Weather forecast on dashboard

requirements.txt

@@ -20,3 +20,4 @@ ocf_blosc2==0.0.4
 zarr==2.14.2
 elexonpy
 fiona==1.9.6
+herbie

src/Weather_forecast_on_dashboard.py

@@ -0,0 +1,133 @@
+import streamlit as st
+import pandas as pd
+import plotly.express as px
+from herbie import FastHerbie
+from datetime import datetime, timedelta, time
+import numpy as np
+from concurrent.futures import ThreadPoolExecutor
+
+@st.cache_data(show_spinner=False)
+def fetch_data_for_init_time(init_time, forecast_date, lat, lon, parameter, model="ifs"):
+    # Adjust the initialization datetime based on the init_time offset in hours
+    init_datetime = forecast_date - timedelta(hours=init_time)  # This will be a datetime object
+
+    FH = FastHerbie([init_datetime], model=model, fxx=range(0, 24, 1), fast=True)
+
+    try:
+        FH.download()  # Ensure the file is downloaded
+    except Exception as e:
+        st.error(f"Failed to download data for initialization time {init_datetime}. Error: {e}")
+        return None, None
+
+    variable_subset = ":10[u|v]" if parameter == "u10:v10" else ":100[u|v]" if parameter == "u100:v100" else "2t"
+
+    try:
+        ds = FH.xarray(variable_subset, remove_grib=False)
+        if isinstance(ds, list):
+            ds = ds[0]
+    except EOFError as e:
+        st.error(f"Failed to read data file for {init_datetime}. Error: {e}")
+        return None, None
+    except Exception as e:
+        st.error(f"An unexpected error occurred while reading data: {e}")
+        return None, None
+
+    return ds, init_datetime
+
+def process_initialization(init_time, forecast_date, lat, lon, parameter, model="ifs"):
+    ds, init_datetime = fetch_data_for_init_time(init_time, forecast_date, lat, lon, parameter, model)
+    if ds is None:
+        return []
+
+    interpolated_data = []
+
+    time_dim = ds['step'] if 'step' in ds.dims else ds['time'] if 'time' in ds.dims else None
+    if time_dim is None:
+        st.error("Neither 'step' nor 'time' dimension is present in the dataset. Unable to proceed with forecast.")
+        return []
+
+    for time_val in time_dim:
+        actual_forecast_time = init_datetime + pd.to_timedelta(time_val.values)
+        if actual_forecast_time.date() == forecast_date.date():
+            if parameter == "u10:v10":
+                u = ds['u10'].interp(latitude=lat, longitude=lon, method="nearest", step=time_val)
+                v = ds['v10'].interp(latitude=lat, longitude=lon, method="nearest", step=time_val)
+                value = np.sqrt(u**2 + v**2)
+            elif parameter == "u100:v100":
+                u = ds['u100'].interp(latitude=lat, longitude=lon, method="nearest", step=time_val)
+                v = ds['v100'].interp(latitude=lat, longitude=lon, method="nearest", step=time_val)
+                value = np.sqrt(u**2 + v**2)
+            else:  # temp
+                value = ds['t2m'].interp(latitude=lat, longitude=lon, method="nearest", step=time_val) - 273.15
+
+            interpolated_data.append({
+                'date_time': actual_forecast_time,
+                'latitude': lat,
