2  Wind buffers - FMS approximation

Estimating wind buffers of FMS tracks, using the JTWC wind radii, since FMS doesn’t provide wind radii.

Trying first using regression.

Code 
%load_ext jupyter_black
%load_ext autoreload
%autoreload 2
Code 
import ocha_stratus as stratus
import pandas as pd
import geopandas as gpd
import numpy as np
import matplotlib.pyplot as plt
import statsmodels.api as sm

from src.constants import *
from src.blob import PROJECT_PREFIX
from src.datasources.ibtracs import expand_quad_col
Code 
query = """
SELECT *
FROM storms.ibtracs_storms
WHERE genesis_basin = 'SP'
"""
with stratus.get_engine(stage="dev").connect() as conn:
    df_storms = pd.read_sql(query, conn)
Code 
query = """
SELECT *
FROM storms.ibtracs_tracks_geo
WHERE basin = 'SP'
"""
with stratus.get_engine(stage="dev").connect() as conn:
    gdf_tracks = gpd.read_postgis(query, conn, geom_col="geometry")
Code 
gdf_tracks = gdf_tracks.merge(df_storms)
Code 
blob_name = (
    f"{PROJECT_PREFIX}/processed/ibtracs/ibtracs_with_usa_radii.parquet"
)
df_tracks_usa = stratus.load_parquet_from_blob(blob_name)
Code 
df_tracks_usa
sid provider valid_time basin nature usa_agency wind_speed gust_speed pressure max_wind_radius last_closed_isobar_radius last_closed_isobar_pressure quadrant_radius_34 quadrant_radius_50 quadrant_radius_64 point_id storm_id longitude latitude
0 1947017S10168 bom 1947-01-20 18:00:00 SP TS jtwc_sh <NA> <NA> <NA> <NA> <NA> <NA> [nan, nan, nan, nan] [nan, nan, nan, nan] [nan, nan, nan, nan] 1f6b08ed-76ce-40dc-802c-de7907891943 None 160.3 -16.2
1 1948011S16140 bom 1948-01-13 18:00:00 SP TS jtwc_sh <NA> <NA> <NA> <NA> <NA> <NA> [nan, nan, nan, nan] [nan, nan, nan, nan] [nan, nan, nan, nan] 8773d4fc-6b39-42ec-8567-f965b981366c None 141.6 -17.9
2 1948011S20157 bom 1948-01-10 12:00:00 SP TS jtwc_sh <NA> <NA> <NA> <NA> <NA> <NA> [nan, nan, nan, nan] [nan, nan, nan, nan] [nan, nan, nan, nan] c633207a-11fb-4896-bae2-e9fdee29ee81 None 157.4 -20.0
3 1948011S20157 bom 1948-01-10 18:00:00 SP TS jtwc_sh <NA> <NA> <NA> <NA> <NA> <NA> [nan, nan, nan, nan] [nan, nan, nan, nan] [nan, nan, nan, nan] dab7824b-3321-4bb1-83f7-fd0c77281e4e None 157.8 -20.0
4 1948011S20157 bom 1948-01-11 00:00:00 SP TS jtwc_sh <NA> <NA> <NA> <NA> <NA> <NA> [nan, nan, nan, nan] [nan, nan, nan, nan] [nan, nan, nan, nan] 1d73d224-9fc5-4fc0-9044-603dfb1695c6 None 158.2 -20.0
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
10827 2025129S08138 bom 2025-05-11 12:00:00 SP TS tcvitals 35 49 1000 22 119 1008 [61.0, 59.0, 26.0, 26.0] [nan, nan, nan, nan] [nan, nan, nan, nan] 4e01d3d8-e391-461e-a268-13633a400425 None 138.1 -8.8
10828 2025129S08138 bom 2025-05-11 18:00:00 SP TS tcvitals 35 49 1000 26 119 1008 [26.0, 59.0, 26.0, 17.0] [nan, nan, nan, nan] [nan, nan, nan, nan] 92344e31-0339-4b86-bf42-a38e784b9bb8 None 137.9 -8.6
10829 2025129S08138 bom 2025-05-12 00:00:00 SP TS tcvitals 30 45 1002 26 119 1010 [nan, nan, nan, nan] [nan, nan, nan, nan] [nan, nan, nan, nan] 776da658-9698-4301-b9c8-79c78ac33d0d None 137.8 -8.5
10830 2025129S08138 bom 2025-05-12 06:00:00 SP TS tcvitals 30 45 1002 26 150 1008 [nan, nan, nan, nan] [nan, nan, nan, nan] [nan, nan, nan, nan] 954e2a0e-fe97-4ef2-898b-f6e1c64a2ec9 None 137.6 -8.3
10831 2025129S08138 bom 2025-05-12 12:00:00 SP TS tcvitals 15 34 1004 52 119 1010 [nan, nan, nan, nan] [nan, nan, nan, nan] [nan, nan, nan, nan] bf1b2ba0-5d5c-443e-8c2b-8fdbfcae39a3 None 137.6 -7.9

