An improved near real-time precipitation retrieval for Brazil

Pfreundschuh, Simon; Ingemarsson, Ingrid; Eriksson, Patrick; Vila, Daniel Alejandro; Calheiros, Alan James P.

doi:https://doi.org/10.5194/egusphere-2022-78

Preprints

https://doi.org/10.5194/egusphere-2022-78

Preprints

06 Apr 2022

Status: this preprint is open for discussion and under review for Atmospheric Measurement Techniques (AMT).

An improved near real-time precipitation retrieval for Brazil

Simon Pfreundschuh¹, Ingrid Ingemarsson¹, Patrick Eriksson¹, Daniel Alejandro Vila², and Alan James P. Calheiros³ Simon Pfreundschuh et al.

Show author details

¹Department of Space, Earth and Environment, Chalmers University of Technology, Gothenburg, Sweden
²Regional office for the Americas, World Meteorological Organization, Asunción, Paraguay
³Coordination of Applied Research and Technological Development, National Institute for Space Research (INPE), São José dos Campos, Brazil

Received: 21 Mar 2022 – Discussion started: 06 Apr 2022

Abstract. Observations from geostationary satellites offer the unique ability to provide spatially continuous coverage at continental scales with high spatial and temporal resolution. Because of this, they are commonly used to complement ground-based measurements of precipitation, whose coverage is often more limited.

We present a novel, neural-network-based, near real-time precipitation retrieval for Brazil based on visible and infrared (VIS/IR) observations from the Advanced Baseline Imager on the Geostationary Operational Environmental Satellite 16. The retrieval, which employs a convolutional neural network to perform Bayesian retrievals of precipitation, was developed with the aims of (1) leveraging the full potential of latest-generation geostationary observations and (2) providing probabilistic precipitation estimates with well-calibrated uncertainties. The retrieval is trained using co-locations with combined radar and radiometer retrievals from the Global Precipitation Measurement (GPM) Core Observatory. Its accuracy is assessed using one month of gauge measurements and compared to the precipitation retrieval that is currently in operational use at the Brazilian Institute for Space Research as well as two state-of-the-art global precipitation products: The Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks Cloud Classification System and the Integrated Multi-Satellite Retrievals for GPM (IMERG). Even in its most basic configuration, the accuracy of the proposed retrieval is similar to that of IMERG, which merges retrievals from VIS/IR and microwave observations with gauge measurements. In its most advanced configuration, the retrieval reduces the mean absolute error for hourly accumulations by 22 % compared the currently operational retrieval, by 50 % for the MSE and increases the correlation by 400 %. Compared to IMERG, the improvements correspond to 15 %, 15 % and 39 %, respectively. Furthermore, we show that the probabilistic retrieval is well calibrated against gauge measurements when differences in a priori distributions are accounted for.

In addition to potential improvements in near real time precipitation estimation over Brazil, our findings highlight the potential of specialized data driven retrievals that are made possible through advances in geostationary sensor technology, the availability of high-quality reference measurements from the GPM mission and modern machine learning techniques. Furthermore, our results show the potential of probabilistic precipitation retrievals to better characterize the observed precipitation and provide more trustworthy retrieval results.

Simon Pfreundschuh et al.

Status: open (until 18 Jun 2022)

Post a comment Subscribe to comment alert

RC1:
'Comment on egusphere-2022-78', Anonymous Referee #1, 07 May 2022 reply
An improved near real-time precipitation retrieval for Brazil

By Pfreundschuh et al.,

The paper presents a convolutional neural network architecture for IR precipitation retrieval over Brazil. The training data are from IR and GPM combined retrievals. The framework is extended such that it can provide uncertainty of the retrievals. The estimates are compared with ground-based gauge data to validate the retrievals. The paper is well written and is of high-quality. I have the following comments.

Major comments:

Validation only with a month of gauge data is not sufficient for clamming those improved results in the abstract. Seasonal to annual validation results are needed to make those claims.

A single storm retrieval is missing. It is imperative to show the output of the algorithm in retrieval of a single or multiple storms and compare the results with the combined GPM retrievals as a reference. One retrieval snapshot speaks very clearly about the skill of the algorithm in reconstructing the training data and retrieve spatial structure of precipitation.

Error metrics are only represented cumulatively. The expectation is that paper presents the quality of retrievals for an individual storm in terms of detection accuracy (e.g., probability of detection, miss) and then focuses on estimation quality metrics at different time scales from a storm scale to monthly and seasonal.

This needs to be clarified whether the training data were only over Brazil or not. If this is the case, then the provided improved statics are not of surprise. This issue needs to be stated in the abstract.

The way the paper explains the Bayesian retrieval is confusing. First, what is the prior distribution? Just obtaining uncertainty of estimates does not mean that the approach is Bayesian, and we can call the distribution a posterior. We can quantify uncertainty in a frequentists sense. It seems that the approach counts the number of retrievals associated with Tbs within bins. Then the bin with maximum is labeled. The problem is then defined as classification problem and the output of the softmax function is considered as the posterior distribution of the retrievals. Even though, I found the approach creative, I am not convinced that it is a Bayesian approach.

It is claimed that spatially aware CNNs provide more accurate retrievals than pixel-level DNNs. The reason is not discussed, and no evidence is provided.

In equation 2, when the prior probability approaches to a small number, the likelihood ratio can be extremely large. The correction numbers in Fig. 4 are too large. Please explain why such a large difference might exist in the retrievals that need such a large correction factor. For correcting probability distribution we can use a simple CDF matching!

The resolution of IR is higher than microwave data. In this sense, you have redundant samples. How were those samples treated in the training?

Explanation of the uncertainty quantification is too complex. Please consider simplifying the text and provide improved explanations.

Some minor comments

Why both the second and third configurations are needed. They are just different in resolution.

Line 160. Provide reasoning.

Line 185”. The range is too wide! The training GPM combined precipitation can range from 0.1 to 200 mm/hr. Why 1000 mm/hr?

Reply
Citation: https://doi.org/10.5194/egusphere-2022-78-RC1

Simon Pfreundschuh et al.

Viewed

Total article views: 215 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
173	36	6	215	6	4

HTML: 173
PDF: 36
XML: 6
Total: 215
BibTeX: 6
EndNote: 4

Views and downloads (calculated since 06 Apr 2022)

Month	HTML	PDF	XML	Total
Apr 2022	130	30	6	166
May 2022	43	6	0	49

Cumulative views and downloads (calculated since 06 Apr 2022)

Month	HTML	PDF	XML	Total
Apr 2022	130	30	6	166
May 2022	43	6	0	49

Viewed (geographical distribution)

Total article views: 204 (including HTML, PDF, and XML) Thereof 204 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 16 May 2022

Short summary

We use methods from the field of artificial intelligence to train an algorithm to predict rain from satellite observations. In contrast to other methods, our algorithm not only predicts the rain but also the uncertainty of the prediction. Using independent measurements from rain gauges, we show that our method performs better than currently available methods and that the provided uncertainty estimates are reliable. Our method makes satellite-based estimates of rain more accurate and reliable.


Total:	0
HTML:	0
PDF:	0
XML:	0