UBC Faculty Research and Publications

Influence of extreme weather disasters on global crop production Lesk, Corey; Rowhani, Pedram; Ramankutty, Navin 2016

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
52383-Lesk_C_et_al_Influence_extreme_weather.pdf [ 521.67kB ]
Metadata
JSON: 52383-1.0376077.json
JSON-LD: 52383-1.0376077-ld.json
RDF/XML (Pretty): 52383-1.0376077-rdf.xml
RDF/JSON: 52383-1.0376077-rdf.json
Turtle: 52383-1.0376077-turtle.txt
N-Triples: 52383-1.0376077-rdf-ntriples.txt
Original Record: 52383-1.0376077-source.json
Full Text
52383-1.0376077-fulltext.txt
Citation
52383-1.0376077.ris

Full Text

 1Influence of extreme weather disasters on global crop production 1 Corey Lesk1, Pedram Rowhani2, and Navin Ramankutty1,3 2 1 Department of Geography, McGill University, Montreal, Canada 3 2 Department of Geography, University of Sussex, Brighton, UK 4 3 Liu Institute for Global Issues and Institute for Resources, Environment and 5 Sustainability, University of British Columbia, Vancouver, Canada   6  7 In recent years, a number of extreme weather disasters (EWDs) have 8 partially or completely damaged regional crop production1–5. While 9 detailed regional accounts of the impacts of EWDs exist, the global scale 10 impacts of droughts, floods, and extreme temperature events on crop 11 production are yet to be quantified. Here we estimate for the first time 12 national cereal production losses across the globe resulting from reported 13 extreme weather events over 1964-2007. We find that droughts and 14 extreme heat events significantly reduced national cereal production by 9-15 10%, while our analysis could not identify a global impact from floods and 16 extreme cold events. Analyzing the underlying processes, we find that 17 production losses due to droughts were associated with a reduction in both 18 harvested area and yields whereas extreme heat mainly decreased cereal 19 yields. Additionally, the results highlight ~7% greater production impacts 20 from more recent droughts and 8-11% more damage in developed 21 countries compared to developing ones. Our findings may help guide 22 agricultural priorities in international disaster risk reduction and 23 adaptation efforts.   24  2In many regions of the world, there have been significant changes in the nature 25 of droughts, floods, and extreme temperature events since the middle of the 20th 26 century6–8. Over agricultural areas, disasters arising from extreme weather can 27 cause significant damage to crops and food system infrastructure, with the 28 potential to destabilize food systems and threaten local to global food security. In 29 recent years, nearly a quarter of all damage and losses from climate-related 30 disasters is on the agricultural sector in developing countries9. With such 31 disasters expected to become more common in the future1,6,7, policy makers 32 need robust scientific information in order to develop effective disaster risk 33 management and adaptation interventions (e.g., infrastructure, technology, 34 management, and insurance) to protect the most vulnerable populations and to 35 ensure global food security.  36  37 Whether an extreme weather event results in a disaster depends not only on the 38 severity of the event itself, but also on the vulnerability and exposure of the 39 human and natural systems that experience it6. Past research has addressed 40 agricultural impacts of specific weather extremes with fixed definitions, such as 41 degree days above some threshold10-15. This approach likely underestimates the 42 crop impacts of EWDs because similar extreme weather events may have 43 differing impacts depending on the vulnerability of the exposed system. 44  45 In this study, we address this bias by using a disaster dataset compiled based on 46 human impact. In addition, we attend to two further limitations of previous work 47 on extreme weather and agriculture. Firstly, several regional empirical studies 48 have highlighted the adverse impacts of extreme heat events on crop yields10–13, 49  3and global modeling efforts have estimated future crop yield declines due to 50 increasing extreme heat stress14,15. But this emphasis on crop yields offers an 51 incomplete picture of agricultural performance and food security because of the 52 potential for compensation or compounding of yield impacts by changes in 53 harvested area16; and because crop production (and not yields) – together with 54 access and utilization – determines food security2,4,7,17,18. Secondly, we seek to 55 investigate the agricultural impacts of often-overlooked extreme weather events, 56 namely floods and extreme cold disasters2,3. Thus, our study is the first, to our 57 knowledge, that takes an empirical approach to estimating the influence of 58 extreme weather disasters on crop area, yields, and production at the global 59 scale. 