10832 rows × 19 columns

Code 
buffer_speeds = [34, 50, 64]
for buffer_speed in buffer_speeds:
    df_tracks_usa = expand_quad_col(
        df_tracks_usa, f"quadrant_radius_{buffer_speed}"
    )
already done for quadrant_radius_34
already done for quadrant_radius_50
already done for quadrant_radius_64
Code 
gdf_tracks_best = gdf_tracks[~gdf_tracks["provisional"]]
Code 
gdf_tracks_best
wind_speed pressure max_wind_radius last_closed_isobar_radius last_closed_isobar_pressure gust_speed sid provider basin nature ... quadrant_radius_64 point_id storm_id geometry atcf_id number season name genesis_basin provisional
120 NaN 994.0 NaN NaN NaN NaN 1907018S13147 bom SP NR ... {NaN,NaN,NaN,NaN} dca3956d-b64f-4775-b43c-2d6636ae4e4a None POINT (146.5 -13) None 4 1907 None SP False
121 NaN 993.0 NaN NaN NaN NaN 1907018S13147 bom SP TS ... {NaN,NaN,NaN,NaN} c6d125e8-2707-4922-a14f-8d48faa6d155 None POINT (145 -15) None 4 1907 None SP False
122 NaN 993.0 NaN NaN NaN NaN 1907018S13147 bom SP TS ... {NaN,NaN,NaN,NaN} 302e2c59-3f41-4439-8ce0-b6063c6aaa16 None POINT (143 -14) None 4 1907 None SP False
123 NaN 993.0 NaN NaN NaN NaN 1907018S13147 bom SP TS ... {NaN,NaN,NaN,NaN} 029dda7b-7ab8-44ad-a268-e8655ea4a003 None POINT (142 -13.5) None 4 1907 None SP False
124 NaN 998.0 NaN NaN NaN NaN 1907018S13147 bom SP TS ... {NaN,NaN,NaN,NaN} 2929807f-457c-4858-80bc-107e1b4772ea None POINT (141.5 -15.6) None 4 1907 None SP False
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
20850 35.0 1000.0 NaN 119.0 1008.0 49.0 2025129S08138 bom SP TS ... {NaN,NaN,NaN,NaN} 780bbeba-8179-46e5-bb1e-904cf7b36c05 None POINT (138.1 -8.8) SH322025 27 2025 None SP False
20851 35.0 1000.0 NaN 119.0 1008.0 49.0 2025129S08138 bom SP TS ... {NaN,NaN,NaN,NaN} 37cae45d-f8a7-491c-9d5d-edd06a2dc2ea None POINT (137.9 -8.6) SH322025 27 2025 None SP False
20852 30.0 1002.0 NaN 119.0 1010.0 45.0 2025129S08138 bom SP TS ... {NaN,NaN,NaN,NaN} d0521216-7f5c-45e9-bef4-00fc97285ee7 None POINT (137.8 -8.5) SH322025 27 2025 None SP False
20853 30.0 1002.0 NaN 150.0 1008.0 45.0 2025129S08138 bom SP TS ... {NaN,NaN,NaN,NaN} c0997f20-6cbc-4e88-823c-b979a0282dac None POINT (137.6 -8.3) SH322025 27 2025 None SP False
20854 15.0 1004.0 NaN 119.0 1010.0 34.0 2025129S08138 bom SP TS ... {NaN,NaN,NaN,NaN} a34c6340-9e66-4a46-96bd-422a1068b485 None POINT (137.6 -7.9) SH322025 27 2025 None SP False

