Skip to main content

Table 2 Mean absolute errors (MAE) of the polar motion (PMX, PMY) prediction [mas] in 2021, using the LS + AR, LSTM, LS + LSTM, and Bulletin A time series from IERS, respectively

From: Improving the accuracy of polar motion prediction using a hybrid least squares and long short-term memory model

No.

PMX

PMY

LS + AR

LSTM

LS + LSTM

Bulletin A

PCT

LS + AR

LSTM

LS + LSTM

Bulletin A

PCT

1

0.257

0.676

0.284

0.269

− 5.6%

0.197

0.466

0.204

0.196

− 4.2%

5

1.603

1.898

1.690

1.563

− 8.1%

1.099

1.108

1.129

1.073

− 5.2%

10

3.046

3.289

3.152

2.892

− 9.0%

1.940

1.812

2.009

1.843

− 9.0%

15

4.593

4.531

4.374

3.988

− 9.7%

2.751

2.598

2.729

2.480

− 10.0%

20

5.751

5.793

5.546

5.033

− 10.2%

3.607

3.509

3.386

3.077

− 10.1%

30

7.995

8.362

7.414

6.809

− 12.1%

5.407

5.501

4.876

4.261

− 14.4%

45

11.483

12.561

10.995

9.227

− 19.2%

8.292

8.879

7.137

5.904

− 20.9%

60

14.915

17.281

13.778

11.260

− 22.4%

11.382

12.764

9.426

7.454

− 26.5%

90

21.950

25.895

15.419

14.619

− 5.5%

17.785

21.144

12.868

11.176

− 15.1%

120

27.506

32.397

15.685

16.819

6.7%

23.904

28.571

14.905

15.500

3.8%

180

32.933

35.620

15.904

19.148

16.9%

31.231

35.412

16.424

22.356

26.5%

270

33.364

27.344

15.926

22.318

28.6%

29.096

27.771

17.218

25.066

31.3%

320

36.116

29.332

16.393

24.096

32.0%

30.430

27.190

16.350

23.980

31.8%

365

37.491

33.456

17.086

25.770

33.7%

31.949

30.263

17.100

25.107

31.9%

  1. The improvement represents the prediction accuracy of LS + LSTM relative to Bulletin A