60  61 We use a statistical method, Superposed Epoch Analysis (also known as 62 compositing, see Methods), to estimate average national per-disaster cereal 63 production losses across the globe due to reported droughts, floods, and 64 temperature extremes from 1964-2007. Additionally we estimate the impacts on 65 cereal yield and harvested area separately to identify processes leading to 66 production losses. Based on ~2800 reported extreme hydro-meteorological 67 disasters collated by the Emergency Events Database EM-DAT19, we find that 68 national cereal production during a drought was significantly reduced by 10.1% 69 on average (95% confidence interval 9.9-10.2%) while years with extreme heat 70 led to national production deficits of 9.1% (8.4-9.5%, Fig. 1a-b). These 71 production deficits were equivalent to roughly six years of production growth, 72 however no significant lasting impact was noted in the years following the 73 disasters. Estimated mean production losses were driven mainly by a 74  4preponderance of disasters with moderate impacts on crops, as opposed to a few 75 extreme cases (Extended Data Fig.1). 76  77 Over 1964-2007, these estimated EWD impacts represent a loss of 1820 million 78 MT due to droughts (approximately equal to the global maize and wheat 79 production in 2013) and 1190 million MT due to extreme heat disasters (more 80 than the global 2013 maize harvest). Over 2000-2007 (the period with the most 81 complete disaster reporting compared to earlier decades), 6.2% of total global 82 cereal production was lost due to EWDs relative to an estimated counterfactual 83 global production without EWD impacts (3.0% to extreme heat and 3.2% to 84 drought).  85  86 Cereal yield declines during EWDs were 5.1% (4.9-5.2%) and 7.6% (7.0-8.1%) 87 for drought and extreme heat, respectively (Fig. 2a). Harvested area dropped 88 4.1% (4.0-4.3%) during droughts but was not significantly affected by extreme 89 heat (Fig. 2b). This may be due to the shorter duration of extreme heat events 90 relative to droughts – while approximately one third of droughts in this study 91 spanned multiple years, all extreme heat events took place within a single year. 92 Droughts may thus be more likely to last long enough to cause complete crop 93 failure and discourage planting while extreme heat disasters, especially outside 94 key crop developmental stages, may impact crop growth and reduce yields 95 without critically damaging harvests.  96  97 Our estimated yield deficits from extreme weather events cannot be directly 98 compared to previous studies of the impact of seasonal mean climate trends over 99  5the same period20 (see Supplementary Discussion). However, we derived a 100 comparable measure to that in Lobell and Field (2007)21, and estimated a yield 101 sensitivity of 6-7% per 1ºC increase in seasonal mean weather associated with 102 extreme heat disasters, which suggests that our observed extreme heat impacts 103 are not necessarily independent from those detected in studies examining 104 changes in seasonal temperatures (Extended Data Figure 4). Methodological 105 differences and uncertainties prevent us from drawing strong conclusions based 106 on this comparison. Our drought impacts, however, seem to be independent of 107 previous estimates that used seasonal weather anomalies (see Supplementary 108 Discussion). 109  110 Our results do not show significant production impacts from extreme cold events 111 and floods (Fig. 1c-d). One potential explanation is that floods tend to occur in 112 the spring in temperate regions as a result of snowmelt and cold weather 113 susceptibility in most agricultural regions is highest outside the growing season, 114 which may render a sizeable portion of the flood and extreme cold disasters 115 analyzed in this study agriculturally irrelevant. The estimated lack of response 116 may also be an artifact of the spatial dimension of these disasters. While drought 117 and extreme temperature affect broad regions, floods are a function of both 118 weather and topography and can be highly localized within a country22. Since 119 this study uses country-level agricultural statistics, one may speculate that a 120 more noticeable flood impact on sub-national production is masked at the 121 national scale. 