GIS Tools, Data, & Methods in the Weekly US Drought Monitor International Workshop on Agromet and GIS Applications for Agricultural Decision Making Jeju, South Korea December 5, 2016 Eric Luebehusen USDA/OCE/WAOB Meteorologist USDM Author
The US Drought Monitor didn t always look like this
May, 1999 - The very first U.S. Drought Monitor!! It was experimental, and developed partially in response to intensifying dryness in the eastern U.S. The map was created in CorelDRAW (basic drawing software). A trip down memory lane 1999 2001 2003 2005 2007 2009 2011 2013
July 7, 1999 (Experimental) Switched map to U.S. only 1999 2001 2003 2005 2007 2009 2011 2013
Aug 11, 1999 (Experimental) This week s map was presented to senior-level government officials at a White House Briefing. They liked it so much 1999 2001 2003 2005 2007 2009 2011 2013
Aug 18, 1999 (Operational!) The following week, it went operational, making this the first official U.S. Drought Monitor! This might have been the fastest Experimental to Operational product in government history! 1999 2001 2003 2005 2007 2009 2011 2013
September 7, 1999 Layout adjusted, still 2 colors. 1999 2001 2003 2005 2007 2009 2011 2013
September 15, 1999 Final 5-color scheme employed 1999 2001 2003 2005 2007 2009 2011 2013
8.2 August, 2003 USDM goes GIS 1999 2001 2003 2005 2007 2009 2011 2013
August 12, 2008 My 1 st USDM! 1999 2001 2003 2005 2007 2009 2011 2013
2008-Current GIS Wx/Hydro data used to aid Dx depiction 1999 2001 2003 2005 2007 2009 2011 2013
2013 National Drought Mitigation Center became central agency for final map production and QC 1999 2001 2003 2005 2007 2009 2011 2013
Late 2013 NDMC changed the final map layout 1999 2001 2003 2005 2007 2009 2011 2013
And today s US Drought Monitor
What makes a drought D1, D2, D3, or D4?
Drought Category Color Frequency Pctile D4, Exceptional Drought: once per 50-100 years 0-2 D3, Extreme Drought: once per 20-50 years 2-5 D2, Severe Drought: once per 10-20 years 5-10 D1, Moderate Drought: once per 5-10 years 10-20 D0, Abnormally Dry: once per 3-5 years 20-30 The drought categories are associated with historical occurrence/likelihood (percentile ranking). This allows us to quantify the drought with objective drought indicators. ----------- It is not anecdotal or subjective, like It s really, really dry!! or I don t remember it ever being this dry we have to be D4!!!
The drought monitor is created/edited in ArcMap GIS.
There are 5 separate drought shapefiles which we edit; they are overlaid on top of each other to give the illusion of one drought map.
One big advantage of editing the drought areas in GIS is the wealth of weather and hydrological data also available in GIS format; we can bring the data directly into the Drought Monitor to assist with the final drought depiction.
Before GIS, authors had to manually triangulate everything.
Initially, there was not much change in the USDM when authors switched from Corel Draw to ArcMap (GIS) in August, 2003. 8.2
July 16, 2002 November 1, 2016 8.2 But now, with new GIS data and methods, drought depiction shows considerable increase in detail. Note the Southeastern U.S. drought, then (before GIS) and now.
ools, Data, and Methodology
This increased drought depiction detail has been made possible by a wide array of hydrological and meteorological data in GIS, spearheaded by efforts at USDA. Here, my USDM GIS editor project shows the wealth of weather and hydrological GIS data available we currently use.
The U.S. NWS provides gridded radar-based precipitation data (AHPS -Advanced Hydrologic Prediction Service), which is available for GIS applications. What is shown here is a Geotiff, created in-house. The raw data is in DBF format and takes a considerable amount of time to download & display
The raw DBF data looks similar, but is comprised of thousands of individual points. USDM authors were handicapped by this data format.
