First version of model

trained on small part of the land , west of Bishkek

meer/Meer.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f8b1fa84aa028104253b6aee5f2e166c986fc79c88c75e694aec2bfb463e3846
+size 75896

meer/mean_values_summary.csv

@@ -0,0 +1,193 @@
+Year,Month,Type,Mean_Value
+2014,01,NDVI,0.03637381
+2014,02,NDVI,-7.648369e-05
+2014,03,NDVI,-0.0102609545
+2014,04,NDVI,0.21833025
+2014,05,NDVI,0.4029976
+2014,06,NDVI,0.39842525
+2014,07,NDVI,0.33484557
+2014,08,NDVI,0.28024414
+2014,09,NDVI,0.2678122
+2014,10,NDVI,0.22297126
+2014,11,NDVI,0.20621522
+2014,12,NDVI,0.018370138
+2014,01,NDWI,-0.07105442
+2014,02,NDWI,-0.026667532
+2014,03,NDWI,-0.0108920885
+2014,04,NDWI,-0.25271535
+2014,05,NDWI,-0.3947968
+2014,06,NDWI,-0.41528398
+2014,07,NDWI,-0.38426134
+2014,08,NDWI,-0.34330943
+2014,09,NDWI,-0.33599624
+2014,10,NDWI,-0.26565516
+2014,11,NDWI,-0.19662789
+2014,12,NDWI,-0.0534919
+2015,01,NDVI,0.015824856
+2015,02,NDVI,0.01410047
+2015,03,NDVI,0.21014738
+2015,04,NDVI,0.38612852
+2015,05,NDVI,0.52105993
+2015,06,NDVI,0.37829387
+2015,07,NDVI,0.31312978
+2015,08,NDVI,0.31475097
+2015,09,NDVI,0.3148233
+2015,10,NDVI,0.28065836
+2015,11,NDVI,0.20766294
+2015,12,NDVI,0.023235604
+2015,01,NDWI,-0.04799539
+2015,02,NDWI,-0.043714006
+2015,03,NDWI,-0.2247919
+2015,04,NDWI,-0.3677456
+2015,05,NDWI,-0.47811177
+2015,06,NDWI,-0.40399328
+2015,07,NDWI,-0.36933282
+2015,08,NDWI,-0.3726222
+2015,09,NDWI,-0.33410805
+2015,10,NDWI,-0.30120784
+2015,11,NDWI,-0.2045634
+2015,12,NDWI,-0.058232486
+2016,01,NDVI,0.031631526
+2016,02,NDVI,0.108674325
+2016,03,NDVI,0.2726109
+2016,04,NDVI,0.45824805
+2016,05,NDVI,0.60704374
+2016,06,NDVI,0.49255392
+2016,07,NDVI,0.44270635
+2016,08,NDVI,0.41391832
+2016,09,NDVI,0.3552221
+2016,10,NDVI,0.31184065
+2016,11,NDVI,0.14771272
+2016,12,NDVI,
+2016,01,NDWI,-0.06354797
+2016,02,NDWI,-0.1311203
+2016,03,NDWI,-0.27396092
+2016,04,NDWI,-0.412583
+2016,05,NDWI,-0.52879846
+2016,06,NDWI,-0.46289545
+2016,07,NDWI,-0.43069345
+2016,08,NDWI,-0.41207507
+2016,09,NDWI,-0.37520212
+2016,10,NDWI,-0.32266992
+2016,11,NDWI,-0.16108795
+2016,12,NDWI,
+2017,01,NDVI,0.028299352
+2017,02,NDVI,-0.0004300331
+2017,03,NDVI,0.15977669
+2017,04,NDVI,0.34166
+2017,05,NDVI,0.5606122
+2017,06,NDVI,0.5247286
+2017,07,NDVI,0.3560093
+2017,08,NDVI,0.3450487
+2017,09,NDVI,0.3213928
+2017,10,NDVI,0.26438868
+2017,11,NDVI,0.16974592
+2017,12,NDVI,
+2017,01,NDWI,-0.06207667
+2017,02,NDWI,-0.025918996
+2017,03,NDWI,-0.18314826
+2017,04,NDWI,-0.33923197
+2017,05,NDWI,-0.497302
+2017,06,NDWI,-0.49217677
+2017,07,NDWI,-0.3974867
