Sixty years of climate livability in China. A look through the magnifying glass of climatically-induced hazards.

It is not something new that the world is experiencing strong changes in the spatio-temporal pattern of rainfall. As the temperature increases due to climate change, extreme cloudbursts have become more and more frequent, negatively affecting the climate livability of our planet. 

This also inevitably leads to an increased frequency of climatically-induced hazards, such as landslides, debris flows, debris floods and floods. As a result, even countries that seemed to be prepared to the occurrence of extreme rainfall events and their cascading effects, have experienced damages to local infrastructure and losses of human lives. The recent example in Germany is one of many, sadly.

Today, I will try to share a brief commentary on what can be observed in China since 1950 studying the national catalogue of climatically-induced hazards. In fact, I have been collaborating for a few years now with Nan Wang, a former PhD student visiting ITC, and now a fully fledged researcher at Northeast Normal University (China). Nan's PhD was centered around a large Chinese project aimed at collecting, storing and examining occurrences of hydro-morphological processes (HMP, any process in the spectrum between debris flows and floods), from 1950's to modern days, across the whole Chinese territory.    

Our experience investigating this unique dataset was mainly directed towards using statistical tools to analyse where, when and how frequently HMPs have occurred. We initially did so by exploring HMPs within specific morphological regions of China. China can be roughly divided into 6 macro areas that internally share some degree of geomorphological homogeneity. These regions are shown in map form and described in Table 1 at this link:

In the same article, which was also our first attempt to explore this dataset, we noted a constant increase of HMP occurrences through China, from 1950 to 2015 (see Fig.1b). And, the same trend applied to the six geomorphological regions mentioned above (see Fig.3). An interesting observation already at this exploratory stage was to note that HMPs primarily occur across China from June to August, and their frequency largely decreases during the rest of the year. This was a consistent trend for the 65 examined years.

Having this concept in mind, we started to ask ourselves how persistent this signal was in time and space. As a result, we explored the use of spatio-temporal clustering procedures, these being described at this link:

There, we were able to extend our previous observations and note that HMPs cluster in specific locations in China (Figure 5). And most importantly, they cluster in time periods of specific durations (Figures 6 and 7), these being shown to drastically decrease in recent years (Figure 9). Moreover, we also noted that these clusters, as intuitively should be, follow the rainfall patterns in space and time (Fig 8 and 10). 

As a result, we asked ourselves how to measure the association between HMPs and rainfall. We tried to do this by implementing rainfall-threshold techniques, these being described at this link,

However, this time we had to face a problem related to our capacity to continuously represent precipitation in space and time. In fact, direct measures of rainfall are typically collected at discrete locations where rain gauges are installed. And, because the rainfall is not smoothly distributed across a given landscape nor across time, simply interpolating the data measured at the weather stations can be misleading.

The alternative we had was to use radar-derived estimates of rainfall. They are continuous by nature. And, although they typically have a relatively coarse resolution, at the scale of the whole China they provide more than enough information. Nevertheless, these indirect measures often underestimate the actual rain one would measure at a site. Therefore, one should first choose the best satellite product, correlate it with rain-gauge based measurements and then correct for the bias/underestimation.

This is exactly what we did, subsetting the spatio-temporal domain to the last 30 years where radar-based rainfall estimates are available. The following effort was to test whether we could predict HMP occurrences by finding a rainfall threshold (Fig. 7 and 8) above which, these processes were present in our spatio-temporal records (Fig.9) and in specific geomorphological settings (Fig. 10 and 11).  

Given that we were able to accurately estimate rainfall thresholds, the next question in the perspective of climate livability is now "how much the spatio-temporal distribution of HMP occurrences will change in a future with even harsher climate changes?". And, "will we be able to still predict these processes accurately enough (assuming we actually and operationally can!) to be useful for disaster risk reduction?".

We are now trying to address the first of the two questions above, while also looking into the number of victims and financial losses induces by HMPs through the recent Chinese history. We are doing so by examining the pattern of extreme rainfall, land use changes, anthropic effects and more. But, this is a story for another time.

I wrote this piece to share a stimulating two-years research experience with all of you. And most importantly, to point out at the outstanding work done by Nan Wang, the researcher who has done all this almost by herself!

More soon....