The New CHIRTS Daily Temperature Product Helps Track a Key Threat to Crop — and Human — Health: Extreme Temperatures
This post is written by Andrew Verdin (Minnesota Population Center, University of Minnesota), Chris Funk (United States Geological Survey & University of California, Santa Barbara Climate Hazards Center), and Juliet Way-Henthorne (University of California, Santa Barbara Climate Hazards Center).
Many crops are sensitive to very hot days, especially during the germination and flowering stages, yet many parts of the world lack dense networks of air temperature observations. For example, the whole continent of Africa might have a couple hundred observations — far too few to see or measure extreme temperatures.
The CHIRTSmax monthly product provides a unique opportunity to translate thermal infrared temperature observations into monthly estimates of maximum temperatures, even in places that don't have nearby weather stations. The new CHIRTS-daily product then provides sub-monthly time series that can be used to identify agricultural impacts.
This blog describes the new CHIRTS-daily product, which will help us track extreme temperatures and the risks these extremes pose for agriculture and health. This new product, described in a forthcoming Scientific Data paper, but available now (here) in an archive that extends from 1983 to 2016, is called the Climate Hazards Center Infrared Temperatures Daily archive (CHIRTS-daily).
This 60°S-70°N high-resolution (0.05°) data product builds on the monthly Climate Hazards Center Infrared Temperatures with Stations Tmax product (CHIRTSmax). The CHIRTSmax (which is discussed here) is a unique new climate record that uses the actual thermal infrared radiation-based surface emissions temperatures, as measured by a quilt of geostationary satellites, to guide monthly Tmax temperature estimates. Enhanced by a high-resolution climatology and a reasonably dense set of Berkeley Earth weather station observations, the CHIRTSmax provides a valuable complement to the CHIRPS rainfall product. The CHIRPS translates the cold cloud component of the geostationary satellite observations into precipitation estimates. The CHIRTSmax translates the warm surface emissions of these observations into Tmax estimates.
Though the archive presently extends only through 2016, the Climate Hazards Center will be developing an operational update system later this year, in collaboration with Ken Knapp at NOAA’s National Centers for Environmental Information.
The potential for cloud contamination makes it difficult to produce a CHIRTSmax daily product based on thermal infrared satellite information, so a daily product has been produced using ERA5 reanalysis fields. The previously mentioned upcoming paper will describe the method and present very promising validation results.
Here, we present two example case studies. These case studies represent typical applications for the CHIRTS-daily maximum temperatures.
The first application uses a 40.6°C temperature threshold. This threshold, often used in human health studies, represents a temperature level at which the human body often has difficulty maintaining adequate cooling of internal organs. The second application uses a threshold of 30°C, a common threshold used to identify agricultural heat stress. The first case study is a global analysis. The second case study focuses on Ethiopia, expanding on supplemental material provided in Funk et al. 2019.
In the first case study, we contrast results from a large set of global daily station data, daily CHIRTSmax estimates, the ERA5 reanalysis, and the Princeton Global Forcing (PGF) archive. The ERA5 and PGF are widely used data sets. The analysis focuses on the climatologically warmest three months for each station. The CHIRTSmax, ERA5, and PGF values have been extracted at the station locations, supporting a one-to-one comparison.
Figure 1 presents a time series plot displaying the average number of days over 40.6°C for the globe and individual regions. What is quite striking is that the ERA5 dramatically underestimates the number of hot days in all regions. The PGF archive, on the other hand, substantially overestimates in Australia and South America. The CHIRTSmax, in all cases, tracks the station-based estimates very closely. Note that the daily GHCN and GSOD data likely contributed to the monthly Berkeley Earth data used in the monthly CHIRTSmax, so Figure 1 should not be interpreted as an independent validation study. Nevertheless, the CHIRTSmax does appear to be well-suited for analyzing trends in temperature extremes, especially in data-sparse regions. It should be noted that Africa stands out as a region with very large increases in the number of very hot days. While all the gridded data sets underestimate the station data change estimate (+5.7 days), the CHIRTS-daily estimate was the closest (+4.7 days). Further analysis of the spatial pattern and health hazards associated with these large increases appears warranted.
We next turn to a typical agro-climatic risk assessment. Building on results presented in the supplemental material of Funk et al. (2019), this case study focuses on July Tmax temperatures in the Amhara province of northern Ethiopia. El Niño-related rainfall deficits in this region contributed to the “the worst drought in 50 years,” and led to widespread crop failures that helped push approximately 11 million people into crisis levels of food insecurity.
These low rainfall levels were also accompanied by exceptionally warm July air temperatures (Fig. 2). The unique nature of the monthly CHIRTSmax archive provides independent satellite-only temperature estimates (CHIRTmax) and station-only temperature estimates (CHTSmax), as well as the blended “best estimate” CHIRTSmax product. Humanitarian relief agencies often use a convergence-of-evidence approach to guide drought assessments. The fact that the independent CHIRTmax and CHTSmax archives both indicated historically extreme air temperatures provide convergent evidence of severe potential crop stress.