The raw DBF data looks similar, but is comprised of thousands of individual points. USDM authors were handicapped by this data format.
The raw DBF data looks similar, but is comprised of thousands of individual points. USDM authors were handicapped by this data format.
The raw DBF data looks similar, but is comprised of thousands of individual points. USDM authors were handicapped by this data format.
The raw DBF data looks similar, but is comprised of thousands of individual points. USDM authors were handicapped by this data format.
The raw DBF data looks similar, but is comprised of thousands of individual points. USDM authors were handicapped by this data format.
The raw DBF data looks similar, but is comprised of thousands of individual points. USDM authors were handicapped by this data format.
The raw DBF data looks similar, but is comprised of thousands of individual points. USDM authors were handicapped by this data format.
The raw DBF data looks similar, but is comprised of thousands of individual points. USDM authors were handicapped by this data format.
The raw DBF data looks similar, but is comprised of thousands of individual points. USDM authors were handicapped by this data format.
The raw DBF data looks similar, but is comprised of thousands of individual points. USDM authors were handicapped by this data format.
The raw DBF data looks similar, but is comprised of thousands of individual points. USDM authors were handicapped by this data format.
The raw DBF data looks similar, but is comprised of thousands of individual points. USDM authors were handicapped by this data format.
The raw DBF data looks similar, but is comprised of thousands of individual points. USDM authors were handicapped by this data format.
The raw DBF data looks similar, but is comprised of thousands of individual points. USDM authors were handicapped by this data format.
The raw DBF data looks similar, but is comprised of thousands of individual points. USDM authors were handicapped by this data format.
To make the data load and display quickly, we convert it from DBF to Geotiff using Python scripts. This is a 2-step process
Wget We use a Windows Scheduler to launch Wget Code, which downloads the raw DBF data from the web to our server. Wget
DBF Geotiff Python Using the same Windows Scheduler, we run Python code to convert the DBF data to geotiff. Python
Wget The pairing of Wget & Python allows our small group at USDA to download and process thousands of data files daily, and is vital to our operations. Python
In the drought GIS application, a fast-loading, easy-to-use GIS precipitation data source in Geotiff format is essential. AHPS gridded radar-based precipitation data ACIS station-based precipitation data
The NWS uses gridded PRISM Normals to provide Precipitation Departures as well. AHPS gridded radar-based precipitation data ACIS station-based precipitation data
AHPS gridded radar-based precipitation data ACIS station-based precipitation data Another product which gets a great deal of attention and airplay within the drought community is the Percent of Normal Precipitation. All of these precipitation products are available over several time frames.
The gridded precipitation product based primarily on radar data provides great detail. But, we still use in-situ measurements to assess drought conditions. AHPS gridded radar-based precipitation data ACIS station-based precipitation data
AHPS gridded radar-based precipitation data ACIS station-based precipitation data We bring in station data covering the same parameter and time period to quality control the gridded data and further support our drought depiction.
The ACIS station data (ACIS Applied Climate Information System) is essential to our GIS efforts. To get this station data into GIS involves several steps, all of which are now automated AHPS gridded radar-based precipitation data ACIS station-based precipitation data
Wget We use Wget to download the raw data files to our server. wget -O Y:\DATA\current_pcpsmry.txt http://drought.rcc-acis.org/pcpsmry/%year%%month%%day%_pcpsmry.txt http://drought.rcc-acis.org/pcpsmry/20161128_pcpsmry.txt Y:\DroughtMonitor\DATA\STATION-TEXT\current_pcpsmry.txt
Raw Data (linked & auto-updates) We use Excel to Import the raw data and Convert it for GIS. GIS-ready data (linked to Raw Data )
ArcMAP points to the GIS-ready Excel Sheet. Once the Symbology is set, the updated data displays as the Excel Sheet is updated, saved, and closed. The down side to GIS Excel: Only 1 instance of any given Excel data sheet can be open at a time.