+2017,08,NDWI,-0.39300305
+2017,09,NDWI,-0.37551898
+2017,10,NDWI,-0.3116423
+2017,11,NDWI,-0.18042384
+2017,12,NDWI,
+2018,01,NDVI,0.043647703
+2018,02,NDVI,0.022244323
+2018,03,NDVI,0.1468368
+2018,04,NDVI,0.38621
+2018,05,NDVI,0.553758
+2018,06,NDVI,0.4577191
+2018,07,NDVI,0.371388
+2018,08,NDVI,0.32523742
+2018,09,NDVI,0.28276566
+2018,10,NDVI,0.24991055
+2018,11,NDVI,0.057450306
+2018,12,NDVI,0.10182946
+2018,01,NDWI,-0.08074473
+2018,02,NDWI,-0.049971007
+2018,03,NDWI,-0.16181277
+2018,04,NDWI,-0.36271462
+2018,05,NDWI,-0.50257903
+2018,06,NDWI,-0.45652497
+2018,07,NDWI,-0.3978019
+2018,08,NDWI,-0.3673624
+2018,09,NDWI,-0.33057702
+2018,10,NDWI,-0.31096348
+2018,11,NDWI,-0.083236896
+2018,12,NDWI,-0.11330218
+2019,01,NDVI,0.12812804
+2019,02,NDVI,0.07167658
+2019,03,NDVI,0.16254286
+2019,04,NDVI,0.3810431
+2019,05,NDVI,0.5881861
+2019,06,NDVI,0.41373178
+2019,07,NDVI,0.33502045
+2019,08,NDVI,0.30443558
+2019,09,NDVI,0.27136824
+2019,10,NDVI,0.2249414
+2019,11,NDVI,0.17058644
+2019,12,NDVI,0.015977545
+2019,01,NDWI,-0.13369952
+2019,02,NDWI,-0.09915775
+2019,03,NDWI,-0.18232904
+2019,04,NDWI,-0.36428666
+2019,05,NDWI,-0.51711845
+2019,06,NDWI,-0.4318895
+2019,07,NDWI,-0.38278067
+2019,08,NDWI,-0.36510453
+2019,09,NDWI,-0.32055837
+2019,10,NDWI,-0.26932168
+2019,11,NDWI,-0.20094076
+2019,12,NDWI,-0.045890413
+2020,01,NDVI,0.02600833
+2020,02,NDVI,0.10948946
+2020,03,NDVI,0.159752
+2020,04,NDVI,0.38456792
+2020,05,NDVI,0.5776242
+2020,06,NDVI,0.45694545
+2020,07,NDVI,0.35098165
+2020,08,NDVI,0.30593276
+2020,09,NDVI,0.27941108
+2020,10,NDVI,0.21813856
+2020,11,NDVI,0.043937486
+2020,12,NDVI,0.047411535
+2020,01,NDWI,-0.0595021
+2020,02,NDWI,-0.13303731
+2020,03,NDWI,-0.18740809
+2020,04,NDWI,-0.36957166
+2020,05,NDWI,-0.51577485
+2020,06,NDWI,-0.46002954
+2020,07,NDWI,-0.40317768
+2020,08,NDWI,-0.34992763
+2020,09,NDWI,-0.31799856
+2020,10,NDWI,-0.2716179
+2020,11,NDWI,-0.067942545
+2020,12,NDWI,-0.08969609
+2021,01,NDVI,0.031039532
+2021,02,NDVI,0.12767474
+2021,03,NDVI,0.17752129
+2021,04,NDVI,0.3315446
+2021,05,NDVI,0.4536556
+2021,06,NDVI,0.3575767
+2021,07,NDVI,0.30110878
+2021,08,NDVI,0.28985253
+2021,09,NDVI,0.26736796
+2021,10,NDVI,0.22649986
+2021,11,NDVI,0.06251545
+2021,12,NDVI,0.13564047
+2021,01,NDWI,-0.069737405
+2021,02,NDWI,-0.14461634
+2021,03,NDWI,-0.21508418
+2021,04,NDWI,-0.33986288
+2021,05,NDWI,-0.44316238
+2021,06,NDWI,-0.39590663
+2021,07,NDWI,-0.35498592
+2021,08,NDWI,-0.34644762
+2021,09,NDWI,-0.3233553
+2021,10,NDWI,-0.27977043
+2021,11,NDWI,-0.099415764
+2021,12,NDWI,-0.15367758

