diff --git a/.gitignore b/.gitignore
index b9e16f4..f506c67 100644
--- a/.gitignore
+++ b/.gitignore
@@ -191,3 +191,4 @@ poetry.toml
 pyrightconfig.json
 
 # End of https://www.toptal.com/developers/gitignore/api/python
+.DS_Store
diff --git a/.gitmodules b/.gitmodules
new file mode 100644
index 0000000..160b587
--- /dev/null
+++ b/.gitmodules
@@ -0,0 +1,3 @@
+[submodule "ECCOv4-py"]
+	path = ECCOv4-py
+	url = git@github.com:ECCO-GROUP/ECCOv4-py.git
diff --git a/ECCOv4-py b/ECCOv4-py
new file mode 160000
index 0000000..83f04e0
--- /dev/null
+++ b/ECCOv4-py
@@ -0,0 +1 @@
+Subproject commit 83f04e03004ba5a15d0b42321fcfc1e56944fe3f
diff --git a/emu_utilities/budget.py b/emu_utilities/budget.py
new file mode 100644
index 0000000..d060e58
--- /dev/null
+++ b/emu_utilities/budget.py
@@ -0,0 +1,174 @@
+from __future__ import annotations
+
+import struct
+import numpy as np
+import xarray as xr
+from pathlib import Path
+from datetime import datetime
+from dateutil.relativedelta import relativedelta
+from typing import TYPE_CHECKING
+from .emu_utilities import EMU
+import re
+if TYPE_CHECKING:
+    from numpy.typing import NDArray
+
+__all__ = ["load_budget"]
+
+
+class EMUBudget(EMU):
+    """Handles loading and processing of EMU budget data."""
+
+    def __init__(self, run_directory: str) -> None:
+        super().__init__(run_directory)
+        self.validate_tool("budg")
+        self.budget_file = self.find_budget_file()
+        self.time = self._build_time_axis()
+
+    def find_budget_file(self) -> Path:
+        files = sorted(self.directory.glob("output/emu_budg.sum_*"))
+        if not files:
+            raise FileNotFoundError("No emu_budg.sum_* file found.")
+        return files[0]
+
+    def _build_time_axis(self) -> NDArray[datetime]:
+        """Construct time axis as datetime array (monthly, starts at Jan 1992)."""
+        with open(self.budget_file, "rb") as f:
+            f.seek(4)  # Skip ibud
+            nmonths = struct.unpack(">l", f.read(4))[0]
+
+        base_date = datetime(1992, 1, 15)
+        return np.array([base_date + relativedelta(months=i) for i in range(nmonths)], dtype="O")
+
+    def _parse_budget_info(self, info_path: Path) -> dict:
+        """Parse the budg.info file and return metadata with separate variable name and units."""
+        meta = {}
+    
+        if not info_path.exists():
+            print(f"Warning: budg.info not found at {info_path}")
+            return meta
+    
+        with open(info_path, "r") as f:
+            for line in f:
+                line = line.strip()
+    
+                if "budget sampled from" in line.lower():
+                    meta["source_path"] = line.split("from")[-1].strip()
+    
+                elif "budget is monthly" in line.lower():
+                    meta["temporal_resolution"] = "monthly"
+    
+                elif re.search(r"budget variable\s*:", line, re.IGNORECASE):
+                    match = re.search(r"budget variable\s*:\s*(.+)", line, re.IGNORECASE)
+                    if match:
+                        full = match.group(1).strip()
+                        # Split into variable name and units
+                        var_match = re.match(r"(.+?)\s*\((.+)\)", full)
+                        if var_match:
+                            meta["budget_name"] = var_match.group(1).strip()
+                            meta["budget_units"] = var_match.group(2).strip() + " / s"
+                        else:
+                            meta["budget_name"] = full
+                            meta["budget_units"] = "unknown"
+    
+                elif "pert_i" in line and "pert_j" in line and "pert_k" in line:
+                    match = re.search(r"pert_i,\s*pert_j,\s*pert_k\s*=\s*(\d+)\s+(\d+)\s+(\d+)", line)
