A Ray of Hope: Near Real-Time Forecasting Tools for the USAPI

Category: Monitoring Drought
Project Team: Pacific Water Resources II
Team Location: NOAA National Centers for Environmental Information – Asheville, North Carolina

GPM Precipitation anomalies for the 2016 March-April-May (MAM) season. Brown areas represent less than normal precipitation. Turquoise areas represent greater than normal precipitation. Image Credit: Pacific Water Resources II team

GPM Precipitation anomalies for the 2016 March-April-May (MAM) season. Brown areas represent less than normal precipitation. Turquoise areas represent greater than normal precipitation. Image Credit: Pacific Water Resources II team

Nicholas Luchetti
Zachary Vozzelli
Kyle Jones

Michael Kruk (Earth Resources Technology (ERT))
John Marra (NOAA Region Climate Services, Director, Pacific Region)

Past/Other Contributors:
Alec Courtright (Center Lead)
Jessica Sutton
Ethan Wright


The U.S. Affiliated Pacific Islands (USAPI) are extremely vulnerable to the precipitation shifts associated with the El Niño Southern Oscillation (ENSO). For example, the 2015-2016 ENSO event caused crippling drought conditions for the USAPI that extended several seasons. In the past, scientists in the region used a spatially-limited, in situ-based ENSO climatology to inform their drought mitigation decisions. To fill this spatial gap, the Pacific Water Resources I team successfully delivered an updated, ENSO-based precipitation climatic reference atlas. The atlas was derived from remotely-sensed data from the National Oceanic and Atmospheric Administration’s Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks-Climate Data Records (PERSIANN-CDR), which provides a 30-year record of daily precipitation at 0.25-degree spatial resolution. While the atlas has been heavily used by scientists in the region following the Pacific Water Resources I project, it is somewhat limited in that it does not provide near real-time precipitation estimates. The Pacific Water Resources II project filled this limitation through the use of near real-time precipitation data from NASA’s Global Precipitation Measurement (GPM) satellite, which provides 30-minute rainfall estimates at 0.1-degree spatial resolution. To fully understand whether satellite-derived rainfall estimates from GPM can be used operationally in a near-real-time anomaly product, an analysis comparing the satellite products (PERSIANN-CDR, GPM) to 27 Global Historical Climate Network Daily (GHCN-D) stations in the west Pacific was completed. Results of this validation study suggest that the PERSIANN-CDR and GPM tend to underestimate the daily precipitation estimates when compared to the GHCN observations. That being said, while the raw station values do not necessarily line up exactly with those from satellite-derived rainfall estimates, the direction of the trends are the same. For example, when the station data suggests periods of dryness, satellite estimates also suggest the same, and vice versa. Therefore, results herein confirm the usefulness of using GPM precipitation estimates to accurately capture the seasonal precipitation trends found across the USAPI. The end results from this project provided a suite of near real-time precipitation forecasting tools that can enhance short-term water resources management.

Previous story | Main Page | Next story

Topic: ,


Daryl Ann Winstead (Mekong River Basin Agriculture) 18-08-2016, 16:23

Interesting project! What were some limitations that the team encountered over the course of the project? Also, what software was used throughout the term? Thanks in advance for your response!

Nick Luchetti 25-08-2016, 11:41

Hi Daryl,

Thanks for the kind words and questions!

Originally, our team was going to use the NOAA CPC CMORPH precipitation data set as the near real-time input, however the data access for that product was a bit challenging. Although a slightly lower resolution, NASA’s GPM product offered the same functionality for the end-goal of this project, and was much easier to access and process. So, I suppose if we had more time, we probably would have figured out the data access to the higher resolution CMORPH data.

Other than that, only limitations we experienced were in the data sets themselves… Satellite derived precipitation estimates tend to underestimate compared to in situ stations….Both PERSIANN-CDR and GPM underestimated daily precipitation across the USAPI, however, PERSIANN-CDR had less of a bias when compared to the GPM. Our team statistically documented this bias, and provided this information via a user guide so that decision maker’s can understand the potential limitations of our tools.

We utilized R, Python, and ArcGIS for data processing and creating the end tools. Additionally, we completed a daily validation comparing in situ stations to both PERSIANN-CDR and GPM estimates (not shown in video)… We utilized ArcGIS and Excel to do this…

Hope that answered your questions!


Nick Luchetti

Emma Baghel 12-08-2016, 12:57

Great job! This looks like fantastic work. Was all of the work necessary to get the results your end-users wanted completed within these 2 terms? If not, what else is left and could you tell me how your partners will take the work you’ve done and expand on it?

Nicholas Luchetti 25-08-2016, 12:09

Hi Emma,

Thanks for the kind words and questions!

Over the past 2 terms, out team’s successfully completed all that was asked of us. However, the USAPI has so many limitations that the amount of application based research that could be done in the region is endless….. As far as creating drought monitoring products through the utilization of satellite-derived precipitation products, our team’s successfully and tremendously enhanced the products they were using out there…. But even with that, you could easily do much much more with satellite-derived precipitation estimates across the region….

I know that a new DEVELOP project is in the works to analyze sea surface heights on a climatological time scale and sub set by ENSO….

Hope that answered your question!


Nick Luchetti


Leave a Reply