+                'longitude': lon,
+                'value': value.compute().item(),
+                'init_time': f"{init_time} hours before"
+            })
+
+    return interpolated_data
+
+def get_forecast(forecast_date, lat, lon, parameter, init_times, model="ifs"):
+    all_data = []
+    with ThreadPoolExecutor() as executor:
+        results = list(executor.map(lambda init_time: process_initialization(init_time, forecast_date, lat, lon, parameter, model), init_times))
+    for result in results:
+        all_data.extend(result)
+    return pd.DataFrame(all_data)
+
+def weather_forecast_page():
+    # Streamlit app
+    st.title("Weather Forecasts from Different Initialization Times")
+
+    parameter_mapping = {
+        "Wind Speed (10m)": "u10:v10",
+        "Wind Speed (100m)": "u100:v100",
+        "Temperature": "2t"
+    }
+
+    parameter_option = st.sidebar.selectbox("Select Parameter", options=list(parameter_mapping.keys()))
+    parameter_value = parameter_mapping[parameter_option]
+    lat = st.sidebar.number_input("Enter Latitude", value=27.035, format="%.6f")
+    lon = st.sidebar.number_input("Enter Longitude", value=70.515, format="%.6f")
+
+    # Use datetime.combine to set the forecast_date to midnight
+    forecast_date = st.sidebar.date_input("Select Forecast Date", datetime.today().date())
+    forecast_date = datetime.combine(forecast_date, time(hour=0))
+
+    # Multiselect for initialization times in hours
+    init_time_options = st.sidebar.multiselect(
+        "Select Initialization Times (in hours before forecast date)",
+        options=[24, 18, 12, 6, 0],
+        default=[24, 12, 0]  # Default selections: 24 hours before, 12 hours before, and current day
+    )
+
+    if st.button("Fetch Data"):
+        if not init_time_options:
+            st.warning("Please select at least one initialization time.")
+        else:
+            st.write("Note: Data loading may take up to 30 seconds. Please be patient...")
+            data = get_forecast(forecast_date, lat, lon, parameter_value, init_times=init_time_options )
+
+            if not data.empty:
+                fig = px.line(
+                    data,
+                    x='date_time',
+                    y='value',
+                    color='init_time',
+                    labels={
+                        'date_time': 'Date and Time',
+                        'value': f'{parameter_option} ({"m/s" if "Wind Speed" in parameter_option else "°C"})',
+                        'init_time': 'Initialization Time'
+                    },
+                    title=f"Forecast of {parameter_option} at ({lat}, {lon}) for {forecast_date.strftime('%Y-%m-%d')}"
+                )
+                st.plotly_chart(fig)
+
+                csv = data.to_csv(index=False)
+                file_name = f"{forecast_date.strftime('%Y-%m-%d')}_{parameter_option}_comparison_data.csv"
+                st.download_button(label="Download data as CSV", data=csv, file_name=file_name, mime='text/csv')
+            else:
+                st.warning("No data was collected.")

