-
Notifications
You must be signed in to change notification settings - Fork 0
/
Part1-4_CDL_S2_Data_Quality_Checks.py
193 lines (133 loc) · 6.67 KB
/
Part1-4_CDL_S2_Data_Quality_Checks.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
# Databricks notebook source
from pyspark.sql.types import *
import pyspark.sql.functions as F
from datetime import datetime, date, timedelta
import plotly.graph_objs as go
# COMMAND ----------
# MAGIC %md #Examine data after processing sentinel-2 (Part-2 Script)
# COMMAND ----------
'''
Note that the final data processing that converts data into rows of time series data for each pixel/year will contain the same data as the Part-2 script, but simply be in a format that directly supports modeling
'''
# COMMAND ----------
df_s2_sampled = spark.read.parquet('dbfs:/FileStore/s2_sampled/s2_dense_test.parquet')
# COMMAND ----------
# for single scene per pixel
def ndvi_calc(b04, b08):
if b08 + b04 == 0:
return None
ndvi = (b08 - b04) / (b08 + b04)
return float(ndvi)
# Register the NDVI UDF
ndvi_udf = F.udf(ndvi_calc, FloatType())
# COMMAND ----------
display(df_s2_sampled)
# COMMAND ----------
display(df_s2_sampled.where(((F.col('lon') == -90.5923870410212) & (F.col('lat')==35.57624322094004) & (F.col('CDL')=='Cotton') & (F.col('year')==2019))).withColumn("NDVI", ndvi_udf(df_s2_sampled["red"], df_s2_sampled["nir"])))
# COMMAND ----------
display(df_s2_sampled.where(((F.col('lon') == -90.57506989498134) & (F.col('lat')==35.575181872687295) & (F.col('CDL')=='Rice') & (F.col('year')==2019))).withColumn("NDVI", ndvi_udf(df_s2_sampled["red"], df_s2_sampled["nir"])))
# COMMAND ----------
display(df_s2_sampled.where(((F.col('lon') == -90.55773411916678) & (F.col('lat')==35.57438554009128) & (F.col('CDL')=='Dbl Crop WinWht/Soybeans') & (F.col('year')==2019))).withColumn("NDVI", ndvi_udf(df_s2_sampled["red"], df_s2_sampled["nir"])))
# COMMAND ----------
display(df_s2_sampled.where(((F.col('lon') == -90.57745082006127) & (F.col('lat')==35.56052757960141) & (F.col('CDL')=='Soybeans') & (F.col('year')==2019))).withColumn("NDVI", ndvi_udf(df_s2_sampled["red"], df_s2_sampled["nir"])))
# COMMAND ----------
display(df_s2_sampled.where(((F.col('lon') == -90.58575725883986) & (F.col('lat')==35.57540310857804) & (F.col('CDL')=='Corn') & (F.col('year')==2019))).withColumn("NDVI", ndvi_udf(df_s2_sampled["red"], df_s2_sampled["nir"])))
# COMMAND ----------
# MAGIC %md #Examine Final form data
# COMMAND ----------
final_s2_df_uri = 'dbfs:/FileStore/s2_sampled/s2_dense_test_final.parquet'
encoded_df = spark.read.parquet(final_s2_df_uri)
