HYSPLIT vs ALOHA preferred distance from source
-
michael.kurz
- Posts: 1
- Joined: February 21st, 2020, 1:29 pm
- Registered HYSPLIT User: Yes
HYSPLIT vs ALOHA preferred distance from source
I'm seeing what appears to be conflicting information regarding preferred minimum distance from the source for using HYSPLIT vs. ALOHA. The HYSPLIT limitations webpage from ARL (https://ready.arl.noaa.gov/hypub/limitations.html) says "For scales less than 1 km the CAMEO/ALOHA model is recommended." But NWS Directive 10-518 defines small scale releases as being 0-10 km from the source, while large scale requiring HYSPLIT is defined as >10 km. Can someone please clear up my confusion -- at what distance does HYSPLIT become the preferred dispersion model, beyond 1 km or beyond 10 km? Thank you.
Re: HYSPLIT vs ALOHA preferred distance from source
The HYSPLIT limitations web page (https://ready.arl.noaa.gov/hypub/limitations.html) has a number of cautions, including the one you mention regarding distances less than 1 km.
In the past, HYSPLIT was not recommended for distances less than 1 km from a source because the minimum time step in the HYSPLIT model is 1 minute, and there was no provision to avoid "leap-frogging" over concentration grid cells. However, the code has been modified so that "leap-frogging" no longer happens. Thus, there is no longer any transport-related reason why HYSPLIT should not be used at distances less than 1 km from the source.
The ALOHA model uses a Gaussian Plume approach -- as opposed to 3D Lagrangian particles or puffs -- and estimates the downwind concentration field from a short-term or continuous release. If one changes any of the inputs to the ALOHA model (e.g., emissions, wind speed, wind direction, and/or atmospheric stability), then the downwind concentrations estimated by ALOHA will change. A Gaussian Plume model uses one set of meteorological inputs at a time -- generally estimated or measured at the source location. As one moves further and further away from the source and if the meteorological conditions change significantly, then the Gaussian Plume model will become less and less accurate.
The CAMEO/ALOHA system also has additional features, like simulation of dense gas physics, that can be very important for short-range transport away from the source, especially within the first kilometer. The CAMEO/ALOHA system also has very useful emissions-estimating algorithms for chemical accidents. HYSPLIT does not. So, even if one is using HYSPLIT, use of the CAMEO/ALOHA system to estimate the emissions going into the HYSPLIT model can be very useful.
Like HYSPLIT, ALOHA's accuracy is dependent on the accuracy of the meteorological data used to drive it. With both models, for example, if the input wind direction used is going in the wrong direction, the models will predict the plume going in that wrong direction.
Gaussian plume models -- like ALOHA -- are generally considered to be useful only out to about 50 km away from the source, at most. And within that 50 km range of applicability , the accuracy may decrease as one gets further and further away from the source. Beyond 50 km, one definitely would not want to use a Gaussian Plume model, but would want to use a model like HYSPLIT that has the ability to account for changing meteorological conditions as one moves further and further away from the source.
In the 0-10 and even the 10 - 50 km ranges downwind, I think it would generally be very useful to consider outputs from both models. The differences in simulation results between the two models can give you useful information about the level of uncertainty in the results.