20071 rows × 23 columns

Code 
sorted(gdf_tracks_best["season"].unique())
[1907,
 1908,
 1909,
 1910,
 1911,
 1912,
 1913,
 1914,
 1915,
 1916,
 1917,
 1918,
 1919,
 1920,
 1921,
 1922,
 1923,
 1924,
 1925,
 1926,
 1927,
 1928,
 1929,
 1930,
 1931,
 1932,
 1933,
 1934,
 1935,
 1936,
 1937,
 1938,
 1939,
 1940,
 1941,
 1942,
 1943,
 1944,
 1945,
 1946,
 1947,
 1948,
 1949,
 1950,
 1951,
 1952,
 1953,
 1954,
 1955,
 1956,
 1957,
 1958,
 1959,
 1960,
 1961,
 1962,
 1963,
 1964,
 1965,
 1966,
 1967,
 1968,
 1969,
 1970,
 1971,
 1972,
 1973,
 1974,
 1975,
 1976,
 1977,
 1978,
 1979,
 1980,
 1981,
 1982,
 1983,
 1984,
 1985,
 1986,
 1987,
 1988,
 1989,
 1990,
 1991,
 1992,
 1993,
 1994,
 1995,
 1996,
 1997,
 1998,
 1999,
 2000,
 2001,
 2002,
 2003,
 2004,
 2005,
 2006,
 2007,
 2008,
 2009,
 2010,
 2011,
 2012,
 2013,
 2014,
 2015,
 2016,
 2017,
 2018,
 2019,
 2020,
 2021,
 2022,
 2023,
 2024,
 2025]
Code 
RADIUS_COL = "quadrant_radius_{speed}_{quad}"
Code 
quads = ["ne", "nw", "se", "sw"]
Code 
df_tracks_merge = gdf_tracks_best.merge(
    df_tracks_usa, on=["sid", "valid_time"], suffixes=("_wmo", "_usa")
)
Code 
df_tracks_merge.columns
Index(['wind_speed_wmo', 'pressure_wmo', 'max_wind_radius_wmo',
       'last_closed_isobar_radius_wmo', 'last_closed_isobar_pressure_wmo',
       'gust_speed_wmo', 'sid', 'provider_wmo', 'basin_wmo', 'nature_wmo',
       'valid_time', 'quadrant_radius_34_wmo', 'quadrant_radius_50_wmo',
       'quadrant_radius_64_wmo', 'point_id_wmo', 'storm_id_wmo', 'geometry',
       'atcf_id', 'number', 'season', 'name', 'genesis_basin', 'provisional',
       'provider_usa', 'basin_usa', 'nature_usa', 'usa_agency',
       'wind_speed_usa', 'gust_speed_usa', 'pressure_usa',
       'max_wind_radius_usa', 'last_closed_isobar_radius_usa',
       'last_closed_isobar_pressure_usa', 'quadrant_radius_34_usa',
       'quadrant_radius_50_usa', 'quadrant_radius_64_usa', 'point_id_usa',
       'storm_id_usa', 'longitude', 'latitude', 'quadrant_radius_34_ne',
       'quadrant_radius_34_nw', 'quadrant_radius_34_se',
       'quadrant_radius_34_sw', 'quadrant_radius_50_ne',
       'quadrant_radius_50_nw', 'quadrant_radius_50_se',
       'quadrant_radius_50_sw', 'quadrant_radius_64_ne',
       'quadrant_radius_64_nw', 'quadrant_radius_64_se',
       'quadrant_radius_64_sw', 'wind_diff'],
      dtype='object')
Code 
for speed in buffer_speeds:
    cols = [RADIUS_COL.format(speed=speed, quad=x) for x in quads]
    df_tracks_merge[
        RADIUS_COL.format(speed=speed, quad="mean")
    ] = df_tracks_merge[cols].mean(axis=1)
Code 
df_tracks_merge
wind_speed_wmo pressure_wmo max_wind_radius_wmo last_closed_isobar_radius_wmo last_closed_isobar_pressure_wmo gust_speed_wmo sid provider_wmo basin_wmo nature_wmo ... quadrant_radius_50_se quadrant_radius_50_sw quadrant_radius_64_ne quadrant_radius_64_nw quadrant_radius_64_se quadrant_radius_64_sw wind_diff quadrant_radius_34_mean quadrant_radius_50_mean quadrant_radius_64_mean
0 NaN NaN NaN NaN NaN NaN 1947017S10168 bom SP TS ... NaN NaN NaN NaN NaN NaN <NA> NaN NaN NaN
1 NaN NaN NaN NaN NaN NaN 1948011S16140 bom SP TS ... NaN NaN NaN NaN NaN NaN <NA> NaN NaN NaN
2 NaN NaN NaN NaN NaN NaN 1948011S20157 bom SP TS ... NaN NaN NaN NaN NaN NaN <NA> NaN NaN NaN
3 NaN NaN NaN NaN NaN NaN 1948011S20157 bom SP TS ... NaN NaN NaN NaN NaN NaN <NA> NaN NaN NaN
4 NaN NaN NaN NaN NaN NaN 1948011S20157 bom SP TS ... NaN NaN NaN NaN NaN NaN <NA> NaN NaN NaN
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
10907 35.0 1000.0 NaN 119.0 1008.0 49.0 2025129S08138 bom SP TS ... NaN NaN NaN NaN NaN NaN 0.0 43.0 NaN NaN
10908 35.0 1000.0 NaN 119.0 1008.0 49.0 2025129S08138 bom SP TS ... NaN NaN NaN NaN NaN NaN 0.0 32.0 NaN NaN
10909 30.0 1002.0 NaN 119.0 1010.0 45.0 2025129S08138 bom SP TS ... NaN NaN NaN NaN NaN NaN 0.0 NaN NaN NaN
10910 30.0 1002.0 NaN 150.0 1008.0 45.0 2025129S08138 bom SP TS ... NaN NaN NaN NaN NaN NaN 0.0 NaN NaN NaN
10911 15.0 1004.0 NaN 119.0 1010.0 34.0 2025129S08138 bom SP TS ... NaN NaN NaN NaN NaN NaN 0.0 NaN NaN NaN