122  123  6Several additional analyses offer more detailed insights into the impacts of these 124 EWDs on cereal production. Cereals in the more technically developed 125 agricultural systems of North America, Europe and Australasia suffered most 126 from droughts, facing on average a 19.9% production deficit compared to 12.1% 127 in Asia, 9.2% in Africa, and no significant impact in Latin America and the 128 Caribbean (overall difference in means p = 0.02, Fig. 3a). This more severe 129 production impact in the developed nations was driven by a substantial yield 130 deficit of 15.9% with no significant reduction in harvested area (Fig. 3b-c). We 131 see three possible explanations for this pattern. First, it may arise from a 132 tendency among lower-income countries to encompass diverse crops and 133 management across many small fields, which may allow for some fields to resist 134 drought better than others. This might reduce the national drought sensitivity 135 compared to higher-income countries, where large-scale monocultures are more 136 dominant. Second, lower-income countries may better resist drought because 137 smallholders tend to employ risk-minimizing strategies compared to the yield-138 maximizing ones prevalent in higher-income countries. Finally, the pattern may 139 relate to generally lower fair-weather yields in lower-income countries. In Asia, 140 we found a significant reduction of 8.8% in harvested area during droughts with 141 no corresponding yield deficit, suggesting that this region has a greater tendency 142 for total crop failure in the event of a drought rather than harvesting with 143 reduced yields16. The production impacts in Africa did not correspond to 144 significant deficits in either yield or harvested area. 145  146 While the production of all three crops was similarly affected by droughts (5-6% 147 deficit each, Fig. 4a), only maize was significantly affected by extreme heat 148  7(11.7% deficit, p = 0.01) (Fig. 4b). Maize was also the only crop with significant 149 yield impacts (12.4%, p = 0.002) (Fig. 4c-d). We are hesitant to draw strong 150 conclusions based on this difference as it may be due to differing variance as well 151 as mean (see Supplementary Discussion).  Furthermore, it may reflect the fact 152 that maize is generally grown during summer months, which have the highest 153 probabilities of extreme heat as defined in EM-DAT, while wheat is grown during 154 the spring. Disaster data with monthly or daily resolution would enable us to 155 investigate whether this apparent susceptibility of maize is a result of differing 156 growing season.  157  158 Finally, more recent droughts (1985-2007) caused cereal production losses 159 averaging 13.7%, greater than the estimated 6.7% during earlier droughts 160 (1964-1984) (p = 0.008, Fig. 5), which may be due to any combination of rising 161 drought severity (although whether drought severity has increased globally is 162 presently debated)23–26, increasing vulnerability27 and exposure to drought6, 163 and/or changing reporting dynamics (Extended Data Figure 3). Sample size 164 limitations prevented us from repeating a regional and temporal analysis for 165 extreme heat. 166  167 Some limitations of our analyses are worth noting. First, we mainly focus on four 168 principal types of EWDs, but follow-up studies should include tropical storms 169 and extreme precipitation and wind events, especially since they may have an 170 increasingly significant impact on agriculture in the context of climate change28. 171 Second, our estimates are biased towards more recent disasters as they are more 172 abundantly reported in EM-DAT than older ones (see Extended Data Figure 3; 173  8Supplementary Discussion). Third, we use EWDs from the EM-DAT database, 174 which collates disasters based on several criteria for significant human impact 175 (see Methods). We may be underestimating the true impact of EWDs if disasters 176 are included mainly based on urban impacts, or if extreme events occurring in 177 sparsely populated areas are less likely to qualify as disasters. Finally, since we 178 observe agricultural impacts at the national level, more dramatic local and 179 regional effects of disasters may be muted (but conversely, finding a signal at the 180 national level highlights the substantial influence of droughts and extreme heat 181 events). Future studies may arrive at a more detailed estimate by using 182 subnational agricultural data, localizing the reported disasters within nations, 183 selecting events taking place during the growing season, and controlling for 184 severity of disasters. Linking the definitions of EWDs used in this study with 185 statistical meteorological definitions will also enable a forecasting of future 186 impacts. 