src/main_india.py

@@ -13,6 +13,7 @@
 from satellite_page import satellite_page
 from status import status_page
 from users import user_page
+from weather_forecast import weather_forecast_page
 
 st.get_option("theme.primaryColor")
 st.set_page_config(layout="wide", page_title="OCF Dashboard")
@@ -26,6 +27,7 @@
         "API Users": user_page,
         "NWP": nwp_page,
         "Satellite": satellite_page,
+        "Weather Forecast": weather_forecast_page, 
     }
 
     demo_name = st.sidebar.selectbox("Choose a page", page_names_to_funcs.keys(), 1)

src/weather_forecast.py

@@ -0,0 +1,222 @@
+import streamlit as st
+import pandas as pd
+import plotly.express as px
+from herbie import FastHerbie
+from datetime import datetime, timedelta, time
+import numpy as np
+from concurrent.futures import ThreadPoolExecutor
+from typing import List, Optional, Tuple
+
+@st.cache_data(show_spinner=False)
+def fetch_data_for_init_time(
+    init_time: datetime, 
+    forecast_date: datetime, 
+    lat: float, 
+    lon: float, 
+    parameter: str, 
+    model: str = "ifs"
+) -> Tuple[Optional['xarray.Dataset'], Optional[datetime]]:
+    """
+    Fetch the weather forecast data for a specific initialization time and forecast date.
+
+    Parameters:
+    - init_time (datetime): The initialization time (usually yesterday at a specific hour).
+    - forecast_date (datetime): The forecast date for which to get the forecast.
+    - lat (float): Latitude of the location for which the forecast is requested.
+    - lon (float): Longitude of the location for which the forecast is requested.
+    - parameter (str): The weather parameter to retrieve. Examples: "u10:v10", "u100:v100", or "2t".
+    - model (str): The weather model to use for fetching data (default: "ifs").
+
+    Returns:
+    - ds (xarray.Dataset or None): The forecast dataset for the specified parameters.
+    - init_time (datetime or None): The initialization time for the dataset.
+    """
+    # Adjust the initialization datetime
+    FH = FastHerbie([init_time], model=model, fxx=range(init_time.hour, 24 + init_time.hour, 1), fast=True)
+
+    try:
+        FH.download()  # Ensure the file is downloaded
+    except Exception as e:
+        st.error(f"Failed to download data for initialization time {init_time}. Error: {e}")
+        return None, None
+
+    # Determine the variable subset based on the selected parameter
+    variable_subset = ":10[u|v]" if parameter == "u10:v10" else ":100[u|v]" if parameter == "u100:v100" else "2t"
+
+    try:
+        ds = FH.xarray(variable_subset, remove_grib=False)
+        if isinstance(ds, list):
+            ds = ds[0]
+    except EOFError as e:
+        st.error(f"Failed to read data file for {init_time}. Error: {e}")
+        return None, None
+    except Exception as e:
+        st.error(f"An unexpected error occurred while reading data: {e}")
+        return None, None
+
+    return ds, init_time
+
+def process_initialization(
+    init_time: datetime, 
+    forecast_date: datetime, 
+    lat: float, 
+    lon: float, 
+    parameter: str, 
+    model: str = "ifs"
+) -> List[dict]:
+    """
+    Process the weather forecast for a specific initialization time, 
+    interpolate data for the given latitude and longitude.
+
+    Parameters:
+    - init_time (datetime): The initialization time for the forecast.
+    - forecast_date (datetime): The forecast date for which the data is requested.
+    - lat (float): Latitude of the location for the forecast.
+    - lon (float): Longitude of the location for the forecast.
+    - parameter (str): The weather parameter to retrieve (e.g., "u10:v10", "u100:v100", or "2t").
+    - model (str): The weather model to use for fetching data (default: "ifs").
+
+    Returns:
+    - interpolated_data (list): A list of dictionaries containing forecast values and metadata.
+    """
+    ds, init_time = fetch_data_for_init_time(init_time, forecast_date, lat, lon, parameter, model)
+    if ds is None:
+        return []
+
+    interpolated_data = []
+
+    # Determine the time dimension
+    time_dim = ds['step'] if 'step' in ds.dims else ds['time'] if 'time' in ds.dims else None
+    if time_dim is None:
+        st.error("Neither 'step' nor 'time' dimension is present in the dataset. Unable to proceed with forecast.")
+        return []
+
+    # Iterate over the forecast steps and interpolate the data
+    for time_val in time_dim:
+        actual_forecast_time = init_time + pd.to_timedelta(time_val.values)
+        if actual_forecast_time.date() == forecast_date.date():
+            if parameter == "u10:v10":
+                u = ds['u10'].interp(latitude=lat, longitude=lon, method="nearest", step=time_val)
+                v = ds['v10'].interp(latitude=lat, longitude=lon, method="nearest", step=time_val)
+                value = np.sqrt(u**2 + v**2)
+            elif parameter == "u100:v100":
+                u = ds['u100'].interp(latitude=lat, longitude=lon, method="nearest", step=time_val)
+                v = ds['u100'].interp(latitude=lat, longitude=lon, method="nearest", step=time_val)
