticeably risen in almost all of the North East and North West zones, with the changes extended down into North Central zone and a small patch in South West zone. The highest rainfall increases – at around 300 millimeters per year – are noted in Borno state. The northern half of Nigeria is much drier than the southern half, so the relative precipitation increases are very large (Figure 3). Parts of Borno state actually saw precipitation increase by 75 percent over the thirty-year period. Here we do not mean that there was a single dry year near 1980 and a single wet year around 2010, but rather the trend, taking into account all years actually indicated a 75-percent-increase of rainfall in that area.

 

Figure 2. Areas in Nigeria with Statistically Significant Trends in Annual Rainfall, 1980-2010, in millimeters

Source: Author, using daily AgMERRA data.
Notes: Results are based on regressions on a time trend at each pixel using 31 years of data. Areas in white did not have a statistically significant time trend at the 10% level using a z-test on the parameter estimate. . Areas outlined in red represent Feed the Future’s Zones of Influence.

 

Normally we think of rainfall increases in dry areas to be a good thing for agriculture – and it may well be in many places. But large increases in rainfall can also be harmful to agriculture, because it can cause floods, make cultivation more challenging, and if accompanied with intensification of rainfall events, can destroy otherwise healthy crops, erode soils and can lead to long-term declines in soil fertility levels.

It is noteworthy that the intersection of the areas of statistically significant climate change for rainfall found in Figure 2 and the areas of statistically significant climate change for temperature found in Figure 1 is the area in the country most affected by civil conflict. While climate is not the main cause of the conflict, it is indeed worth considering whether climate change had a role in creating conditions that were ripe for initiating conflict or exacerbating the conflict.

 

Figure 3. Relative change in normal annual rainfall, 1980-2010, percent

Source: Author, using daily AgMERRA data.
Notes: Results are based on regressions on a time trend at each pixel using 31 years of data. . Areas outlined in red represent Feed the Future’s Zones of Influence.

 

This note pointed to a few climate trends that can be found in readily available data. Much more could be done with daily gridded historical climate data to discover important climate trends that impact agriculture and people. These might include looking at shocks such as changes in onset of rains; significant pauses in rains; heat waves; changes in night-time temperature during the growing season; and changes in rainfall intensity. With such data, it would be possible to better understand which shocks are important for that particular country, and would allow for planning for how to help projects and the country deal with those shocks.