+                    if match:
+                        meta["i"] = int(match.group(1))
+                        meta["j"] = int(match.group(2))
+                        meta["k"] = int(match.group(3))
+    
+                elif "budget model grid location" in line.lower():
+                    meta["location_str"] = line.split("-->")[-1].strip()
+    
+                elif "iobjf =" in line:
+                    match = re.search(r"iobjf\s*=\s*(\d+)", line)
+                    if match:
+                        meta["iobjf (multiplier)"] = int(match.group(1))
+    
+                elif "ifunc =" in line:
+                    match = re.search(r"ifunc\s*=\s*(\d+)", line)
+                    if match:
+                        meta["ifunc"] = int(match.group(1))
+    
+        return meta
+
+    def load_budget_data(self) -> tuple[list[str], NDArray[np.float32]]:
+        """Read EMU budget binary file and return names and 2D array [nvar, ntime]."""
+        with open(self.budget_file, "rb") as f:
+            ibud = struct.unpack(">l", f.read(4))[0]
+            if not (1 <= ibud <= 5):
+                raise ValueError(f"Invalid budget ID: {ibud}")
+            ibud -= 1
+
+            nmonths = struct.unpack(">l", f.read(4))[0]
+            emubudg_name = []
+            emubudg = np.zeros((0, nmonths), dtype=np.float32)
+
+            while True:
+                fvar_bytes = f.read(12)
+                if len(fvar_bytes) < 12:
+                    break
+                fvar = fvar_bytes.decode("utf-8").strip()
+                emubudg_name.append(fvar)
+
+                fdum = np.array(struct.unpack(f">{nmonths}f", f.read(4 * nmonths)), dtype=np.float32)
+                emubudg = np.vstack((emubudg, fdum))
+
+        return emubudg_name, emubudg, ibud
+
+    def make_dataset(self) -> xr.Dataset:
+        names, emubudg, ibud = self.load_budget_data()
+        nmonths = len(self.time)
+        nvar = len(names)
+
+        lhs = emubudg[1, :]
+        rhs = emubudg[2, :].copy()
+        for i in range(3, nvar):
+            rhs += emubudg[i, :]
+
+        # adv
+        adv = np.zeros(nmonths, dtype=np.float32)
+        for i in range(nvar):
+            if "adv" in names[i]:
+                adv += emubudg[i, :]
+
+        # mix
+        mix = np.zeros(nmonths, dtype=np.float32)
+        for i in range(nvar):
+            if "mix" in names[i]:
+                mix += emubudg[i, :]
+
+        # frc
+        frc = np.zeros(nmonths, dtype=np.float32)
+        for i in range(nvar):
+            if names[i] not in ["dt", "lhs"]:
+                if all(k not in names[i] for k in ["adv", "mix", "lhs", "dt"]):
+                    frc += emubudg[i, :]
+
+        # Build dataset
+        data_vars = {
+            "lhs": (["time"], lhs),
+            "rhs": (["time"], rhs),
+            "advection": (["time"], adv),
+            "mixing": (["time"], mix),
+            "surface_forcing": (["time"], frc),
+        }
+
+        coords = {"time": self.time}
+        ds = self.create_base_dataset(data_vars, coords)
+
+        info_files = list(self.directory.glob("output/budg.info"))
+
+        if info_files:
+            meta = self._parse_budget_info(info_files[0])
+            ds.attrs.update(meta)
+        else:
+            print("Warning: No budg.info file found.")
+
+        return ds
+
+
+def load_budget(run_directory: str) -> xr.Dataset:
+    """Load EMU budget data as an xarray Dataset."""
+    emu = EMUBudget(run_directory)
+    return emu.make_dataset()
diff --git a/emu_utilities/tracer.py b/emu_utilities/tracer.py
index ac542d9..0bd823c 100644
--- a/emu_utilities/tracer.py
+++ b/emu_utilities/tracer.py
@@ -4,9 +4,13 @@
 