187  188 Overall there are four main conclusions from our study. First, over the period 189 1964-2007 drought and extreme heat events substantially damaged national 190 agricultural production across the globe. Within the framework of this study, no 191 impact on agriculture was identified from floods and extreme cold events. 192 Second, drought reduced cereal yield as well as completely damaged crops while 193 extreme heat only affected yield, reflecting clear differences in the processes 194 leading to overall production impacts. Third, this study highlights an important 195 temporal dimension to these impacts. While the damage to cereal production is 196 considerable, this impact is only short term as agricultural output rebounds and 197 continues its growth trend after the global average disaster. Additionally, we 198  9show that recent droughts had a larger impact on cereal production than earlier 199 ones. Finally, our regional and crop specific analysis finds that developed nations 200 suffer most from these extreme events.  201  202 Present climate projections suggest that extreme heat events will be increasingly 203 common and severe in the future1. Droughts are likely to become more frequent 204 in some regions, though significant uncertainty persists in the projections6. This 205 study, by highlighting the important historical impacts of these extreme events 206 on agriculture, emphasizes the urgency with which the global cereal production 207 system must adapt to extremes in a changing climate. Understanding the key 208 processes leading to such crop losses enables an informed prioritization of 209 disaster risk reduction and adaptation interventions to better protect the most 210 vulnerable farming systems and the populations dependent on them.  211  212   213  10References 214 1. Battisti, D. S. & Naylor, R. L. Historical warnings of future food insecurity 215 with unprecedented seasonal heat. Science 323, 240–244 (2009). 216 2. WFP. Pakistan flood impact assessment. (2010). 217 3. Gu, L. et al. The 2007 Eastern US Spring Freeze: Increased Cold Damage in 218 a Warming World. Bioscience 58, 253 (2008). 219 4. Barriopedro, D., Fischer, E. M., Luterbacher, J., Trigo, R. M. & García-220 Herrera, R. The hot summer of 2010: redrawing the temperature record 221 map of Europe. Science 332, 220–4 (2011). 222 5. Coumou, D. & Rahmstorf, S. A decade of weather extremes. Nat. Clim. 223 Chang. 2, 491–496 (2012). 224 6. IPCC. Managing the Risks of Extreme Events and Disasters to Advance 225 Climate Change Adaptation. (Cambridge University Press, 2012). 226 7. UNISDR. The Pocket GAR 2013 - From shared risk to shared value: the 227 business case for disaster risk reduction. (2013). 228 8. WMO. Atlas of mortality and economic losses from weather, climate and 229 water extremes (1970-2012). (2014). 230 9. FAO. The impact of natural hazards and disasters on agriculture and food 231 and nutrition security - A call for action to build resilient livelihoods. (Food 232 and Agriculture Organization of the United Nations, 2015). 233 10. Lobell, D. B. et al. The critical role of extreme heat for maize production in 234 the United States. Nat. Clim. Chang. 3, 1–5 (2013). 235 11. Lobell, D. B., Sibley, A. & Ivan Ortiz-Monasterio, J. Extreme heat effects on 236 wheat senescence in India. Nat. Clim. Chang. 2, 186–189 (2012). 237 12. Lobell, D. B., Bänziger, M., Magorokosho, C. & Vivek, B. Nonlinear heat 238 effects on African maize as evidenced by historical yield trials. Nat. Clim. 239 Chang. 1, 42–45 (2011). 240 13. Moriondo, M., Giannakopoulos, C. & Bindi, M. Climate change impact 241 assessment: the role of climate extremes in crop yield simulation. Clim. 242 Change 104, 679–701 (2010). 243 14. Teixeira, E. I., Fischer, G., van Velthuizen, H., Walter, C. & Ewert, F. Global 244 hot-spots of heat stress on agricultural crops due to climate change. Agric. 245 For. Meteorol. 170, 206–215 (2013). 