In the NWS document (NWS 10-518) you referenced -- current version (as of Feb 28, 2020) at https://www.nws.noaa.gov/directives/sym ... 18curr.pdf -- I found the following, and I have added some comments in {brackets} building on the ideas discussed above:
(1) page A-6: HYSPLIT should not be used at distances less than 1 km from the source. {This was primarily based on the leap-frogging issue discussed above, related to the 1-minute minimum time step in HYSPLIT. This issue has been fixed, and so, the primary reason for this caution is no longer applicable.}
(2) page 4: "For small scale (0-10 km), short duration (less than one hour) releases, many local jurisdictions including emergency responders use ALOHA (part of the Computer-Aided Management of Emergency Operations (CAMEO) software suite) to forecast the dispersion impact." {This doesn't mean that HYSPLIT is not potentially useful within this range.}
(3) page 5: "In the event of a large atmospheric release (horizontal extent usually >10 km lasting more than an hour) of some chemical, biological or radioactive contaminant, an emergency manager may call the WFO and request atmospheric transport and dispersion information." {This doesn't mean that ALOHA is not potentially useful at distances greater than 10 km.}
In the past, HYSPLIT was not recommended for distances less than 1 km from a source because the minimum time step in the HYSPLIT model is 1 minute, and there was no provision to avoid "leap-frogging" over concentration grid cells. However, the code has been modified so that "leap-frogging" no longer happens. Thus, there is no longer any transport-related reason why HYSPLIT should not be used at distances less than 1 km from the source.
The ALOHA model uses a Gaussian Plume approach -- as opposed to 3D Lagrangian particles or puffs -- and estimates the downwind concentration field from a short-term or continuous release. If one changes any of the inputs to the ALOHA model (e.g., emissions, wind speed, wind direction, and/or atmospheric stability), then the downwind concentrations estimated by ALOHA will change. A Gaussian Plume model uses one set of meteorological inputs at a time -- generally estimated or measured at the source location. As one moves further and further away from the source and if the meteorological conditions change significantly, then the Gaussian Plume model will become less and less accurate.
The CAMEO/ALOHA system also has additional features, like simulation of dense gas physics, that can be very important for short-range transport away from the source, especially within the first kilometer. The CAMEO/ALOHA system also has very useful emissions-estimating algorithms for chemical accidents. HYSPLIT does not. So, even if one is using HYSPLIT, use of the CAMEO/ALOHA system to estimate the emissions going into the HYSPLIT model can be very useful.
Like HYSPLIT, ALOHA's accuracy is dependent on the accuracy of the meteorological data used to drive it. With both models, for example, if the input wind direction used is going in the wrong direction, the models will predict the plume going in that wrong direction.
Gaussian plume models -- like ALOHA -- are generally considered to be useful only out to about 50 km away from the source, at most. And within that 50 km range of applicability , the accuracy may decrease as one gets further and further away from the source. Beyond 50 km, one definitely would not want to use a Gaussian Plume model, but would want to use a model like HYSPLIT that has the ability to account for changing meteorological conditions as one moves further and further away from the source.
In the 0-10 and even the 10 - 50 km ranges downwind, I think it would generally be very useful to consider outputs from both models. The differences in simulation results between the two models can give you useful information about the level of uncertainty in the results.
In the NWS document (NWS 10-518) you referenced -- current version (as of Feb 28, 2020) at https://www.nws.noaa.gov/directives/sym ... 18curr.pdf -- I found the following, and I have added some comments in {brackets} building on the ideas discussed above:
(1) page A-6: HYSPLIT should not be used at distances less than 1 km from the source. {This was primarily based on the leap-frogging issue discussed above, related to the 1-minute minimum time step in HYSPLIT. This issue has been fixed, and so, the primary reason for this caution is no longer applicable.}
(2) page 4: "For small scale (0-10 km), short duration (less than one hour) releases, many local jurisdictions including emergency responders use ALOHA (part of the Computer-Aided Management of Emergency Operations (CAMEO) software suite) to forecast the dispersion impact." {This doesn't mean that HYSPLIT is not potentially useful within this range.}
(3) page 5: "In the event of a large atmospheric release (horizontal extent usually >10 km lasting more than an hour) of some chemical, biological or radioactive contaminant, an emergency manager may call the WFO and request atmospheric transport and dispersion information." {This doesn't mean that ALOHA is not potentially useful at distances greater than 10 km.}
Last edited by MarkCohen on March 3rd, 2020, 10:28 am, edited 1 time in total.
Reason: Had not realized that the leap-frogging issue had been addressed.
Reason: Had not realized that the leap-frogging issue had been addressed.