# COMMAND ----------
## Decode the encoded columns to verify all worked....
def decode_bands(bands_bytes):
ints = [int.from_bytes(bands_bytes[i:i+2], 'big') for i in range(0, len(bands_bytes), 2)]
return ",".join(str(i) for i in ints)
decode_bands_udf = F.udf(decode_bands, StringType())
decoded_df = encoded_df.withColumn('decoded_bands', decode_bands_udf(F.col('bands')))
def decode_tiles(tiles_bytes):
return tiles_bytes.decode('UTF-8')
decode_tiles_udf = F.udf(decode_tiles, StringType())
decoded_df = decoded_df.withColumn('decoded_tiles', decode_tiles_udf(F.col('tiles')))
def decode_img_dates(img_dates_bytes):
ints = [int.from_bytes(img_dates_bytes[i:i+2], 'big') for i in range(0, len(img_dates_bytes), 2)]
return ",".join((date(1970, 1, 1) + timedelta(i)).strftime('%Y-%m-%d') for i in ints)
decode_img_dates_udf = F.udf(decode_img_dates, StringType())
decoded_df = decoded_df.withColumn('decoded_img_dates', decode_img_dates_udf(F.col('img_dates')))
def decode_scl_vals(scl_vals_bytes):
ints = [int.from_bytes(scl_vals_bytes[i:i+1], 'big') for i in range(0, len(scl_vals_bytes), 1)]
return ",".join(str(i) for i in ints)
decode_scl_vals_udf = F.udf(decode_scl_vals, StringType())
decoded_df = decoded_df.withColumn('decoded_scl_vals', decode_scl_vals_udf(F.col('scl_vals')))
# def decode_string_utf8(binary_data):
# return binary_data.decode('UTF-8')
# decode_string_utf8_udf = udf(decode_string_utf8, StringType())
# decoded_df = decoded_df.withColumn('decoded_CDL', decode_string_utf8_udf(F.col('CDL')))
# COMMAND ----------
# for multiple scenes per pixel
def calculate_ndvi(decoded_bands):
bands_list = decoded_bands.split(',')
# Assuming B8 is the NIR band and B4 is the Red band
B8 = [float(bands_list[i]) for i in range(7, len(bands_list), 12)]
B4 = [float(bands_list[i]) for i in range(3, len(bands_list), 12)]
ndvi = [(nir - red) / (nir + red) for nir, red in zip(B8, B4) if (nir + red) != 0]
return ",".join(str(i) for i in ndvi)
calculate_ndvi_udf = F.udf(calculate_ndvi, StringType())
decoded_df = decoded_df.withColumn("ndvi", calculate_ndvi_udf(F.col("decoded_bands")))
# COMMAND ----------
display(decoded_df)
# COMMAND ----------
# row_idx = 1
# row = decoded_df.limit(row_idx+1).collect()[row_idx]
# row = decoded_df.where(((F.col('lon') == -91.73448361213347) & (F.col('lat')==34.00006583641735) & (F.col('year')==2021))).first()
# row = decoded_df.where(((F.col('lon') == -90.28821275706181) & (F.col('lat')==36.34484935337164) & (F.col('year')==2021))).first()
# row = decoded_df.where(((F.col('lon') == -90.64483039822385) & (F.col('lat')==36.46194643796142) & (F.col('CDL')=='Corn'))).first()
row = decoded_df.where(((F.col('CDL') == 'Corn'))).sample(False, .01).first()
print(row['lat'],',',row['lon'], ',', row['year'])
dates = row["decoded_img_dates"].split(',')
ndvi_values = [float(v) for v in row["ndvi"].split(',')]
scl_values = [int(v) for v in row["decoded_scl_vals"].split(',')]
# Convert SCL numbers to string equivalents for the plot
SCL_str_mappings = {
0: "No Data",
1: "Saturated or defective pixel",
2: "Topographic casted shadows",
3: "Cloud shadows",
4: "Vegetation",
5: "Not-vegetated",
6: "Water",
7: "Unclassified",
8: "Cloud medium probability",
9: "Cloud high probability",
10: "Thin cirrus",
11: "Snow or ice",
}
SCL_color_mappings = {
0: "black",
1: "red",
2: "rgb(50, 50, 50)", # very dark grey
3: "saddlebrown", # dark brown
4: "green",
5: "darkgoldenrod", # dark yellow
6: "blue",
7: "darkgrey",
8: "grey",
9: "white",
10: "rgb(0, 191, 255)", # very bright blue
11: "rgb(255, 20, 147)", # very bright pink
}
colorscale = [
[i / (len(SCL_color_mappings) - 1), color] for i, color in enumerate(SCL_color_mappings.values())
]
scl_strings = [SCL_str_mappings[v] for v in scl_values]
# trace = go.Scatter(x=dates, y=ndvi_values, mode="markers", text=scl_strings, marker=dict(color=scl_values, colorscale="Viridis"))
trace = go.Scatter(
x=dates, y=ndvi_values, mode="markers", text=scl_strings,
marker=dict(color=scl_values, colorscale=colorscale, showscale=False, cmin=0, cmax=len(SCL_color_mappings)-1)
)
layout = go.Layout(
title="NDVI Time Series",
xaxis=dict(title="Date"),
yaxis=dict(title="NDVI"),
)
fig = go.Figure(data=[trace], layout=layout)
fig.show()