246  1115. Deryng, D., Conway, D., Ramankutty, N., Price, J. & Warren, R. Global crop 247 yield response to extreme heat stress under multiple climate change 248 futures. Environ. Res. Lett. 9, 034011 (2014). 249 16. Iizumi, T. & Ramankutty, N. How do weather and climate influence 250 cropping area and intensity? Glob. Food Sec. 4, 46–50 (2014). 251 17. Johnstone, S. & Mazo, J. Global warming and the Arab Spring. Survival 252 (Lond). 53, 11–17 (2011). 253 18. Welton, G. The impact of Russia’s 2010 grain export ban. Oxfam Policy 254 Pract. Agric. Food L. 11, 76–107 (2011). 255 19. CRED. EM-DAT: The OFDA/CRED International Disaster Database. (2011). 256 at <www.emdat.be> 257 20. Lobell, D. B., Schlenker, W. & Costa-Roberts, J. Climate Trends and Global 258 Crop Production Since 1980. Science 616, (2011). 259 21. Lobell, D. B. & Field, C. B. Global scale climate–crop yield relationships and 260 the impacts of recent warming. Environ. Res. Lett. 2, 014002 (2007). 261 22. Thornton, P. K., Ericksen, P. J., Herrero, M. & Challinor, A. J. Climate 262 variability and vulnerability to climate change: a review. Glob. Chang. Biol. 263 1–16 (2014). doi:10.1111/gcb.12581 264 23. Dai, A. Increasing drought under global warming in observations and 265 models. Nat. Clim. Chang. 3, 52–58 (2012). 266 24. Sheffield, J., Wood, E. F. & Roderick, M. L. Little change in global drought 267 over the past 60 years. Nature 491, 435–8 (2012). 268 25. Trenberth, K. E. et al. Global warming and changes in drought. Nat. Clim. 269 Chang. 4, 17–22 (2014). 270 26. Greve, P. et al. Global assessment of trends in wetting and drying over land. 271 Nat. Geosci. 7, 716–721 (2014). 272 27. Lobell, D. B., Roberts, M. J., Schlenker, W., Braun, N., Little, B. B., Rejesus, R. 273 M. & Hammer, G. L. Greater sensitivity to drought accompanies maize yield 274 increase in the U.S. Midwest. Science 344, 516-9 (2014). 275 28. Gornall, J. et al. Implications of climate change for agricultural productivity 276 in the early twenty-first century. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 365, 277 2973–89 (2010). 278 29. Brad Adams, J., Mann, M. E. & Ammann, C. M. Proxy evidence for an El 279 Niño-like response to volcanic forcing. Nature 426, 274–8 (2003). 280  1230. FAO. FAOSTAT. (2014). at <http://faostat3.fao.org>  281  282 Supplementary Information is linked to the online version of the paper at 283 www.nature.com/nature 284 Acknowledgements. We thank R. Below who is in charge of the EM-DAT project 285 at the Centre for Research on the Epidemiology of Disasters for sharing the data. 286 We thank C. Champalle for testing the original idea using data over East Africa in 287 a class project. This research was supported by a Discovery Grant from the 288 Natural Science and Engineering Research Council of Canada to N.R. 289 Author contribution. This research was designed and coordinated by N.R. All 290 authors performed analyses, discussed the results, and wrote the manuscript.  291 Author information. Reprints and permissions information is available at 292 www.nature.com/reprints. The authors declare no competing financial interests. 293 Readers are welcome to comment on the online version of the paper. 294 Correspondence and requests for materials should be addressed to N.R. 295 (navin.ramankutty@ubc.ca) 296   297  13Figure 1. Influence of extreme weather disasters on national cereal 298 production. Normalized production composites for (a) drought, (b) extreme 299 heat, (c) flood, and (d) extreme cold disasters over 7-year windows centered on 300 the disaster year (blue lines). Box plots depict the distributions of 1000 false-301 disaster control composites, with red crosses denoting extreme outliers. 302 Production during drought and extreme heat years was 10.1% and 9.1% below 303 the control mean, while no significant production signal was detected for floods 304 or extreme cold. Production resumed normal levels immediately following 305 drought and extreme heat events. The increasing trend in production over the 7-306 year window reflects the observed growth trend. 307  308 Figure 2. Influence of extreme weather disasters on national cereal yields 309 and harvested area. Yield (blue) and harvested area (red) composites for (a) 310 drought and (b) extreme heat, with significant points (those lying beyond the 311 control box plot whiskers) marked by stars (box plots not shown for clarity). 