The temperature, precipitation and soil moisture drivers of extreme hot-dry compound events and rapid dry-to-wet seesaw events are modelled for New Zealand using tri- and bi-variate copulas (respectively), as well as a more conventional approach of analysing each variable separately. Copula-based direct modelling of the joint variation between these variables reveals that the conventional approach leads predominantly to the underestimation of return periods for these extreme events.

Flash floods are becoming more frequent in mountainous regions due to heavier rainstorms. To protect people and property, we are working to better understand local hydrology and improve the efficiency of early warning systems for urban flooding in Lesser Himalayas. By combining community knowledge, low-cost technology, we can enhance understanding of flood dynamics and strengthen preparedness in mountains. This work is a step toward building resilience by bridging science and community insight.

We combine two globally capable glacier evolution models to include processes that are typically neglected but thought to control tropical glacier retreat (e.g. sublimation). We apply the model to Peru's Vilcanota-Urubamba Basin. The model captures observed glacier mass changes,but struggles with surface albedo dynamics. Projections show glacier mass shrinking to 17 % or 6 % of 2000 levels by 2100 under moderate- and high-emission scenarios, respectively.

The transport of dissolved organic carbon (DOC) from land into streams is changing due to climate change. We used a multi-year dataset of DOC and predictors of DOC in a subarctic stream to find out how transport of DOC varied between seasons and between years. We found that the way DOC is transported varied strongly seasonally, but year-to-year differences were less apparent. We conclude that the mechanisms of transport show a higher degree of interannual consistency than previously thought.

Floods continue to be a wicked problem that require developing early warning systems with plausible assumptions of risk behaviour, with more targeted conversations with the community at risk. Through this paper we advocate the use of a SMART approach to encourage bottom-up initiatives to develop inclusive and purposeful early warning systems that benefit the community at risk by engaging them at every step of the way along with including other stakeholders at multiple scales of operations.