10912 rows × 56 columns

Code 
def calc_r2(y_pred, y_true, k):
    n = len(y_true)
    ss_res = np.sum((y_true - y_pred) ** 2)
    ss_tot = np.sum((y_true - np.mean(y_true)) ** 2)
    r2 = 1 - ss_res / ss_tot
    r2_adj = 1 - (1 - r2) * (n - 1) / (n - k - 1)
    return r2, r2_adj
Code 
col_sets = (
    ("wind", ["wind_speed_wmo"]),
    ("wind_pres", ["wind_speed_wmo", "pressure_wmo"]),
)
dicts = []
df_reg = df_tracks_merge.dropna(subset=["wind_speed_wmo", "pressure_wmo"])
df_reg = df_reg[df_reg["wind_speed_wmo"] >= 0]
conversion_factor = 1
for speed in buffer_speeds:
    cutoff_speed = speed * conversion_factor
    for quad in quads + ["mean"]:
        target_col = RADIUS_COL.format(speed=speed, quad=quad)
        df_reg[target_col] = df_reg[target_col].fillna(0)
        for cutoff in [True, False]:
            if cutoff:
                df_model = df_reg[df_reg["wind_speed_wmo"] >= cutoff_speed]
            else:
                df_model = df_reg
            for col_set in col_sets:
                var_cols = col_set[1]
                X = df_model[var_cols]
                y = df_model[target_col]
                X = sm.add_constant(X)
                model = sm.OLS(y, X).fit()

