Nonlocality of tropical cyclone activity in the climate system

This is also part of the CliSAP- 2 Research Topic B4 Regional storms and their marine impacts for which Dr. Frauke Feser (Helmholtz-Zentrum Geesthacht HZG) and Thomas Frisius are the coordinators. Results were presented in several joint meetings.

Nonlocality of tropical cyclone activity refers to the observation that the frequency of tropical cyclones depend rather on relative SST  (local minus average tropical SST) than on local SST. We have been studied the mechanism for the first time in idealized model simulations. The model set up with two continents and two interjacent oceans has been used for experiments with the atmospheric climate model PlaSim at a spectral resolution T170. In the reference experiment REF SST profiles of the two ocean basins are identical while for the experiment WARMBASIN (COLDBASIN) the northern hemisphere basin of the eastern ocean includes a warm (cold) SST anomaly up to 2.5K. The sensitivity experiments WARMBASIN and COLDBASIN reveal a significant nonlocality effect, i.e. the tropcial cyclone activity in the unmodified basins changes contrary to the SST anomaly of the modified basins (see the figure). The model code and a description how to repeat these experiments will be provided below.

We have extended these preliminary idealized experiments by using realistic continents and found a similar response by heating of the Pacific SST. Furthermore, we want to analyse the sensitivity of TC activity to land surface properties. Two master students (Abubakar Abdullah and Kenneth Ayadiani) work on these issues.

Participating researchers: Thomas Frisius, Daria Schönemann, Abubakar Abdullah (master student) and Kenneth Ayadiani (master student)


  • Frisius, T., and S.M.A. Abdullah, 2017: Nonlocality of tropical cyclone activity in idealized climate simulations. Submitted to JAMES.

Figure: Detected hurricane tracks in the experiments a) REF, b) WARMBASIN and c) COLDBASIN. The color indicates intensity according to the Saffir-Simpson scale (see legend) and continents are masked with grey color.

Description of the Planet Simulator experiments

An idealized land-sea distribution composed of two continents and two oceans (see Figure) is used to simulate nonlocality effects in tropical cyclone activity. The two continents extend meridionally from 90S to 90N while zonally they reach from 180W to 90W and 0E to 90E, respectively. The experiments do neither include an annual cycle nor a diurnal cycle and the SST of the two-ocean basins is prescribed with a time independent field.

The model starts from rest for a 10 year simulation at the low resolution T42 for each individual model experiment. This time period proved to be sufficient for the establishment of a quasi-steady state climate. The final state of this spin-up run forms the initial state of a five-year simulation at the high resolution T170. The outcome of the high resolution run forms the basis of the analysis. The five experiments REF, WARMBASIN, COLDBASIN, GLOBALWARMING and GLOBALCOOLING have been performed. WARMBASIN and COLDBASIN includie a positive and a negative SST anomaly of 2.5 K in the northeastern basin (see Figure), respectively. In the experiments GLOBALWARMING and GLOBALCOOLING the SST has been uniformly increased and decreased by 2.5K, respectively.

The source code of PlaSim including files for initialization and boundary conditions for each experiment can be downloaded by following the links:

The textfile guidance.txt contains a short guidance for the implementation of the experiments.

Hurricane tracks

Hurricane tracks have been computed from the model output. For identification the following rules have been applied:

  • The streamfunction value at the local minimum (maximum) must undershoot (exceed) the average values of the surrounding grid points within a certain search radius 5° by the increment 1×106 m2/s.
  • The maximum horizontal wind speed at 1000hPa within the search radius 5° is referred to as intensity and must exceed the threshold 12m/s.
  • The magnitude of horizontal wind at 300hPa averaged over the grid points within the search radius 5° (upper layer wind) must be below the threshold 20m/s.
  • The convective precipitation rate averaged over the grid points within the search radius 5° must exceed the threshold 0.36 mm/h.

The identified convective cyclones are connected to individual cyclone tracks. For this purpose it is necessary to compare the identified cyclone positions of adjacent 6 hourly output time steps. The nearest neighbors continue the cyclone track if their distance below the tracking radius 5°. A filter removes the cyclone tracks that cannot be classified as hurricane tracks. The filter criteria are:

  • The intensity must exceed the threshold 25m/s at one or more locations along the cyclone track
  • The upper layer wind must be below the threshold 8m/s at the location where the cyclone track begins.
  • The cyclone track must begin at a latitude having an absolute value below 30°.
  • At the time of maximum intensity a local maximum of 300 hPa temperature must exist within the search radius 2.5° having the and exceed the average 300 hPa temperature of the grid points within the search radius 5° by the difference 1K.
  • The cyclone track must at least extend over a time period of 24 hrs .

 Hurricane tracks of the various experiments (ASCII):