On the Robustness of Annual Daily Precipitation Maxima Estimates Over Monsoon Asia

Abstract

Understanding precipitation extremes over Monsoon Asia is vital for water resource management and hazard mitigation, but there are many gaps and uncertainties in observations in this region. To better understand observational uncertainties, this study uses a high-resolution validation dataset to assess the consistency of the representation of annual daily precipitation maxima (Rx1day) over land in 13 observational datasets from the Frequent Rainfall Observations on Grids (FROGS) database. The FROGS datasets are grouped into three categories: in situ-based and satellite-based with and without corrections to rain gauges. We also look at three sub-regions: Japan, India, and the Maritime Continent based on their different station density, orography, and coastal complexity. We find broad similarities in spatial and temporal distributions among in situ-based products over Monsoon Asia. Satellite products with correction to rain gauges show better general agreement and less inter-product spread than their uncorrected counterparts. However, this comparison also reveals strong sub-regional differences that can be explained by the quantity and quality of rain gauges. High consistency in spatial and temporal patterns are observed over Japan, which has a dense station network, while large inter-product spread is found over the Maritime Continent and India, which have sparser station density. We also highlight that while corrected satellite products show improvement compared to uncorrected products in regions of high station density (e.g., Japan) they have mixed success over other regions (e.g., India and the Maritime Continent). In addition, the length of record available at each station can also affect the satellite correction over these poorly sampled regions. Results of the additional comparison between all considered datasets and the sub-regional high resolution dataset remain the same, indicating that the overall quality of the station network has implications for the reliability of the in situ-based products derived and also the satellite products that use a correction to in situ data. Given these uncertainties in observations, there is no single best dataset for assessment of Rx1day in Monsoon Asia. In all cases we recommend users understand how each dataset is produced in order to select the most appropriate product to estimate precipitation extremes to fit their purpose.

More intense daily precipitation in CORDEX‐SEA regional climate models than their forcing global climate models over Southeast Asia

“…Subsequently, the quantiles related to higher precipitation amounts from all datasets increasingly diverge from each other, with maximum variance observed in the highest quantiles (greater than the 99th percentile). This inter-comparison of seasonal climatologies in daily precipitation is in line with the previous regional studies on mean (Juneng et al, 2016;Tangang et al, 2020) or extreme precipitation (Nguyen et al, 2020), highlighting the substantial uncertainties among different observations over the region.…”

“…Note that we did not select the newest version of APHRODITE (i.e., V1901) due to its shorter time period (1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005) compared to its predecessor (i.e., V1101; 1951-2007) utilized here. In addition, Nguyen et al (2020) also noted the limited difference between APHRODITE V1901 and V1101 in the representation of extremes, which gives further confidence in using the V1101 version here.…”

“…At the global scale, observational uncertainties can be as large as inter-model spread (Herold et al, 2016) while at regional scales the ranking of a particular model can be substantially changed by the observational reference data employed particularly for precipitation (Kotlarski et al, 2019). Focusing on the Maritime Continent, Nguyen et al (2020) found large inconsistencies in extreme precipitation intensity and inter-annual variability among various observational products due to a poor station network. Therefore, uncertainties related to the reference data should ideally be taken into account when assessing climate model performance in the present-day climate over SEA.…”

“…A challenge for evaluating the performance of a climate model relates to observational uncertainties. Indeed, many studies have highlighted large differences between different products of observations at the global (Donat et al ., 2016; Herold et al ., 2016, 2017; Alexander et al ., 2020; Bador et al ., 2020a) and regional (e.g., Europe [Kotlarski et al ., 2014; Prein et al ., 2016]; SEA [Nguyen et al ., 2020]) scales. At the global scale, observational uncertainties can be as large as inter‐model spread (Herold et al ., 2016) while at regional scales the ranking of a particular model can be substantially changed by the observational reference data employed particularly for precipitation (Kotlarski et al ., 2019).…”

Introducing and Evaluating the Climate Hazards Center IMERG with Stations (CHIMES): Timely Station-Enhanced Integrated Multisatellite Retrievals for Global Precipitation Measurement

“…In the U.S., Europe, and Central America, IMERGlate means were 14 to 34% too large, while CHIRP2 means were within -2 to 3 percent (Table 2). Recent evaluations of IMERG (Nguyen et al 2020) indicate that this may relate to the overestimation of the wettest precipitation events.…”