 import numpy as np
 import xarray as xr
+import sys
 
-from .emu_utilities import EMU, find_time_from_file_names
-from .resample import llc_compact_to_tiles
+# import the ECCOv4 py library 
+sys.path.insert(0,'../ECCOv4-py')
+import ecco_v4_py as ecco
+
+from .emu_utilities import EMU
 
 if TYPE_CHECKING:
     from numpy import datetime64
@@ -40,43 +44,6 @@ def __init__(self, run_directory: str, mean: bool) -> None:
             raise ValueError(f"Expected EMU tool 'trcr', but got '{self.tool}' from directory: {self.run_name}")
         self.mean = mean
 
-    def find_time(self) -> NDArray[datetime64]:
-        """Extract timestamps from tracer file names.
-
-        Returns:
-            Array of datetime64 objects representing available timestamps.
-        """
-        pattern = "output/ptracer_mon_mean.*.data" if self.mean else "output/ptracer_mon_snap.*.data"
-        return np.array(find_time_from_file_names(self.directory, pattern))
-
-    def load_data(self, trcr_files) -> NDArray[np.float32]:
-        """Load tracer data from binary files and apply appropriate scaling.
-
-        Applies vertical and horizontal area weights to correctly represent
-        the tracer concentration in each cell.
-
-        Args:
-            trcr_files: List of tracer data files to load.
-
-        Returns:
-            Structured array of tracer data with applied weights.
-        """
-        data = np.full(
-            (self.time.size, self.nr, self.ntiles, self.ny // self.ntiles, self.nx), np.nan, dtype=np.float32
-        )
-
-        for i, trcr_file in enumerate(trcr_files):
-            with open(trcr_file, "rb") as f:
-                full_data = np.fromfile(f, dtype=">f4").astype(np.float32)
-            data[i] = llc_compact_to_tiles(full_data.reshape((self.nr, self.ny, self.nx)))
-            for k in range(self.nr):
-                # Apply vertical and horizontal area weights to correctly represent tracer concentration
-                data[i, k, :, :, :] = (
-                    data[i, k, :, :, :] * self._coordinate_factory.drf[k] * self._coordinate_factory.hfacc[k, :, :, :]
-                )
-
-        return data
-
     def make_dataset(self) -> xr.Dataset:
         """Create an xarray Dataset from tracer data.
 
@@ -87,25 +54,14 @@ def make_dataset(self) -> xr.Dataset:
             Dataset containing tracer data with appropriate coordinates and metadata.
         """
         if self.mean:
-            trcr_files = list(self.directory.glob("output/ptracer_mon_mean.*.data"))
+            ds = ecco.load_ecco_vars_from_mds(self.directory.glob("output"), 
+                                mds_grid_dir = self.directory.glob("temp"), 
+                                mds_files = "ptracer_mon_mean")
         else:
-            trcr_files = list(self.directory.glob("output/ptracer_mon_snap.*.data"))
-        time_unsorted = self.find_time()
-        sort_idx = np.argsort(time_unsorted)
-        self.time = time_unsorted[sort_idx]
-        trcr_files = [trcr_files[i] for i in sort_idx]
-
-        data = self.load_data(trcr_files)
-
-        coords = self._coordinate_factory.create_tile_coordinates(include_z=True, include_time=True, times=self.time)
-        data_vars = {
-            "tracer": (["time", "k", "tile", "j", "i"], data),
-        }
-        ds = self.create_base_dataset(data_vars, coords)
-
-        # Apply ocean mask to exclude land areas
-        mask = self._coordinate_factory.create_mask(include_z=True)
-        ds = ds.where(mask > 0)
+            ds = ecco.load_ecco_vars_from_mds(self.directory.glob("output"), 
+                                mds_grid_dir = self.directory.glob("temp"), 
+                                mds_files = "ptracer_mon_mean")
+            ds = ds.rename({"TRAC01":"tracer"})
 
         # Calculate depth-integrated tracer values
         ds["tracer_depth_integrated"] = (ds["tracer"] * mask).sum(dim="k", min_count=1)