                X_pred = df_reg[var_cols]
                X_pred = sm.add_constant(X_pred)
                df_reg["pred"] = model.predict(X_pred)
                if cutoff:
                    df_reg["pred"] = df_reg.apply(
                        lambda row: 0
                        if row["wind_speed_wmo"] <= cutoff
                        else row["pred"],
                        axis=1,
                    )
                df_reg["pred"] = df_reg["pred"].apply(
                    lambda x: 0 if x < 0 else x
                )
                r2, r2_adj = calc_r2(
                    df_reg["pred"], df_reg[target_col], k=len(var_cols)
                )
                dicts.append(
                    {
                        "col_set": col_set[0],
                        "cutoff": cutoff,
                        "speed": speed,
                        "quad": quad,
                        "r2": r2,
                        "r2_adj": r2_adj,
                        "summary": model.summary(),
                        "params": model.params,
                    }
                )
        # for plot_col in ["wind_speed_wmo", "pressure_wmo"]:
        #     df_reg.plot.scatter(
        #         x=plot_col,
        #         y=target_col,
        #         alpha=0.1,
        #     )
Code 
df_reg[["wind_speed_wmo", "pressure_wmo"]].corr()
wind_speed_wmo pressure_wmo
wind_speed_wmo 1.000000 -0.967436
pressure_wmo -0.967436 1.000000
Code 
df_r2 = pd.DataFrame(dicts)
Code 
df_r2[df_r2["quad"] == "mean"]
col_set cutoff speed quad r2 r2_adj summary params
16 wind True 34 mean 0.124057 0.123973 OLS Regression ... const 0.330641 wind_speed_wmo 0...
17 wind_pres True 34 mean 0.130680 0.130513 OLS Regression ... const -482.626782 wind_speed_wmo ...
18 wind False 34 mean 0.134587 0.134503 OLS Regression ... const -9.683708 wind_speed_wmo 0...
19 wind_pres False 34 mean 0.136667 0.136501 OLS Regression ... const -450.050374 wind_speed_wmo ...
36 wind True 50 mean 0.216839 0.216764 OLS Regression ... const -9.616171 wind_speed_wmo 0...
37 wind_pres True 50 mean 0.247978 0.247833 OLS Regression ... const -829.599372 wind_speed_wmo ...
38 wind False 50 mean 0.223813 0.223739 OLS Regression ... const -11.52552 wind_speed_wmo ...
39 wind_pres False 50 mean 0.234731 0.234584 OLS Regression ... const -328.345480 wind_speed_wmo ...
56 wind True 64 mean 0.278334 0.278264 OLS Regression ... const -10.795898 wind_speed_wmo ...
57 wind_pres True 64 mean 0.326307 0.326178 OLS Regression ... const -628.360791 wind_speed_wmo ...
58 wind False 64 mean 0.258214 0.258143 OLS Regression ... const -6.618191 wind_speed_wmo 0...
59 wind_pres False 64 mean 0.262354 0.262212 OLS Regression ... const -73.380294 wind_speed_wmo ...
Code 
df_r2[
    (df_r2["col_set"] == "wind_pres")
    & df_r2["cutoff"]
    & (df_r2["quad"] == "mean")
]["summary"].apply(display)
OLS Regression Results
Dep. Variable: quadrant_radius_34_mean R-squared: 0.061
Model: OLS Adj. R-squared: 0.061
Method: Least Squares F-statistic: 224.0
Date: Mon, 29 Sep 2025 Prob (F-statistic): 5.84e-95
Time: 17:01:08 Log-Likelihood: -36349.
No. Observations: 6859 AIC: 7.270e+04
Df Residuals: 6856 BIC: 7.272e+04
Df Model: 2
Covariance Type: nonrobust
coef std err t P>|t| [0.025 0.975]
const -482.6268 127.422 -3.788 0.000 -732.413 -232.840
wind_speed_wmo 1.0338 0.116 8.934 0.000 0.807 1.261
pressure_wmo 0.4695 0.124 3.791 0.000 0.227 0.712
Omnibus: 940.345 Durbin-Watson: 0.135
Prob(Omnibus): 0.000 Jarque-Bera (JB): 1370.908
Skew: 1.078 Prob(JB): 2.05e-298
Kurtosis: 3.391 Cond. No. 2.13e+05


Notes:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 2.13e+05. This might indicate that there are
strong multicollinearity or other numerical problems.
OLS Regression Results
Dep. Variable: quadrant_radius_50_mean R-squared: 0.098
Model: OLS Adj. R-squared: 0.098
Method: Least Squares F-statistic: 225.9
Date: Mon, 29 Sep 2025 Prob (F-statistic): 7.26e-94
Time: 17:01:09 Log-Likelihood: -19030.
No. Observations: 4151 AIC: 3.807e+04
Df Residuals: 4148 BIC: 3.809e+04
Df Model: 2
Covariance Type: nonrobust
coef std err t P>|t| [0.025 0.975]
const -829.5994 72.275 -11.478 0.000 -971.296 -687.902
wind_speed_wmo 1.1480 0.072 16.049 0.000 1.008 1.288
pressure_wmo 0.7925 0.070 11.348 0.000 0.656 0.929
Omnibus: 832.942 Durbin-Watson: 0.191
Prob(Omnibus): 0.000 Jarque-Bera (JB): 1446.699
Skew: 1.307 Prob(JB): 0.00
Kurtosis: 4.236 Cond. No. 1.91e+05


Notes:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 1.91e+05. This might indicate that there are
strong multicollinearity or other numerical problems.
OLS Regression Results
Dep. Variable: quadrant_radius_64_mean R-squared: 0.138
Model: OLS Adj. R-squared: 0.137
Method: Least Squares F-statistic: 168.1
Date: Mon, 29 Sep 2025 Prob (F-statistic): 1.80e-68
Time: 17:01:10 Log-Likelihood: -8574.2
No. Observations: 2107 AIC: 1.715e+04
Df Residuals: 2104 BIC: 1.717e+04
Df Model: 2
Covariance Type: nonrobust
coef std err t P>|t| [0.025 0.975]
const -628.3608 46.989 -13.372 0.000 -720.511 -536.211
wind_speed_wmo 0.8599 0.050 17.215 0.000 0.762 0.958
pressure_wmo 0.5955 0.045 13.151 0.000 0.507 0.684
Omnibus: 291.564 Durbin-Watson: 0.261
Prob(Omnibus): 0.000 Jarque-Bera (JB): 421.367
Skew: 1.041 Prob(JB): 3.17e-92
Kurtosis: 3.681 Cond. No. 1.46e+05