Climate hazards typically involve this type of climate shock with underlying vulnerability and exposure. Figure 2 represents this schematically. Vulnerability is represented by the gap between median and 20th percentile World Bank per capita incomes. Despite increases in agricultural productivity, the number of extremely food insecure Ethiopians has increased. This increase may be due to a series of recent climate shocks, combined with an increased price-spike vulnerability, for poorer Ethiopians. Figure 2 also represents the third dimension of climate hazards — exposure — using United Nation estimates of Ethiopian population. Between 1993 and 2070, Ethiopia may experience a five-fold increase in population, with the number of Ethiopians increasing from about 50 to 250 million people.
While the ~+2.5°C temperature anomaly shown in Fig. 2 appears concerning, it is difficult to interpret temperature anomalies from an agronomic perspective. Plants typically respond to the actual temperature values. In colder areas, crops may actually benefit from warmer temperatures. In hot areas, temperature increases can result in increased wilting and moisture loss. This dependence means that an accurate background climatology can improve the utility of climate hazard assessments. Figure 3 shows long-term average Tmax values for the CHIRTS-daily, the University of East Anglia’s Climatic Research Unit product and the ERA5 data set. The 0.05° CHIRTS product clearly represents Ethiopia’s complex orographic influences. We find some of the steepest temperature gradients on Earth, with highland areas having mean maximum temperatures ranging from 15 to 20°C, while nearby lowland areas may have mean values of greater than 37°C. The Moving Window Regression modeling process used to construct the CHIRTS background climatology uses satellite thermal infrared mean fields as a predictor, and these fields help capture these complex gradients. These patterns are not captured well by the coarse CRU product. The physically based ERA5 reanalysis captures the overall pattern with reasonable fidelity, but many fine details are missed. Furthermore, as noted above in the global case study, there appears to be a consistent tendency of the ERA5 to underestimate the magnitude of the mean maximum temperatures, making it an inappropriate product for estimating the hazards associated with temperature extremes.
For agricultural impact assessments, days over a 30°C threshold is a common metric of heat stress. The bar plot shown in Figure 4 shows a time series of this metric for consecutive Julys in Amhara province — a critical crop growing region strongly impacted in 2015. This time series gives us a meaningful basis for assessing agricultural temperature-related impacts. First, note that this time series is quite variable. In some years, only 10 percent of the pixel-days exceeded 30°C. In 2015, on the other hand, almost 30 percent of the pixel-years exceeded 30°C. This suggests widespread temperature impacts that negatively impacted crop production. Capturing such impacts from space may be very valuable.
We conclude with a brief 2070 climate change “projection” for Amhara, based on a perturbed version of the observed CHIRTS-daily data set. Our goal here is to simply provide a representative usage case, not to perform a detailed climate change assessment.
In general, it is not unreasonable to assume that future air temperatures may be represented by current air temperatures (CUR) perturbed by a smoothly varying set of increases (DELTA). Furthermore, let CUR be a data set with high spatial and temporal resolution, like CHIRTS-daily. DELTA may be derived from climate models such as those represented in the Phase 5 or 6 Coupled Model Intercomparison Projects (CMIP5 or CMIP6). Here, we use a single DELTA value of +1.3°C, which is the change in the Phase 5 multi-model ensemble Tmax average, for Amhara in July, between 2070 and 2020, based on the 6 Wm-2 Representative Concentration Pathway. Perturbing our data set with this value and recalculating the fraction of pixel-days warmer than 30°C results in the black line and filled circles in Fig. 7. By 2070, Ethiopia is likely to hold ~250 million people (Fig. 2), and the average heat stress fraction may be around 23%, a level exceeded in only the most extreme months in the observational record.
We know that climate change is creating more extreme temperatures. Now, satellites are helping us see these extremes even in the most remote places on the planet. Without these systems and products like CHIRTS-daily, we're blind.
Funk, C., Peterson, P., Landsfeld, M., Pedreros, D., Verdin, J., Shukla, S., ... & Michaelsen, J. (2015). The climate hazards infrared precipitation with stations—a new environmental record for monitoring extremes. Scientific data, 2(1), 1-21.
Funk, C., Harrison L., Shukla S., Hoell A., Korecha D., Magadzire T., Husak G., and Galu G., 2016, Assessing the contributions of local and east Pacific warming to the 2015 droughts in Ethiopia and Southern Africa, Bulletin of the American Meteorological Society, December 2016, S75-S77, doi:10.1175/BAMS-16-0167.1.
Funk, C., Peterson, P., Peterson, S., Shukla, S., Davenport, F., Michaelsen, J., …, & Rowland, J. (2019). A High-Resolution 1983-2016 T max Climate Data Record Based on Infrared Temperatures and Stations by the Climate Hazard Center. Journal of Climate, 32(17), 5639-5658.
Verdin A, Funk C, Peterson P, Landsfeld M, Tuholske C, Grace K (2020) Development and validation of the CHIRTS-daily quasi-global high-resolution daily temperature data set, Scientific Data, In Revision.