Archive for June, 2023

Improved Drift Model and Search Recommendations for MH370

Recommended Search Areas Surrounding 34.23°S, 93.79°S

After three years of work, Bobby Ulich and I are publishing a very detailed drift model for predicting the point of impact of MH370. The most likely point of impact (POI) is estimated to be along the 7th arc at 34.0°S latitude, which is around 76 km (41 NM) northeast of the last estimated position (LEP) of 34.23°S, 93.79°E from our previous study (UGIB 2020). While the current study primarily uses drift modeling, the previous study primarily used satellite, weather, and aircraft performance data. The consistency of these two studies lends support to the results. Because of uncertainties in the drift model, the present study estimates that it is possible that debris drifted from an origin along the 7th arc between 30°S and 36°S.

The study incorporates the following data sets, assumptions, and methodologies:

  • Seventeen reliable MH370 debris reports with unique locations and finding dates
  • 86,400 drift trials predicted by CSIRO using the BRAN2015 ocean model for the flaperon and for generic debris
  • Windage for generic MH370 debris varied over a range
  • A Bayesian statistic for the localization error of BRAN2015 drift tracks
  • Errors in the BRAN2015 ocean current and wind speeds
  • Minimum and maximum reporting delays of recovered debris based on barnacle encrustation
  • Maximum likelihood estimation theory
  • Processing methods for computing a PDF and estimating its uncertainty
  • Bayesian statistics for accommodating a large range of allowable transport speed of debris
  • Validation studies through non-blind, partially blind, and blind tests

The figure above shows the subsea search recommendations from the current study. The search area is subdivided into the following zones, in order of search priority:

Zone 1: Shown as the white rectangle, the area was largely searched before, but some areas were missed due to missing or low-quality data, including areas with challenging terrain. We also consider that one or more contacts previously dismissed may be individual parts from MH370 rather than the expected field of debris. Within Zone 1 is Zone 1A, shown by the red racetrack, which defines the limits of an impact before 00:21:07, which is when a log-in to the Inflight Entertainment (IFE) server was expected. Zone 1B is the part of Zone 1 that is not within Zone 1A.

Zone 2: Shown as the purple racetrack, this area extends to +/- 70 NM from the 7th Arc, and represents the likely limits to a glide after fuel exhaustion.

Zone 3: Shown as the large green racetrack, this area extends to +/- 140 NM from the 7th arc, and represents the extreme limits of a long glide after fuel exhaustion.

CSIRO shared with us two sets of 86,400 predicted trial drifter tracks using the ocean model BRAN2015. Each trial was assumed to have originated at a unique location near Arc 7, in an array with a fairly uniform areal density and within 25 km of Arc 7 between 8°S and 44°S latitude. The location (latitude, longitude) of each trial drifter was predicted by CSIRO at one-day intervals up to 1,027 days after crash. One set of 86,400 trials used the drift parameters CSIRO determined for the flaperon by sea trials with a cut-down flaperon. The other set was for non-flaperon debris and assumed 1.2% windage to account for Stoke’s Drift, which is shown in this video for all 86,400 trials:

Representative drift paths from our estimated point of impact (POI) to the 17 debris locations used in the analysis is shown in the following figure:

The paper “Improved Prediction of MH370 Crash Location Based on Drift Modeling of Floating Debris” by Bobby Ulich, Ph.D. and Victor Iannello, Sc.D., can be downloaded here.


I’d like to personally show my gratitude to Bobby Ulich, who has made this drift study a priority for three years, and was the driving force for its completion. We also gratefully acknowledge the many valuable contributions of David Griffin and by CSIRO, who made this work possible. In addition, we thank Don Thompson, Michael Exner, and Henrik Rydberg for reviewing an earlier version of the paper.

Posted in Aviation | 514 Comments »