312 Drought was associated with significant deficits in both yield and harvested area 313 (5.1 and 4.1%), while extreme heat revealed only significant yield impacts of 314 7.6% with no significant effect on harvested area. 315  316 Figure 3. A regional analysis of the influence of drought. Regional composites 317 of (a) production, (b) yield, and (c) harvested area for drought, with significant 318 points (those lying beyond the control box plot whiskers) marked by stars (box 319 plots not shown for clarity). P-values reflect significance of differences between 320 regions in drought-year response (Kruskal-Wallis test). The drought-year 321 normalized production is 7.8 and 10.7% lower in developed Western countries 322  14than in Asia and Africa, a difference driven by a significantly greater yield deficit. 323 Meanwhile, the Latin America and Caribbean region exhibits no significant 324 response to drought. 325  326 Figure 4. The influence of drought and extreme heat on maize, rice, and 327 wheat. a-f, Drought and extreme heat composites of production, yield, and 328 harvested area for maize (blue), rice (red), and wheat (green), with significant 329 points (those lying beyond the control box plot whiskers) marked by stars (box 330 plots not shown for clarity). P-values reflect significance of differences between 331 crops in disaster-year response (Kruskal-Wallis test). Maize production 332 responds more to extreme heat than wheat and rice, an effect driven by a 333 substantial yield deficit.  334  335 Figure 5. A temporal analysis of the influence of drought. Production 336 composites for (a) earlier (1964-1984) versus (b) later (1985-2007) droughts, 337 with boxplots of 100 respective control composites. In later instances, mean 338 drought-year production losses were greater (13.7%) than in earlier instances 339 (6.7%; p = 0.008, Kruskal-Wallis test). 340  341 Extended Data Figure 1. Distributions of individual responses to drought 342 and extreme heat. Histograms of disaster-year differences from means of 1000 343 resampled controls for (a-c) drought and (d-f) extreme heat. A preponderance of 344 moderately negative values (falling towards the right of the red shaded areas) 345 underlies the negative mean disaster year signals, with a limited influence of 346 extreme cases (those at the left of the red shaded areas). 347  15 348 Extended Data Figure 2. The influence of sample size on estimated disaster 349 impacts. Estimated mean 16-cereal aggregated production deficit for (a) 350 extreme heat and (b) drought in 200 sub-samples with size of (1, 2, … , n) 351 (points). Dotted grey line shows the final estimated mean production deficit 352 (9.1% for extreme heat, 10.1% for drought). The majority of initial variability at 353 low sample sizes dissipates into the mean at well below the actual sample size 354 (n=39 for extreme heat, n=247 for drought). 355  356 Extended Data Figure 3. Time-series of the number of extreme heat and 357 drought disasters per year from the EM-DAT database. The EM-DAT 358 database is based on a compilation of disaster reports gathered from various 359 organizations including United Nations agencies, governments, and the 360 International Federation of Red Cross and Red Crescent Societies. The time-361 series of reported disasters per year exhibits an increasing trend, likely the 362 result of more complete disaster reporting in more recent decades with a 363 possible contribution from increasing disaster incidence. There is also large 364 inter-annual variability in the number of events. 365  366 Extended Data Figure 4. Seasonal weather anomalies of drought and 367 extreme heat disasters in EM-DAT. Normalized composite mean growing 368 season temperature for (a) extreme heat and (b) drought, and (c) total 369 precipitation for drought. Box plots depict the distributions of 1000 false-370 disaster control composites, with red crosses denoting extreme outliers. Extreme 371 heat events correspond to seasonal temperature anomalies of 1.2ºC, while 372  16drought years have only 0.15ºC warmer temperatures, with no significant 373 precipitation anomaly.   374  375 Extended Data Table 1: Statistical significance of individual crop analysis. 376 Percent of points on control composites less than EWD composites for individual 377 crop analysis, 1000 control replicates total. 378  379 Extended Data Table 2: Statistical significance of 16-cereal aggregate 380 analysis. Percent of points on control composites less than EWD composites for 381 16-cereal aggregate, 1000 control replicates total. 