in

Potential for surprising heat and drought events in wheat-producing regions of USA and China

Potential for surprising heat and drought events in wheat-producing regions of USA and China
Announcement


UNSEEN evaluation in Midwestern USA and Northeastern China

In the US midwest, the UNSEEN ensemble shows a steady rise in maximum temperatures that are possible over time; the interquartile range of the ensemble results used to fall below 30 °C, and now the upper end of the interquartile range approaches 35 °C (Fig. 1). Historically, the maximum temperatures recorded for the past 40 years have been lower than the extremes produced by the UNSEEN ensemble. The highest values in the large ensemble for recent years reach 40 °C, while the highest values in the observational dataset are around 37 °C.

Announcement
Fig. 1: UNSEEN events in the US study region.

Historical observations of temperature and precipitation in March–May in USA Midwest winter wheat producing region (blue crosses), overlaid on gray boxplots of the UNSEEN large ensemble. Boxplots visualize the illustrated as the median, interquartile range, 1.5x interquartile range and outliers. Plots are for the following variables: (a) Maximum temperature, (b) Number of days above the “stress” threshold of 27.8 °C, c number of days above the “enzyme breakdown” threshold of 32.8 °C, and (d) Total precipitation.

The number of days that exceed critical heat thresholds have also been increasing in both the observed and modeled datasets for the US midwest. The UNSEEN ensemble contains discrete events that are well beyond the observed record, including one event with more than 20 days exceeding the “enzyme breakdown” threshold.

There is no clear trend in rainfall in the observed or simulated datasets of March–May in the US midwest (Fig. 1d). The 2014 historical drought is close to the most extremely dry events simulated in the UNSEEN dataset, although there are a few UNSEEN events that are drier than this historical record. Such events could negatively impact wheat yields, as happened in 2014. In Kansas in 2014, the wheat monitor reported that “wheat condition declined all month and, by the end of May, 62% of the crop was reported to be in very poor to poor condition, compared to 47% at the beginning of the month and 45% last year”28 The yield per harvested acre was the lowest since 199528 News reports from local public radio explained that “persistent drought, harsh winds and below normal winter temperatures, combined with already low sub-soil moisture levels, have decimated the winter wheat crop in Kansas, Oklahoma, and Texas. These States make up the heart of the wheat belt—even with drought-affected low yields last year, they still produced one-third of the national winter wheat crop”29.

In China, results are similar (Fig. 2). Maximum temperatures in March–May show an increase with time, and the large ensemble includes many unprecedented events. This includes temperatures in the high 30 s, while the historical record is closer to 35 °C. The number of “stress” days and “enzyme breakdown” days are both increasing, with UNSEEN possibilities of more than 10 days in which the “enzyme breakdown” threshold is exceeded in one season.

Fig. 2: As in Fig. 1, for the China winter wheat region.
figure 2

Historical observations of temperature and precipitation in March–May (blue crosses), overlaid on gray boxplots of the UNSEEN large ensemble. Boxplots visualize the illustrated as the median, interquartile range, 1.5x interquartile range and outliers. Plots are for the following variables: (a) maximum temperature, (b) number of days above the “stress” threshold of 27.8 °C, (c) number of days above the “enzyme breakdown” threshold of 32.8 °C, and (d) total precipitation. Note that plot (b) did not pass the fidelity test and therefore should be interpreted with caution, as the kurtosis of the observed data is outside the 95th percentile of the UNSEEN ensemble.

The UNSEEN ensemble also contains several record-breaking drought events that have lower rainfall than ever observed in the region. These are physically plausible events that are drier than what has been historically observed. Hotter and drier summers can improve sowing and harvesting conditions and reduce the risk of waterlogging. However, should these periods coincide with sensitive crop development phases e.g. flowering and grain filling, this can result in relatively lower yield outcomes.

Increasing probability of extremes

Over time, the UNSEEN ensemble demonstrates a discernible change in the likelihood of extremely hot temperatures in both USA and China regions. Figure 3 plots the extreme value distribution fitted to the observations and the UNSEEN ensemble for maximum temperature in the March–May season. In both cases, maximum temperatures are higher now than in the 1980s, with a 1-in-100 year event in 1981 happening on average more often than every 6 years in 2020 in the US midwest. The simulated change is slightly less in northeast China, with a 1-in-100-year event in 1981 happening on average approximately every 16 years in 2020 (Fig. 3a, d). This translates into a 1% chance of the event happening in 1981, moving to a 17% (USA) and 6% (China) chance of happening in the year 2020.

Fig. 3: Extreme value distributions for daily maximum temperatures.
figure 3

Temperatures are in March–May for the US midwest (top row) and northeast China (bottom row). a, d Return period of seasonal maximum temperatures in 1981 and 2020. GEV fits are plotted for observations using dotted lines and light shading to indicate the uncertainty estimates. GEV fits are plotted for the UNSEEN ensemble using solid lines and dark shading for uncertainty estimates. All GEV fits are non-stationary distributions with covariates for the years 1981 and 2020. The magnitude of the 100-year event is indicated with a black horizontal line. b, e UNSEEN ensemble in gray overlaid with observations in blue, with a non-stationary 2-year return period estimation for each dataset. c, f UNSEEN ensemble in gray overlaid with observations in blue, with a non-stationary 100-year return period estimation for each dataset. The 95th percentile confidence interval is plotted using shading for the observations in blue. Statistical uncertainty is estimated as 95% confidence intervals based on the normal approximation.

