Comparing deuterium excess to large-scale precipitation recycling models in the tropics

Stephen Cropper,Jolanda J E Theeuwen,Arie Staal,Obbe A Tuinenburg,Kurt Solander,Chonggang Xu,Brent D Newman

doi:10.1038/s41612-021-00217-3

Stephen Cropper, Jolanda J E Theeuwen + Show 5 more

Open Access

https://doi.org/10.1038/s41612-021-00217-3

Copy DOI

Abstract

Precipitation recycling is essential to sustaining regional ecosystems and water supplies, and it is impacted by land development and climate change. This is especially true in the tropics, where dense vegetation greatly influences recycling. Unfortunately, large-scale models of recycling often exhibit high uncertainty, complicating efforts to estimate recycling. Here, we examine how deuterium excess (d-excess), a stable-isotope quantity sensitive to recycling effects, acts as an observational proxy for recycling. While past studies have connected variability in d-excess to precipitation origins at local or regional scales, our study leverages >3000 precipitation isotope samples to quantitatively compare d-excess against three contemporary recycling models across the global tropics. Using rank-correlation, we find statistically significant agreement (bar tau = 0.52 to 0.70) between tropical d-excess and recycling that is strongly mediated by seasonal precipitation, vegetation density, and scale mismatch. Our results detail the complex relationship between d-excess and precipitation recycling, suggesting avenues for further investigation.

Highlights

Growing water scarcity, changing terrestrial hydrologic cycles, and their impacts on ecological function have generated considerable interest in the origins of precipitation[1,2]
We hypothesize that (1) recycling ratios generated from a representative model should correlate with the spatial and temporal variations in d-excess (H1), given the sensitivity of d-excess to the phase changes that occur during precipitation recycling; (2) the particle tracking models will better predict d-excess by capturing a more representative set of land evaporation sources (H2) because unlike the Mass Balance method (RRR), the particle tracking approach to computing the recycling ratio (LRR) theoretically accounts for all global sources of land evaporation; and (3) increased spatial coverage of d-excess sampling data will improve model-observation agreement (H3), as the mismatch in scales between point-like d-excess observations and large-scale recycling model estimates is resolved by averaging over larger datasets
Before introducing d-excess as an observational proxy for recycling, we first compared the three precipitation recycling models considered in this study, Mass Balance (RRR)[14], WAM2layers (LRR)[17], and UTrack (LRR)[18], across the Köppen-Geiger (KG) climate subzones that occur in the tropics[33]

Summary

INTRODUCTION

Growing water scarcity, changing terrestrial hydrologic cycles, and their impacts on ecological function have generated considerable interest in the origins of precipitation[1,2]. We hypothesize that (1) recycling ratios generated from a representative model should correlate with the spatial and temporal variations in d-excess (H1), given the sensitivity of d-excess to the phase changes that occur during precipitation recycling; (2) the particle tracking models will better predict d-excess by capturing a more representative set of land evaporation sources (H2) because unlike the Mass Balance method (RRR), the particle tracking approach to computing the recycling ratio (LRR) theoretically accounts for all global sources of land evaporation; and (3) increased spatial coverage of d-excess sampling data will improve model-observation agreement (H3), as the mismatch in scales between point-like d-excess observations and large-scale recycling model estimates is resolved by averaging over larger datasets.

RESULTS

DISCUSSION

METHODS

CODE AVAILABILITY