Percent of normal rainfall data is great but how do we convert this to a drought category?? AHPS gridded radar-based precipitation data ACIS station-based precipitation data
This is accomplished thru the Standardized Precipitation Index (SPI). This puts the precipitation into a historical distribution, which is associated with the corresponding percentile and drought category. This data is supplied in Shapefile format and is easy to work with. D0 D1 D2 AHPS gridded radar-based precipitation data ACIS station-based precipitation data D3 D4
And like the station data, we also have a gridded SPI to use for our drought work. This data is produced (by NC State, R. Ward, et al) in Geotiff format, making it much easier to work with. AHPS gridded radar-based precipitation data ACIS station-based precipitation data
There is also a gridded SPEI, which accounts for temperature and evaporative losses, for our drought work. This data is produced in Geotiff format as well, though we are using it with caution. AHPS gridded radar-based precipitation data ACIS station-based precipitation data
Data provided by USGS in shapefile format In-situ data also includes streamflow percentiles from the U.S. Geological Survey. Streamflows are a good early indicator of developing drought and impacts.
Data provided by USDA/NRCS (csv) and CA Dept of Water Resources (web -> Excel Table) Data converted using Excel and linked to GIS Drought in the western U.S. operates on much longer time scales and having a strong seasonal component. At USDA, we are downloading, converting, and plotting the monthly reservoir data in GIS.
Data provided by USDA/NRCS (csv) and linked directly to GIS The western U.S. relies heavily on spring snowmelt for water, so we monitor the snowpack during the course of the cold season.
The USDM authors use a large amount of derived data to help with the drought work. Here, the weekly satellitederived Vegetation Health Index is shown, courtesy of Dr. Felix Kogan, et al @ NOAA/NESDIS (geotiff). Data provided weekly by NOAA/NESDIS in geotiff format
We don t have a nation-wide network of uniform soil moisture sensors, so we rely on modeled soil moisture (available as a Geotiff). Data provided daily by NOAA/EMC
Other products are coming online for US Drought Monitor authors to evaluate, such as the Evaporative Stress Index (ESI). Data provided weekly by NOAA/CDC in ascii format
If you ve ever edited shapefiles in GIS, you know it can be a very difficult task. What follows is a very brief sample of how we do it
I recently demonstrated on an USDA Webinar the thankless task of editing shapefiles (in super-fast motion!)
This is a rapid-fire tour!
This is a rapid-fire tour! Each week takes ~ 20-40 hours of your time when authoring.
GIS has not just made the map more accurate and detailed, but there are also a host of secondary derived products using the GIS data.
At USDA, Harlan Shannon has been producing crops in drought products since 2006. Here, wheat areas in drought is shown.
Harlan also extracts the data by date (top left) and by state (bottom right)
The ability to extract county-level drought information has also led USDA s Farm Service Agency to use the USDM as a tool for triggering financial aid to producers.
This has been a source of much angst and stress for USDM authors. You ll see why on the next slide. Using GIS, FSA determines whether any part of a county is touched by the corresponding trigger (primary counties); adjacent counties are included as a buffer. 8 consecutive weeks 1 week 1 week http://www.fsa.usda.gov/programs-and-services/disaster-assistance-program/disaster-designation-information/index
Nearly $6 billion dollars in payouts since 2011. Does not include this year s figures.
The National Drought Mitigation Center has likewise taken GIS data and developed web pages for regional-, state-, and county-level data extraction. http://droughtmonitor.unl.edu/
ools, Data, and Methodology
As the data, methodology, expertise, and process have evolved, so has the U.S. Drought Monitor. GIS has been a BIG part of this evolution.
It can be a thankless job at times, but a learning experience on many levels. Assessing the drought on a weekly basis involves countless hours of the 11 volunteer authors, hundreds of local and regional experts, and GIS data and software.
July 16, 2002 November 1, 2016 8.2 GIS has made our drought work more detailed and accurate, though the amount of data at times can be too much of a good thing!
Eric Luebehusen USDA/OCE/WAOB Meteorologist USDM Author