382  383 Extended Data Table 3: Statistical significance of regional analysis. Percent 384 of points on control composites less than EWD composites for 16-cereal 385 aggregate by region, 1000 control replicates total.  386  387 Extended Data Table 4: Sample sizes for individual crop and 16-cereal 388 aggregate analyses. 389  390 Extended Data Table 5: Sample sizes for regional analysis. 391  392 Extended Data Table 6: Kruskal-Wallis assumptions test results for group 393 comparison analyses. 394  395  396  397  17Methods 398 Superposed Epoch Analysis (SEA) is used to isolate an average EWD response 399 signal using time series of national agricultural production data and EWDs. SEA 400 is a statistical approach that has been used to enhance the signal (i.e., influence 401 of particular events) in time-series data, while reducing noise due to extraneous 402 variables29. The EWDs are compiled from the Emergency Events Database EM-403 DAT19 and consist of 2184 floods, 497 droughts, 138 extreme heat events, and 404 194 extreme cold events from 177 countries over the period 1964-2007. EM-405 DAT collects information on a reported disaster if at least ten people died, a state 406 of emergency was declared, international assistance was called, or at least 100 407 people were either injured, made homeless, or required immediate assistance19. 408 Disaster reports are gathered from various organizations including United 409 Nations agencies, governments, and the International Federation of Red Cross 410 and Red Crescent Societies20. The agricultural data consist of country-level total 411 production, average yield, and total harvested area data for 16 cereals30, 412 covering the 177 countries in the set of EWDs from 1961 to 2010.  413  414 From the time-series of agricultural data, we extracted shorter sets of time-415 series using a seven-year window centered on the year of occurrence of each 416 EWD, with three years of data preceding and following each EWD. The data were 417 normalized to the average of the three years preceding and following the event 418 to remove the absolute magnitude of national data from the signal. For multi-419 year droughts, we averaged across all drought years to produce a single disaster 420 year datum. For a three-year drought, for example, the seven-year window 421 became a nine-year window with seven data points (with the middle three years 422  18being averaged and assigned to year 0). The seven-year sets of EWD time series 423 were then centered on the disaster year and averaged year-wise to yield single 424 composited time-series of production, yield, and harvested area for each EWD 425 type (a total of 12 composited time series). The averaging thus strengthens the 426 signal at the central year of EWD occurrence, while also cancelling the noise in 427 the non-disaster years preceding and following the event.  428  429 During compositing, points on individual time-series co-occurring with another 430 disaster in the set were excluded from the mean. This procedure resulted in 431 variable sample size across the seven years of the composites. For brevity, we 432 have here presented mean sample sizes across all years; complete tabulated 433 sample sizes are displayed in Extended Data Tables 4-5. Our composited mean 434 estimate does not seem to be influenced by outliers (see Extended Data Figure 1 435 and Supplementary Discussion). The signal-to-noise strength will certainly 436 depend on the sample size, and we performed an analysis to estimate the 437 influence of sample size (see Extended Data Tables 4 and 5, Extended Data Figure 438 2, and Supplementary Discussion).  439  440 In addition to average per-disaster estimates, we also calculated aggregate 441 production losses over specific time periods. For each extreme heat or drought 442 event, we first applied the average per-disaster percentage loss estimate 443 (different values for extreme heat or drought) to the average national production 444 across the six adjacent non-disaster years. We then computed the aggregate 445 drought or heat related global production loss for each year by summing the 446 production losses for each event over the given time period. We estimated the 447  19percentage of global production lost to the EWDs relative to an estimated 448 counterfactual global production in a world without EWDs (the latter being the 449 sum of observed global production plus the estimated production loss).  