In both case study locations, however, the best-fitting extreme value distribution for the observational dataset is a stationary GEV fit. This is in contrast to a non-stationary fit for the UNSEEN ensemble. If one were to simply extrapolate a nonstationary GEV fit from observational data in the USA study region (Fig. 3a, dotted line), for example, one would estimate lower values and much larger uncertainties as compared to the dynamically consistent UNSEEN ensemble, thus representing the strength of this type of analysis.

Assuming that the model is accurately representing the range of today’s climate, this could indicate that both regions have been “lucky” in recent years, and both regions have not experienced the full range of high temperatures that are now possible in today’s climate. In fact, these regions have been selected for wheat production partially because of favorable climate conditions in the past, and the critical thresholds were essentially boundaries. That is no longer the case, and extreme temperatures are much more likely. Recent memory of temperature extremes is on the lower end of the distribution of plausible extremes for today’s climate, especially in the US midwest where the difference between the observed and UNSEEN trend is highest. According to the UNSEEN ensemble, an event that would have been a 1-in-100 year maximum temperature event for March–May in the US midwest in 1981 is now a 1-in-6 year event. Other studies have similarly detected long term positive trends in temperature in both regions, with some attribution to anthropogenic climate change30,31,32,33,34. Whereas attribution studies compare the current climate with a pre-industrial climate, here, we are able to discern trends in the recent decades which might be of value in referencing people’s recent lived experiences.

Record-breaking extreme heat and drought

As expected given the stochastic nature of weather, we find that historical weather observations are limited in their range compared to a large ensemble of plausible weather outcomes. The UNSEEN ensemble contains a variety of heat and drought events for each location that would break historical records.

Extreme heat and extreme dryness are not independent of each other, often occurring simultaneously as a result of blocking weather patterns. Therefore, we plot the relationship between extreme heat and dryness in each location in Fig. 4. In both regions, extreme heat is strongly associated with dryness, and very wet events do not co-occur with extreme heat.

Announcement
Fig. 4: Extreme heat vs extreme dryness.
figure 4

Cumulative precipitation plotted against the number of days crossing the “enzyme breakdown” threshold in March–May for (a) USA and (b) China. UNSEEN ensemble members are plotted in gray, overlaid with observations in blue. The historical record for number of “enzyme breakdown” days is plotted as a blue horizontal line; any gray UNSEEN events above that line are record-breaking.

If there is a record-breaking hot season in which the number of days above the enzyme breakdown threshold is higher than experienced in the past (higher than the blue line in Fig. 4), it is likely to also be a dry season. In the US midwest, the UNSEEN ensemble produces 161 record-breaking seasons with high temperatures, and while most of them are relatively dry, 14% of them have extremely low rainfall that is less than the worst drought on record, the drought of 2014. This also applies in the other direction, of the 31 UNSEEN drought events that are worse than the worst drought experienced in the last 40 years, 71% of these events also have record-breaking heat.

In China, the results are similar; 63% of the UNSEEN record-breaking drought events are also record-breaking heat events in terms of the frequency of days above the enzyme breakdown threshold. We can imagine event-based storylines of extreme heat and extremely low rainfall that would cause unprecedented impacts at the intersection of these two hazards. Higher temperatures also produce higher evaporation rates, which can further reduce water availability for agriculture, beyond the record-breaking low precipitation.

Individual extreme events

One of the primary advantages of analyzing a large ensemble of physically plausible events is that the ensemble allows users to examine the drivers and physical contributing factors for specific extremes that have never happened in the observed record. In Fig. 5, we plot composites of geopotential height (GPH) anomalies and wind anomalies at the 500 mb pressure level, to analyze the most extreme events from the larger ensemble.

Fig. 5: Pressure and wind anomalies for UNSEEN events.
figure 5

Composites of geopotential height and wind anomalies at 500 mb associated with the most extreme events in the UNSEEN ensemble for the study region. Each plot is a composite of the 10 seasons of (a) highest precipitation, (b) lowest precipitation, and (c) highest number of enzyme breakdown days in each study area. The first row depicts the USA study area, and the second row the China study area, both delineated with a black box. Individual plots for each of the 10 events used to make these composites are available in the Supplementary Information.

In the USA region, the 10 driest March–May seasons in the SEAS5 ensemble (Fig. 5bi) are dominated by northerly and westerly wind anomalies. Such wind anomalies pull dry air from the continental USA to the study region, limiting precipitation. These are produced by positive seasonal anomalies of geopotential height to the west and south of the study region. In most of these 10 UNSEEN events, the positive anomalies are concentrated in the southwest of the USA, but there are individual events where the positive anomalies extend more widely across the USA (see Fig. SI11 for plots of anomalies associated with individual UNSEEN events).