Notes:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 1.46e+05. This might indicate that there are
strong multicollinearity or other numerical problems. 
17    None
37    None
57    None
Name: summary, dtype: object
Code 
df_r2.iloc[3]["summary"]
OLS Regression Results
Dep. Variable: quadrant_radius_34_ne R-squared: 0.139
Model: OLS Adj. R-squared: 0.139
Method: Least Squares F-statistic: 838.2
Date: Mon, 29 Sep 2025 Prob (F-statistic): 0.00
Time: 16:50:54 Log-Likelihood: -53939.
No. Observations: 10411 AIC: 1.079e+05
Df Residuals: 10408 BIC: 1.079e+05
Df Model: 2
Covariance Type: nonrobust
coef std err t P>|t| [0.025 0.975]
const -354.9920 92.342 -3.844 0.000 -536.000 -173.984
wind_speed_wmo 1.0720 0.077 13.916 0.000 0.921 1.223
pressure_wmo 0.3354 0.090 3.716 0.000 0.158 0.512
Omnibus: 3093.480 Durbin-Watson: 0.223
Prob(Omnibus): 0.000 Jarque-Bera (JB): 8622.195
Skew: 1.582 Prob(JB): 0.00
Kurtosis: 6.141 Cond. No. 2.16e+05


Notes:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 2.16e+05. This might indicate that there are
strong multicollinearity or other numerical problems.
Code 
r2
0.26235404579165633
Code 
WINSTON_SID
'2016041S14170'
Code 
df_tracks_merge["season"].unique()
array([1947, 1948, 1949, 1950, 1952, 1954, 1955, 1956, 1960, 1961, 1962,
       1963, 1964, 1966, 1967, 1968, 1969, 1970, 1971, 1972, 1973, 1974,
       1975, 1976, 1977, 1978, 1979, 1980, 1981, 1982, 1983, 1984, 1985,
       1986, 1987, 1988, 1989, 1990, 1991, 1992, 1993, 1994, 1995, 1996,
       1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007,
       2008, 2009, 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018,
       2019, 2020, 2021, 2022, 2024, 2023, 2025])
Code 
df_tracks_merge["wind_diff"] = (
    df_tracks_merge["wind_speed_wmo"] - df_tracks_merge["wind_speed_usa"]
)
Code 
df_tracks_merge["wind_diff"].hist()

Code 
df_tracks_merge[df_tracks_merge["wind_diff"].abs() > 10][
    ["sid", "valid_time", "wind_speed_wmo", "wind_speed_usa", "wind_diff"]
]
sid valid_time wind_speed_wmo wind_speed_usa wind_diff
166 1968094S08156 1968-04-07 12:00:00 55.0 70 -15.0
171 1968094S08156 1968-04-09 12:00:00 60.0 45 15.0
172 1968094S08156 1968-04-09 18:00:00 65.0 40 25.0
173 1968094S08156 1968-04-10 00:00:00 70.0 35 35.0
174 1968094S08156 1968-04-10 06:00:00 65.0 30 35.0
... ... ... ... ... ...
6367 2007136S10158 2007-05-20 18:00:00 30.0 15 15.0
6368 2007136S10158 2007-05-21 00:00:00 30.0 15 15.0
10394 2023036S12163 2023-02-08 12:00:00 40.0 51 -11.0
10397 2023036S12163 2023-02-09 06:00:00 50.0 64 -14.0
10731 2023316S09167 2023-11-16 00:00:00 35.0 -1 36.0

1409 rows × 5 columns

Code 
df_tracks_merge.set_index("sid").loc[WINSTON_SID][
    ["wind_speed_wmo", "wind_speed_usa"]
]
wind_speed_wmo wind_speed_usa
sid
2016041S14170 25.0 25
2016041S14170 25.0 25
2016041S14170 25.0 25
2016041S14170 30.0 30
2016041S14170 35.0 35
... ... ...
2016041S14170 40.0 40
2016041S14170 40.0 40
2016041S14170 40.0 40
2016041S14170 40.0 40
2016041S14170 40.0 40

64 rows × 2 columns