450  451 The significance-testing procedure involved setting up a “control” estimate by 452 randomly resampling the agricultural data using sets of fictitious disasters with 453 randomly-generated years and countries of occurrence. The fictitious EWD time 454 series were averaged as for the true ones to yield composited ‘control’ time 455 series, and the entire process was repeated 1000 times. We quantified EWD-year 456 deficits in production, yield, and harvested area by subtracting the true EWD 457 time series from the mean of the controls. Excluding randomly generated 458 disasters that happened to be real disasters systematically raised the impact 459 estimates by ~1%; to present a more conservative and rigorous detection of the 460 disaster signal, we elected not to exclude such pseudo-disasters. Note that we 461 chose not to de-trend the time series before compositing to remove technology-462 driven growth, but rather simply estimate the disaster signal as difference from 463 control (see Fig. 1). We estimated the 95% confidence intervals for our point 464 estimates of impacts using an approach similar to a delete-one jackknife 465 resampling method (see Supplementary Discussion). 466  467 The percent significance of each estimate of the EWD composites relative to 468 control was estimated as the percentage of 1000 control points less than the 469 EWD composite estimate for each year. Points with estimated significance of 470 <0.5% or >99.5% were considered significant deficits and surpluses, 471 respectively, corresponding to a two-tailed 99% confidence level. While we 472  20chose a two-tailed approach for robustness, we found no significant surpluses. 473 The significant points appear as stars in Figures 2-4, while for Figures 1 and 5 we 474 present the EWD composites with the distribution of controls represented as an 475 array of box-and-whisker plots for a visual representation of significance. The 476 complete tabulated percent significance values are presented in Extended Data 477 Tables 1-3. 478  479 The earlier-versus-later analysis for droughts was performed by applying the 480 SEA procedure to the set of droughts divided roughly equally into earlier and 481 later halves. Similarly, the regional analysis was conducted by repeating SEA for 482 full set of disasters divided into four regional groupings, and the by-crop 483 composites were obtained by repeating SEA on the full disaster sets using crop-484 specific agricultural data from FAO30. Statistical significance of differences 485 between crop-specific, regional, and earlier-versus-later composites was 486 assessed using the Kruskal-Wallace test. We applied a quadratic transformation 487 to the data for comparison to equalize variance between groups (verified using 488 Levene’s test), and used non-parametric tests when comparing groups as normal 489 assumptions were not met (see Supplementary Discussion). 490  491 Code availability. All the core programs including codes to perform superposed 492 epoch analysis and the various statistics described in this paper are available on 493 Github (https://github.com/nramankutty/SEA-code). 494  495  496 Drought (n = 222) Extreme Heat (n = 32)Extreme Cold (n = 51)Flood (n = 756)Normalized CompositeNormalized Compositea bc d1.21.11.00.90.81.21.11.00.90.81.21.11.00.90.810.1% 9.1%Year from Event Year from Event1.21.11.00.90.8-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3Year from Eventa bYear from EventDrought (n = 222) Extreme Heat (n = 32)YieldHarvested Area1 2 30-1-2-30.941.021.041.060.980.960.92Normalized Composite1 2 30-1-2-30.941.021.041.060.980.960.9211Production Yield Harvested AreaYear from Event Year from EventYear from Event1-3 2 30-1-2 1-3 2 30-1-2p = 0.02 p = 0.002 p = 0.13a b cNormalized Composite1-3 2 30-1-21.11.0510.950.90.850.81.11.0510.950.90.850.81.11.0510.950.90.850.8Eur., N.A., Aus. (n = 28)Asia (n = 32)Africa (n = 125)L.Am. & Carib. (n = 37)Production Yield Harvested AreaDroughta b cd e f1-3 2 30-1-2Normalized Composite1.110.9ExtremeHeat1.21.110.90.81.21.110.90.81.21.110.90.81-3 2 30-1-2 1-3 2 30-1-21.110.91.110.9p = 0.61 p = 0.68 p = 0.65p = 0.01 p = 0.002 p = 0.18Year from Event Year from Event Year from EventMaize (n = 28)Rice (n = 16)Wheat (n = 32)Maize (n = 208)Rice (n = 171)Wheat (n = 234)a bEarlier (1964-1984)n = 126Later (1985-2007)n = 121Year from Event Year from Event1.110.91.110.9-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3Normalized Composite

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.52383.1-0376077/manifest

Comment

Related Items