Wind anomaly patterns are similar for the hottest seasons (Fig. 5ci); seasons with the largest number of hot days above the enzyme breakdown threshold are characterized by large regions of high pressure over the study area and to the southwest (Fig. 5ci). There are likely land-atmosphere feedbacks that can strengthen the heating effects during anomalously dry events35. In contrast, the wettest events (Fig. 5ai) have seasonal wind anomalies from the south and east, bringing moisture from the Gulf of Mexico and Atlantic (Fig. SI11).

In the China study region, wind anomalies are also critical to generating the extremely wet, dry, and hot events in Fig. 5, second row. The driest seasons (Fig. 5bii) modeled in the SEAS5 ensemble had wind anomalies from the north and west, bringing air over land towards the study area. This was associated with a low-pressure zone to the northeast of the study area. Hot seasons with the highest number of days above the enzyme breakdown threshold (Fig. 5cii) had similar wind anomalies as the low-precipitation events, and they also show a low pressure region to the northeast of the study area.

In eastern China, the 10 wettest seasons have wind anomalies from the opposite direction, coming from the south and east, bringing moisture to the study region (Fig. 5aii). These very wet seasons are associated with strong high-pressure regions to the northeast of the study area, generating clockwise winds that pull moisture from over the oceans to the winter wheat region. In July 2021, there was an extreme rainfall event in Henan province, once of the regions within our study area, and subsequent meteorological analyses identified that this was indeed caused by winds coming from the east, bringing moisture to the region36 similar to the synthetic events pictured in Fig. 5a.

In both regions, wind anomalies coming from land are associated with hot/dry seasons, and the opposite direction of wind anomalies over water are associated with extremely wet seasons, as might be expected. However, the patterns of geopotential height anomalies that produce such wind anomalies have some variety in their general shape, size, and location. Atmospheric anomalies associated with individual UNSEEN events are plotted in Supplementary Figs. 1015. For example, while northwesterly wind anomalies in the China winter wheat region are associated with a low-pressure zone to the northeast, this zone is larger in some realizations (e.g. Fig. SI 14B), and extends further south in other realizations (e.g. Fig. SI 14A and SI 14F). Therefore, meteorologists and climate scientists can be alert to several different varieties of the same pattern, which can all produce the wind anomalies that are associated with extremely hot/dry conditions in the region that produces winter wheat.

Compound events

The UNSEEN approach can be used to detect whether the likelihood of simultaneous extremes in both regions is higher than would be expected from random chance. In the observational datasets, there are no correlations between the two study regions of USA and China for maximum temperatures or total precipitation in the March–May season. In the UNSEEN ensemble, total precipitation remains uncorrelated between the two regions, but there is a small correlation for maximum daily temperature. The TXx correlation is 0.06 with 95% confidence intervals of 0.03-0.09. This is likely due to the influence of climate change on extreme temperatures globally, which affects both regions. See Supplementary Fig. 9 for a scatterplot of the temperatures.

While there is not a strong relationship between the two regions, there are individual UNSEEN events that do happen to produce simultaneous extremes in both locations. We identified the top 250 ensemble members for each study region that produced the greatest numbers of enzyme breakdown days, and there are 10 ensemble members that overlap in those two lists, producing extreme heat simultaneously in both locations. Figure 6 illustrates a composite of the geopotential height and wind anomalies associated with these 10 events, which are extreme in both study regions. This represents a dynamically consistent event-based storyline of a co-occuring event in both locations. The composite seems to be associated with a zonal wavenumber-3 disturbance in the higher latitude circulation, creating high pressure systems over both study areas. This compound event simultaneously creates the conditions seen in Fig. 5ci (USA) and 5cii (China) (Table 1).

Fig. 6: Pressure and wind anomalies for compound events.
figure 6

Composites of geopotential height and wind anomalies at 500 mb associated with concurrent extreme events in the UNSEEN ensemble in both study regions. Each plot is a composite of the 10 seasons that produced an extreme number of enzyme breakdown days in both study areas. The two study areas are delineated by gray boxes. Individual plots for each of the 10 events used to make these composites are available in Supplementary Fig. 16.

Table 1 100-year return periods for the daily maximum temperature in March–May, for 1981 and 2020 in both study regions, as simulated by the UNSEEN ensemble.
Announcement

Plots of individual compound events are available in Supplementary Fig. 16. The ensemble that generated the most extreme compound event was an UNSEEN ensemble member in 2018 (Fig. SI 16J), which shows similar atmospheric patterns to this composite. It simulated an event that had an regional average of 12.9 enzyme breakdown days in the USA study region (the observed record is 8.5), and in China, this event produced a regional average of 5.2 enzyme-breakdown days (observed record of 2.9).



Source link

Announcement

What do you think?

Written by

Announcement
Announcement

Leave a Reply

Your email address will not be published. Required fields are marked *

Announcement
Anti-War Protesters Picket White House Correspondents’ Dinner – Mother Jones

Anti-War Protesters Picket White House Correspondents’ Dinner – Mother Jones

Fatal crash reported on Palomares Road in Castro Valley

Fatal crash reported on Palomares Road in Castro Valley