+                value = np.sqrt(u**2 + v**2)
+            else:  # Temperature (in Kelvin, converted to Celsius)
+                value = ds['t2m'].interp(latitude=lat, longitude=lon, method="nearest", step=time_val) - 273.15
+
+            interpolated_data.append({
+                'date_time': actual_forecast_time,
+                'latitude': lat,
+                'longitude': lon,
+                'value': value.compute().item(),
+                'init_time': init_time.strftime('%Y-%m-%d %H:%M')
+            })
+
+    return interpolated_data
+
+def get_forecast(
+    forecast_date: datetime, 
+    lat: float, 
+    lon: float, 
+    parameter: str, 
+    init_times: List[datetime], 
+    model: str = "ifs"
+) -> pd.DataFrame:
+    """
+    Get the weather forecast data for multiple initialization times.
+
+    Parameters:
+    - forecast_date (datetime): The date for which the forecast is requested.
+    - lat (float): Latitude of the location for the forecast.
+    - lon (float): Longitude of the location for the forecast.
+    - parameter (str): The weather parameter to retrieve (e.g., "u10:v10", "u100:v100", or "2t").
+    - init_times (list): A list of datetime objects representing initialization times.
+    - model (str): The weather model to use for fetching data (default: "ifs").
+
+    Returns:
+    - pd.DataFrame: A Pandas DataFrame containing the forecast data for the selected parameters and times.
+    """
+    all_data = []
+    with ThreadPoolExecutor() as executor:
+        results = list(executor.map(lambda init_time: process_initialization(init_time, forecast_date, lat, lon, parameter, model), init_times))
+    for result in results:
+        all_data.extend(result)
+    return pd.DataFrame(all_data)
+
+def weather_forecast_page() -> None:
+    """
+    The main page for displaying weather forecasts with user inputs.
+
+    - Allows users to input latitude and longitude.
+    - Select the forecast date.
+    - Choose initialization times based on yesterday's forecast generation times.
+    - Fetch and display forecast data for the chosen initialization times and location.
+    """
+    # Streamlit app
+    st.title("Weather Forecasts from Different Initialization Times")
+
+    # Parameter selection
+    parameter_mapping = {
+        "Wind Speed (10m)": "u10:v10",
+        "Wind Speed (100m)": "u100:v100",
+        "Temperature": "2t"
+    }
+    parameter_option = st.sidebar.selectbox("Select Parameter", options=list(parameter_mapping.keys()))
+    parameter_value = parameter_mapping[parameter_option]
+
+    # Latitude and longitude inputs
+    lat = st.sidebar.number_input("Enter Latitude", value=27.035, format="%.6f")
+    lon = st.sidebar.number_input("Enter Longitude", value=70.515, format="%.6f")
+
+    # Forecast date selection
+    forecast_date = st.sidebar.date_input("Select Forecast Date", datetime.today().date())
+    forecast_date = datetime.combine(forecast_date, time(hour=0))
+
+    # Create initialization times for yesterday at 00:00, 06:00, 12:00, and 18:00
+    yesterday = forecast_date - timedelta(days=1)
+    init_times = [
+        yesterday.replace(hour=0),
+        yesterday.replace(hour=6),
+        yesterday.replace(hour=12),
+        yesterday.replace(hour=18)
+    ]
+
+    # Multi-selection for initialization times
+    init_time_options = st.sidebar.multiselect(
+        "Select Initialization Times",
+        options=[time.strftime('%Y-%m-%d %H:%M') for time in init_times],
+        default=[time.strftime('%Y-%m-%d %H:%M') for time in init_times]  # Default selects all times
+    )
+    selected_init_times = [datetime.strptime(time, '%Y-%m-%d %H:%M') for time in init_time_options]
+
+    # Fetch and display the data
+    if st.button("Fetch Data"):
+        if not selected_init_times:
+            st.warning("Please select at least one initialization time.")
+        else:
+            st.write("Note: Data loading may take up to 30 seconds. Please be patient...")
+            data = get_forecast(forecast_date, lat, lon, parameter_value, selected_init_times, model="ifs")
+
+            if not data.empty:
+                fig = px.line(
+                    data,
+                    x='date_time',
+                    y='value',
+                    color='init_time',
+                    labels={
+                        'date_time': 'Date and Time',
+                        'value': f'{parameter_option} ({"m/s" if "Wind Speed" in parameter_option else "°C"})',
+                        'init_time': 'Initialization Time'
+                    },
+                    title=f"Forecast of {parameter_option} at ({lat}, {lon}) for {forecast_date.strftime('%Y-%m-%d')}"
+                )
+                st.plotly_chart(fig)
+
+                # Downloadable CSV
+                csv = data.to_csv(index=False)
+                file_name = f"{forecast_date.strftime('%Y-%m-%d')}_{parameter_option}_comparison_data.csv"
+                st.download_button(label="Download data as CSV", data=csv, file_name=file_name, mime='text/csv')
+            else:
+                st.warning("No data was collected.")