meer/model_creation.py

@@ -0,0 +1,96 @@
+# -*- coding: utf-8 -*-
+"""
+Created on Sun Mar 24 14:50:30 2024
+
+@author: Metinov Adilet
+"""
+
+import rasterio
+import os
+import numpy as np
+import pandas as pd
+from sklearn.model_selection import train_test_split
+from tensorflow.keras.models import Sequential
+from tensorflow.keras.layers import Dense
+from tensorflow.keras.optimizers import Adam
+
+def calculate_mean_value(src, crop_window):
+    left, top, width, height = crop_window
+    image_data = src.read(1, window=rasterio.windows.Window(left, top, width, height))
+    return np.mean(image_data)
+
+# Инициализируем список для сохранения результатов
+results = []
+
+# Путь к директории с изображениями
+image_directory = 'C:\\Users\\Administrator\\Desktop\\Data\\'
+months = ['01', '02', '03', '04', '05', '06', '07', '08', '09', '10', '11', '12']
+crop_window = (0, 0, 1200, 1200)
+
+# Цикл по годам
+for year in range(2014, 2022):  # 2022, потому что range не включает конечное значение
+    # Цикл для NDVI и NDWI
+    for prefix in ['NDVI', 'NDWI']:
+        for month in months:
+            image_path = os.path.join(image_directory, f'{prefix}_Image_{year}_{month}.tif')
+            if os.path.exists(image_path):
+                with rasterio.open(image_path) as src:
+                    mean_value = calculate_mean_value(src, crop_window)
+                    results.append({'Year': year, 'Month': month, 'Type': prefix, 'Mean_Value': mean_value})
+            else:
+                print(f'Image for {year}_{month} not found.')
+
+# Создаем DataFrame из списка результатов
+results_df = pd.DataFrame(results)
+
+# Сохраняем результаты в CSV файл
+output_path = os.path.join(image_directory, 'mean_values_summary.csv')
+results_df.to_csv(output_path, index=False)
+
+print(f'Results saved to {output_path}')
+
+# Загружаем данные
+data_df = pd.read_csv(output_path)
+
+# Проверяем на наличие nan в столбце 'Mean_Value' и заменяем на соответствующие значения
+for i, row in data_df.iterrows():
+    if pd.isnull(row['Mean_Value']):
+        if row['Type'] == 'NDVI':
+            data_df.at[i, 'Mean_Value'] = 0.2
+        elif row['Type'] == 'NDWI':
+            data_df.at[i, 'Mean_Value'] = -0.2
+
+# Преобразуем 'Type' с помощью one-hot encoding
+type_dummies = pd.get_dummies(data_df['Type'], prefix='Type')
+data_df = pd.concat([data_df.drop('Type', axis=1), type_dummies], axis=1)
+
+# Преобразование всех данных в float32
+data_df['Year'] = data_df['Year'].astype('float32')
+data_df['Month'] = data_df['Month'].astype('float32')
+for column in type_dummies.columns:
+    data_df[column] = data_df[column].astype('float32')
+
+# Создаем признаки (X) и цель (y), убедимся, что все данные в нужном формате
+X = data_df.drop('Mean_Value', axis=1).astype('float32')
+y = data_df['Mean_Value'].astype('float32')
+
+# Разделяем данные на обучающую и тестовую выборки
+X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
+# Определение модели
+model = Sequential([
+    Dense(64, activation='relu', input_shape=(X_train.shape[1],)),
+    Dense(64, activation='relu'),
+    Dense(1)
+])
+
+# Компиляция модели
+model.compile(optimizer=Adam(learning_rate=0.001), loss='mean_squared_error')
+
+# Обучение модели
+history = model.fit(X_train, y_train, validation_split=0.2, epochs=100, batch_size=32)
+
+# Оценка модели на тестовых данных
+loss = model.evaluate(X_test, y_test)
+print(f'Test loss: {loss}')
+# Сохранение обученной модели
+model.save('C:\\Users\\Administrator\\Desktop\\Data\\Model\\Meer.h5')